
Early detection of the risk of sarcopenia at younger ages is crucial for implementing preventive strategies, fostering healthy muscle development, and minimizing the negative impact of sarcopenia on health and aging. In this study, we propose a novel sarcopenia risk detection technique that combines surface electromyography (sEMG) signals and empirical mode decomposition (EMD) with machine learning algorithms. First, we recorded and preprocessed sEMG data from both healthy and at-risk individuals during various physical activities, including normal walking, fast walking, performing a standard squat, and performing a wide squat. Next, electromyography (EMG) features were extracted from a normalized EMG and its intrinsic mode functions (IMFs) were obtained through EMD. Subsequently, a minimum redundancy maximum relevance (mRMR) feature selection method was employed to identify the most influential subset of features. Finally, the performances of state-of-the-art machine learning (ML) classifiers were evaluated using a leave-one-subject-out cross-validation technique, and the effectiveness of the classifiers for sarcopenia risk classification was assessed through various performance metrics. The proposed method shows a high accuracy, with accuracy rates of 0.88 for normal walking, 0.89 for fast walking, 0.81 for a standard squat, and 0.80 for a wide squat, providing reliable identification of sarcopenia risk during physical activities. Beyond early sarcopenia risk detection, this sEMG-EMD-ML system offers practical values for assessing muscle function, muscle health monitoring, and managing muscle quality for an improved daily life and well-being.
Citation: Konki Sravan Kumar, Daehyun Lee, Ankhzaya Jamsrandoj, Necla Nisa Soylu, Dawoon Jung, Jinwook Kim, Kyung Ryoul Mun. sEMG-based Sarcopenia risk classification using empirical mode decomposition and machine learning algorithms[J]. Mathematical Biosciences and Engineering, 2024, 21(2): 2901-2921. doi: 10.3934/mbe.2024129
[1] | Luyao Li, Licheng Fang, Huan Liang, Tengda Wei . Observer-based event-triggered impulsive control of delayed reaction-diffusion neural networks. Mathematical Biosciences and Engineering, 2025, 22(7): 1634-1652. doi: 10.3934/mbe.2025060 |
[2] | Mingxia Gu, Zhiyong Yu, Haijun Jiang, Da Huang . Distributed consensus of discrete time-varying linear multi-agent systems with event-triggered intermittent control. Mathematical Biosciences and Engineering, 2024, 21(1): 415-443. doi: 10.3934/mbe.2024019 |
[3] | Yilin Tu, Jin-E Zhang . Event-triggered impulsive control for input-to-state stability of nonlinear time-delay system with delayed impulse. Mathematical Biosciences and Engineering, 2025, 22(4): 876-896. doi: 10.3934/mbe.2025031 |
[4] | Yunxia Wei, Yuanfei Zhang, Bin Hang . Construction and management of smart campus: Anti-disturbance control of flexible manipulator based on PDE modeling. Mathematical Biosciences and Engineering, 2023, 20(8): 14327-14352. doi: 10.3934/mbe.2023641 |
[5] | Siyu Li, Shu Li, Lei Liu . Fuzzy adaptive event-triggered distributed control for a class of nonlinear multi-agent systems. Mathematical Biosciences and Engineering, 2024, 21(1): 474-493. doi: 10.3934/mbe.2024021 |
[6] | Yongli Yan, Fucai Liu, Teng Ren, Li Ding . Nonlinear extended state observer based control for the teleoperation of robotic systems with flexible joints. Mathematical Biosciences and Engineering, 2024, 21(1): 1203-1227. doi: 10.3934/mbe.2024051 |
[7] | Dong Xu, Xinling Li, Weipeng Tai, Jianping Zhou . Event-triggered stabilization for networked control systems under random occurring deception attacks. Mathematical Biosciences and Engineering, 2023, 20(1): 859-878. doi: 10.3934/mbe.2023039 |
[8] | Tianqi Yu, Lei Liu, Yan-Jun Liu . Observer-based adaptive fuzzy output feedback control for functional constraint systems with dead-zone input. Mathematical Biosciences and Engineering, 2023, 20(2): 2628-2650. doi: 10.3934/mbe.2023123 |
[9] | Wenjing Wang, Jingjing Dong, Dong Xu, Zhilian Yan, Jianping Zhou . Synchronization control of time-delay neural networks via event-triggered non-fragile cost-guaranteed control. Mathematical Biosciences and Engineering, 2023, 20(1): 52-75. doi: 10.3934/mbe.2023004 |
[10] | Tingting Cai, Yuqian Wang, Liang Wang, Zongying Tang, Jun Zhou . Analysis and event-triggered control for a stochastic epidemic model with logistic growth. Mathematical Biosciences and Engineering, 2023, 20(2): 2243-2260. doi: 10.3934/mbe.2023105 |
Early detection of the risk of sarcopenia at younger ages is crucial for implementing preventive strategies, fostering healthy muscle development, and minimizing the negative impact of sarcopenia on health and aging. In this study, we propose a novel sarcopenia risk detection technique that combines surface electromyography (sEMG) signals and empirical mode decomposition (EMD) with machine learning algorithms. First, we recorded and preprocessed sEMG data from both healthy and at-risk individuals during various physical activities, including normal walking, fast walking, performing a standard squat, and performing a wide squat. Next, electromyography (EMG) features were extracted from a normalized EMG and its intrinsic mode functions (IMFs) were obtained through EMD. Subsequently, a minimum redundancy maximum relevance (mRMR) feature selection method was employed to identify the most influential subset of features. Finally, the performances of state-of-the-art machine learning (ML) classifiers were evaluated using a leave-one-subject-out cross-validation technique, and the effectiveness of the classifiers for sarcopenia risk classification was assessed through various performance metrics. The proposed method shows a high accuracy, with accuracy rates of 0.88 for normal walking, 0.89 for fast walking, 0.81 for a standard squat, and 0.80 for a wide squat, providing reliable identification of sarcopenia risk during physical activities. Beyond early sarcopenia risk detection, this sEMG-EMD-ML system offers practical values for assessing muscle function, muscle health monitoring, and managing muscle quality for an improved daily life and well-being.
As computer technology and machinery manufacturing technology develops, people are expecting more and more from production automation. Since the last century, the manipulator system has gradually replaced human beings to complete the dangerous and repetitive work in various fields [1,2,3,4]. At the same time, the manipulator system can also significantly improve the production efficiency [5,6,7]. The flexible manipulator has better performance of high stability, high precision, high efficiency and low energy consumption than the traditional rigid manipulator [8,9,10,11]. Consequently, it is more adapted to the complex and changeable working environments in various fields. For example, for the sake of improving the automation level of agricultural production, the flexible manipulator system is adopted to pick fruit and vegetable crops in the field of agriculture, so as to further ensure the safety of agricultural products and improve the production efficiency in the process of processing and production. However, the flexible manipulator is characterized by its complex structure, low control accuracy, difficult control, etc. These defects may lead to vibration of the flexible manipulator system, which greatly affects the stability of the actual production. Thus, improving its stiffness and control accuracy, and suppressing the vibration of the flexible manipulator system have become the focus of current research.
By referring to the literature, we can know that the manipulator system with special flexible structure is a typical infinite dimensional distributed parameter system [12,13,14]. Most of the existing studies of the manipulators are based on the ordinary differential equation (ODE) dynamic models [15,16,17,18]. Nevertheless, these ODE models limit the system to a few key patterns, greatly affecting system performance [19]. For getting the accurate description of the flexible manipulator systems, the model cannot be constructed only through a single ODE; otherwise, there will be spillover instability [20]. Therefore, it is necessary to introduce partial differential equations (PDEs) in the flexible connection systems. At present, there are some research achievements on flexible systems described by PDEs. In [21], for the sake of the achievement of control goals, a boundary controller with input backlash is constructed based on the infinite-dimensional dynamic model. For the single flexible link manipulator system in [22], a sliding mode boundary controller is designed based on the adaptive radial basis function (RBF) neural network (NN) to drive the joint to the required position and quickly suppress the vibration on the beam. Then, an adaptive fault-tolerant control method is raised by using RBFNN and LaSalle's invariance principle to solve the failure problem of the actuator of the single-link flexible manipulator in [23].
Over the past two decades, systems modeled by PDEs have attracted more and more researchers because of their wide application in various fields, and numerous methods have been reported [24,25,26,27,28]. However, these results [21,22,23,24,25,26,27,28] all ignored the constraint problem. In fact, many real-world systems are limited by constraints in various ways [29]. It is possible that such constraints are due to physical restriction of systems, or caused by the requirements of safe operation [30]. Motivated by progress in constraints, lots of state constraint problems have been researched for ODE systems [31,32,33]. With the rise of the research on PDE systems, some scholars also put their attention to the problem of state constraints of PDE flexible mechanical systems. In [34], a class of flexible riser systems with backlash modeled by PDEs is considered. In order to solve its position and velocity constraints, logarithmic BLF is used. In [35], the state feedback control problem of moving vehicle-mounted manipulator modeled by PDE with output constraint is studied. Under the action of the designed control scheme, the position control and vibration suppression are effectively improved. For the uncertain PDE flexible manipulator system in [36], a NN fault-tolerant control scheme under state constraints is proposed. In the design process, the tangent BLF is utilized to handle the constraint problem, and get a good control.
In addition to the state constraint problem, in today's society, production resources are also tight. While meeting the quality of control, saving resources has become an important aspect that needs to be consider. In recent years, the event triggered control [37,38,39,40], as an effective method that can not only achieve control objectives, but also save resources, has raised the broad interest of all researchers. The event triggered control is a control mechanism of sampling on demand. System resources can only be used when necessary, and can meet the expected control performance indicators. In [41], a collaborative design scheme consisted of switching event triggering mechanism and mode dependent adaptive control law is proposed which solves the mismatch problem and avoids the Zeno behavior. In [42], for nonlinear uncertain systems, besides the design methods on the basis of fixed threshold strategy and relative threshold strategy, a new switching threshold strategy is proposed. However, the above results are only applicable to the system modeled by ordinary differential methods. When these methods are directly applied to the control system modeled by partial differential methods, it may lead to the failure of control strategy, and even bring huge losses to practical engineering. In addition, in the actual production and life, many control systems need to be modeled by partial differential method to achieve better control effect. Among them, the flexible manipulator system modeled by partial differential method is widely used in [43,44,45]. Therefore, in order to make efficient use of resources, the event-triggered control of flexible manipulator systems modeled by PDEs under state constraints is a significant topic of study that has inspired our own research.
It can be seen from the above analysis that although researchers have put forward many research results for flexible manipulator system, there are still some limitations. Therefore, the event-triggered control of a PDE flexible manipulator with constraints will be taken as the research object in this paper, and the control goal of saving communication resources will be achieved by designing event-triggered controllers. On the premise of achieving the stable performance of the system, good vibration suppression effect of the flexible manipulator will be maintained. On account of the above discussion, the innovation of this article is given below: when dealing with the state constraint of PDE flexible manipulator system, an event trigger control strategy is introduced.
In this paper, an event-triggered control design problem is studied for flexible manipulator systems with full state constraints. Under frameworks of adaptive backstepping control design technique, an event-triggered control scheme is proposed for flexible manipulator systems. The main contribution of the paper is summarized as follows:
1) In this paper, the design problem of event-triggered control is studied for flexible manipulator system with full state constraints and an event-triggered control method is proposed. Different from the constraint control scheme in [31,32], the event-triggered control strategy proposed in this paper can save unnecessary control signal transmission and improve the system performance.
2) An event-triggered mechanism with relative threshold is designed, and the control signal update is event-driven under well-established event-triggered strategy. The proposed event-triggered control scheme effectively reduces the communication burden in the controller-to-the-actuator channel and still ensures the system stability, and it achieves the control objective.
The main contents of Sections 2 to 6 are as follows: Section 2 is the partial differential system model, and it gives the assumption and control objectives. The design procedure of the event trigger controllers based on Tan-BLF and backstepping technique is introduced in Section 3. Section 4 is the system stability analysis process. In Section 5, the effectiveness of the proposed method is further demonstrated with the help of a simulation example. Finally, the conclusion is given in Section 6.
Notations. To simplify and differentiate, notations (A)r=∂(A)/∂(A)∂r∂r, (˙A)=∂(A)/∂(A)∂τ∂τ throughout this paper. In the same way, (A)rr means ∂2(A)/∂2(A)∂r∂r2, (A)rrr=∂3(A)/∂3(A)∂r∂r3 and (A)rrrr=∂4(A)/∂4(A)∂r∂r4, (¨A) =∂2(A)/=∂2(A)∂τ∂τ2. In addition, (A)T stands for transposition of (A).
Based on the Hamiltonian principle [37], the dynamic model of the flexible manipulator system is solved as follows
∫t2t1(ϵEk−ϵEp+ϵW)dt=0 | (1) |
where ϵ(A) means the variation of (A). The expressions of kinetic energy Ek, potential energy Ep and work W produced in the operation of the system are respectively listed as
Ek=12Ih˙ϱ2(τ)+12∫L0ℓ˙Y2(r,τ)dr+12m˙Y2(X,τ) | (2) |
Ep=12∫L0EIζ2rr(r,τ)dr | (3) |
W=Φ(τ)ϱ(τ)+O(τ)Y(X,τ) | (4) |
where Ih stands for the hub inertia; ϱ(τ) represents the joint angle; ℓ and Y are the mass per unit length and the arc length at r of the flexible manipulator, respectively, where Y(r,τ)=rϱ(τ)+ζ(r,τ); the mass of the payload is m; the bending stiffness is denoted by EI; the manipulator length and the connecting rod vibration deflection at r are expressed by X and ζ(r,τ); the torque input of the joint motor and the force input of the actuator are represented with Φ(τ) and O(τ), respectively.
Combined with the Hamiltonian principle, through a series of derivations, the system PDE model can be written as follows:
ℓ¨Y(r,τ)=−EIζrrrr(r,τ) | (5) |
Φ(τ)=Ih¨ϱ(τ)−EIζrr(0,τ) | (6) |
O(τ)=m¨Y(X,τ)−EIζrrr(X,τ) | (7) |
ζ(0,τ)=ζr(0,τ)=ζrr(X,τ)=0 | (8) |
Furthermore, ϱ(τ) and ζ(X,τ) are the system outputs, and they meet ϱ(τ)<kd1 and ζ(X,τ)<kd2 with kd1 and kd2 being constants. There are two constants kc1 and kc2 such that following formulas hold:
|z1(0)|=|ϱ(0)−ϱd|<kc1 | (9) |
|z3(0)|=|ζ(X,0)−ζd(X,0)|<kc2 | (10) |
where z1=ϱ(τ)−ϱd, ϱd is the ideal angle position, and ϱd is a constant, and z3=ζ(X,τ)−ζd(X,τ), ζd(X,τ) means the required vibration.
Assumption 1 [46]. Suppose that the parameters ζrr(0,τ) and ζrrr(X,τ) are attainable.
Control objective: The event-triggered controllers are designed to realize the following control objectives:
1) suppresses the vibration of the manipulator and stabilizes it in the desired position.
2) the joint angle ϱ(τ) and boundary vibration diversion ζ(X,τ) are confined within the constraints.
3) the system signals are all bounded.
4) it can effectively avoid the occurrence of the Zeno behavior.
The following Eqs (11) and (12) are the system boundary errors:
z1(τ)=ϱ(τ)−ϱd | (11) |
z2(τ)=˙ϱ(τ)−μ(τ) | (12) |
z3(τ)=ζ(X,τ)−ζd(X,τ) | (13) |
z4(τ)=˙ζ(X,τ)−ϑ(τ) | (14) |
where μ(τ)=−k1z1(τ) and ϑ(τ)=−k3z3(τ) are virtual controls with ˙ϱd=0 and ˙ζd=0, k1>0 and k3>0.
Taking the derivative of (11)–(14), and combining (5)–(8), one has
˙z1(τ)=z2(τ)+μ(τ)−˙ϱd=z2(τ)−k1z1(τ) | (15) |
˙z2(τ)=(Φ(τ)+EIζrr(0,τ))/Ih−˙μ(τ) | (16) |
˙z3(τ)=z4(τ)+ϑ(τ)−˙ζd(X,τ)=z4(τ)−k3z3(τ) | (17) |
˙z4(τ)=(O(τ)+EIζrrr(X,τ)−ρX(Φ(τ)+EIζrr(0,τ)))/m−˙ϑ(τ) | (18) |
where ρ=m/mIhIh.
Choose the following Lyapunov function:
V1(τ)=12log(k2c1k2c1−z21(τ)) | (19) |
Then, taking the derivative of V1(τ) based on ˙z1(τ), one gets
˙V1(τ)=−k1z21(τ)k2c1−z21(τ)+z1(τ)z2(τ)k2c1−z21(τ) | (20) |
In order to eliminate the z1(τ)z2(τ) in (20), the Lyapunov function V2(τ) is selected in the following form:
V2(τ)=V1(τ)+12Ihz22(τ) | (21) |
The derivative of V2(τ) along time is
˙V2(τ)=−k1z21(τ)k2c1−z21(τ)+z1(τ)z2(τ)k2c1−z21(τ)+z2(τ)(Φ(τ)+EIζrr(0,τ)−Ih˙μ(τ)) | (22) |
The boundary controller is designed as
Φ(τ)=−EIζrr(0,τ)+Ih˙μ(τ)−z1(τ)k2c1−z21(τ)−k2z2(τ) | (23) |
where k2>0 is a constant. Substituting (23) into (22) yields
˙V2(τ)=−k1z21(τ)k2c1−z21(τ)−k2z22(τ) | (24) |
Construct the following Lyapunov function, V3(τ), as
V3(τ)=V2(τ)+12log(k2c2k2c2−z23(τ)) | (25) |
Then, from (17), the ˙V3(τ) can be obtained as
˙V3(τ)=˙V2(τ)−k3z23(τ)k2c2−z23(τ)+z3(τ)z4(τ)k2c2−z23(τ)=−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)+z3(τ)z4(τ)k2c2−z23(τ) | (26) |
Select V4(τ) as
V4(τ)=V3(τ)+12mz24(τ) | (27) |
Then, the differential coefficient of V4(τ) is
˙V4(τ)=˙V3(τ)+mz4(τ)˙z4(τ)=−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)+z3(τ)z4(τ)k2c2−z23(τ)+z4(τ)(O(τ)+EIζrrr(X,τ)−ρX(Φ(τ)+EIζrr(0,τ))−m˙ϑ(τ)) | (28) |
The desired boundary controller is designed as
O(τ)=−EIζrrr(X,τ)+ρX(Φ(τ)+EIζrr(0,τ))+m˙ϑ(τ)−z3(τ)k2c2−z23(τ)−k4z4(τ) | (29) |
where k4>0 is a constant.
From (29) and (28), one has
˙V4(τ)=−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)−k2z24(τ) | (30) |
The event trigger mechanism is proposed so that the communication resources are commendably reduced.
Under the event-triggering mechanism, the boundary control strategy is designed as follows
Φ0(τ)=ℏ1(τk),∀τ∈[τk,τk+1) | (31) |
τk+1=inf{τ∈R||e1(τ)|⩾δ1|Φ0(τ)|+κ1} | (32) |
O0(τ)=ℏ2(τs),∀τ∈[τs,τs+1) | (33) |
τs+1=inf{τ∈R||e2(τ)|⩾δ2|O0(τ)|+κ2} | (34) |
where e1(τ)=ℏ1(τ)−Φ0(τ), e2(τ)=ℏ2(τ)−O0(τ), κ1, κ2, 0<δ1<1, 0<δ2<1 are all positive design parameters. τk,k∈Z+, and τs,s∈Z+ are the moments when the event is triggered. The times will be respectively marked as τk+1 and τs+1 whenever (32) and (34) are triggered, and the control values Φ0(τk+1) and O0(τs+1) will be applied to the system.
Design ℏ1(τ) and ℏ2(τ) as follows:
ℏ1(τ)=−(1+δ1)[ϕ1(τ)tanh(z2(τ)ϕ1(τ)a1)+ˉκ1tanh(z2(τ)ˉκ1a1)] | (35) |
ℏ2(τ)=−(1+δ2)[ϕ2(τ)tanh(z4(τ)ϕ2(τ)a2)+ˉκ2tanh(z4(τ)ˉκ2a2)] | (36) |
where
ϕ1(τ)=−EIζrr(0,τ)+Ih˙μ(τ)−z1(τ)k2c1−z21(τ)−k2z2(τ) |
ϕ2(τ)=−EIζrrr(X,τ)+ρX(Ih˙μ(τ)−z1(τ)k2c1−z21(τ)−k2z2(τ))+m˙ϑ(τ)−z3(τ)k2c2−z23(τ)−k4z4(τ) |
ai>0,i=1,2, and ˉκi>κi/κi(1−δi)(1−δi),i=1,2.
According to (32) and (34), it holds that |ℏ1(τ)−Φ0(τ)|<δ1|Φ0(τ)|+κ1 and |ℏ2(τ)−O0(τ)|<δ2|O0(τ)|+κ2, for ∀τ∈[τk,τk+1) and ∀τ∈[τs,τs+1). Then, there are parameters λ1(τ)⩽1 λ2(τ)⩽1, η1(τ)⩽1 and η2(τ)⩽1, such that
ℏ1(τ)=(1+λ1(τ)δ1)Φ0(τ)+λ2(τ)κ1 | (37) |
ℏ2(τ)=(1+η1(τ)δ2)O0(τ)+η2(τ)κ2 | (38) |
Then, we get
Φ0(τ)=ℏ1(τ)1+λ1(τ)δ1−λ2(τ)κ11+λ1(τ)δ1 | (39) |
O0(τ)=ℏ2(τ)1+η1(τ)δ2−η2(τ)κ21+η1(τ)δ2 | (40) |
Further, ˙V2(τ) and ˙V4(τ) can be rewritten as
˙V2(τ)=−k1z21(τ)k2c1−z21(τ)+z1(τ)z2(τ)k2c1−z21(τ)+z2(τ)(ℏ1(τ)1+λ1(τ)δ1−λ2(τ)κ11+λ1(τ)δ1+EIζrr(0,τ)−Ih˙μ(τ)) | (41) |
Note that, for ∀n∈R and ai>0,i=1,2, one has −ntanh(n/naiai)⩽0. Then, based on (35), one has
z2(τ)ℏ1(τ)⩽0 | (42) |
In addition, due to λ1(τ)⩽1 and λ2(τ)⩽1, (41) and (42) hold
z2(τ)ℏ1(τ)1+λ1(τ)δ1⩽z2(τ)ℏ1(τ)1+δ1 | (43) |
|λ2(τ)κ11+λ1(τ)δ1|⩽κ11−δ1 | (44) |
Thus, according to (35), one gets
z2(τ)ℏ1(τ)1+λ1(τ)δ1⩽−z2(τ)ℏ1(τ)tanh(z2(τ)ℏ1(τ)a1)−z2(τ)ˉκ1tanh(z2(τ)ˉκ1a1) | (45) |
Both adding and subtracting |z2(τ)ˉκ1| and z2(τ)ℏ1(τ), one obtains
z2(τ)ℏ1(τ)1+λ1(τ)δ1⩽|z2(τ)ℏ1(τ)|−z2(τ)ℏ1(τ)tanh(z2(τ)ℏ1(τ)a1)+|z2(τ)ˉκ1|−z2(τ)ˉκ1tanh(z2(τ)ˉκ1a1)−|z2(τ)ˉκ1|+z2(τ)ℏ1(τ) | (46) |
Consider the property of tanh(·) that
0⩽|D|−Dtanh(Dγ)⩽0.2785γ | (47) |
with D∈R and γ>0. Therefore, one has
z2(τ)ℏ1(τ)1+λ1(τ)δ1⩽0.557a1−|z2(τ)ˉκ1|+z2(τ)ℏ1(τ) | (48) |
Substituting (48) into (47), one has
˙V2(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)+0.557a1−|z2(τ)ˉκ1|−z2(τ)λ2(τ)κ11+λ1(τ)δ1 | (49) |
Then, based on (44), this leads to
˙V2(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)+0.557a1−|z2(τ)ˉκ1|+|z2(τ)κ11−δ1| | (50) |
Further consider ˉκ1>κ1/κ1(1−δ1)(1−δ1), this leads to
−|z2(τ)ˉκ1|+|z2(τ)κ11−δ1|⩽0 | (51) |
Then, ˙V2(τ) is further expressed as
˙V2(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)+0.557a1 | (52) |
and ˙V3(τ) can be rewritten as
˙V3(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22+0.557a1−k3z23(τ)k2c2−z23(τ)+z3(τ)z4(τ)k2c2−z23(τ) | (53) |
In the same way, consider (40), and ˙V4(τ) can be rewritten as
˙V4(τ)=˙V3(τ)+mz4(τ)˙z4(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)+z3(τ)z4(τ)k2c2−z23(τ) +0.557a1+z4(τ)(ℏ2(τ)1+η1(τ)δ2−ρX(Φ(τ)+EIζrr(0,τ))+EIζrrr(X,τ)−m˙ϑ(τ)−η2(τ)κ21+η1(τ)δ2) | (54) |
Note that when a2>0 for ∀n∈R, −ntanh(n/na2a2)⩽0 is always true. Thus, from (36), it can be sure that
z4(τ)ℏ2(τ)⩽0 | (55) |
Since |ηi|⩽1,i=1,2, it can be seen that
z4(τ)ℏ2(τ)1+η1(τ)δ2⩽z4(τ)ℏ2(τ)1+δ2 | (56) |
|η2(τ)κ21+η1(τ)δ2|⩽κ21−δ2 | (57) |
Further, according to (36), one has
z4(τ)ℏ2(τ)1+η1(τ)δ2⩽−z4(τ)ℏ2(τ)tanh(z4(τ)ℏ2(τ)a2)−z4(τ)ˉκ2tanh(z4(τ)ˉκ2a2) | (58) |
Then, both adding and subtracting |z4(τ)ˉκ2| and z4(τ)ℏ2(τ) on the right side of (58), it holds that
z4(τ)ℏ2(τ)1+η1(τ)δ2⩽|z4(τ)ℏ2(τ)|−z4(τ)ℏ2(τ)tanh(z4(τ)ℏ2(τ)a2)+|z4(τ)ˉκ2|−z4(τ)ˉκ2tanh(z4(τ)ˉκ2a2)−|z4(τ)ˉκ2|+z4(τ)ℏ2(τ) | (59) |
Because of the property in (47), it is further known that
z4(τ)ℏ2(τ)1+η1(τ)δ2⩽0.557a2−|z4(τ)ˉκ2|+z4(τ)ℏ2(τ) | (60) |
Substituting (60) into (54), one has
˙V4(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)−k4z24(τ)+0.557a1+0.557a2−|z4(τ)ˉκ2|−z4(τ)η2(τ)κ21+η1(τ)δ2 | (61) |
Then, based on (57), this leads to
˙V4(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)−k4z24(τ)+0.557a1+0.557a2−|z4(τ)ˉκ2|+|z4(τ)κ21−δ2| | (62) |
Further consider ˉκ2>κ2/κ2(1−δ2)(1−δ2), and one has
−|z4(τ)ˉκ2|+|z4(τ)κ21−δ2|⩽0 | (63) |
Then, ˙V4(τ) is further expressed as
˙V4(τ)⩽−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)−k4z24(τ)+0.557a1+0.557a2 | (64) |
We can get the theorem result as below according to the above analysis process.
Theorem 1: Consider the flexible manipulator system as shown in (5)–(8), under Assumption 1, and design the event-triggered controllers in (35) and (36). Then the presented approach guarantees that 1) the vibration of the manipulator is effectively restrained and stabilized, 2) all the signals displaying in the closed-loop system are bounded, 3) the joint angle ϱ(τ) and the boundary vibration diversion ζ(X,τ) fulfill the constraint conditions ϱ(τ)<kd1 and ζ(X,τ)<kd2, respectively, and 4) the system can effectively avoid the occurrence of the Zeno behavior.
Proof:
The Barrier Lyapunov function is considered as
V(τ)=V4(τ)=12log(k2c1k2c1−z21(τ))+12Ihz22(τ)+12log(k2c2k2c2−z23(τ))+12mz24(τ) | (65) |
On the basis of the above analysis, one obtains
˙V(τ)≤−k1z21(τ)k2c1−z21(τ)−k2z22(τ)−k3z23(τ)k2c2−z23(τ)−k4z24(τ)+0.557a1+0.557a2≤−cV(τ)+d | (66) |
where c=min(2k1,2k2/2k2IhIh,2k3,2k4/2k4mm), and d=0.557a1+0.557a2.
Multiplying ecτ on both sides of (66), and integrating (66) over [0,τ], one can obtained that
V(τ)≤(V(0)−dc)e−cτ+dc | (67) |
According to (66) and (67), the boundedness of errors zi,i=1,2,3,4 is known. Meanwhile, since ϱd, μ, ζd(X,τ) and ϑ are bounded, according to (11)–(14), one gets that ϱ(τ), ˙ϱ(τ), ζ(X,τ) and ˙ζ(X,τ) are also bounded. Similarly, according to (35) and (36), the boundedness of ℏ1(τ) and ℏ2(τ) are obviously proved. Considering e1=ℏ(τ)−νd and e3=ρ(X,τ)−ρd(X,τ), we can get |ϱ(τ)|=|z1+ϱd|⩽|z1|+|ϱd| and |ζ(X,τ)|=|z3+ζd(X,τ)| ⩽|z3|+|ζd(X,τ)|. According to (9) and (10), it holds that |ϱ(τ)|⩽kc1+|ϱd|<kd1 and |ζ(X,τ)|⩽kc2+|ζd(X,τ)|<kd2, which means that system states are within their constraint bounds.
Recall the definition of ei(τ),i=1,2, i.e., e1(τ)=ℏ1(τ)−Φ0(τ), e2(τ)=ℏ2(τ)−U0(τ), where Φ0(τ)=ℏ1(τk) for ∀τ∈[τk,τk+1), and O0(τ)=ℏ2(sk) for ∀τ∈[τs,τs+1). Then, one has
˙e1(τ)=˙ℏ1(τ)−˙ℏ1(τk) |
˙e2(τ)=˙ℏ2(τ)−˙ℏ2(τs) |
Here ˙ℏ1(τk) and ˙ℏ2(τs) can be regarded as constants at the time interval [τk,τk+1) and [τs,τs+1), which means that ˙ℏ1(τk)=0 and ˙ℏ2(τs)=0. Then, we have
ddτ|e1|=sign(e1)˙e1⩽|˙ℏ1| | (68) |
ddτ|e2|=sign(e2)˙e2⩽|˙ℏ2| | (69) |
According to the definition of ℏ1(τ) and ℏ2(τ) in (35) and (36), we know that ˙ℏ1(τ) and ˙ℏ2(τ) are the functions of zi,i=1,2,3,4. From the result before, all the system signals are bounded, so ˙ℏ1(τ) and ˙ℏ2(τ) are bounded. Then, we assume that ˙ℏ1(τ)⩽ˉℏ1, ˙ℏ2(τ)⩽ˉℏ2 with ˉℏ1 and ˉℏ2 being constants. In addition, e1(τk)=0, e2(τs)=0 and limτ→τk+1e1(τk+1)=δ1|Φ0(τ)|+κ1, limτ→τs+1e2(τs+1)=δ2|O0(τ)|+κ2. By integrating (68) and (69) on their both sides, one gets τk+1−τk⩾T=(δ1|Φ0(τ)|+κ1)/(δ1|Φ0(τ)|+κ1)ˉℏ1ˉℏ1 and τs+1−τs⩾T=(δ2|O0(τ)|+κ2)/(δ2|O0(τ)|+κ2)ˉℏ2ˉℏ2. Thus, the Zeno behavior can be avoided.
Theorem 1 is demonstrated integrally.
In order to verify the effectiveness of the control strategy designed in this paper, a system simulation based on (5)–(8) is considered. The system parameters are selected as follows: EI=10Nm2, X=1m, ℓ=0.5kgm−1, m=2kg, and Ih=1kgm2. The other related parameters are chosen as ϱd=0.03, ζd=0, k1=10, k2=10, k3=10, k4=10, κ1=1, κ2=1, δ1=0.15, δ2=0.15, ˉκ1=2, ˉκ2=2, a1=0.4, a2=0.4. In order to further compare with other control methods, the proportional differential (PD) control ℏ1(τ)=−2.5ζrr(0,τ)−1.5ϱ(τ)−1.5˙ϱ(τ), ℏ2(τ)=−2.5ζrrr(X,τ)−1.5ζ(X,τ)−1.5˙ζ(X,τ) is proposed in this paper. The simulation results are given as Figures 1–9.
Figure 1 shows the system vibration deflection without any control. It is obvious that the manipulator moves freely with large amplitude. Figure 2 shows the system vibration deflection under the action of the event trigger controller. It can be seen from the figure that the amplitude of the manipulator becomes gentle within a short time, forming an obvious contrast with Figure 1. Figure 3 is the trajectory of boundary vibration deflection ζ(X,τ) of the system with (curve) or without (dotted line) control. It can be seen that when the system does not apply any control, ζ(X,τ) changes periodically, and the amplitude is changed greatly, which will damage the flexible manipulator system and reduce the working accuracy. In the case of control, the boundary vibration deflection ζ(X,τ) gradually tends to be stable, which greatly reduces the system loss. Figures 4 and 5 respectively indicate the trajectories of state junction angle ϱ(τ) and state boundary deflection ζ(X,τ). From the figures we can know, that ϱ(τ) and ζ(X,τ) remain within the constraint boundaries. In the meantime, the tracking performances of ϱ(τ) and ζ(X,τ) are good. It can be clearly seen that based on the adopted control solution; they are adjusted to the expected value. Figure 6 is the trajectories of event trigger controllers ℏ1(τ) and ℏ2(τ). Figure 7 is the time interval for triggering events, which indicates that the Zeno phenomenon is successfully avoided. The displacement and the boundary output changes under state constraints with PD control are shown in Figures 8 and 9. Compared with the simulation results in the previous case, it is obvious that the control strategy proposed in this paper is smoother and more effective than PD control. Obviously, they are bounded. From all the analysis so far, we conclude the following result. The control objectives of this paper can be realized under the action of the proposed control strategy.
A vibration suppression control algorithm with event-triggered mechanism is proposed for manipulator system described by PDEs with state constraints. State constraints and event-triggered problems are considered simultaneously in the course controller design. Based on the BLF and relative threshold strategy, the method we developed reduces the cost of information transmission and guarantees that the system signals of the under consideration are all bounded. The elastic vibration of flexible manipulator is well suppressed, and the constrained states do not break the constraint bounds. In addition, the Zeno phenomenon is successfully avoided. Simulation results show the validity of the proposed algorithm. At a future date, the proposed scheme can be further studied and spread in other partial differential practical systems with DoS attacks like in [47].
The paper is supported by the National Key Research and Development Program of China (2018YFF0300605).
The authors declare that they have no conflict of interest.
[1] |
A. Cruz-Jentoft, G. Bahat, J. Bauer, Y. Boirie, O. Bruyère, T. Cederholm, et al., Sarcopenia: Revised European consensus on definition and diagnosis, Age Ag., 48 (2019), 16–31. https://doi.org/10.1093/ageing/afy169 doi: 10.1093/ageing/afy169
![]() |
[2] |
R. A. Fielding, B. Vellas, W. J. Evans, S. Bhasin, J. E. Morley, A. B. Newman, et al., Sarcopenia: An undiagnosed condition in older adults. Current consensus Definition: Prevalence, etiology, and consequences. International Working Group on Sarcopenia, J. Am. Med. Dir. Assoc., 12 (2011), 249–256. https://doi.org/10.1016/j.jamda.2011.01.003 doi: 10.1016/j.jamda.2011.01.003
![]() |
[3] |
I. Janssen, Evolution of sarcopenia research, Appl. Physiol. Nutr. Metab., 35 (2010), 707–712. https://doi.org/10.1139/h10-067 doi: 10.1139/h10-067
![]() |
[4] |
A. Dawson, E. Dennison, Measuring the musculoskeletal aging phenotype, Maturitas, 93 (2016), 13–17. https://doi.org/10.1016/j.maturitas.2016.04.014 doi: 10.1016/j.maturitas.2016.04.014
![]() |
[5] |
C. Beaudart, R. Rizzoli, O. Bruyère, J. Y. Reginster, E. Biver, Sarcopenia: Burden and challenges for public health, Arch. Public Health, 72 (2014). https://doi.org/10.1186/2049-3258-72-45 doi: 10.1186/2049-3258-72-45
![]() |
[6] |
C. Beaudart, E. McCloskey, O. Bruyère, M. Cesari, Y. Rolland, R. Rizzoli, et al., Sarcopenia in daily practice: Assessment and management, BMC Geriatr., 16 (2016). https://doi.org/10.1186/s12877-016-0349-4 doi: 10.1186/s12877-016-0349-4
![]() |
[7] |
M. Cho, S. Lee, S. Song, A review of Sarcopenia Pathophysiology, diagnosis, treatment and future direction, J. Korean Med. Sci, 37 (2022). https://doi.org/10.3346/jkms.2022.37.e146 doi: 10.3346/jkms.2022.37.e146
![]() |
[8] |
D. Albano, C. Messina, J. Vitale, L. M. Sconfienza, Imaging of sarcopenia: Old evidence and new insights, Eur. Radiol., 30 (2020), 2199–2208. https://doi.org/10.1007/s00330-019-06573-2 doi: 10.1007/s00330-019-06573-2
![]() |
[9] |
G. Guglielmi, F. Ponti, M. Agostini, M. Amadori, G. Battista, A. Bazzocchi, The role of DXA in sarcopenia, Ag. Clin. Exp. Res., 28 (2016), 1047–1060. https://doi.org/10.1007/s40520-016-0589-3 doi: 10.1007/s40520-016-0589-3
![]() |
[10] |
P. Tandon, M. Mourtzakis, G. Low, L. Zenith, M. Ney, M. Carbonneau, et al., Comparing the variability between measurements for sarcopenia using magnetic resonance imaging and computed tomography imaging, Am. J. Transplant., 16 (2016), 2766–2767. https://doi.org/10.1111/ajt.13832 doi: 10.1111/ajt.13832
![]() |
[11] |
K. Feng, J. Ji, Q. Ni, A novel gear fatigue monitoring indicator and its application to remaining useful life prediction for spur gear in intelligent manufacturing systems, Int. J. Fatigue, 168 (2023), 107459. https://doi.org/10.1016/j.ijfatigue.2022.107459 doi: 10.1016/j.ijfatigue.2022.107459
![]() |
[12] |
K. Feng, J. Ji, K. Wang, D. Wei, C Zhou, Q Ni, A novel order spectrum-based Vold-Kalman filter bandwidth selection scheme for fault diagnosis of gearbox in offshore wind turbines, Ocean Eng., 266 (2022), 112920. https://doi.org/10.1016/j.oceaneng.2022.112920 doi: 10.1016/j.oceaneng.2022.112920
![]() |
[13] |
K. Feng, J. Ji, Q. Ni, Y Li, W Mao, L Liu, A novel vibration-based prognostic scheme for gear health management in surface wear progression of the intelligent manufacturing system, Wear, 522 (2023), 204697. https://doi.org/10.1016/j.wear.2023.204697 doi: 10.1016/j.wear.2023.204697
![]() |
[14] |
S. Zhao, J. Liu, Z. Gong, Y. S. Lei, X. OuYang, C. C. Chan, et al., Wearable physiological monitoring system based on electrocardiography and electromyography for upper limb rehabilitation training, Sensors, 20 (2020), 4861. https://doi.org/10.3390/s20174861 doi: 10.3390/s20174861
![]() |
[15] |
S. Prabu, K. Srinivas, B. K. Rani, R. Sujat, B. D. Parameshachari, Prediction of muscular paralysis disease based on hybrid feature extraction with machine learning technique for COVID-19 and post-COVID-19 patients, Pers. Ubiquit. Comput., 27 (2023), 831–844. https://doi.org/10.1007/s00779-021-01531-6 doi: 10.1007/s00779-021-01531-6
![]() |
[16] |
I. Campanini, C. Disselhorst-Klug, W. Z. Rymer, R. Merletti, Surface EMG in clinical assessment and neurorehabilitation: Barriers limiting its use, Front. Neurol., 11 (2020), 934. https://doi.org/10.3389/fneur.2020.00934 doi: 10.3389/fneur.2020.00934
![]() |
[17] |
M. Al-Ayyad, H. A. Owida, R. De Fazio, B. Al-Naami, P. Visconti, Electromyography monitoring systems in rehabilitation: A review of clinical applications, wearable devices and signal acquisition methodologies, Electronics, 12 (2023), 1520. https://doi.org/10.3390/electronics12071520 doi: 10.3390/electronics12071520
![]() |
[18] |
R. Habenicht, G. Ebenbichler, P. Bonato, S. Ziegelbecker, L. Unterlerchner, P. Mair, et al., Age-specific differences in the time-frequency representation of surface electromyographic data recorded during a submaximal cyclic back extension exercise: a promising biomarker to detect early signs of sarcopenia, J. NeuroEng. Rehabil., 8 (2020). https://doi.org/10.1186/s12984-020-0645-2 doi: 10.1186/s12984-020-0645-2
![]() |
[19] |
A. Leone, G. Rescio, A. Manni, P. Siciliano, A. Caroppo, Comparative analysis of supervised classifiers for the evaluation of sarcopenia using a sEMG-based platform, Sensors, 22 (2022), 2721. https://doi.org/10.3390/s22072721 doi: 10.3390/s22072721
![]() |
[20] |
J. M. Jasiewicz, J. H. Allum, J. W. Middleton, A. Barriskill, P. Condie, B. Purcell, et al., Gait event detection using linear accelerometers or angular velocity transducers in able-bodied and spinal-cord injured individuals, Gait Posture, 24 (2006), 502–509. https://doi.org/10.1016/j.gaitpost.2005.12.017 doi: 10.1016/j.gaitpost.2005.12.017
![]() |
[21] | M. Halaki, G. Karen, Normalization of EMG signals: To normalize or not to normalize and what to normalize to?, in computational intelligence in electromyography analysis–a perspective on current applications and future challenges (ed Ganesh R. Naik), InTech, (2012). https://doi.org/10.5772/49957 |
[22] | E. H. Norden, S. Zheng, L. R. Steven, M. C. Wu, H. H. Shih, Q. N. Zheng, et al., The empirical mode decomposition and the hilbert spectrum for nonlinear and non-stationary time series analysis, in Proceedings: Mathematical, Physical and Engineering Sciences, 454 (1998), 903–995. https://doi.org/10.1098/rspa.1998.0193 |
[23] |
J. Too, A. R. Abdullah, N. M. Saad, Classification of Hand Movements based on Discrete Wavelet Transform and Enhanced Feature Extraction, Int. J. Adv. Comput. Sci. Appl., 10 (2019). https://doi.org/10.14569/ijacsa.2019.0100612 doi: 10.14569/ijacsa.2019.0100612
![]() |
[24] |
S. A. Christopher, I. MdRasedul, A comprehensive study on EMG feature extraction and classifiers, J. Biomed. Eng. Biosci., 1 (2018). https://doi.org/10.32474/oajbeb.2018.01.000104 doi: 10.32474/oajbeb.2018.01.000104
![]() |
[25] |
P. Qin, X. Shi, Evaluation of feature extraction and classification for lower limb motion based on SEMG signal, Entropy, 22 (2020), 852. https://doi.org/10.3390/e22080852 doi: 10.3390/e22080852
![]() |
[26] |
C. Ding, H. Peng, Minimum redundancy feature selection from microarray gene expression data, J. Bioinform. Comput. Biol., 3 (2015), 185–205. https://doi.org/10.1142/s0219720005001004 doi: 10.1142/s0219720005001004
![]() |
[27] |
T. M. Cover, P. D. Hart, Nearest neighbor pattern classification, IEEE Trans. Inf. Theory, 13 (1967), 21–27. https://doi.org/10.1109/tit.1967.1053964 doi: 10.1109/tit.1967.1053964
![]() |
[28] |
M. Hall, A decision Tree-Based attribute weighting filter for naive bayes, In Springer eBooks, 2007, 59–70. https://doi.org/10.1007/978-1-84628-663-6_5 doi: 10.1007/978-1-84628-663-6_5
![]() |
[29] |
T. K. Ho, Random decision forests, Proceedings of 3rd International Conference on Document Analysis and Recognition, 1 (1995), 278-282. doi: 10.1109/ICDAR.1995.598994 doi: 10.1109/ICDAR.1995.598994
![]() |
[30] | T. Chen, C. Guestrin, XGBoost: A scalable tree boosting system, in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016,785–794. https://doi.org/10.1145/2939672.2939785 |
[31] |
F. Murtagh, Multilayer perceptrons for classification and regression, Neurocomputing, 2 (1991), 183–197. https://doi.org/10.1016/0925-2312(91)90023-5 doi: 10.1016/0925-2312(91)90023-5
![]() |
1. | Yunxia Wei, Yuanfei Zhang, Bin Hang, Construction and management of smart campus: Anti-disturbance control of flexible manipulator based on PDE modeling, 2023, 20, 1551-0018, 14327, 10.3934/mbe.2023641 | |
2. | Zhiyong Wei, Qingchun Zheng, Peihao Zhu, Wenpeng Ma, Jieyong Deng, A New Fuzzy Backstepping Control Based on RBF Neural Network for Vibration Suppression of Flexible Manipulator, 2024, 14, 2076-3417, 6054, 10.3390/app14146054 | |
3. | Yang Zhang, Liang Zhao, Jyotindra Narayan, Adaptive control and state error prediction of flexible manipulators using radial basis function neural network and dynamic surface control method, 2025, 20, 1932-6203, e0318601, 10.1371/journal.pone.0318601 |