Study Of Fixed-time Neurodynamic Time-Varying Optimization Algorithms And Their Application To Battery Energy Storage System

Time-varying optimization problems,as one of the hot problems in optimization theory,are widely used in smart grids,communication systems,biomedical and robotics applications.In recent years,this problem has received more and more attention from scholars,and a lot of algorithms have been proposed for solving it.However,these algorithms have more room for improvement in terms of accuracy,convergence speed,and anti-interference capability.In view of this,this paper designs a class of fixed-time neurodynamic time-varying algorithms and combines them with the theory related to stability analysis to guarantee that the algorithms converge to the optimal solution of the time-varying optimization problem in fixed time.Finally,the algorithm is also applied to the problem of optimal power acquisition for each battery pack in battery energy storage system,demonstrating the practical value of the algorithm.Finally,the algorithms are also applied to the state-of-charges balance problem in battery energy storage system to show their practical value.The main research of this paper is as follows:(1)For the centralized time-varying convex optimization problem,a fixed-time timevarying optimization algorithm is designed by modifying the prediction-correction algorithm,which has faster convergence speed.In addition,considering the anti-interference capability of the algorithm,an integration-enhanced fixed-time time-varying optimization algorithm is designed based on the integration-enhanced zeroed neural network model and the sliding mode control technique.The algorithm inherits the anti-interference ability of the integration-enhanced zeroed neural network and the same fixed-time convergence property.In addition,the algorithm is systematically analyzed by convex optimization theory and stability theory in this paper to ensure the reliability of the algorithm.Finally,the effectiveness and superiority of the algorithm are further verified by simulation comparison experiments.(2)For the distributed time-varying convex optimization problem,the centralized algorithm proposed in this paper is modified by combining the cooperative control scheme for multi-intelligent body systems.Two distributed fixed-time time-varying optimization algorithms,the distributed optimization-cooperative algorithm and the distributed cooperative-optimization algorithm,are designed respectively.Both distributed algorithms inherit the strong anti-interference ability and fixed-time convergence property of the centralized algorithm.However,they have more advantages than the centralized algorithms in terms of computational efficiency,privacy,and scalability.At the same time,both distributed algorithms have their own advantages and disadvantages.The distributed optimization-cooperative algorithm requires less communication capability among the intelligences,but the applicability of the algorithm is narrow.The distributed cooperativeoptimization algorithm increases the communication burden among the agents to a certain extent,but breaks some restrictions of the optimization-collaborative algorithm on the second-order derivatives of the objective function,which greatly expands the applicability of the algorithm.The effectiveness and superiority of the two distributed algorithms proposed in this paper are also analyzed at the levels of theoretical proof and simulation experiments,respectively.(3)For the optimal power acquisition problem of battery pack in battery energy storage system,this paper fully considers the time-varying factor of optimal power and the safety of the system.A time-varying optimization problem with constraints is modeled.For this problem,the penalty function method is used to deal with the constraints,and the distributed algorithm designed in this paper is combined to solve the problem.Compared with the existing research work,the algorithm in this paper has more advantages in terms of applicability,convergence speed and anti-interference capability.Finally,the effectiveness and superiority of this algorithm are further verified by two sets of simulation experiments.