Loading [MathJax]/jax/element/mml/optable/GeneralPunctuation.js
Research article

A precise solution to the shortest path optimization problem in graphs using Z-numbers

  • Communication networks are exposed to internal or external risks that can affect all or part of the system. The most important components that form the infrastructure of these systems are routers, which act as nodes. In the field of graph theory, there are sophisticated techniques that can be used to optimize the path of a packet as it travels through various routers from its origin to its destination. A notable example of such an algorithm is Dijkstra's algorithm, which is designed to efficiently determine the shortest path. The algorithm works under the assumption that the system operates under ideal conditions. Real-time systems can perform better if risk factors and optimal conditions are taken into account. The relationship between the nodes can be expressed by various metrics such as distance, delay, and bandwidth. The aforementioned metrics facilitate the calculation of the optimal path, with the ultimate objective of achieving low-latency networks characterized by rapid response times. Round-trip time (RTT) can be employed as a metric for measuring enhancements in a range of latency types, including those associated with processing, transmission, queuing, and propagation. The use of Z-numbers was employed in this study to incorporate risk into the optimal path metric. RTT was the preferred metric and reliability was represented by fuzzy linguistic qualifiers. A comparison of several scenarios was shown using a numerical example of a communication network. It is expected that this study will have a significant impact on the evolution from models that consider only ideal conditions to real-time systems that include risks using Z-numbers.

    Citation: Nurdoğan Güner, Halit Orhan, Tofigh Allahviranloo, Bilal Usanmaz. A precise solution to the shortest path optimization problem in graphs using Z-numbers[J]. AIMS Mathematics, 2024, 9(11): 30100-30121. doi: 10.3934/math.20241454

    Related Papers:

    [1] Haiping Ren, Ziwen Zhang, Qin Gong . Estimation of Shannon entropy of the inverse exponential Rayleigh model under progressively Type-Ⅱ censored test. AIMS Mathematics, 2025, 10(4): 9378-9414. doi: 10.3934/math.2025434
    [2] Muhammad Akram, Sumera Naz, Gustavo Santos-García, Muhammad Ramzan Saeed . Extended CODAS method for MAGDM with 2-tuple linguistic T-spherical fuzzy sets. AIMS Mathematics, 2023, 8(2): 3428-3468. doi: 10.3934/math.2023176
    [3] Peeyush Garg, Pradeep Kumar Gautam, Amit Kumar Verma, Gnananandh Budi . Deformation and failure analysis of heterogeneous slope using nonlinear spatial probabilistic finite element method. AIMS Mathematics, 2024, 9(10): 26339-26370. doi: 10.3934/math.20241283
    [4] Yulong Li, Long Zhou, Fengjie Geng . Dynamics on semi-discrete Mackey-Glass model. AIMS Mathematics, 2025, 10(2): 2771-2807. doi: 10.3934/math.2025130
    [5] Saima Rashid, Fahd Jarad, Taghreed M. Jawa . A study of behaviour for fractional order diabetes model via the nonsingular kernel. AIMS Mathematics, 2022, 7(4): 5072-5092. doi: 10.3934/math.2022282
    [6] Roderick Edwards, Michelle Wood . Branch prioritization motifs in biochemical networks with sharp activation. AIMS Mathematics, 2022, 7(1): 1115-1146. doi: 10.3934/math.2022066
    [7] Juxia Xiao, Ping Yu, Zhongzhan Zhang . Weighted composite asymmetric Huber estimation for partial functional linear models. AIMS Mathematics, 2022, 7(5): 7657-7684. doi: 10.3934/math.2022430
    [8] Qiang Zhao, Chao Zhang, Jingjing Wu, Xiuli Wang . Robust and efficient estimation for nonlinear model based on composite quantile regression with missing covariates. AIMS Mathematics, 2022, 7(5): 8127-8146. doi: 10.3934/math.2022452
    [9] Bahar Acay, Ramazan Ozarslan, Erdal Bas . Fractional physical models based on falling body problem. AIMS Mathematics, 2020, 5(3): 2608-2628. doi: 10.3934/math.2020170
    [10] Zijing Ye, Shihe Xu, Xuemei Wei . Analysis of a free boundary problem for vascularized tumor growth with a time delay in the process of tumor regulating apoptosis. AIMS Mathematics, 2022, 7(10): 19440-19457. doi: 10.3934/math.20221067
  • Communication networks are exposed to internal or external risks that can affect all or part of the system. The most important components that form the infrastructure of these systems are routers, which act as nodes. In the field of graph theory, there are sophisticated techniques that can be used to optimize the path of a packet as it travels through various routers from its origin to its destination. A notable example of such an algorithm is Dijkstra's algorithm, which is designed to efficiently determine the shortest path. The algorithm works under the assumption that the system operates under ideal conditions. Real-time systems can perform better if risk factors and optimal conditions are taken into account. The relationship between the nodes can be expressed by various metrics such as distance, delay, and bandwidth. The aforementioned metrics facilitate the calculation of the optimal path, with the ultimate objective of achieving low-latency networks characterized by rapid response times. Round-trip time (RTT) can be employed as a metric for measuring enhancements in a range of latency types, including those associated with processing, transmission, queuing, and propagation. The use of Z-numbers was employed in this study to incorporate risk into the optimal path metric. RTT was the preferred metric and reliability was represented by fuzzy linguistic qualifiers. A comparison of several scenarios was shown using a numerical example of a communication network. It is expected that this study will have a significant impact on the evolution from models that consider only ideal conditions to real-time systems that include risks using Z-numbers.



    Overweight and obesity is a chronic metabolic disease, which caused by many factors. It is an important factor leading to diabetes mellitus, cardiac and cerebral vessels, and metabolic syndrome (MS), which have become a rapidly growing epidemic worldwide[1,2]. According to the Body Mass Index (BMI)[3] standard published by the United Nations World Health Organization (WHO), the number of overweight and obese people in China has approached a quarter of the total population, and obesity not only seriously endangers people's physical and mental health, but places a huge burden on the economy as well as the society [4]. Research on weight loss and health can help prevent metabolic diseases, and has been widely employed in other fields such as soldiers' physical reserve and combat effectiveness in the military and prediction of human life limit in natural disasters[5,6,7].

    Many researchers employ clinical experiments and mathematical methods to investigated weight loss. It was found that dieting, exercise, drug and other means are effective approaches to lose weight. Slight obesity can reduce the weight by controlling diet and increasing exercise, while it is necessary for severe obesity to reduce weight by taking medicine[8,9,10,11]. Different weight loss methods have different weight loss effects. Therefore, considering the need of human energy intake to maintain metabolism, based on the principle of human energy conservation, a series of weight loss mathematical models are established. From a macro perspective, the relationship between energy consumption and weight loss have been studied [12,13,14,15,16,17]. For example, if the obese lose weight under semi-starvation and complete starvation, the relationship between weight loss and energy demand have been found, and the conclusion has been drawn that people with higher body fat rate would survive longer under complete starvation [12,13]. From the established mathematical model of the relation between body weight and caloric intake[14], it has been known that a direct relationship between body weight and caloric intake has been proposed. Thus, weight loss can be achieved by reducing body fat. Different weight-loss schemes will directly affect the effect of weight-loss. Through model calculation and numerical simulation, the change rule of weight-loss under different schemes are predicted. It has been found that the effects of weight-loss have been achieved by dieting, exercise and other weight-loss methods[15,16,17]. In recent years, drug weight loss has also been an essential method for people who who diet and exercise less obviously[18]. Nakai used statistical analysis of clinical data and numerical fits of diet pills to give a relationship between drug dose, changes in excretion fat and weight loss[19]. From the microcosmic point of view, clinical practice has also studied the impact of the changes of biochemical indicators of human body in the process of weight loss on health. It is found that weight loss not only can improve blood glucose, blood lipid and protein values of three major nutrients and improve the metabolic capacity of the human body, but there are differences in the process time of adjusting the three major nutrients in the way of weight loss[20,21,22,23,24,25].

    Based on the data analysis and biological principles of weight loss through dieting, this paper establishes a mathematical model of metabolic regulation in the process of weight loss. The change trend of three major nutrients in the body under different ways of weight loss is analyzed. The effects of different weight loss schemes on the index values of blood glucose, serum lipid, protein and ketone body are simulated, and optimized the weight loss schemes by comparing and analysing the differences of each index. This paper provides a theoretical basis for guiding healthy weight loss.

    This paper is organized as follows. In Section 2, the model of body metabolism regulation in the process of weight loss is established and the change trend of model solution is analyzed qualitatively. Sufficient conditions for the feasibility of three weight loss methods of dieting, dietary exercise and drug are given respectively. In Section 3, some parameters of the model are identified by numerical simulation, and the differences of weight loss methods are analyzed. The conclusion is presented in Section 4.

    In the process of weight loss, the intake of nutrients and energy consumption have changed, which will lead to change in fasting blood glucose, body fat and protein value and other trace elements[24,25]. Next, take the experiment of the weight loss by dieting as an example to analyze the changes of three major nutrients metabolic indicators. iCarbonX-Israel conducted a seven-day weight-loss experiment on seven volunteers, collected data on protein, antibody group, metabolic group, biochemical group, and weight, and studied the regulation and health utility of the human weight loss mechanism[26]. The following part of the data of three major nutrients in the process of dieting are selected to analyze the effects of weight loss on the metabolism of carbohydrate, fat and protein in human body. The index data of carbohydrate, fat and protein metabolism of seven volunteers during the period of fasting are selected, and the changes of blood glucose, body fat rate and total protein are simulated by Excel, as shown in Figures 13.

    Figure 1.  Trend chart of blood glucose change.
    Figure 2.  Trend chart of body fat.
    Figure 3.  Trend chart of serum total protein.

    As can be seen from Figure 1, the initial blood glucose values of the seven volunteers were different, but the changes were similar and the overall trend of fluctuations declined. As can be seen from Figure 2, the body fat of the seven volunteers decreased on the first day, and the next day showed an upward or slow downward trend. The overall body fat rate showed a fluctuating downward trend. As can be seen from Figure 3, the total serum protein of seven volunteers fluctuated within a certain range, but most of the volunteers showed a downward trend on the first day and an upward trend on the second day. Some volunteers showed the opposite situation, but the overall trend was slightly downward.

    The above figures reflect the three major nutrients metabolic indicators of the volunteers during the dieting period, and the overall changes show a decrease in fluctuations which indicate that the body maintains the balance between the indicators through its own regulation.

    Food is the main source of substances needed to maintain normal metabolic energy. Carbohydrates, fats and proteins in nutrients are the basic substances to maintain human energy. Their conversion rules follow the principle of tricarboxylic acid cycle[27,28]. When food intake is uneven, the transformation between carbohydrate and fat is a two-way process. When the body absorbs too much carbohydrates, it can be converted into fat and stored in the body. In the process of converse decomposition of fat, a small amount of carbohydrate will be produced, and the derivative ketone body will be produced at the same time. If the ketone body is piled up too much in the human body, it will easy to cause ketoacidosis [29]. The conversion of protein and carbohydrate is a one-way process, carbohydrates do not convert to proteins. When carbohydrate metabolism is impaired, protein can be converted to carbohydrate. According to the requirement of biological energy, the three major nutrients and biochemical indicators should be within the standard range when the human body is in a healthy state. Otherwise, the human body will be in a sub-healthy state, and even life-threatening[29,30]. The metabolic transition diagram of carbohydrate, fat and protein in the human body is shown in Figure 4.

    Figure 4.  Metabolic transition between carbohydrates, fat and protein.

    Weight loss methods will affect the transformation of biochemical indicators, in which exercise can help fat consumption, dieting will reduce the absorption of various substances and drugs can inhibit fat absorption. According to the mechanism analysis of the interaction of the three nutrients[30] in Figure 4, the following biological hypotheses are made in the process of weight loss.

    (1) Carbohydrates do not convert to protein, and fat and protein do not convert to each other.

    (2) Based on the structure of human body, the three major nutrients have a certain rate of self-production and consumption. When the human body intakes too much carbohydrates, it will increase the storage of fat and consume part of fat and carbohydrate in sports.

    (3) Weight loss drugs can inhibit the body's absorption of fat.

    (4) The conversion rate and self-production rate of the three major nutrients in the body are limited.

    (5) The three major nutrients have normal value range. If the three nutrients beyond the normal range of human body, the human body will present a sub-health state, even life-threatening. The endpoint value of this range is defined as the critical value of human body limit.

    (6) When fat is decomposed into carbohydrate, ketone body will be formed. The human body has certain ability to remove ketone body. Because of individual differences, the formation rate and clearance rate of ketone body are different, which are limited. There exists a safety threshold in the ketone body. If the value of the ketone body exceeds the critical value, it will easy to suffer from ketoacidosis.

    Based on the above hypotheses, we can get the conversion relationship between the three major nutrients and ketone body in the process of weight loss, as shown in Figure 5. The parameters of Figure 5 are shown in Table 1.

    Figure 5.  Transformation of carbohydrate, fat and protein.
    Table 1.  Model (1) related parameters.
    ParametricParameter mean Parametric Parameter mean
    x1(t) Value of carbohydrate in human body at time t x2(t) Value of fat in human body at time t
    x3(t) Value of protein in human body at time t x4(t) Value of human ketone body at time t
    μ1(t) Absorption rate of carbohydrate intake at time t μ2(t) Absorption rate of fat intake at time t
    μ3(t) Absorption rate of protein intake at time t k1(t) Rate of protein convert into carbohydrate at time t
    k2(t) Metabolic transformation rate of protein except carbohydrate and fat at time t k3(t) Metabolic rate of carbohydrates at time t
    k4(t) Metabolic transformation rate of fat except carbohydrate and protein at time t k5(t) Rate of carbohydrate convert into fat at time t
    k6(t) Rate of fat consumption at time t caused by exercise k7(t) Rate of carbohydrate consumption at time t caused by exercise
    k(t) Rate of ketone body production from fat decomposition in the human body at time t υ(t) Rate of clearance of ketone body from the human body at time t
    w1(t) Basic constant rate of blood glucose in the human body w2(t) Basic constant rate of body fat in the human body
    w3(t) Basic constant rate of protein in the human body ε(t) Rate of fat convert into carbohydrate at time t
    δ(t) Drug inhibits the absorption rate of body fat intake at time t

     | Show Table
    DownLoad: CSV

    Based on the above data analysis and biological hypotheses (1)-(6), a mathematical model for the regulation of three major nutrients and the ketone body during weight loss are established.

    { x(t)=A(t)x(t)+w(t) x(0)=x0 (1)

    where

    w(t)=(w1,w2,w3,k(t)x2(0))T,x(t)=(x1(t),x2(t),x3(t),x4(t))T
    A(t)=(h1(t)ε(t)k1(t)0k5(t)h2(t)0000h3(t)00k(t)0υ(t))
    h1(t)=μ1(t)k3(t)k5(t)k7(t),h2(t)=μ2(t)ε(t)k4(t)k6(t)δ(t)
    h3(t)=μ3(t)k1(t)k2(t) (2)

    According to the biological hypotheses of the three major nutrients metabolism in human body (1)-(6), the following basic assumptions are made for model (1) parameters

    (H1) wi (i=1,2,3) are all non-negative constants, ks(t) (s=1,2,3,4) are known continuous non-negative bounded functions. That is to say, there exist positive numbers ˉks(t), ks(t) (s=1,2,3,4) such that 0<ˉks(t)ks(t)ks.

    (H2) The normal ranges of the three major nutrients xj(t) (j=1,2,3,4) are

    Λ={xj(t) (j=1,2,3,4) | 0<aixi(t)bi, 0x4(t)c (i=1,2,3)}

    and the initial values of the three major nutrients are satisfied xi(0)>ai, 0x4(0)<c.

    (H3) There exist positive numbers mr, nr (r=1,2) such that m1k(t)m2, n1υ(t)n2.

    Weight loss can achieve the expected goal of healthy weight loss by adjusting diet, exercise and dosage of weight-loss drug. While diet, exercise and dosage of weight-loss drug are limited. That is to say μi(t) (i=1,2,3), kh(t) (h=5,6,7), δ(t), ε(t) affect the law of change of xj(t) (j=1,2,3,4), and there exist positive numbers ε, δ, ˉμi, μi, kh (i=1,2,3; h=5,6,7) such that 0¯μi<μi(t)μi, 0δ(t)δ, 0kh(t)kh and 0ε(t)ε.

    Let qi(ξ)<pi(ξ)<0 (ξ=(ˉμi,μi,kh,δ,ε)) be negative numbers which are determined by the values of ˉμi, μi, kh, δ and ε. The following research contents is completed under the conditions (H1) - (H3).

    Theorem 1. If model (1) satisfies the following conditions

    (1) gi(t)<0 (i=1,2,3), where g1(t)=h1(t)+k5(t), g2(t)=h2(t)+ε(t), g3(t)=h3(t)+k1(t),

    (2) |h1(t)h1(0)|+max (|ε(t),k1(t)|)<|h1(0)|,

    (3) |k5(t)|+max (|h2(t)h2(0)|,|h3(t)h3(0)|)<min (|h2(0)|,|h3(0)|),

    then its solution is bounded, and there are two cases as follows.

    Case 1 When 0<¯μiμi(t)μi, qi(ξ)hi(t)gi(t)pi(ξ), wiqi(ξ)ai (i=1,2,3), 1p1(ξ)(w1+w2k1w3p3(ξ))b1, 1p2(ξ)(w1+w2k1w3p3(ξ))b2, w3p3(ξ)b3 and m2n1(x2(0)+w2q2(ξ))c, then limtt0xj(s)dstΛ (j=1,2,3,4).

    Case 2 When 0μi(t)¯μi, gi(t)qi(ξ) (i=1,2,3), 1qr(ξ)(2i=1wik1w3q3(ξ))<ar (r=1,2), w3q3(ξ)<a3 and m1n2(x2(0)+1q2(ξ)(2i=1wik1w3q3(ξ)))>c, then limtt0xj(s)dstΛ (j=1,2,3,4).

    Proof. Model (1) is decomposed into the following two sub-models

    y(t)=ˉA(t)y(t)+ˉw. (3)
    x4(t)=k(t)(x2(0)x2(t))υ(t)x4(t). (4)

    where

    y(t)=(x1(t),x2(t),x3(t))T,ˉw(t)=(w1,w2,w3)T
    ˉA(t)=(A11(t)A12(t)A21(t)A22(t)). (5)
    A11(t)=h1(t),A12(t)=(ε(t),k1(t)), A21(t)=(k5(t),0)T, A22(t)=diag(h2(t),h3(t))

    By means of the freeze coefficient method in [31], the homogeneous linear equation corresponding to model (3) is frozen as follows

    y(t)=diagˉA(0)y(t)+(ˉA(t)diagˉA(0))y(t) (6)

    It follows from formula (5) that

    \begin{equation} \|A_{11}(t)-A_{11}(0)\|\leq|h_{1}(t)-h_{1}(0)| , \; \|A_{12}(t)\|\leq\max (|\varepsilon(t)|, |k_1(t)|) , \; \|A_{21}(t)\|\leq|k_5(t)|) \end{equation} (7)
    \begin{equation} \|A_{22}(t)-A_{22}(0)\|\leq\max (|h_{2}(t)-h_{2}(0)|, |h_{3}(t)-h_{3}(0)|) \end{equation}

    We computer

    \begin{equation} \int^{t}_{0}{\|e^{A_{11}(0)(t-\tau)}\|[\|A_{11}(\tau)-A_{11}(0)\|+\|A_{12}(\tau)\|]{\rm d}\tau} \\ \leq\int^{t}_{0}{e^{h_{1}(0)(t-\tau)}(|h_{1}(t)-h_{1}(0)|+\max (|\varepsilon(t)|, |k_1(t)|)){\rm d}\tau} \end{equation} (8)
    \begin{equation} \; \; \; \; \; \; \leq|h_1(0)|\int^{t}_{0}{e^{h_{1}(0)(t-\tau)}{\rm d}\tau}\leq -\frac{|h_1(0)|}{h_1(0)} = 1\; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \end{equation}

    and

    \begin{equation} \int^{t}_{0}{\|e^{A_{22}(0)(t-\tau)}\|[\|A_{21}(\tau)\|+\|A_{22}(\tau)-A_{22}(0)\|]{\rm d}\tau}\; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \\ \; \; \; \; \; \; \; \; \; \; \; \; \; \leq\int^{t}_{0}{\max (e^{h_{2}(0)(t-\tau)}, e^{h_{3}(0)(t-\tau)})(|k_5(t)|+\max(|\bar{h}_2(t)-\bar{h}_2(0)|, |h_3(t)-h_3(0)|)){\rm d}\tau} \\\leq\min(|h_2(0)|, |h_3(0)|)\int^{t}_{0}{\max (e^{h_{2}(0)(t-\tau)}, e^{h_{3}(0)(t-\tau)}){\rm d}\tau}\; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \\ \leq -\frac{\min(|h_2(0)|, |h_3(0)|)}{\max(h_2(0), h3(0))} = 1\; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \; \end{equation} (9)

    as shown in [31], the solution of model (1) is stable and bounded.

    In order to discuss the changing trend of the solution of model (1), the four equations in model (1) are integrated on both sides.

    \begin{equation} \left\{ \begin{aligned} & \ \frac{x_1(t)-x_1(0)}{t} = \frac{\int^{t}_{0}{h_1(s)x_1(s){\rm d}s}}{t}+\frac{\int^{t}_{0}{\varepsilon(s)x_2(s){\rm d}s}}{t}+\frac{\int^{t}_{0}{k_1(s)x_3(s){\rm d}s}}{t}+w_1\\ &\ \frac{x_2(t)-x_2(0)}{t} = \frac{\int^{t}_{0}{h_2(s)x_2(s){\rm d}s}}{t}+\frac{\int^{t}_{0}{k_5(s)x_1(s){\rm d}s}}{t}+w_2\\ &\ \frac{x_3(t)-x_3(0)}{t} = \frac{\int^{t}_{0}{h_3(s)x_3(s){\rm d}s}}{t}+w_3\\ &\ \frac{x_4(t)-x_4(0)}{t} = \frac{\int^{t}_{0}{k(s)(x_2(0)-x_2(s)){\rm d}s}}{t}+\frac{\int^{t}_{0}{\upsilon(s)x_4(s){\rm d}s}}{t}\\ \end{aligned} \right. \end{equation} (10)

    If the condition of Case 1 holds, the following results are obtained

    From h_3(t)\leq g_3(t)\leq p_3(\xi) , the third equation of formula (10) can be scaled down to the following formula

    \begin{equation} \frac{\int^{t}_{0}{x_3(s){\rm d}s}}{t}\leq\frac{1}{p_3(\xi)}(\frac{x_3(t)-x_3(0)}{t}-w_3) \end{equation} (11)

    Taking the limit of both sides of inequality (11), we have

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_3(s){\rm d}s}}{t}\leq \frac{w_3}{p_3(\xi)}\leq b_3 \end{equation} (12)

    By condition (H _1 ) and h_i(t)\leq g_i(t)\leq p_i(\xi) (i = 1, 2) , the first two equations of formula (10) are added together and then scaled down to get

    \begin{equation} \sum\limits_{i = 1}^{2}\frac{p_i(\xi)\int^{t}_{0}{x_i(s){\rm d}s}}{t} \geq \sum\limits_{i = 1}^{2}(\frac{x_i(t)-x_i(0)}{t})-\frac{k_1\int^{t}_{0}{x_3(s){\rm d}s}}{t}-\sum\limits_{i = 1}^{2}w_i \end{equation} (13)

    By substituting inequality (12) into inequality (13) and taking the limit of both sides of inequality (13) yield

    \begin{equation} \lim\limits_{t \to 0}\frac{\int^{t}_{0}{x_i(s){\rm d}s}}{t}\leq-\frac{1}{p_i(\xi)}(w_1+w_2-\frac{k_1w_3}{p_3(\xi)})\leq b_i\ (i = 1, 2) \end{equation} (14)

    Then by h_3(t) > q_3(\xi) , the three equations in (10) can be reduced separately to get

    \begin{equation} \frac{q_{\rm i}\int^{t}_{0}{x_i(s)ds}}{t}\leq\frac{x_i(t)-x_i(0)}{t}-w_i\ (i = 1, 2, 3) \end{equation} (15)

    Taking the limit of both sides of inequality (15), we have

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_i(s){\rm d}s}}{t}\geq-\frac{w_i}{q_i}\geq a_i\ (i = 1, 2, 3) \end{equation} (16)

    From formulas (16) and (H _3 ), the fourth equation of formula (10) can be scaled down to get

    \begin{equation} \frac{x_4(t)-x_4(0)}{t}\leq m_2(x_2(0)-x_2(t))-\frac{n_1\int^{t}_{0}{x_4(s){\rm d}s}}{t}\end{equation} (17)

    Taking the limit of both sides of inequality (17), we have

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_4(s){\rm d}s}}{t}\leq\frac{m_2}{n_1}(x_2(0)+\frac{w_i}{q_i})\leq c \end{equation} (18)

    From formula (12), (14), (16) and (18), we get

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_i(s){\rm d}s}}{t}\in\Lambda\ (i = 1, 2, 3, 4) \end{equation}

    If the condition of Case 2 holds, the following results are obtained

    Take the limit on both sides of the inequality that is obtained by substituting g_3(t) < q_3(\xi) into the third equation of equation (10), we have

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_3(s){\rm d}s}}{t} \lt -\frac{w_3}{q_3(\xi)} \lt a_3 \end{equation} (19)

    By g_i(t)\leq p_i(\xi) (i = 1, 2) and condition (H _1 ), the first two equations of formula (10) are added together and then scaled down to get

    \begin{equation} \sum\limits_{i = 1}^{2}\frac{q_i(\xi)\int^{t}_{0}x_i(s){\rm d}s}{t} \gt \sum\limits_{i = 1}^{2}(\frac{x_i(t)-x_i(0)}{t})-\frac{k_1\int^{t}_{0}x_3(s){\rm d}s}{t}-w_1-w_2 \end{equation} (20)

    By substituting inequality (19) into inequality (20) and taking the limit of both sides of inequality (20) yield

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_i(s){\rm d}s}}{t} \lt -\frac{1}{q_i(\xi)}(w_1+w_2-\frac{k_1w_3}{q_3(\xi)}) \lt a_i\; \; (i = 1, 2) \end{equation} (21)

    From formula (21) and (H _3 ), the fourth equation of formula (10) can be scaled down to get

    \begin{equation} \frac{x_4(t)-x_4(0)}{t} \gt m_2(x_2(0)-x_2(t))-\frac{n_1\int^{t}_{0}{x_4(s){\rm d}s}}{t} \end{equation} (22)

    Taking the limit of both sides of inequality (22), we have

    \begin{equation} \lim\limits_{t \to 0}\frac{\int^{t}_{0}{x_4(s){\rm d}s}}{t} \gt \frac{m_2}{n_2}(x_2(0)-\frac{1}{q_2(\xi)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi)})) \gt c \end{equation} (23)

    From formula (19), (21) and (23), we get

    \begin{equation} \lim\limits_{t \to \infty}\frac{\int^{t}_{0}{x_j(s){\rm d}s}}{t}\not\in\Lambda\; (j = 1, 2, 3, 4). \end{equation}

    Model (1) is used as the basic model to study the changes of three major nutrients and ketone body under weight loss of diet, exercise and drug.

    If glycogen decomposition is not enough to maintain blood glucose balance, the energy of the body can be provided through the cooperation of fat and protein in the case of dieting and non exercise weight loss [30]. Without supplementary energy, the following assumptions can be made for model (1).

    (1) The absorption rate of carbohydrates, fats and proteins from food is \mu_i(t) (i = 1, 2, 3) .

    (2) Fat will not be converted into carbohydrate, that is k_5(t) = 0 .

    (3) The rate of fat and carbohydrate consumption caused by exercise is k_6(t) = k_7(t) = 0 .

    (4) The absorption rate of fat inhibited by drugs is \delta(t) = 0 .

    Based on assumptions (1)-(4) and the modified model (1), we get the following metabolic regulation model during weight loss through dieting.

    \begin{equation} \left\{ \begin{aligned} & \ \mathit{\boldsymbol{x}}^{\prime}(t) = \mathit{\boldsymbol{A}}_1(t)\mathit{\boldsymbol{x}}(t)+\mathit{\boldsymbol{w}}(t)\\ &\ \mathit{\boldsymbol{x}}(0) = \mathit{\boldsymbol{x}}_0\\ \end{aligned} \right. \end{equation} (24)

    where

    \begin{equation} \mathit{\boldsymbol{A}}_1(t) = \left( \begin{array}{cccc} -k_3(t) & \varepsilon(t) & k_1(t) & 0\\ 0 & -\varepsilon(t)-k_4(t) & 0 & 0\\ 0 & 0 & -k_1(t)-k_2(t)& 0\\ 0 & -k(t) & 0& -\upsilon(t)\\ \end{array} \right) \end{equation}

    Knowing from assumptions (1)-(4), the parameters of formula (2) can be rewritten as

    \begin{equation} h_1(t) = -k_3(t), \ h_2(t) = -(\varepsilon(t)+k_4(t)), \ h_3(t) = -(k_1(t)+k_2(t)) \end{equation}

    therefore, the parameters of condition (1) in Theorem 1 are as follows

    \begin{equation} g_1(t) = h_1(t)\leq-\bar{k}_3, \ g_2(t) = -k_4(t)\leq-\bar{k}_4, \ g_3(t) = -k_2(t)\leq-\bar{k}_2 \end{equation}

    The above results satisfy condition 1 of Theorem 1, and the Corollary 1 is the following.

    Corollary 1. If model (24) satisfies conditions |k_3(0)-k_3(t)|+\max(|\varepsilon(t)|, |k_1(t)|) < k_3(0) and \max(|\varepsilon(0)+k_4(0)-\varepsilon(t)-k_4(t)|, \ |k_1(0)+k_2(0)-k_1(t)-k_2(t)|) < \max(\varepsilon(0)+k_4(0), \ k_1(0)+k_2(0)) , when \frac {w_1}{\bar{k}_3} < a_1 , \frac {w_2}{\bar{k}_4} < a_2 , \frac {w_3}{\bar{k}_2} < a_3 , \frac{m_2}{n_1}(x_2(0)+\frac{w_2}{\bar{k}_4}) > c , then \lim_{t \to \infty}\frac{\int^{t}_{0}{x_j(s){\rm d}s}}{t}\not\in\Lambda\ (j = 1, 2, 3, 4).

    Proof. According to the given conditions, model (24) conforms to the condition of Theorem 1. Similarly, from the proof of Case 2 of Theorem 1, we can get

    \begin{equation} \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_1(s){\rm d}s} \lt \frac{w_1}{\bar{k}_3(\xi)} \lt a_1, \ \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_2(s){\rm d}s} \lt \frac{w_2}{\bar{k}_4(\xi)} \lt a_2 \end{equation}
    \begin{equation} \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}s} \lt \frac{w_3}{\bar{k}_2(\xi)} \lt a_3, \ \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}s} \gt \frac{m_2}{n_1}(x_2(0)-\frac{w_2}{\bar{k}_4(t)}) \gt c \end{equation}

    therefore

    \begin{equation} \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}s}\not\in\Lambda\ (j = 1, 2, 3, 4). \end{equation}

    Corollary 1 shows that prolonged dieting can lead to three major nutrients in the body below the normal range. At the same time, with the increase of the demand of transforming fat into carbohydrate, the ketone body produced by fat metabolism will continue to increase to the highest safe critical value, and the human body is more likely to suffer from ketoacidosis.

    A certain amount of exercise will promote fat and carbohydrate metabolism when considering the way of exercise and dietary weight loss. In the process of fat decomposition, a small amount of carbohydrate will be produced. Excessive carbohydrate intake translates into fat. When the body's own energy consumption is much greater than the energy intake of the diet, some of the protein will also be converted into carbohydrate, but will not be converted into fat [29].

    The following assumptions are made for model (1)

    (1) Conversion rate of dietary intake of carbohydrate, fat and protein are \mu_i(t)\neq0 (i = 1, 2, 3) .

    (2) The wear rate of carbohydrate consumption and fat consumption caused by exercise are k_6(t)\neq0 , k_7(t)\neq0 .

    (3) The absorption rate of drugs inhibiting fat in human diet is \delta(t) = 0 .

    Under assumptions (1)-(3), the following metabolic regulation model can be obtained by modifying model (1).

    \begin{equation*} \left\{ \begin{aligned} & \ \mathit{\boldsymbol{x}}^{\prime}(t) = \mathit{\boldsymbol{A}}_2(t)\mathit{\boldsymbol{x}}(t)+\mathit{\boldsymbol{w}}(t)\\ &\ \mathit{\boldsymbol{x}}(0) = \mathit{\boldsymbol{x}}_0\\ \end{aligned} \right. \end{equation*} (25)

    where

    \begin{equation*} \mathit{\boldsymbol{A}}_2(t) = \left( \begin{array}{cccc} h_1(t) & \varepsilon(t) & k_1(t) & 0\\ k_5(0) & \bar{h}_2(t) & 0 & 0\\ 0 & 0 & h_3(t)& 0\\ 0 & -k(t) & 0& -\upsilon(t)\\ \end{array} \right) \end{equation*} (26)

    As is known by assumptions (1)-(3), the parameters of formula (26) are the same as in the formula (2), except for \bar{h}_2(t) = \mu_2(t)-\varepsilon(t)-k_4(t)-k_6(t) . Since there is no drug effect, then \bar{h}_2(t)\geq h_2(t) . In this case, the parameter g_2(t) of condition (1) of Theorem 1 becomes g_2(t) = \bar{h}_2(t)-\varepsilon(t) , and the other parameters remain unchanged. The parameters q_i(\xi_1)\leq p_i(\xi_1) (\xi_1 = (\bar{\mu}_i, \mu_i, k_h, \varepsilon)) are still negative and only depend on the values \bar{\mu}_i , \mu_i (i = 1, 2, 3) , k_h\ (h = 5, 6, 7) and \varepsilon . Thus, the corollary is the following.

    Corollary 2. If model (25) satisfies conditions (1)-(2) of Theorem 1, and \max(|\bar{h}_2(t)-\bar{h}_2(0)|, |h_3(t)-h_3(0)|)+|k_5(t)| < \max(|\bar{h}_2(0)|, |h_3(0)|) , then the solution of model (25) has the following two cases

    Case 1 When \bar{\mu}_i\leq \mu_i(t)\leq \mu_i , -\frac{w_i}{q_i(\xi_1)}\geq a_i\ (i = 1, 2, 3) , q_s(\xi_1)\leq h_s(t)\leq g_s(t)\leq p_s(\xi_1) (s = 1, 3) , q_2(\xi_1)\leq h_2(t)\leq\bar{g}_2(t)\leq p_2(\xi_1) , -\frac{1}{p_r(\xi_1)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{p_3(\xi_1)})\leq b_r (r = 1, 2) , -\frac{w_3}{p_3(\xi_1)}\leq b_3 and \frac{m_2}{n_1}(x_2(0)+\frac{w_2}{q_2(\xi_1)})\leq c , then \lim_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}s}\in\Lambda (j = 1, 2, 3, 4) .

    Case 2 When 0 < \mu_i(t) < \bar{\mu} , g_i(t) < q_i(\xi_1) (i = 1, 3) , -\frac{1}{q_r(\xi_1)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_1)}) < a_r (r = 1, 2) , -\frac{w_3}{q_3(\xi_1)} < a_3 and \frac{m_1}{n_2}(x_2(0)+\frac {1}{q_2(\xi_1)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_1)})) > c , then \lim_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}(s)}\not\in\Lambda (j = 1, 2, 3, 4) .

    Proof. According to the given conditions, model (25) satisfies the condition of Theorem 1. Similarly, from the proof of Case 1 of Theorem 1, we get

    a_r\leq -\frac{w_r}{q_r(\xi_1)}\leq \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_r(s){\rm d}s}\leq-\frac{1}{p_r(\xi_1)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_1)})\leq b_r\; \; (r = 1, 2)
    a_3\leq -\frac{w_3}{q_3(\xi_1)}\leq\lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}s}\leq -\frac{w_3}{p_3(\xi_1)}\leq b_3
    \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}s}\leq\frac{m_2}{n_1}(x_2(0)+\frac{w_2}{q_2(\xi_1)})\leq c

    then \lim_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}(s)}\in\Lambda (j = 1, 2, 3, 4) .

    Proceeding as in the proof of Case 2 of Theorem 1, we have

    \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_r(s){\rm d}s} \lt -\frac{1}{q_r(\xi_1)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_1)}) \lt a_r\; \; (r = 1, 2)
    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)} \lt -\frac{w_3}{q_3(\xi_1)} \lt a_3
    \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_4(s)ds} \gt \frac{m_1}{n_2}(x_2(0)+\frac {1}{q_2(\xi_1)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_1)})) \gt c

    then \lim_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_j(s)d(s)}\not\in\Lambda (j = 1, 2, 3, 4) .

    It can be seen from Corollary 2 that healthy weight loss can be achieved when the amount of diet and exercise satisfy Case 1. If the amount of diet and exercise satisfy Case 2, long-term weight loss will lead to the imbalance of the three major nutrients in the human body, so that the three nutrients are lower than the normal range. Therefore, It is necessary to control energy intake and amount of exercise reasonably to achieve the expected weight loss effect.

    For the people with severe obesity and those who are not sensitive to weight loss of exercise or dieting, they should be treated with drugs. Gastrointestinal lipase inhibitors can effectively inhibit the absorption of fat, reduce the intake of calories, decrease the gluconeogenesis, cut down the conversion rate of fat to carbohydrate, and ultimately achieve the goal of weight loss [32,33]. Depending on the human body's consideration of the role of those drugs in the process of non-exercise weight loss, the following modifications and assumptions can be made to model (1).

    (1) The absorption rate of fat inhibited by drugs is \delta(t)\neq0 .

    (2) The rate of fat conversion into carbohydrate under the effect of diet pills changed to \tilde{\varepsilon}(t) to satisfy \tilde{\varepsilon}(t)\leq\tilde{\varepsilon}(t) < \varepsilon .

    (3) Dietary intake parameters \mu_i\neq0\; (i = 1, 2, 3) , and the consumption rate of fat and carbohydrate caused by exercise became k_6(t) = k_7(t) = 0 .

    Under assumptions (1)-(3), the following metabolic regulation models can be obtained in the course of weight loss drugs.

    \begin{equation*} \left\{ \begin{aligned} & \ \mathit{\boldsymbol{x}}^{\prime}(t) = \mathit{\boldsymbol{A}}_3(t)\mathit{\boldsymbol{x}}(t)+\mathit{\boldsymbol{w}}(t)\\ &\ \mathit{\boldsymbol{x}}(0) = \mathit{\boldsymbol{x}}_0\\ \end{aligned} \right. \end{equation*} (27)

    where

    \begin{equation*} \mathit{\boldsymbol{A}}_3(t) = \left( \begin{array}{cccc} \tilde{h}_1(t) & \tilde{\varepsilon}_(t) & k_1(t) & 0\\ k_5(t) & \tilde{h}_2(t) & 0 & 0\\ 0 & 0 & h_3(t)& 0\\ 0 & -k(t) & 0& -\upsilon(t)\\ \end{array} \right) \end{equation*} (28)

    As is known by assumptions (1)-(3), the parameters of formula (28) are \tilde{h_1}(t) = \mu_1(t)-k_3(t)-k_5(t) and \tilde{h_2}(t) = \mu_2(t)-k_4(t)-\tilde{\varepsilon}(t)-\delta(t) , except that h_3(t) is the same as the parameter of formula (2). Since there is no exercise effect, then \tilde{h}_1(t) \geq h_1(t) and \tilde{h}_2(t) \geq h_2(t) .

    The parameters g_1(t) and g_2(t) of condition (1) of Theorem 1 become g_1(t) = \tilde{h}_1(t)+k_5(t) and g_2(t) = \tilde{h}_2(t)+\tilde{\varepsilon}(t) . The parameters q_i(\xi_2)\leq p_i(\xi_2) (\xi_2 = (\bar{\mu_i}, \mu_i, k_5, \tilde{\varepsilon}, \delta)) are still negative and only depend on the values of \bar{\mu_i} , \mu_i\; (i = 1, 2, 3) , k_5 , \tilde{\varepsilon} and \delta . Thus, the corollary is the following.

    Corollary 3. If model (27) satisfies conditions (1) of Theorem 1, and |\tilde{h}_1(t)-\tilde{h}_1(0)|+\max(|\tilde{\varepsilon}(t)|, |k_1(t)|) < |\tilde{h}_1(0)| , \max(|\tilde{h}_2(t)-\tilde{h}_2(0)|, |h_3(t)-h_3(0)|)+|k_5(t)| < \min(|\tilde{h}_2(0), h_3(0)|) , then the solution of model (27) has the following two cases

    Case 1 When \bar{\mu}_i\leq \mu_i(t)\leq \mu_i , -\frac{w_i}{q_i(\xi_2)}\geq a_i\ (i = 1, 2, 3) , q_r(\xi_2)\leq \tilde{h}_r(t)\leq g_r(t)\leq p_r(\xi_2) (r = 1, 2) , q_3(\xi_2)\leq h_3(t)\leq g_3(t)\leq p_3(\xi_2) , -\frac{1}{p_r(\xi_2)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{p_3(\xi_2)})\leq b_r (r = 1, 2) , -\frac{w_3}{p_3(\xi_2)}\leq b_3 and \frac{m_2}{n_1}(x_2(0)+\frac{w_2}{q_2(\xi_2)})\leq c , then \lim_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}}\in\Lambda (j = 1, 2, 3, 4) .

    Case 2 When 0 < \mu_i(t) < \bar{\mu_i} , g_i(t) < q_i(\xi_2) , -\frac{1}{q_r(\xi_2)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_2)}) < a_r (r = 1, 2) , -\frac{w_3}{q_3(\xi_2)} < a_3 , \frac{m_1}{n_2}(x_2(0)+\frac {1}{q_2(\xi_2)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_2)})) > c , then \lim_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}s}\not\in\Lambda (j = 1, 2, 3, 4) .

    Proof. According to the given conditions, model (27) satisfies the condition of Theorem 1. Similarly, from the proof of Case 1 of Theorem 1, we get

    a_r\leq -\frac{w_r}{q_r(\xi_2)}\leq \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_r(s){\rm d}s}\leq-\frac{1}{p_r(\xi_2)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{p_3(\xi_2)})\leq b_r\; (r = 1, 2)
    a_3\leq\frac{w_3}{q_3(\xi_2)}\leq\lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}s}\leq -\frac{w_3}{p_3(\xi_2)}\leq b_3
    \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}s}\leq\frac{m_2}{n_1}(x_2(0)+\frac{w_2}{q_2(\xi_2)})\leq c

    then \lim_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}(s)}\in\Lambda (j = 1, 2, 3, 4) .

    From the proof of Case 2 of Theorem 1, we have

    \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_r(s){\rm d}s} \lt -\frac{1}{q_r(\xi_2)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_2)}) \lt a_r\; \; (r = 1, 2)
    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)} \lt -\frac{w_3}{q_3(\xi_2)} \lt a_3
    \lim\limits_{t \to \infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}s}\geq \frac{m_1}{n_2}(x_2(0)+\frac {1}{q_2(\xi_2)}(\sum\limits_{i = 1}^{2}w_i-\frac{k_1w_3}{q_3(\xi_2)})) \gt c

    then \lim_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_j(s){\rm d}(s)}\not\in\Lambda (j = 1, 2, 3, 4) .

    According to Corollary 3, when model (3) satisfies the condition of Case 1, it can achieve healthy weight loss. Otherwise, Case 2 is established and the three major nutrients will be below the normal range for a limited time. As there is no exercise, the ability of fat consumption is reduced. Since weight loss with drugs is a short-term effect, in the long run, the three major nutrients are in the normal range for a longer period of time than weight loss with exercise. The weight loss effect of the combination of drugs and exercise is more reasonable.

    The normal range of human blood glucose is from 60 mg/dL to 80 mg/dL. If it is lower than the lower limit of 45 mg/dL or higher than the upper limit of 120 mg/dL, the risk of late blood glucose symptoms and diabetes mellitus will low. The normal value of total protein content in human serum is 60-80 g/L. If the total protein of human body is lower than 60 g/L, the power of immunity will decrease. According to the proportion of body fat weight to total body weight, the normal body fat of adults ranges from 15% to 25%. The body fat more than the upper limit 25% can be diagnosed as obesity. If the body fat is below the lower limit of 8%, it will cause human dysfunction. When the safety threshold of ketone body in human body is 3 mmol/L, exceeding this standard will lead to ketoacidosis [29]. According to the interview with the sixth volunteer who participated in the 7-day fasting and weight loss experiment, when the blood glucose dropped to the range of 45-60 mg/dl, the volunteer did not appear hypoglycemia and other discomfort symptoms. In this section, for the sixth volunteer, the normal ranges of the three major nutrients and ketone body in the human body are selected.

    \Lambda = \{x_j(t)\ (j = 1, 2, 3, 4)|\ 45\leq x_1(t)\leq 120, \ 8\leq x_2(t)\leq 25,
    \; \; \; \; \; \; \; \; \; 60\leq x_3(t)\leq 80, \ 0\leq x_3(t)\leq 3\ (i = 1, 2, 3, 4)\}

    In the process of weight loss, the values of the three major nutrients metabolic indicators can be obtained by real-time blood glucose collection and blood routine. By using the metabolic regulation model in the process of weight loss, we can quantitatively analyze the changes of biochemical indexes of the three major nutrients, predict the impact of different weight loss methods on the health of the body, and guide the correct weight loss methods.

    Firstly, based on the actual three days nutrient detection data of seven days fasting and weight loss, some parameter values of model (1) are estimated, and then the effects of different weight loss methods on the three major nutrient changes are analyzed according to the theorem conditions. According to the actual data, the initial values of the three major nutrients are x_1(0) = 73, \ x_2(0) = 29.2 and x_3(0) = 62 respectively.

    1. Parameter estimation

    The carbohydrate, fat and protein consumed by the body during weight loss through dieting are used to maintain metabolism. When the blood glucose decreases, the conversion rate of fat and protein will gradually increase, which makes the blood glucose present a dynamic equilibrium state. It may be advisable to consider the conversion rate k_i(t) and \varepsilon(t) as time-dependent periodic functions in the following forms

    \begin{equation} \left\{ \begin{aligned} & \ k_i(t) = a_i+b_isin(t)\ (i = 1, 2, 3, 4)\\ &\ \varepsilon(t) = a+bsin(t)\\ \end{aligned} \right. \end{equation} (29)

    By substituting formula (29) into model (24), we obtain the partial parameter values of model (24), which are estimated by the data for 7 days of carbohydrate, fat and protein metabolism indexes of the 6th volunteer [26] by using numerical approximation algorithm [34]. Then we get

    \begin{equation*} \begin{aligned} & k_1(t) = 0.001, \ k_2(t) = 0.02+0.01sin{t}, \ k_3(t) = 0.084+0.006sin{t}\\ &\ k_4(t) = 0.086+0.004sin{t}, \ \varepsilon(t) = 0.0025sin{t}, w_i = 0.3\ (i = 1, 2, 3) \end{aligned} \end{equation*} (30)

    From the literature[35,36], we choose k(t) = 0.0034 and \upsilon(t) = 0.025 . The theoretical values of blood glucose, body fat, total protein and ketone body can be calculated numerically for model (24) and plotted in the same graph with the actual data collected at 9 o'clock per day. The simulated curves of blood glucose, body fat, total protein and ketone body for 7 days are obtained as shown in Figure 6 (1)-(3). (The solid line represents the model fitting curve, and the dashed line represents the experimental data fitting curve) The simulated curve of theoretical value of ketone body is shown in Figure 6 (4). (The solid line in the figure represents the model fitting curve)

    Figure 6.  Changing curve of three major nutrient metabolic indicators in the first week.

    The relative error value of fitting can be calculated from the actual values and the theoretical values of the numerical simulation in Figure 6 (1)-(3) as shown in Table 2.

    Table 2.  Relative error between fitting result and actual value.
    Relative error of index Before weight loss First day Second day Third day Fourth day Fifth day Sixth day Seventh day
    Relative error of blood glucose 0 0.027 0.038 0.018 0.009 0.009 0.003 0.016
    Relative error of total protein 0 0.016 0.004 0.006 0.013 0.016 0.003 0.003
    Relative error of body fat rate 0.01 0.025 0.009 0.007 0.02 0.008 0 0.0007

     | Show Table
    DownLoad: CSV

    From Table 2, the overall average error of fitting is 0.01, which shows that the fitting effect is good. Figure 6 (1) shows that blood glucose fluctuate between 50 mg/dL and 60 mg/dL after two days. Figure 6 (2) shows that body fat decreases from 29.7% before weight loss to 26.7% on the 7th day. Figure 6 (3) shows that total protein fluctuate within the normal range. Figure 6 (4) shows that ketone body index rises rapidly after the first day of the fasting and increase to 1.47 mmol/L on the 7th day. During the 7-day weight loss period, the blood glucose and body fat of the volunteers show a downward trend of cyclical fluctuation, but basically keep within the safe range.

    2. Numerical simulation of three major nutrients and ketone body

    The parameters are selected by formula (30) are substituted into the conclusion of Corollary 1 to verify the validity, and model (24) satisfies

    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_1(s){\rm d}(s)}\leq8, \ \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_2(s){\rm d}(s)}\leq4.6
    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)}\leq30, \ \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}(s)} \gt 3

    The above results indicate that the three major nutrients and ketone body indexes of the body are not within the safe value range in the long-term weight loss. Explain that the body's three major nutrients and ketone body indicators are not in the safe range after long-term dieting. This result shows that dieting should not be carried out for a long time. The model (24) is used to simulate the predictive curves of blood glucose, total protein, body fat and ketone body for 7 to 30 days as shown in Figure 7. The numerical results are shown in Table 3.

    Figure 7.  Predictive curve of three major nutrient metabolic indicators.
    Table 3.  Description table of three nutrients and ketone body changes.
    Time (Day) Blood glucose (mg/dL) Body fat (%) Total protein (g/L) Ketone body (mmol/L) Characteristic description of weight loss
    7 54.6 26.7 62 1.47 Within the normal range
    9 49 14 57.4 2.5 Total protein below the lower limit of normal value
    13 49 8 55 3.1 Total protein and ketone body beyond normal range
    23 44 5 42 7 Not within the normal range

     | Show Table
    DownLoad: CSV

    Suggestions for weight loss by dieting: According to the above analysis, the 13th day after losing weight by dieting will be harmful to health, and the behavior of losing weight by fasting should be stopped. Otherwise, on the 23rd day, life may be endangered.

    Considering the weight loss mode of exercise and diet, k_j(t)\ (j = 2, 3, 4) and \varepsilon(t) in model (25) are still regarded as time-dependent periodic functions. k_j(t)\ (j = 2, 3, 4) , \varepsilon(t) and w_i\ (i = 1, 2, 3) selection are the same as the functions in formula (30). When the human body ingests food, the absorption rate of the three major nutrients is gradually increased, and its absorption rate will gradually decrease with the metabolism and energy consumption[37,38]. The absorption rate \mu_i(t)\ (i = 1, 2, 3) of the three major nutrients can also be regarded as a time-dependent periodic function. The parameters \mu_i(t) , \bar{\mu_i} , \mu_i (i = 1, 2, 3) and k_h(t)\ (h = 5, 6, 7) for exercise and diet are shown in Table 4.

    Table 4.  Description table of three nutrients and ketone body changes.
    Case \mu_1(t) \mu_2(t) \mu_3(t) k_5(t) k_6(t) k_7(t) \mu_1 \mu_2 \mu_3 \bar{\mu}_1 \bar{\mu}_2 \bar{\mu}_3
    1 0.086
    8+0.0
    143
    sint
    0.08
    4+0.0
    05
    sint
    0.02
    145+
    0.015
    75sint
    0.01
    69
    0.0
    09
    0.0
    00
    8
    0.0
    10
    11
    0.0
    89
    0.0
    37
    2
    0.0
    70
    4
    0.0
    72
    4
    0.0
    05
    7
    2 0.03 0.005 0.002 0.01
    69
    0.0
    47
    0.027 0.03 0.03 0.005 0.005 0.002 0.002

     | Show Table
    DownLoad: CSV

    Two cases of the selected parameters in Table 4 are respectively substituted into model (25), and the conclusion of Corollary 2 is valid. The parameters of two Cases in Table 4 are substituted into the model (25), and the conclusion of Corollary 2 is valid. The numerical simulation of model (25) can be obtained in the following two cases. The numerical simulation of model (25) can be obtained under two Cases.

    In Case 1, the solution of model (25) satisfies the following relationship

    45.45\leq\lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_1(s){\rm d}(s)}\leq105, \ 10\leq\lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_2(s){\rm d}(s)}\leq25
    60\leq\lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)}\leq69.8, \ 0 \lt \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}(s)} \lt 3

    In Case 1, the weight loss can keep the three major nutrients and ketone body indicators within the normal range. The trend curves of the numerical simulation of model (25) in Case 1 are shown in Figure 8 (1)-(4).

    Figure 8.  Changing curve of three major nutrient metabolic indicators under Case 1.

    Figure 8 (1) shows that blood glucose fluctuated within the range of 80 mg/dL-120 mg/dL after 123 days. Figure 8 (2) shows that body fat is within the range of 18%-20%. The total protein in Figure 8 (3) is always within the range of 60 g/L-80 g/L. Figure 8 (4) shows that the values of ketone body continue to rise within the safe range.

    In Case 2, the solution of model (25) satisfies the following relationship

    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_1(s){\rm d}(s)} \lt 11, \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_2(s){\rm d}(s)} \lt 8, \ \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)} \lt 38, \ \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}(s)} \gt 3

    In Case 2, the weight loss ultimately excludes the three major nutrients and ketone body indicators from the safe range. The trend curves of the numerical simulation of model (25) in Case 2 are show in Figure 9 (1)-(4). The changes of three major nutrients and ketone body are described in Table 5.

    Figure 9.  Changing curve of three major nutrient metabolic indicators under Case 2.
    Table 5.  Description table of three major nutrients and ketone body changes.
    Time (Day) Blood glucose (mg/dL) Body fat (%) Total protein (g/L) Ketone body (mmol/L) Characteristic description of weight loss
    40 55 8 61 3.2 Body fat and ketone body beyond normal range
    100 44 6.7 59.5 5 Blood glucose and total protein below the lower limit of normal value
    200 30 4.9 58.9 6 Not within the normal range

     | Show Table
    DownLoad: CSV

    Exercise and dietary weight loss suggestion: The effect of weight loss can be achieved by reasonably controlling diet and exercise, while the three major nutrients can be controlled within the safe range. If the diet intake too little (too much) and exercise too much (too little), which can not achieve the expected goal of weight loss, it will also damage to your health, and may even endanger your life. This method is better than diet and weight loss.

    Considering the way of the weight loss with drugs and without exercise, the parameters k_j(t)\; (j = 1, 2, 3, 4) , k_j(t) and w_i\; (i = 1, 2, 3) in model (27) are still regarded as periodic functions related to time, and their selection are the same as the functions in formula (30). According to [39], the absorption rate of fat blocked by drugs is \delta(t) = 0.3\mu_2(t) , and the rate of fat conversion to blood glucose is \tilde{\varepsilon}(t) = 0.3\varepsilon(t) . Two groups of different parameters are selected, as shown in Table 6.

    Table 6.  Parameter selection.
    Case \mu_i(t)\ (i=1, 2, 3), \bar{\mu_i}, k_5(t) \tilde{\varepsilon}(t)=0.3\varepsilon(t) \delta(t)=0.3\mu_2(t)
    1 Consistent with the parameters in Table 4 0.0012 0.0252+0.0015sint
    2 0.0015

     | Show Table
    DownLoad: CSV

    Two Cases of the selected parameters in Table 6 are substituted into model (27) to verify the validity of the conclusion of Corollary 3, and the numerical simulation can be obtained under two cases.

    In Case 1, the solution of model (27) satisfies the following relation

    52\leq\lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_1(s){\rm d}(s)}\leq119.6, \ 11\leq\lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_2(s){\rm d}(s)}\leq25
    60\leq\lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)}\leq69.8, \ 0 \lt \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}(s)} \lt 3

    In Case 1, the weight loss can keep the body's three major nutrients and ketone body indicators within the normal range. The trend curves of the numerical simulation of model (27) in Case 1 are shown in Figure 10 (1)-(4).

    Figure 10.  Changing curve of three major major nutrient metabolic indicators under Case 1.

    Figure 10(1) shows that blood glucose fluctuates within the range of 45 mg/dL-120 mg/dL. Figure 10(2) shows that body fat is maintained between 18% and 22%. Figure 10(3) shows that the total protein is always within the range of 60 g/L-80 g/L. Figure 10(4) shows that the values of ketone body continue to rise within the safe range.

    In Case 2, the solution of model (27) satisfies the following relationship

    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_1(s){\rm d}(s)} \lt 20, \ \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_2(s){\rm d}(s)} \lt 8
    \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_3(s){\rm d}(s)} \lt 38, \ \lim\limits_{t\to\infty}\frac{1}{t}\int^{t}_{0}{x_4(s){\rm d}(s)} \gt 3

    In Case 2, the weight loss ultimately makes the three major nutrients and ketone body indicators out of the safe range. The trend curves of the numerical simulation of model (27) in Case 2 are shown in Figure 11 (1)-(4). The changes of three major nutrients and ketone body are described in Table 7.

    Figure 11.  Changing curve of three major nutrient metabolic indicators under Case 2.
    Table 7.  Description table of three major nutrients and ketone body changes.
    Time (Day) Blood glucose (mg/dL) Body fat (%) Total protein (g/L) Ketone body (mmol/L) Characteristic description of weight loss
    40 60 8.9 61.5 3 Ketone body reaches the upper limit of normal value
    150 44 10 59.5 4.5 Blood glucose and total protein below the lower limit of normal value, and the ketone body beyond normal range.
    300 36 6 59 5.1 Blood glucose, total protein and ketone body are not in normal range

     | Show Table
    DownLoad: CSV

    Weight loss with drugs and without exercise:Under the conditions of Case 1, the three major nutrients and ketone body can be controlled within the safe range. Under the conditions of Case 2, the weight loss with drugs unsafe to lose weight. At the same time, the effect of long-term weight loss is not obvious, especially, the medication may also produce digestive symptoms[40]. For patients with mild obesity, it is recommended to take exercise weight loss to reduce weight. For those who with severe obesity and who are not sensitive to exercise or dieting weight loss, adjuvant medication can be used.

    Based on the biological principles and experimental data of weight loss by dieting, the metabolic regulation models under different weight loss models were established. Based on the normal values of the three major nutrients, the feasible conditions of weight loss by dieting, exercise and drugs were given by using the changing trend of the model solution. By using the numerical simulation method, the rationality of the model, the effectiveness of different weight-loss methods and the impact of different ways of weight reduction on the health of volunteers were analyzed, which provided theoretical reference for guiding weight loss.

    The different weight-loss methods are effective. The way of weight loss with proper diet and exercise is better than that without food and drug. Short time dieting can achieve safe weight loss, while long time dieting will affect health, which is dangerous. Drug weight loss can achieve safe weight loss with proper diet, but it has certain side effects, which is not recommended for patients with mild obesity. Therefore, the patients with different obesity should choose the appropriate way to reduce weight according to their own obesity situation.

    This model did not consider the metabolism of different age and gender groups in different ways of weight loss and the impact of weight loss on health, which is important in the actual process of weight loss, and the weight loss model will be further studied.

    This work was supported by the project of Nature Scientific Foundation of Heilongjiang Province (A2016004).

    The authors declare there is no conflict of interest.



    [1] L. A. Zadeh, Fuzzy sets, Inf. Control, 8 (1965), 338–353. https://doi.org/10.1016/S0019-9958(65)90241-X doi: 10.1016/S0019-9958(65)90241-X
    [2] L. A. Zadeh, The concept of a linguistic variable and its application to approximate reasoning–-Ⅰ, Inf. Sci., 8 (1975), 199–249. https://doi.org/10.1016/0020-0255(75)90036-5 doi: 10.1016/0020-0255(75)90036-5
    [3] L. A. Zadeh, A note on Z-numbers, Inf. Sci., 181 (2011), 2923–2932. https://doi.org/10.1016/j.ins.2011.02.022 doi: 10.1016/j.ins.2011.02.022
    [4] B. Kang, D. Wei, Y. Li, Y. Deng, Decision making using Z-numbers under uncertain environment, J. Comput. Infor. Syst., 8 (2012), 2807–2814.
    [5] P. Patel, S. Rahimi, E. Khorasani, Applied Z-numbers, 2015 Annual Conf. North American Fuzzy Inf. Proc. Soc., 5 (2015), 1–6. https://doi.org/10.1109/NAFIPS-WConSC.2015.7284154 doi: 10.1109/NAFIPS-WConSC.2015.7284154
    [6] Z. Ren, H. Liao, Y. Liu, Generalized Z-numbers with hesitant fuzzy linguistic information and its application to medicine selection for the patients with mild symptoms of the COVID-19, Comput. Industr. Engin., 145 (2020), 106517. https://doi.org/10.1016/j.cie.2020.106517 doi: 10.1016/j.cie.2020.106517
    [7] T. Zamali, M. A. Lazim, Multi-criteria decision making based on Z-Number valuation for uncertain information, 2021 2nd Int. Conf. Artif. Intell. Data Sci., 2 (2021), 1–4. https://doi.org/10.1109/AiDAS53897.2021.9574134 doi: 10.1109/AiDAS53897.2021.9574134
    [8] B. Kang, D. Wei, Y. Li, Y. Deng, A method of converting Z-number to classical fuzzy number, J. Infor. Comput. Sci., 9 (2012), 703–709.
    [9] R. A. Aliev, A. Alizadeh, R. R. Aliyev, O. H. Huseynov, The arithmetic of Z-Numbers: Theory and applications, Singapore: World Scientific, 2015. https://doi.org/10.1142/9575
    [10] R. A. Aliev, A. V. Alizadeh, O. H. Huseynov, The arithmetic of discrete Z-numbers, Infor. Sci., 290 (2015), 134–155. http://dx.doi.org/10.1016/j.ins.2014.08.024 doi: 10.1016/j.ins.2014.08.024
    [11] R. A. Aliev, O. H. Huseynov, L. M. Zeinalova, The arithmetic of continuous Z-numbers, Inf. Sci., 373 (2016), 441–460. https://doi.org/10.1016/j.ins.2016.08.078 doi: 10.1016/j.ins.2016.08.078
    [12] A. S. A. Bakar, A. Gegov, Multi-layer decision methodology for ranking Z-numbers, Int. J. Comput. Intell. Syst., 8 (2015), 395–406. https://doi.org/10.1080/18756891.2015.1017371 doi: 10.1080/18756891.2015.1017371
    [13] R. A. Aliev, O. H. Huseynov, R. Serdaroglu, Ranking of Z-numbers and its application in decision making, Int. J. Inf. Tech. Decision Making, 15 (2016), 1503–1519. https://doi.org/10.1142/S0219622016500310 doi: 10.1142/S0219622016500310
    [14] S. Ezadi, T. Allahviranloo, New multi-layer method for Z-number ranking using hyperbolic tangent function and convex combination, Int. Autom. Soft Comput., 2017, 1–7. https://doi.org/10.1080/10798587.2017.1367146 doi: 10.1080/10798587.2017.1367146
    [15] S. Ezadi, T. Allahviranloo, S. Mohammadi, Two new methods for ranking of Z-numbers based on sigmoid function and sign method, Int. J. Intell. Syst., 33 (2018), 1476–1487. https://doi.org/10.1002/int.21987 doi: 10.1002/int.21987
    [16] Online content: OSPF protocol analysis, 1991. Available from: https://www.rfc-editor.org/rfc/rfc1245.
    [17] Y. Deng, Y. Chen, Y. Zhang, S. Mahadevan, Fuzzy Dijkstra algorithm for shortest path problem under uncertain environment, Appl. Soft Comput., 12 (2012), 1231–1237. https://doi.org/10.1016/j.asoc.2011.11.011 doi: 10.1016/j.asoc.2011.11.011
    [18] K. K. Mishra, Dijkstra's Algorithm for solving fuzzy number Shortest Path Problem, Malaya J. Matematik, 8 (2019), 714–719.
    [19] A. Candra, M. A. Budiman, K. Hartanto, Dijkstra's and a-star in finding the shortest path: A tutorial, J. 2020 Int. Conf. Data Sci. Artif. Intell. Business Anal., 58 (2020), 28–32. http://dx.doi.org/10.1109/DATABIA50434.2020.9190342 doi: 10.1109/DATABIA50434.2020.9190342
    [20] M. Akram, A. Habib, J. C. R. Alcantud, An optimization study based on Dijkstra algorithm for a network with trapezoidal picture fuzzy numbers, Neural Comput. Appl., 33 (2021), 1329–1342. https://doi.org/10.1007/s00521-020-05034-y doi: 10.1007/s00521-020-05034-y
    [21] S. S. Biswas, Z-Dijkstra's algorithm to solve shortest path problem in a Z-Graph, Oriental J. Comput. Sci. Techn., 10 (2017), 180–186. http://dx.doi.org/10.13005/ojcst/10.01.24 doi: 10.13005/ojcst/10.01.24
    [22] P. Veerammal, Fuzzy Z-Number shortest path problem using Dijkstra algorithm, Periodico di Mineral., 91 (2022), 942–950.
    [23] A. Papoulis, Random variables and stochastic processes, New York: McGraw Hill, 1965.
    [24] M. D. Springer, The algebra of random variables, New York: Wiley, 1979.
    [25] B. Bede, Fuzzy analysis, Heidelberg: Springer, 2013. https://doi.org/10.1007/978-3-642-35221-8_8
    [26] E. Eljaoui, S. Melliani, L. S. Chadli, Multiplication operation and powers of trapezoidal fuzzy numbers, Cham: Springer, 2019. https://doi.org/10.1007/978-3-030-02155-9_11
    [27] H. Nasseri, Fuzzy numbers: Positive and nonnegative, Int. Mathem. Forum, 3 (2008), 1777–1780.
    [28] B. Kang, Y. Deng, R. Sadiq, Total utility of Z-number, Appl. Intell., 48 (2018), 703–729. https://doi.org/10.1007/s10489-017-1001-5 doi: 10.1007/s10489-017-1001-5
    [29] K. R. Saoub, Graph theory: An introduction to proofs, algorithms, and applications, Florida: CRC Press, 2021.
    [30] E. W. Dijkstra, A note on two problems in connexion with graphs, New York: Association for Computing Machinery, 2022. https://doi.org/10.1145/3544585.3544600
    [31] A. Atary, A. Bremler-Barr, Efficient round-trip time monitoring in OpenFlow networks, IEEE INFOCOM 2016 35th Annual IEEE Int. Conf. Comput. Commun., 31 (2016), 1–9. https://doi.org/10.1109/INFOCOM.2016.7524501 doi: 10.1109/INFOCOM.2016.7524501
    [32] G. Martínez, J. A. Hernández, P. Reviriego, P. Reinheimer, Round trip time (rtt) delay in the internet: Analysis and trends, IEEE Net., 38 (2023), 280–285. https://doi.org/10.1109/MNET004.2300008 doi: 10.1109/MNET004.2300008
    [33] M. T. Keller, W. T. Trotter, Applied combinatorics, California: Open Textbook Library, 2017.
    [34] D. Rachmawati, L. Gustin, Analysis of Dijkstra's algorithm and A* algorithm in shortest path problem, J. Phys. Confer. Series, 1566 (2020), 012061. http://dx.doi.org/10.1088/1742-6596/1566/1/012061 doi: 10.1088/1742-6596/1566/1/012061
    [35] B. Erkayman, E. Gundogar, A. Yılmaz, An integrated fuzzy approach for strategic alliance partner selection in third-party logistics, Sci. World J., 2012 (2012), 486306. http://dx.doi.org/10.1100/2012/486306 doi: 10.1100/2012/486306
    [36] M. Ebrat, R. Ghodsi, Construction project risk assessment by using adaptive-network-based fuzzy inference system: An empirical study, KSCE J. Civil Engin., 18 (2014), 1213–1227. https://doi.org/10.1007/s12205-014-0139-5 doi: 10.1007/s12205-014-0139-5
  • This article has been cited by:

    1. Mantana Chudtong, Andrea De Gaetano, Food choices and body weight changes: A mathematical model analysis, 2025, 22, 1551-0018, 1790, 10.3934/mbe.2025065
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1233) PDF downloads(108) Cited by(0)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog