Research article

Voltage stability enhancement in grid-connected microgrid using enhanced dynamic voltage restorer (EDVR)

  • Received: 19 October 2020 Accepted: 21 December 2020 Published: 07 January 2021
  • Microgrid (MG) has extensive properties to overcome common problems of local distribution system. Some of those problems are generation and demand difference, blackout and brownout, environmental concerns due to burning of natural resources in power stations (indirectly), and reliability issues. Research on microgrid is being conducted to enhance its features to mitigate power quality (PQ) problems associated with distribution system. Voltage sag and swell have been major power quality problems for decades, loads in distribution system are heavily affected due to these power quality problems. In the distribution system, microgrid and power quality compensation strategy should be existed in order to ensure reliability and voltage sag/swell mitigation. Dynamic voltage restorer (DVR) is comprehensive power electronics based Flexible Alternating Current Transmission System (FACTS) device, it is third-generation FACTS device as its control scheme selection flexibility and power line coupling approach make it advance when compare to first and second-generation FACTS devices. In this paper, an Enhanced Dynamic Voltage Restorer (EDVR) is presented to efficiently mitigate voltage sag/swell in grid connected microgrid. On the one side, the presence of microgrid structure ensures reliability of distribution system for local loads on the other side, EDVR ensures voltage sag/swell free power supply for loads. The control strategy of EDVR is based on enhanced synchronous reference frame (ESRF) approach and fuzzy technique system. ESRF is specially design for fast and precise operation of EDVR whereas; fuzzy technique system is responsible for standardized voltage supply for local loads. DC link voltage of EDVR is effectively regulated with the help of proposed control scheme at the time of voltage sag/swell compensation. Stability analysis of ESRF control has been done using modeling of VSC and eigenvalue analysis system. Simulation results on MATLAB/Simulink verified the performance of EDVR under presented control approach hence the specific loads in distribution system are more secure under proposed microgrid system with EDVR.

    Citation: Ahsan Iqbal, Ayesha Ayoub, Asad Waqar, Azhar Ul-Haq, Muhammad Zahid, Syed Haider. Voltage stability enhancement in grid-connected microgrid using enhanced dynamic voltage restorer (EDVR)[J]. AIMS Energy, 2021, 9(1): 150-177. doi: 10.3934/energy.2021009

    Related Papers:

    [1] Xianyi Li, Xingming Shao . Flip bifurcation and Neimark-Sacker bifurcation in a discrete predator-prey model with Michaelis-Menten functional response. Electronic Research Archive, 2023, 31(1): 37-57. doi: 10.3934/era.2023003
    [2] Mengxin He, Zhong Li . Dynamic behaviors of a Leslie-Gower predator-prey model with Smith growth and constant-yield harvesting. Electronic Research Archive, 2024, 32(11): 6424-6442. doi: 10.3934/era.2024299
    [3] Jiange Dong, Xianyi Li . Bifurcation of a discrete predator-prey model with increasing functional response and constant-yield prey harvesting. Electronic Research Archive, 2022, 30(10): 3930-3948. doi: 10.3934/era.2022200
    [4] Yichao Shao, Hengguo Yu, Chenglei Jin, Jingzhe Fang, Min Zhao . Dynamics analysis of a predator-prey model with Allee effect and harvesting effort. Electronic Research Archive, 2024, 32(10): 5682-5716. doi: 10.3934/era.2024263
    [5] Yuan Tian, Hua Guo, Wenyu Shen, Xinrui Yan, Jie Zheng, Kaibiao Sun . Dynamic analysis and validation of a prey-predator model based on fish harvesting and discontinuous prey refuge effect in uncertain environments. Electronic Research Archive, 2025, 33(2): 973-994. doi: 10.3934/era.2025044
    [6] Kerioui Nadjah, Abdelouahab Mohammed Salah . Stability and Hopf bifurcation of the coexistence equilibrium for a differential-algebraic biological economic system with predator harvesting. Electronic Research Archive, 2021, 29(1): 1641-1660. doi: 10.3934/era.2020084
    [7] San-Xing Wu, Xin-You Meng . Hopf bifurcation analysis of a multiple delays stage-structure predator-prey model with refuge and cooperation. Electronic Research Archive, 2025, 33(2): 995-1036. doi: 10.3934/era.2025045
    [8] Érika Diz-Pita . Global dynamics of a predator-prey system with immigration in both species. Electronic Research Archive, 2024, 32(2): 762-778. doi: 10.3934/era.2024036
    [9] Yuan Tian, Yang Liu, Kaibiao Sun . Complex dynamics of a predator-prey fishery model: The impact of the Allee effect and bilateral intervention. Electronic Research Archive, 2024, 32(11): 6379-6404. doi: 10.3934/era.2024297
    [10] Jiani Jin, Haokun Qi, Bing Liu . Hopf bifurcation induced by fear: A Leslie-Gower reaction-diffusion predator-prey model. Electronic Research Archive, 2024, 32(12): 6503-6534. doi: 10.3934/era.2024304
  • Microgrid (MG) has extensive properties to overcome common problems of local distribution system. Some of those problems are generation and demand difference, blackout and brownout, environmental concerns due to burning of natural resources in power stations (indirectly), and reliability issues. Research on microgrid is being conducted to enhance its features to mitigate power quality (PQ) problems associated with distribution system. Voltage sag and swell have been major power quality problems for decades, loads in distribution system are heavily affected due to these power quality problems. In the distribution system, microgrid and power quality compensation strategy should be existed in order to ensure reliability and voltage sag/swell mitigation. Dynamic voltage restorer (DVR) is comprehensive power electronics based Flexible Alternating Current Transmission System (FACTS) device, it is third-generation FACTS device as its control scheme selection flexibility and power line coupling approach make it advance when compare to first and second-generation FACTS devices. In this paper, an Enhanced Dynamic Voltage Restorer (EDVR) is presented to efficiently mitigate voltage sag/swell in grid connected microgrid. On the one side, the presence of microgrid structure ensures reliability of distribution system for local loads on the other side, EDVR ensures voltage sag/swell free power supply for loads. The control strategy of EDVR is based on enhanced synchronous reference frame (ESRF) approach and fuzzy technique system. ESRF is specially design for fast and precise operation of EDVR whereas; fuzzy technique system is responsible for standardized voltage supply for local loads. DC link voltage of EDVR is effectively regulated with the help of proposed control scheme at the time of voltage sag/swell compensation. Stability analysis of ESRF control has been done using modeling of VSC and eigenvalue analysis system. Simulation results on MATLAB/Simulink verified the performance of EDVR under presented control approach hence the specific loads in distribution system are more secure under proposed microgrid system with EDVR.


    The interaction between two or more species has been an important and interesting issue in biology and ecology since the famous Lotka-Volterra model [1,2] was proposed. The interaction between different species will generate rich interesting dynamics of biological species, and exhibit the complexity and diversity [3,4]. Amensalism, as a typical type of interaction between the species, has been intensively considered in the last decades. Amensalism describes a basic biological interaction in nature, where one species inflicts harm on another not affected by the former, which means that it does not receive any costs or benefits to itself. The first pioneer work for the investigations of amensalism model is due to Sun [5], who, in 2003, proposed the following two-species amensalism model:

    {dxdt=r1x(k1xαyk1),dydt=r2y(k2yk2), (1.1)

    where x=x(t) and y=y(t) represent the population densities of two species at time t, respectively; r1 and k1 describe the intrinsic growth rate and the carrying capacity of the first species, respectively; r2 and k2 describe the intrinsic growth rate and the carrying capacity of the second species, respectively; α describes the effect of the second species on the first species. They explored the stability properties of all possible equilibria in system (1.1).

    Setting a11=r1k1, a12=r1αk1 and a22=r2k2, we can see that model (1.1) can be rewritten as

    {dxdt=x(r1a11xa12y),dydt=y(r2a22y), (1.2)

    where all parameters r1, r2, a11, a12, and a22 are positive real numbers.

    Ever since the first amensalism model was presented, the complicated dynamics of amensalism models have been studied extensively (see [6,7,8,9,10,11,12,13,14,15] and the references cited therein). For example, in [6], Guan and Chen considered a two-species amensalism model with Beddington-DeAngelis functional response and gave some comprehensive bifurcation and global dynamics of this system. Recently, Luo et al. [12] proposed an amensalism model with Holling-Ⅱ functional response and weak Allee effect for the first species, and discussed its local dynamics and global structure.

    In the real world, from the point of view of human needs, the management of renewable resources, the exploitation of biological resources, and the harvesting of populations are commonly practiced in forestry, fishery, and wildlife management [16,17,18]. Hence, it is necessary to introduce and to consider the harvesting of species in population models. During the last decade, population models with harvesting and the role of harvesting in the management of renewable resources have received significant attention from the researchers [19,20,21,22,23,24]. Generally speaking, there are three types of harvesting: 1) constant harvesting [25]; 2) linear harvesting [25]; 3) non-linear harvesting [26,27]. As is well known, non-linear harvesting is more realistic from biological and economical points of view [28]. In [26], Clark first proposed the non-linear harvesting term h(E,x)=qExcE+lx, which is called Michaelis-Menten type functional form of catch rate, here q is the catchability coefficient, E is the external effort devoted to harvesting, c and l are constants. After that, many scholars started to focus attention on the influence of the Michaelis-Menten type harvesting on the population systems [28,29,30,31,32,33,34]. For example, a reaction-diffusion predator-prey model with non-local delay and Michaelis-Menten type prey-harvesting was investigated by Zhang et al. [32], they obtained that the discrete and non-local delays are responsible for a stability switch in this system, and a Hopf bifurcation occurs as the delays pass through a critical value. In [33], Hu and Cao discussed a predator-prey system with the nonlinear Michaelis-Menten type predator harvesting and gave a detailed analysis of the stability and bifurcation for this system.

    These studies have shown that harvesting activity has important influence to dynamics of population systems and the harvested models can exhibit richer dynamics compared to the model with no harvesting. However, seldom did scholars consider the dynamical behaviors of an amensalism model with a harvesting term. Accordingly, inspired by the previous works, based on the system (1.2), we now introduce the Michaelis-Menten type harvesting to the second species, then we propose the following amensalism model:

    {dxdt=x(r1a11xa12y),dydt=y(r2a22y)qEycE+ly, (1.3)

    where the parameters r1, r2, a11, a12, and a22 are positive constants, and this term qEycE+ly represents Michaelis-Menten harvesting.

    Setting ¯t=r1t, ¯x=a11r1x, ¯y=a12r1y, α=a12qElr21, δ=r2r1, β=a22a12, γ=a12cElr1, and dropping the bars, then system (1.3) is transformed into

    {dxdt=x(1xy),dydt=y(δβy)αyγ+y. (1.4)

    The layout of this paper is arranged as follows. In the next section, the global dynamics including positivity and boundedness of solutions, number of equilibria, local asymptotical stability, codimension one bifurcations, dynamical behaviors near infinity, closed orbits analysis and global phase portraits are shown for system (1.4). Further, in Section 3, we also obtain the global dynamics of system (1.1) without harvesting term. Numerical simulations and discussions are displayed by Section 4 for demonstrating the theoretical results and the impact of harvesting term. Finally, a brief conclusion is presented in Section 5.

    In this section, we mainly investigate the global dynamics of system (1.4), which include the positivity and boundedness of solutions, the existence and local stability analysis of equilibria, all possible bifurcation behaviors and the global structure of system (1.4).

    Here, we give the positivity and boundedness of the solutions of model (1.4) in the region R2+={(x,y):x0,y0}.

    Lemma 2.1. All solutions of model (1.4) with positive initial value are positive for all t0.

    Proof. Solving model (1.4) with positive initial condition (x(0),y(0)) gives the result:

    x(t)=x(0)[expt0(1x(s)y(s))ds]>0,y(t)=y(0)[expt0(δβy(s)αγ+y(s))ds]>0.

    Therefore, we can easily see that each solution of model (1.4) starting from the positive initial condition (x(0),y(0)) still remains in the first quadrant.

    Lemma 2.2. All solutions of system (1.4) with positive initial value are bounded in region Ω={(x(t),y(t)):0x(t)<1 and 0y(t)<δβ} for the large enough time t.

    Proof. If the initial value is selected in R2+, each solution of model (1.4) remains positive from the above result. One can see

    ˙x<x(1x)

    from the first equation of model (1.4). And a standard comparison theorem shows that

    limtsupx(t)<1.

    Also, the following inequality

    ˙y<βy(δβy)

    is obtained from the second equation of system (1.4). It yields

    limtsupy(t)<δβ.

    Hence, there exists a sufficiently large time T, such that 0x(t)<1 and 0y(t)<δβ for t>T. In summary, all the solutions of system (1.4) are always bounded. Thus we complete the proof.

    Theorem 2.1. System (1.4) always has two boundary equilibria E0(0,0) and E1(1,0) for all positive parameters. For the existence of other equilibria, we have

    1) For the possible positive equilibria:

    i) if α>α, then system (1.4) has no positive equilibria;

    ii) if α=α and 0<δβγ<2β, then there is a unique positive equilibrium E33(1y3,y3);

    iii) if α<α<α, when α<δ+δγββγ and δβγ>0, or α=δ+δγββγ and 0<δβγ<2β, there is one positive equilibrium E31(1y1,y1); when α>δ+δγββγ and 0<δβγ<2β, there are two distinct positive equilibria E31(1y1,y1) and E32(1y2,y2);

    iv) if α=α and 0<δβγ<β, then E31 coincides with E1 and there is one positive equilibrium E32(1y2,y2), where y2=δβγβ;

    v) if 0<α<α and α>δ+δγββγ, then there is one positive equilibrium E32(1y2,y2).

    2) For the other possible boundary equilibria:

    i) if α>α, then system (1.4) has no other boundary equilibria; there is one boundary equilibrium E22(0,y2).

    ii) if α=α and δβγ>0, then there is a boundary equilibrium E23(0,y3);

    iii) if α<α<α and δβγ>0, then there are two distinct boundary equilibria E21(0,y1) and E22(0,y2);

    iv) if α=α and δβγ>0, then E21 coincides with E0 and there is one boundary equilibrium E22(0,y2), where y2=δβγβ;

    v) if 0<α<α, then

    Where α=(δ+βγ)24β, α=δγ, y1=δβγ(δ+βγ)24αβ2β, y2=δβγ+(δ+βγ)24αβ2β, and y3=δβγ2β.

    Proof. It is easy to see that system (1.4) always possesses two boundary equilibria given by E0(0,0) and E1(1,0) for all positive parameters.

    1) If x0 and y0, system (1.4) has a positive equilibrium which satisfies

    {x=1y,βy2(δβγ)y+αδγ=0. (2.1)

    For the positive equilibria, y must satisfy 0<y<1. Let Δ denote the discriminant of the second equation of (2.1), namely,

    Δ=(δ+βγ)24αβ, (2.2)

    and let

    α=(δ+βγ)24β. (2.3)

    Obviously, when 0<α<α, the second equation of (2.1) has two roots

    y1=δβγΔ2β  and  y2=δβγ+Δ2β.

    When α=α, the second equation of (2.1) has one root

    y3=δβγ2β.

    Let

    k(y)=βy2(δβγ)y+αδγ,

    and denote by

    α=δγ. (2.4)

    Then we have

    H1) if α>α, then k(0)>0;

    H2) if α=α, then k(0)=0;

    H3) if α<α, then k(0)<0.

    In addition, 0<αα, and α=α if and only if δ=βγ.

    Based on the above analysis, combining with H1), H2) and H3), we can derive that:

    ⅰ) If α>α, then Δ<0 which implies k(y)=0 has no real roots. So system (1.4) has no positive equilibria.

    ⅱ) If α=α, then Δ=0, so k(y)=0 has a unique real roots y3=δβγ2β, combing with the condition 0<y3<1, we get 0<δβγ<2β. In this situation, there exists a unique positive equilibrium E33(1y3,y3).

    ⅲ) If α<α<α, then Δ>0 and k(0)>0, implying k(y)=0 two positive real roots y1 and y2 if δβγ>0. Moreover, when k(1)<0, namely, α<δ+δγββγ, we have 0<y1<1<y2, that is, there is one positive point E31(1y1,y1); when k(1)=0 and the symmetry axis of k(y) is y=δβγ2β<1, namely, α=δ+δγββγ and 0<δβγ<2β, we have 0<y1<y2=1, which means there is one positive point E31(1y1,y1); when k(1)>0 and the symmetry axis of k(y) is y=δβγ2β<1, namely, α>δ+δγββγ and 0<δβγ<2β, we have 0<y1<y2<1, that is there are two different positive equilibria E31(1y1,y1) and E32(1y2,y2).

    ⅳ) If α=α, then Δ>0 and k(0)=0, hence k(y)=0 has two real roots y1=0 and 0<y2=δβγβ<1 if 0<δβγ<β, thus E31 coincides with E1 and there is one positive equilibrium E32(1y2,y2).

    ⅴ) If α<α, then Δ>0 and k(0)<0, so there are y1<0 and y2>0. When k(1)>0, i.e., α>δ+δγββγ, we get one positive root 0<y2<1, thus there exits one positive equilibrium E32(1y2,y2).

    The proof of the corresponding boundary equilibria in 2) is similar to that of 1), and hence omitted here. This completes the proof of Theorem 2.1.

    In the following, we focus on the local stability of each equilibrium of system (1.4).

    Theorem 2.2. For the boundary equilibria E0 and E1, the following statements are true.

    1) If 0<α<α, then E0 is a hyperbolic unstable node and E1 is a hyperbolic saddle.

    2) If α>α, then E0 is a hyperbolic saddle and E1 is a hyperbolic stable node.

    3) If α=α, then

    i) when δβγ, E0 and E1 are both saddle-nodes;

    ii) when δ=βγ, E0 is a non-hyperbolic saddle and E1 is a non-hyperbolic stable node.

    Proof. The Jacobian matrix of system (1.4) evaluated at any equilibrium is

    J(x,y)=(2x+1yx0H(x,y)), (2.5)

    where H(x,y)=δ2βyαγ(γ+y)2.

    For the equilibrium E0, the Jacobian matrix is

    J(E0)=(100δαγ), (2.6)

    and the two eigenvalues of J(E0) are λ1(E0)=1>0 and λ2(E0)=δαγ. Obviously, if 0<α<α, then λ2(E0)>0, so E0 is a hyperbolic unstable node. If α>α, then λ2(E0)<0, thus E0 is a hyperbolic saddle. If α=α, then λ2(E0)=0, this means that the equilibrium E0 is non-hyperbolic, so it is hard to directly judge its type from their eigenvalues. We further discuss its stability properties by applying Theorem 7.1 in Chapter 2 in [35].

    In order to change system (1.4) into a standard form, we expand system (1.4) in power series up to the fourth order around the origin

    {dxdt=xx2xyx+P(x,y),dydt=(αγ2β)y2αγ3y3+αγ4y4+Q0(y)Q(x,y), (2.7)

    where Q0(y) represents a power series with the terms yi (i5). From dxdt=0, we obtain that there is a unique implicit function x=φ0(y)=0 such that φ0(y)+P(φ0(y),y)=0 and φ0(0)=φ0(0)=0. Then substituting x=φ0(y)=0 into the second equation of (2.7), we get that

    dydt=(αγ2β)y2αγ3y3+αγ4y4+O(|y|5). (2.8)

    Because α=α=δγ, the coefficient at y2 is δγβ. By Theorem 7.1 in [35], we have when δβγ, the equilibrium E0 is a saddle-node.

    When δ=βγ, (2.8) becomes

    dydt=βγy3+βγ2y4+O(|y|5). (2.9)

    Employing the notations of Theorem 7.1 in Chapter 2 in [35], we have m=3 and am=βγ<0, so the equilibrium E0 is a non-hyperbolic saddle.

    For the equilibrium E1, the Jacobian matrix is

    J(E1)=(110δαγ). (2.10)

    The two eigenvalues of the above matrix J(E1) are λ1(E1)=1<0 and λ2(E1)=δαγ. Clearly, if α<α, then λ2(E1)>0, so E1 is a hyperbolic saddle. If α>α, then λ2(E1)<0, then E1 is a hyperbolic stable node. If α=α, then λ2(E1)=0, the equilibrium E1 is non-hyperbolic and its stability cannot be given directly from the eigenvalues. In order to determinate the stability of E1, we translate E1 to the origin by the translation (¯x,¯y)=(x1,y) and expand system (1.4) in power series up to the fourth order around the origin, which makes system (1.4) to be the following form:

    {d¯xdt=¯x¯y¯x¯y¯x2,d¯ydt=(αγ2β)¯y2αγ3¯y3+αγ4¯y4+Q1(¯y), (2.11)

    where Q1(¯y) represents a power series with terms ¯yi (i5).

    To transform the Jacobian matrix into a standard form, we use the invertible translation

    (uv)=(1101)(¯x¯y), (2.12)

    then system (2.11) becomes

    {dudt=uu2+uv+(αγ2β)v2αγ3v3+αγ4v4+Q1(v),dvdt=(αγ2β)v2αγ3v3+αγ4v4+Q1(v). (2.13)

    By introducing a new time variable τ=t, we get

    {dudτ=u+u2uv(αγ2β)v2+αγ3v3αγ4v4Q1(v)u+P(u,v),dvdτ=(βαγ2)v2+αγ3v3αγ4v4Q1(v)Q(u,v). (2.14)

    Based on the implicit function theorem, from dudτ=0, we can deduce a unique function

    u=φ1(v)=(αγ2β)v2+(αγ2αγ3β)v3+(αγ2αγ3+αγ4β(αγ2β)2)v4+,

    which satisfies φ1(0)=φ1(0)=0 and φ1(v)+P(φ1(v),v)=0. Then substituting it into the second equation of (2.14), we have

    dvdτ=(βαγ2)v2+αγ3v3αγ4v4+O(|v|5). (2.15)

    From α=δγ it follows that the coefficient at v2 is βδγ. Thus, by using Theorem 7.1 in Chapter 2 in [35], we get that when δβγ, the equilibrium E1 is a saddle node.

    When δ=βγ, (2.15) can be written as

    dvdτ=βγv3βγ2v4+O(|v|5).

    By Theorem 7.1 in [35] again, we obtain m=3 and am=βγ>0. Then E0 is a non-hyperbolic unstable node. Hence, E1 is a non-hyperbolic stable node due to that we have used the transformation τ=t. Accordingly, Theorem 2.2 is proved.

    Theorem 2.3. For the boundary equilibria E21, E22 and E23, the following statements are true.

    1) Assume α=α and δβγ>0, then there exists the boundary equilibrium E23(0,y3). Moreover,

    i) If δβγ>2β, then E23 is a saddle-node, which includes a stable parabolic sector.

    ii) If 0<δβγ<2β, then E23 is a saddle-node, which includes an unstable parabolic sector.

    iii) If δβγ=2β, then E23 is non-hyperbolic.

    2) Assume α<α<α and δβγ>0, then there are two boundary equilibria E21(0,y1) and E22(0,y2). Moreover,

    i) If α<δ+δγββγ, then E21 is a hyperbolic unstable node and E22 is a hyperbolic stable node.

    ii) If α>δ+δγββγ and δβγ>2β, then E21 is a hyperbolic saddle and E22 is a hyperbolic stable node.

    iii) If α>δ+δγββγ and 0<δβγ<2β, then E21 is a hyperbolic unstable node and E22 is a hyperbolic saddle.

    iv) If α=δ+δγββγ and δβγ>2β, then E21 a saddle-node, which includes an unstable parabolic sector and E22 is a hyperbolic stable node.

    v) If α=δ+δγββγ and 0<δβγ<2β, then E21 a hyperbolic unstable node and E22 is a saddle-node, which includes a stable parabolic sector.

    3) Assume α=α and δβγ>0, then there is one boundary equilibrium E22(0,y2), where y2=δβγβ. Moreover,

    i) If 0<δβγ<β, then E22 is a hyperbolic saddle.

    ii) If δβγ=β, then E22 is a saddle-node, which includes a stable parabolic sector.

    iii) If δβγ>β, then E22 is a hyperbolic stable node.

    4) Assume 0<α<α, then there is one boundary equilibrium E22(0,y2).

    i) If α<δ+δγββγ, then E22 is a hyperbolic stable node.

    ii) If α=δ+δγββγ, then E22 is a saddle-node, which includes a stable parabolic sector.

    iii) If α>δ+δγββγ, then E22 is a hyperbolic saddle.

    Proof. For the equilibrium E2i(i = 1, 2, 3), the Jacobian matrix is

    J(E2i)=(1yi00H(xi,yi)), (2.16)

    where H(xi,yi)=δ2βyiαγ(γ+yi)2. The two eigenvalues of the above matrix J(E2i) are λ1(E2i)=1yi and λ2(E2i)=H(xi,yi).

    Because (xi,yi) satisfies δβyiαγ+yi=0, we can derive

    H(xi,yi)=δ2βyiαγ(γ+yi)2=2βyiγ+yi(yiδβγ2β).

    Then

    λ2(E21)=H(x1,y1)=2βy1(γ+y1)2(y1δβγ2β)=2βy1(γ+y1)2Δ2β>0,
    λ2(E22)=H(x2,y2)=2βy2(γ+y2)2(y2δβγ2β)=2βy2(γ+y2)2Δ2β<0

    and

    λ2(E23)=H(x3,y3)=2βy3(γ+y3)2(y3δβγ2β)=0.

    In addition, for λ1(E2i)=1yi, we discuss it in the following four cases.

    Case 1. α=α and δβγ>0.

    If δβγ>2β, then y3>1, which implies λ1(E23)=1y1<0, so E23 is non-hyperbolic. In order to determinate the stability of the equilibrium E23, we translate E23 to the origin by letting (¯x,¯y)=(x,yy3), and expand the system in power series up to the third order around the origin, under which system (1.4) can be transformed into

    {d¯xdt=(1y3)¯x¯x¯y¯x2,d¯ydt=c0¯y+c1¯y2+c2¯y3+Q2(¯y), (2.17)

    where c0=δ2βy3αγ(γ+y3)2=0, c1=αγ(γ+y3)3β, c2=αγ(γ+y3)4, and Q2(y) represents for a power series with terms ¯yi (i4).

    Then introducing a new time variable τ=(1y3)t, we get

    {d¯xdτ=¯x11y3¯x¯y11y3¯x2¯x+P(¯x,¯y),d¯ydτ=c11y3¯y2+c21y3¯y3+11y3Q2(¯y)Q2(¯x,¯y). (2.18)

    We see that the coefficient at ¯y2 is c11y3>0. Hence, from Theorem 7.1 in Chapter 2 of [35], we have E23 is a saddle-node, which includes a stable parabolic sector and this parabolic sector is on the upper half-plane.

    If δβγ<2β, then y3<1, which means λ1(E23)=1y1>0. Same analysis as the above we can get E23 is also a saddle-node, which includes an unstable parabolic sector and the parabolic sector is on the lower half-plane.

    If δβγ=2β, then y3=1, which means λ1(E23)=1y1=0, so E23 is a non-hyperbolic critical point with two zero eigenvalues.

    Case 2. α<α<α and δβγ>0.

    1) If α<δ+δγββγ, then y1<1<y2, that implies λ1(E21)=1y1>0 and λ1(E22)=1y2<0, so E21 is a hyperbolic unstable node and E22 is a hyperbolic stable node;

    2) If α>δ+δγββγ and δβγ>2β, then 1<y1<y2, implying λ1(E21)=1y1<0 and λ1(E22)=1y2<0, thus E21 is hyperbolic saddle and E22 is hyperbolic stable node;

    3) If α>δ+δγββγ and δβγ<2β, then y1<y2<1, that means λ1(E21)=1y1>0, λ1(E22)=1y2>0, hence E21 is a hyperbolic unstable node and E22 is a hyperbolic saddle;

    4) If α=δ+δγββγ and δβγ>2β, then 1=y1<y2, which means λ1(E21)=1y1=0 and λ1(E22)=1y2<0, so E21 is non-hyperbolic and E22 is a hyperbolic stable node.

    In order to determinate the stability of the equilibrium E21, we translate the equilibrium E21 to the origin by the translation (¯x,¯y)=(x,y1), and expand system (1.4) in power series up to the third order around the origin, which makes system (1.4) to be the following form:

    {d¯xdt=¯x¯y¯x2,d¯ydt=d0¯y+d1¯y2+d2¯y3+Q3(¯y), (2.19)

    where d0=δ2βαγ(γ+1)2, d1=αγ(γ+1)3β, d2=αγ(γ+1)4, and Q3(¯y) represents for a power series with terms ¯yi satisfying i4.

    Then introducing a new time variable τ=d0t, we get

    {d¯xdτ=1d0¯x¯y1d0¯x2Q(¯x,¯y),d¯ydτ=¯y+d1d0¯y2d2d0¯y3+1d0Q3(¯y)¯y+P(¯x,¯y). (2.20)

    From d¯ydτ=0, we can derive a unique implicit function ¯y=ϕ(¯x)=0, which satisfies ϕ(0)=ϕ(0)=0 and ϕ(¯x)+P(¯x,ϕ(¯x))=0. Substituting ¯y=ϕ(¯x)=0 into the first equation of (2.20), we have that

    d¯xdτ=1d0¯x2.

    The coefficient at ¯x2 is

    1d0=γ+1δβγ2β<0.

    By Theorem 7.1 in Chapter 2 of [35], we know E21 is a saddle-node, which includes an unstable parabolic sector and this parabolic sector is on the left half-plane.

    5) If α=δ+δγββγ and δβγ<2β, then y1<y2=1, implying λ1(E21)=1y1>0 and λ1(E22)=1y2=0, therefore E21 a hyperbolic unstable node and E22 is non-hyperbolic. Similarly to the proof of E21 in 4), we can get that E22 is a saddle-node, which includes a stable parabolic sector.

    Case 3. α=α and δβγ>0.

    If δβγ>β, then y2>1, that is λ1(E22)=1y2<0, so E22 is a hyperbolic stable node. If δβγ<β, then y2<1, that is λ1(E22)=1y2>0, thus E22 is a hyperbolic saddle. If δβγ=β, then y2=1, that is λ1(E22)=1y2=0, thus E22 is a non-hyperbolic. Similar to the proof of E21 in 4), we can obtain E22 is a saddle-node, which includes a stable parabolic sector.

    Case 4. α<α.

    If α>δ+δγββγ, then y2<1, which implies λ1(E22)=1y2>0, thus E22 is a hyperbolic saddle. If α<δ+δγββγ, then y2>1, which means λ1(E22)=1y2<0, thus E22 is a hyperbolic stable node. If α=δ+δγββγ, then y2=1, which implies λ1(E22)=1y2=0, thus E22 non-hyperbolic. Similar to the proof of 4) in Case 2, we can derive E22 is a saddle-node, which includes a stable parabolic sector.

    This completes the proof of Theorem 2.3.

    Theorem 2.4. For the positive equilibria E31, E32 and E33, the following statements are true.

    1) If α<α<α, when α<δ+δγββγ and δβγ>0, or α=δ+δγββγ and 0<δβγ<2β, or α>δ+δγββγ and 0<δβγ<2β, then E31 is a hyperbolic saddle.

    2) If 0<α<α and α>δ+δγββγ, or α=α and 0<δβγ<β, or α<α<α, α>δ+δγββγ and 0<δβγ<2β, then E32 is a hyperbolic stable node.

    3) If α=α and 0<δβγ<2β, then E33 is a saddle-node.

    Proof. For the equilibrium E3i(i = 1, 2, 3), the Jacobian matrix is

    J(E3i)=(yi1yi10H(xi,yi)), (2.21)

    where H(xi,yi)=δ2βyiαγ(γ+yi)2. The two eigenvalues of the above matrix J(E3i) are λ1(E3i)=yi1<0 and λ2(E3i)=H(xi,yi). From the analysis of the proof of Theorem 2.3, it follows that λ2(E31)>0, λ2(E32)<0, and λ2(E33)=0.

    Hence, when α<α<α, α<δ+δγββγ and δβγ>0, or α<α<α, α=δ+δγββγ and 0<δβγ<2β, or α<α<α, α>δ+δγββγ and 0<δβγ<2β, we have λ1(E31)<0 and λ2(E31)>0, which means E31 is a hyperbolic saddle.

    When 0<α<α and α>δ+δγββγ, or α=α and 0<δβγ<β, or α<α<α, α>δ+δγββγ and 0<δβγ<2β, there exists the positive equilibrium E32 with two eigenvalues λ1(E32)<0 and λ2(E32)<0, hence E32 is a hyperbolic stable node.

    When α=α and 0<δβγ<2β, there exists the positive equilibrium E33 with two eigenvalues λ1(E33)<0 and λ2(E33)=0. Then we further to study the stability of E33 by transforming E33 to the origin by the translation (¯x,¯y)=(xx3,yy3) and expand system (1.4) in power series up to the third order around the origin, which makes system (1.4) to be the following form:

    {d¯xdt=x3¯xx3¯y¯x¯y¯x2,d¯ydt=c0+c1¯y+(αγ(γ+y3)3β)¯y2αγ(γ+y3)4¯y3+Q4(¯y), (2.22)

    where c0=δy3β(y3)2α+αγγ+y3=0, c1=δ2βy3αγ(γ+y3)2=0, and Q4(¯y) stands for a power series with terms ¯yi (i4).

    In order to transform the Jacobian matrix into a standard form, we use the invertible translation

    (uv)=(1101)(¯x¯y), (2.23)

    then system (2.22) can be expressed as

    {dudt=x3u+uvu2+(αγ(γ+y3)3β)v2αγ(γ+y3)4v3+Q4(v),dvdt=(αγ(γ+y3)3β)v2αγ(γ+y3)4v3+Q4(v). (2.24)

    Then introducing a new time variable τ=x3t, we get

    {dudτ=u1x3uv+1x3u21x3(αγ(γ+y3)3β)v2+αγx3(γ+y3)4v31x3Q4(v)    u+P(u,v),dvdτ=1x3(αγ(γ+y3)2β)v2+αγx3(γ+y3)4v31x3Q4(v)Q(u,v). (2.25)

    According to the implicit function theorem, from dudτ=0, we can derive a unique function

    u=φ4(v)=1x3(αγ(γ+y3)3β)v2+1(x3)2(αγ(γ+y3)3β)v2αγx3(γ+y3)4v3+,

    which satisfies φ4(0)=φ4(0)=0 and φ4(v)+P(φ4(v),v)=0. Substituting it into the second equation of (2.25), we have that

    dvdτ=1x3(αγ(γ+y3)2β)v2+αγx3(γ+y3)4v3+O(|v|4).

    The coefficient at the term v2 is

    1x3(βαγ(γ+y3)2)=1x3(β(δβy3)γγ+y3)=β(δβγ)x3(δ+βγ)>0.

    Therefore, on basis of Theorem 7.1 in [35], the equilibrium E33 of system (1.4) is a saddle-node.

    The proof of Theorem 2.4 is finished.

    For readers' convenience, the local dynamical properties of equilibria of system (1.4) are totally summarized in Table 1. It can be clearly seen that the number and stability of equilibria are different according to the value range of the parameters. Therefore, we divide (α,β,γ,δ)R4+ into the following twelve regions.

    R11={(α,β,γ,δ)R4+:α=α,δ=βγ, or α=α,δ=βγ or α>α};R12={(α,β,γ,δ)R4+:α=α,δ<βγ};R2a={(α,β,γ,δ)R4+:0<α<α,αδ+δγββγ};R2b={(α,β,γ,δ)R4+:α=α,δβγβ};R22={(α,β,γ,δ)R4+:α=α,δβγ2β};R3a={(α,β,γ,δ)R4+:0<α<α,α>δ+δγββγ};R3b={(α,β,γ,δ)R4+:α=α,0<δβγ<β};R32={(α,β,γ,δ)R4+:α<α<α,αδ+δγββγ,δβγ>2β};R33={(α,β,γ,δ)R4+:α=α,0<δβγ<2β};R41={(α,β,γ,δ)R4+:α<α<α,α<δ+δγββγ,δ>βγ};R42={(α,β,γ,δ)R4+:α<α<α,α=δ+δγββγ,0<δβγ<2β};R5={(α,β,γ,δ)R4+:α<α<α,α>δ+δγββγ,0<δβγ<2β}.
    Table 1.  Equilibria of system (1.4) in finite planes.
    Conditions Equilibra
    0<α<α α<δ+δγββγ E0(u),E1(sd),E22(s)
    α=δ+δγββγ E0(u),E1(sd),E22(sn)
    α>δ+δγββγ E0(u),E1(sd),E22(sd),E32(s)
    α=α 0<δβγ<β E0(sn),E1(sn),E22(sd),E32(s)
    δβγ=β E0(sn),E1(sn),E22(sn)
    δβγ>β E0(sn),E1(sn),E22(s)
    δ=βγ E0(sd),E1(s)
    δ<βγ E0(sn),E1(sn)
    α<α<α α<δ+δγββγ δβγ>0 E0(sd),E1(s),E21(u),E22(s),E31(sd)
    α=δ+δγββγ 0<δβγ<2β E0(sd),E1(s),E21(u),E22(sn),E31(sd)
    δβγ>2β E0(sd),E1(s),E21(sn),E22(s)
    α>δ+δγββγ 0<δβγ<2β E0(sd),E1(s),E21(u),E22(sd),E31(sd),E32(s)
    δβγ>2β E0(sd),E1(s),E21(sd),E22(s)
    α=α 0<δβγ<2β E0(sd),E1(s),E23(sn),E33(sn)
    δβγ=2β E0(sd),E1(s),E23(nh)
    δβγ>2β E0(sd),E1(s),E23(sn)
    δβγ=0 E0(sd),E1(s)
    α>α E0(sd),E1(s)
    Note: where "sd", "sn", "u", "s" and 'nh" represent saddle, saddle-node, unstable node, stable node and non-hyperbolic point, respectively.

     | Show Table
    DownLoad: CSV

    For R1=R11R12, system (1.4) possesses two equilibria E0 and E1. In R11, E0 is a saddle and E1 is a stable node. In R12, E0 and E1 represent saddle-nodes. And specifically, in R1, E1 is stable in the first quadrant. Therefore, the qualitative properties of these two equilibria can be seen in Figure 1(a).

    Figure 1.  The qualitative properties of the equilibria for R1, R2, R3, R4 and R5.

    For R2=R21R22, there are three equilibria for system (1.4). In R21=R2aR2b, system (1.4) admits E0 (unstable node or saddle-node), E1 (saddle or saddle-node) and E22 (stable node or saddle-node). In particular, in R21, E0 is unstable and E22 is stable in the first quadrant. In R22, there are E0 (saddle), E1 (stable node) and E23 (non-hyperbolic or saddle-node). The qualitative properties of these equilibria in R2 are displayed by Figure 1(b), (c).

    For R3=R31R32R33, there exist four equilibria for system (1.4). Considering R31=R3aR3b, there are E0 (unstable node or saddle-node which is unstable in the first quadrant), E1 (saddle or saddle-node), E22 (saddle) and E32 (stable node). Considering R32, system (1.4) admits E0 (saddle), E1 (stable node), E21 (saddle or saddle-node) and E22 (stable node). Considering R33, there exist E0, E1, E23 and E33, which are saddle, stable node, saddle-node and saddle-node. The detailed image descriptions of the qualitative properties of the equilibria in R3 are illustrated by Figure 1(d)(f).

    For R4=R41R42, system (1.4) has five equilibria E0 (saddle), E1 (stable node), E21 (unstable node), E22 (stable node or saddle-node) and E31 (saddle). Particularly, E22 is stable in the first quadrant. As shown in Figure 1(g), the qualitative properties of these five equilibria are given.

    For R5, system (1.4) possesses six equilibria E0, E1, E21, E22, E31 and E32, which represent saddle, stable node, unstable node, saddle, saddle and stable node, respectively. Hence, the qualitative properties of these six equilibria are displayed by Figure 1(h).

    In this section, all possible bifurcations of system (1.4) will be found out. Moreover, the feasible conditions for the occurence of those bifurcations will be given.

    Theorem 2.5. System (1.4) undergoes a saddle-node bifurcation at equilibrium E33 when the parameters satisfy the restriction α=αSN=(δ+βγ)24β along with the conditions 0<δβγ<2β and α>δ+δγββγ as given in Theorem 2.1.

    Proof. Now we will verify the transversality condition for the occurrence of a saddle-node bifurcation at α=αSN by utilizing the Sotomayor's theorem [36]. From Section 2.2, we have the two eigenvalues of J(E33) are λ1(E33)<0 and λ2(E33)=0. Denote V and W the eigenvectors corresponding to the eigenvalue λ2(E33) for the matrices J(E33) and J(E33)T, respectively. Simple computations yield

    V=(v1v2)=(11) and W=(w1w2)=(01).

    Furthermore, we can obtain

    Fα(E33;αSN)=(0yγ+y)(E33;αSN)=(0δβγδ+βγ) (2.26)

    and

    D2F(E33;αSN)(V,V)=(2F1x2v21+22F1xyv1v2+2F1y2v222F2x2v21+22F2xyv1v2+2F2y2v22)(E33;αSN)=(02β+2αSNγ(γ+y3)3). (2.27)

    Obviously, when 0<δβγ<2β, the vectors V and W satisfy the transversality conditions

    WTFα(E33;αSN)=δβγδ+βγ0

    and

    WT[D2F(E33;αSN)(V,V)]=2β+2αSNγ(γ+y3)3=2β(βγδ)δ+βγ0.

    Hence, by Sotomayor's theorem, when α=αSN, system (1.4) undergoes a saddle-node bifurcation at non-hyperbolic critical point E33. The number of positive equilibria of system (1.4) changes from zero to two as α passes from the right of α=αSN to the left. Thus, the proof of Theorem 2.5 is completed.

    Theorem 2.6. System (1.4) undergoes a transcritical bifurcation at the boundary equilibrium E1 when the parameters satisfy the conditions α=αTC=δγ and 0<δβγ<2β.

    Proof. Now we also apply Sotomayor's theorem [36] to prove the transversality condition for the occurrence of transcritical bifurcation at E1 as α=αTC. From Section 2.2, we know that λ1(E1)<0 and λ2(E1)=0. Let V and W represent the two eigenvectors corresponding to the zero eigenvalue λ2(E1) of the matrices J(E1) and J(E1)T, respectively, then they are given by

    V=(v1v2)=(11) and W=(w1w2)=(01).

    Furthermore, we can obtain

    Fα(E1;αTC)=(0yγ+y)(E1;αTC)=(00), (2.28)
    DFα(E1;αTC)V=(000γ(γ+y)2)(11)(E1;αTC)=(01γ), (2.29)

    and

    D2F(E1;αTC)(V,V)=(2F1x2v21+22F1xyv1v2+2F1y2v222F2x2v21+22F2xyv1v2+2F2y2v22)(E1;αTC)=(02β+2δγ).

    Clearly, the vectors V and W satisfy the transversality conditions

    WTFα(E1;αTC)=0,WT[DFα(E1;αTC)V]=1γ0

    and

    WT[D2F(E1;αTC)(V,V)]=2β+2δγ0  for 0<δβγ<2β.

    Therefore, by Sotomayor's theorem, if δβγ, system (1.4) experiences a transcritical bifurcation at non-hyperbolic critical point E1 as the parameter α varies through the bifurcation value α=αTC.

    Thus, the proof of Theorem 2.6 is completed.

    In this subsection, we pay some attention to the global structure of the plane nonlinear dynamic system (1.4).

    For the sake of the research of the global dynamics for system (1.4), we have to investigate the qualitative properties of equilibria at infinity, which is important and useful to study the behavior of orbits when |x|+|y| .

    Firstly, using Poincaré transformation of Chapter 5 in [35]:

    x=1z,y=uz,ds=dtz,

    system (1.4) can be transformed into

    {duds=(1β)u2+(δ1)uz+uαuz2u+γz,dzds=uz+zz2. (2.30)

    In the first quadrant of u-z plane, setting z=0, system (2.30) has one boundary equilibrium e(1β1,0) if β>1 and no boundary equilibrium if β1 on the nonnegative u-axis.

    Proceed to the next step, we apply the second type of Poincaré transformation:

    x=vz,y=1z,ds=dtz,

    system (1.4) becomes

    {dvds=v2+(1δ)vz+(β1)v+αvz21+γz,dzds=δz2+βz+αz31+γz. (2.31)

    In the nonnegative cone of v-z plane, there exist two boundary equilibria e1(0,0) and e2(β1,0) if β>1 and only one e1(0,0) if β1 for system (2.31).

    In this paper, of concern is the global dynamics of system (1.4) when β>1. The case β1 is similar, hence, it is omitted.

    Theorem 2.7. In system (2.30), e(1β1,0) is a saddle. In system (2.31), e1(0,0) is an unstable node and e2(β1,0) is a saddle.

    Proof. The Jacobian matrix at e(1β1,0) is evaluated as follows:

    J(e)=(1δ1β10ββ1),

    obviously, which has two eigenvalues λ1(e)=1<0 and λ2(e)=ββ1>0. Hence, e is a saddle in system (2.30).

    Similarly, the corresponding Jacobian matrices of system (2.31) evaluated at e1(0,0) and e2(β1,0) are

    J(e1)=(β100β)  and  J(e2)=(1β(δ1)(β1)0β).

    Clearly, e1 is an unstable node and e2 is a saddle in system (2.31). This proof of Theorem 2.7 is completed.

    From the point of view of Poincaré transformation, we get

    x1z=+, yuz=+, x1v1z1=0,y11z1=+, x2v2z2=+, y21z2=+.

    Consequently, e and e2 are equivalent to an infinite singular point I for system (1.4), e1 is equivalent to an infinite singular point I1 for system (1.4). By combining the Poincaré transformation of Chapter 5 in [35] and Theorem 2.7, we illustrate the dynamical behaviors near infinity of model (1.4) with Figure 2.

    Figure 2.  The dynamics of system (1.4) near infinity.

    We, here, for further research of the global structure of system (1.4), need to discuss the existence of closed orbits in phase space.

    Theorem 2.8. No closed orbit exists for model (1.4) in R2+.

    Proof. On the contrary, assume that model (1.4) exhibits a closed orbit in R2+. From the discussion in Theorem 2.1, we can deduce that all positive equilibria will lie on invariant lines y=δβγΔ2β or y=δβγ+Δ2β. On the basis of Theorem 4.6 of [35], in the interior of the closed orbit, system (1.4) must admit a positive equilibrium. As such, the closed orbit has two points of intersection with y=δβγΔ2β or y=δβγ+Δ2β, which is contrary to the existence and uniqueness of solutions. Therefore, there is no closed orbit for model (1.4). We end the proof of Theorem 2.8.

    System (1.4), according to the aforementioned Theorem 2.7, admits two equilibria I and I1 at infinity, which stand for a saddle and an unstable node, respectively. A step further, we derive that model (1.4) exists no closed orbit in R2+ from the above Theorem 2.8. It now follows that all possible cases of global phase portraits for model (1.4) can be given.

    For the purpose of simplicity, ξI and ω(ξI) are utilized to refer the unstable manifold of I and the ω-limit set of ξI in the nonnegative cone of R2, respectively.

    For (α,β,γ,δ) R1, it is easy to be checked that ω(ξI) is E1. See Figure 3(a) for the global phase portrait of model (1.4) related to the case R1.

    Figure 3.  The global phase portraits of system (1.4).

    For (α,β,γ,δ) R2, we can easily verify that ω(ξI) is E22 in R21 and ω(ξI) is E23 in R22. See Figure 3(b), (c) for the global phase portraits of model (1.4) related to the cases R21 and R22.

    For (α,β,γ,δ) R3, we analyze the global phase portraits of model (1.4) from the three subregions R31, R32 and R33, respectively. When (α,β,γ,δ) R31, we can verify that ω(ξI) is E32. The global phase portraits of model (1.4) related to R31 are given by Figure 3(d), (e). When (α,β,γ,δ) R32, ω(ξI) is E22. The global phase portrait of model (1.4) related to R32 is shown by Figure 3(f). When (α,β,γ,δ) R33, it is clear that ω(ξI) is E33 and the global phase portraits of model (1.4) in this region are shown in Figure 3(g), (h).

    For (α,β,γ,δ) R4, it is clear that ω(ξI) is E22. See Figure 3(i) for the global phase portrait of model (1.4) related to the case R4.

    For (α,β,γ,δ) R5, it is also clear that ω(ξI) is E32. And the global phase portraits of model (1.4) in R5 are shown in Figure 3(j), (k).

    Now, we mainly consider the dynamic properties of all possible equilibria and bifurcation behaviors of model (1.1). More, in R2+, we analyze the global structure of this model and illustrate our analysis with the global phase portraits.

    To facilitate this, system (1.1) takes the following form

    {dxdt=x(1xy),dydt=y(δβy). (3.1)

    On account of the dynamic properties of system (3.1) corresponding to system (1.1), we study the dynamics of system (3.1) in the following.

    As it has been shown in [5], the existence and stability properties of equilibria of system (3.1) are clearly summarized in Table 2. Next, we divide (δ,β)R2+ into two regions as follows.

    G1=G11G12={(δ,β)R2+:0<β<δ}{(δ,β)R2+:δ=β};G2={(δ,β)R2+:0<δ<β}.
    Table 2.  Local dynamical properties of equilibria.
    Equilibrium Existence conditions Type
    E0(0,0) (always exists) unstable node
    E1(1,0) (always exists) saddle
    E2(0,δβ) 0<δ<β saddle
    δ=β non-hyperbolic (stable in R2+)
    0<β<δ stable node
    E3(1δβ,δβ) 0<δ<β stable node

     | Show Table
    DownLoad: CSV

    Consider (δ,β)G1, system (3.1) has three equilibria E0, E1 and E2. Specifically, E0 and E1 represent an unstable node and a saddle. E2 is a stable node in G11 and non-hyperbolic in G12, which is stable in the first quadrant. See Figure 4(a), the qualitative properties of these three equilibria are illustrated.

    Figure 4.  The qualitative properties of the equilibria for G1 and G2.

    Consider (δ,β)G2, system (3.1) admits four equilibria E0, E1, E2 and E3, which are unstable node, saddle, saddle and stable node, respectively. See Figure 4(b), the qualitative properties of these four equilibria are illustrated.

    Here, we will analyze some of all the possible bifurcation scenarios for model (3.1). According to the previous methods applied in Section 2.3, a heuristic discussion similar to the one used by Theorem 2.6 can easily get the conditions for a transcritical bifurcation. Therefore, the following theorem can be directly obtained.

    Theorem 3.1 System (3.1) undergoes a transcritical bifurcation at E2 when the parameters satisfy the condition δ=β.

    In order to investigate the global structure of system (3.1), the methods proposed by Section 2.4 can be extended. By applying Poincaré transformation and analysis of closed orbit, Figure 5(a) describes the existence and stability of the singular points at infinity, as well as all the boundary equilibria. In addition, Figure 5(b)(d) describes minutely the global phase portraits in R2+ similarly.

    Figure 5.  (a) The dynamics of (3.1) near infinity; (b) The global phase portrait of (3.1) in G1; (c) The global phase portrait of (3.1) in G1; (d) The global phase portrait of (3.1) in G2.

    Some numerical examples, in this section, are proposed so as to check the obtained parameter conditions and theoretical results of systems (1.4) and (3.1). Further, we are interested to show the role of harvesting on the local dynamical properties.

    Example 4.1. We consider parameter values as below: α = 2, β = 1.2, γ = 0.8, δ = 2.5. Simple calculations lead to α=α, δβγβ and δ>β, which means both systems (1.4) and (3.1) have three boundary equilibria. The numerical phase portraits related to this case are shown in Figures 6(a) and 7(a). The impact of harvesting is shown by Figure 8.

    Figure 6.  The numerical phase portraits of system (1.4): (a) α = 2, β = 1.2, γ = 0.8, δ = 2.5; (b) α = 1.85, β = 0.3, γ = 2.5, δ = 0.8; (c) α = 0.8, β = 2.5, γ = 0.4, δ = 2.
    Figure 7.  The numerical phase portraits of system (3.1): (a) β = 1.2, δ = 2.5; (b) β = 0.3, δ = 0.8; (c) β = 2.5, δ = 2.
    Figure 8.  The impact of Michaelis-Menten type harvesting on the second species.

    Example 4.2. Some parameter values are taken α = 1.85, β = 0.3, γ = 2.5 and δ = 0.8. In this case, parameters satisfy the conditions 0<α<α and α>δ+δγββγ. Consequently, system (1.4) has a positive equilibrium (global asymptotically stable) and three boundary equilibria. From δ>β, we can effortlessly observe there exists no positive equilibrium for model (3.1). See Figures 6(b) and 7(b), the numerical phase portraits corresponding to this situation are presented. The impact of harvesting is illustrated with curves in Figure 9.

    Figure 9.  The impact of Michaelis-Menten type harvesting on the second species.

    Example 4.3. We select parameter values as follows: α = 0.8, β = 2.5, γ = 0.4, δ = 2. For these parameters, we derive α=α and 0<δβγ<2β. Therefore, system (1.4) possesses the same types of equilibria as the above Example 4.2. Whereas, there is a positive equilibrium (global asymptotically stable) in system (3.1) for δ<β. See Figures 6(c) and 7(c), the corresponding numerical phase portraits are presented. The impact of harvesting is illustrated with curves in Figure 10.

    Figure 10.  The impact of Michaelis-Menten type harvesting on the second species.

    It can be seen that the harvesting effect has a significant role on the population densities. From a biological point of view, the permanence and extinction of species are influenced to differential extent by the harvesting effect on the second species, for instance, greatly slowing the extinction of the first species (Figure 8) or guaranteeing the permanence of the system by preserving the first species from extinction (Figure 9). In addition, both species need much more time to reach their stable coexistence state due to the harvesting effect, which is shown in Figure 10.

    In this article, by combining local and global dynamical analysis, we have investigated the dynamics of the amensalism system with Michaelis-Menten type harvesting for the second species. In order to further explore the influence of harvesting effect, a complete qualitative analysis of the system with no harvesting is also performed. We could show some differences for the model (1.4) with harvesting on the second species and the system (3.1) with no harvesting.

    1) When harvesting is present in the system, it has shown that model (1.4) has more different types of equilibria. System (3.1) has at most four equilibria including one interior point in R2+ and the unique interior point (if it exists) of system (3.1) is globally stable. However, when the Michaelis-Menten type harvesting is on the second species, we shown that the model (1.4) has at most six equilibria including two interior points in R2+ and the interior point E31 is a saddle.

    2) The complex existence of equilibria leads to more bifurcation behaviors of model (1.4). Two kinds of bifurcation, under some parameter restrictions, saddle-node bifurcation and transcritical bifurcation will occur for system (1.4), while there exhibits only one saddle-node bifurcation for the system with no harvesting.

    3) More types of equilibria and more bifurcation behaviors can induce potentially dramatic changes to the dynamics of the system. By the classifications of global phase portraits of model (1.4), we find that more complex dynamics occur for the appearance of harvesting on the second species. Such as, for model (1.4) with harvesting, the first species survives permanently rather than extinction or both species take a lot longer to approach the coexistence state.

    These results reveal that the dynamical properties of system (1.4) get more complicated and richer compared to the system with no harvesting. Furthermore, it would also be interesting to study system (3.1) with both the first species and the second species with harvesting terms since we usually harvest, or would like to harvest, both populations.

    The first author was supported by the National Natural Science Foundation of China (No.12001503) and the third author was supported by the Project of Beijing Municipal Commission of Education (KM 202110015001).

    The authors declare there is no conflicts of interest.



    [1] Zia MF, Benbouzid M, Elbouchikhi E, et al. (2020) Microgrid transactive energy: Review, architectures, distributed ledger technologies, and market analysis. IEEE Access 8: 19410–19432. doi: 10.1109/ACCESS.2020.2968402
    [2] Han Y, Ning X, Yang P, et al. (2019) Review of power sharing, voltage restoration and stabilization techniques in hierarchical controlled DC microgrids. IEEE Access 7: 149202–149223. doi: 10.1109/ACCESS.2019.2946706
    [3] Wu X, Xu Y, He J, et al. (2019) Pinning-based hierarchical and distributed cooperative control for AC microgrid clusters. DIEEE Trans Power Elect, 1–20.
    [4] Natesan C, Kumar AS, Palani P, et al. (2014) Survey on microgrid: power quality improvement techniques. ISRN Otolaryngology. Available from: https://www.oalib.com/paper/3089465.
    [5] Rocabert J, Luna A, Blaabjerg F, et al. (2012) Control of power converters in AC microgrids. IEEE Trans Power Electron 27: 4734–4749. doi: 10.1109/TPEL.2012.2199334
    [6] Frolov V, Thakurta PG, Backhaus S, et al. (2019) Operations and uncertainty-aware installation of FACTS devices in a large transmission system. IEEE Trans contral Network Syst 6: 961–970. doi: 10.1109/TCNS.2019.2899104
    [7] Nascimentoa S, Gouvêa M (2016) Voltage stability enhancement in power systems with automatic facts device allocation. Energy Procedia 107: 60–67. doi: 10.1016/j.egypro.2016.12.129
    [8] Wvong D, Mihirig AM (1986) Catastrophe theory applied to transient stability assessment of power systems. IEE Proceed 133: 314–318. doi: 10.1049/ip-d.1986.0052
    [9] Yorino N, Priyadi A, Kakui H, et al. (2010) A new method for obtaining critical clearing time for transient stability. IEEE Trans Power Syst 25: 1620–1626. doi: 10.1109/TPWRS.2009.2040003
    [10] Pradhan M, Mishra MK (2019) Dual P-Q theory based energy-optimized dynamic voltage restorer for power quality improvement in a distribution system. IEEE Trans Ind Elect 66: 2946–2955. doi: 10.1109/TIE.2018.2850009
    [11] Ghahremani E, Kamwa I (2012) Optimal placement of multiple-type FACTS devices to maximize power system loadability using a generic graphical user interface. IEEE Trans Power Syst, 1–15.
    [12] Prodanovixc M, Green TC (2006) High-quality power generation through distributed control of a power park microgrid. IEEE Trans Ind Elect 53: 1471–1482. doi: 10.1109/TIE.2006.882019
    [13] He L, Li Y, Guerrero JM, et al. (2020) A comprehensive inertial control strategy for hybrid AC/DC microgrid with distributed generations. IEEE Trans Smart Grid 11: 1737–1747. doi: 10.1109/TSG.2019.2942736
    [14] Nejabatkhah F, Li YW, Tian H (2019) Power quality control of smart hybrid AC/DC microgrids: An overview. IEEE Access 7: 52295–52318. doi: 10.1109/ACCESS.2019.2912376
    [15] Shalukho AV, Lipuzhin IA, Voroshilov AA (2019) Power quality in microgrids with distributed generation. 2019 International ural conference on electrical power engineering (UralCon), 54–58.
    [16] Ovaskainen M, Rni J, Leinonen A (2019) Superposed control strategies of a BESS for power exchange and microgrid power quality improvement. IEEE International Conference on Environment and Electrical Engineering; IEEE Industrial and Commercial Power Systems Europe. Merus Power Dynamics Oy, Nokia, Finland.
    [17] Baimel D, Belikov J, Guerrero JM, et al. (2017) Dynamic modeling of networks, microgrids, and renewable sources in the dq0 reference frame: A survey. IEEE Access 5: 21323–21335. doi: 10.1109/ACCESS.2017.2758523
    [18] Srivatchana NS, Rangarajanb P, Rajalakshmib S (2015) Control scheme for power quality improvement in Islanded. Procedia Technol 21: 212–215. doi: 10.1016/j.protcy.2015.10.090
    [19] Dong T, Li L, Key ZM (2013) A combined system of APF and SVC for power quality improvement in microgrid. Power Engineering and Automation Conference (PEAM). Available from: https: //ieeexplore.ieee.org/document/6612554.
    [20] Priyavarthini S, Kathiresan AC, Nagamani C, et al. (2018) PV-fed DVR for simultaneous real power injection and sag/swell mitigation in a wind farm. IET Power Electron 11: 2385–2395. doi: 10.1049/iet-pel.2018.5123
    [21] Nikoobakht A, Aghaei J, Parvania M, et al. (2018) Contribution of FACTS devices in power systems security using MILP-based OPF. IET Gener, Transm Distrib 12: 3744–3755. doi: 10.1049/iet-gtd.2018.0376
    [22] Roldan-Pxerez J, Garcxıa-Cerrada A, Ochoa-Gimenez M, et al. (2019) Delayed-signal-ancellation-based sag detector for a dynamic voltage restorer in distorted grids. IEEE Trans Sus Energy 10: 2015–2027. doi: 10.1109/TSTE.2018.2877505
    [23] Wang J, Xing Y, Wu H, et al. (2019) A novel dual-DC-port dynamic voltage restorer with reduced-rating integrated DC–DC converter for wide-range voltage sag compensation. IEEE Trans Power Electron 34: 7437–7449. doi: 10.1109/TPEL.2018.2882534
    [24] Li P, Xie L, Han J, et al. (2018) A new voltage compensation philosophy for dynamic voltage restorer to mitigate voltage sags using three-phase voltage ellipse parameters. IEEE Trans Power Electron 33: 1154–1166. doi: 10.1109/TPEL.2017.2676681
    [25] Rawat MS, Vadhera S (2016) Comparison of FACTS devices for transient stability enhancement of multi machine power system. 2016 International Conference on Microelectronics, Computing and Communications. Available from: https://ieeexplore.ieee.org/document/7522419.
    [26] Ogunboyo PT, Tiako R, Davidson IE (2018) Effectiveness of dynamic voltage restorer for unbalance voltage mitigation and voltage profile improvement in secondary distribution system. Can J Elect Comput Eng 41: 105–114.
    [27] Maharashtra S (2013) Enhancement of voltage profile using dynamic voltage restorer. Int J Adv Res Elect, Elect Instrum Eng (IJAREEIE) 2: 5871–5876.
    [28] Ramasamy AK, Iyer RK, Chandaramuthy VK, et al. (2005) Dynamic voltage restorer for voltage sag compensation. IEEE PEDS, 1289–1294.
    [29] Hagha MT, Shakera A, Sohrabia F, et al. (2017) Fuzzy-based controller for DVR in the presence of DG. Procedia Comput Sci 120: 684–690. doi: 10.1016/j.procs.2017.11.296
    [30] Razak HNA, Said DM, Ahmad N (2018) Improved analysis of DVR performance for voltage sag mitigation. ELEKTRIKA-J Elect Eng 17: 36–40. doi: 10.11113/elektrika.v17n3.110
    [31] Rao SS, Krishna PSR, Babu S (2017) Mitigation of voltage sag, swell and THD using dynamic voltage restorer with photovoltaic system. International Conference on Algorithms, Methodology, Models and Applications in Emerging Technologies (ICAMMAET), 1–7.
    [32] Ma H, Du C (2015) Multi-objective coordinated design of TCSC and SVC for improving transient stability fifth. International Conference on Instrumentation and Measurement, Computer, Communication and Control, 246–250.
    [33] Zheng ZX, Xiao XY, Chen XY, et al. (2018) Performance evaluation of a MW-class SMES-BES DVR system for mitigation of voltage quality disturbances. IEEE Trans Ind Appl 54: 3090–3099. doi: 10.1109/TIA.2018.2823259
    [34] Remya VK, Parthiban P, Ansal V, et al. (2018) Dynamic voltage restorer (DVR)–A review. J Green Eng 8: 519–572. doi: 10.13052/jge1904-4720.844
    [35] Riddhi FB (2018) Study on compensation of voltage sag and voltage swell by using DVR (Dynamic Voltage Restorer). IEEE International Conference on Current Trends toward Converging Technologies, Coimbatore, 1–12.
    [36] Tiwari SP, Sharma PK (2017) Synchronous reference frame theory for active power filter. Int J Eng Sci Comput 7. Available from: https://www.researchgate.net/publication/335602113_synchronous_reference_frame_theory_for_active_power_filter.
    [37] EIAdany M, EIDesouky A, Sallam A (2018) Power system transient stability: An algorithm for assessment and enhancement based on catastrophe theory and FACTS devices. IEEE Access 6: 26424–26437. doi: 10.1109/ACCESS.2018.2834906
    [38] Dash SK, Ray PK (2018) Power quality improvement utilizing PV fed unified power quality conditioner based on UV-PI and PR-R controller. CPSS Trans Power Electron Appl 3: 243–253. doi: 10.24295/CPSSTPEA.2018.00024
    [39] Bhushan S, Rakhonde A, Bobade CM (2017) Harmonic mitigation using modified synchronous reference frame theory. IRJET 4: 2340–2345.
    [40] Zheng Z, Xia X, Huang C, et al. (2019) Enhancing transient voltage quality in a distribution power system with SMES-based DVR and SFCL. IEEE Trans Appl Supercond 29: 1–5.
    [41] Zhai H, Zhuo F, Zhu C, et al. (2020) An optimal compensation method of shunt active power filters for system-wide voltage quality improvement. IEEE Trans Power Electron 67: 1270–1281. doi: 10.1109/TED.2020.2967242
    [42] Karthikeyan A, Gururaj D, Krishna A, et al. (2010) Dual role CDSC-based dual vector control for effective operation of DVR with harmonic mitigation. IEEE Trans Power Electron 66: 4–13.
    [43] Singh B, Solanki J (2009) A comparison of control algorithms for dstatcom. IEEE Trans Power Electron 56: 2738–2745.
    [44] Choudante SD, Bhole AA (2018) A review: Voltage stability and power flow improvement by using UPFC controller. 2018 International conference on computation of power, energy, information and communication (ICCPEIC), 462–465.
    [45] Upadhyay P, Singh N, Yadav S (2018) Voltage quality compensation of DFIG with series DVR (SDVR) under three phase fault. International Conference on Electronics, Materials Engineering and Nano-Technology. Available from: https://ieeexplore.ieee.org/document/8465397.
    [46] Biricik S, Komurcugil H, Tuyen ND, et al. (2019) Protection of sensitive loads using sliding mode controlled three-phase DVR with adaptive notch filter. IEEE Trans Power Electron 66: 5465–5475.
    [47] Kamarposhti MA, Lesani H (2010) Comparison between parallels and series FACTS devices on static voltage stability using MLP index. International symposium on power electronics, electrical drives, automation and motion, 257–261.
    [48] Prakash Y, Sankar S (2014) Power quality improvement using DVR in power system. Power and energy systems: Towards sustainable energy (PESTSE), 1–6.
    [49] Zong S, Fan G, Yang X (2019) Double voltage rectification modulation for bidirectional DC/DC resonant converters for wide voltage range operation. IEEE Trans Power Electron 34: 6510–6521. doi: 10.1109/TPEL.2018.2875816
    [50] Zia MF, Benbouzid M, Elbouchikhi E, et al. (2020) Microgrid transactive energy: Review, architectures, distributed ledger technologies, and market analysis. IEEE Access 8: 19410–19432. doi: 10.1109/ACCESS.2020.2968402
    [51] Lu Y (2018) Adaptive-fuzzy control compensation design for direct adaptive fuzzy control. IEEE Trans Fuzzy Syst 26: 3222–3231. doi: 10.1109/TFUZZ.2018.2815552
    [52] Meng L, Guerrero JM (2017) Optimal power quality service in multi-bus microgrid systems. 2017 IEEE Power & Energy Society General Meeting. Available from: https://ieeexplore.ieee.org/document/8274169.
    [53] Zhong Q, Yao W, Lin L, et al. (2018) Data analysis and applications of the power quality monitoring. International Conference on Power System Technology (POWERCON), 6–8.
    [54] Omar AI, Shady HE, Aleem A, et al. (2019) An improved approach for robust control of dynamic voltage restorer and power quality enhancement using grasshopper optimization algorithm. ISA Trans 95: 110–129. doi: 10.1016/j.isatra.2019.05.001
    [55] Talwariya A, Singh P, Kolhe ML, et al. (2020) Fuzzy logic controller and game theory based distributed energy resources allocation. AIMS Energy 8: 474–492. doi: 10.3934/energy.2020.3.474
    [56] Poullikkas A, Papadouris S, Kourtis G, et al. (2014) Storage solutions for power quality problems in cyprus' electricity distribution network. AIMS Energy 2: 1–17. doi: 10.3934/energy.2014.1.1
    [57] Ahmed eldessouky and hossam gabbar (2016) SVC control enhancement applying self-learning fuzzy algorithm for islanded microgrid. AIMS Energy 4: 363–378. doi: 10.3934/energy.2016.2.363
    [58] Patnaik RK, Mishra SP, Patra JP (2015) Active and reactive power based feedback linearization control technique for grid-connected voltage source converter. Power Communication and Information Technology Conference (PCITC), 290–296.
  • This article has been cited by:

    1. Qianqian Li, Ankur Jyoti Kashyap, Qun Zhu, Fengde Chen, Dynamical behaviours of discrete amensalism system with fear effects on first species, 2023, 21, 1551-0018, 832, 10.3934/mbe.2024035
    2. Yajie Sun, Ming Zhao, Yunfei Du, Bifurcations, chaos analysis and control in a discrete predator–prey model with mixed functional responses, 2024, 17, 1793-5245, 10.1142/S1793524523500286
    3. Qun Zhu, Fengde Chen, Zhong Li, Lijuan Chen, Global Dynamics of Two-Species Amensalism Model with Beddington–DeAngelis Functional Response and Fear Effect, 2024, 34, 0218-1274, 10.1142/S0218127424500755
    4. MING ZHAO, YAJIE SUN, YUNFEI DU, COMPLEX DYNAMICAL BEHAVIORS OF A DISCRETE MODIFIED LESLIE–GOWER PREDATOR–PREY MODEL WITH PREY HARVESTING, 2025, 33, 0218-3390, 101, 10.1142/S0218339024500451
    5. Manoj Kumar Singh, Arushi Sharma, 2025, 3283, 0094-243X, 040017, 10.1063/5.0265020
  • Reader Comments
  • © 2021 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(4602) PDF downloads(580) Cited by(17)

Figures and Tables

Figures(36)  /  Tables(3)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog