Research On Finite State Feedback And Time-delay In Structural Active Control Based On PSO-BP Algorithm

Active control algorithm has made great progress,but at present there are still some feasibility problems from the practical point of view,in which this research focuses on two problems.Firstly,active control system takes a certain amount of time for real-time data processing calculation and exerting of control force,and the resulting time-delay problem will make phase of control force exerted by actuator be different,which will not only reduce the performance of the control system,but also lead to the disorder of the power system.Secondly,active control algorithm generally uses full state feedback,but the degrees of freedom of civil engineering structures generally have a large number,so it is uneconomical and unrealistic to set sensors on all degrees of freedom of the structure.Active control shows great advantages in the control algorithm,but due to the above mentioned reasons the research of active control technology is more concentrated in the field of theory and its engineering applications are very few,so it is necessary to investigate the method of solving these problems.This research takes a high-rise building and a footbridge for entering platform as engineering background,and investigates the use of two kinds of intelligent algorithms which is BP algorithm and Particle Swarm Optimization(PSO)algorithm to solve this two problems.Specifically,the research focuses on the following aspects:(1)BP algorithm is currently the most widely used neural network,but the problem that the randomness of initial weight and threshold of BP neural network is easy to produce local minimum and slow convergence speed,and the research investigates the use of Particle Swarm Optimization(PSO)algorithm to optimize the initial weight and threshold of BP neural network,then the PSO-BP neural network which has better network performance is established to solve the problems behind.(2)aiming at the demand for structural full state feedback problem in active control algorithm,the PSO-BP network is used to establish the identifier of the acceleration response of the full degree of freedom based on the acceleration response of the known part of the structure,then transforms the structural acceleration response predicted by PSO-BP network into velocity and displacement response by Time Domain Integral method and Wavelet Transform to Remove Baseline Drift method.The above is the design method of the PSO-BP neural network state observer based on partial acceleration response feedback.(3)aiming at the time-delay problem in active control algorithm,the PSO-BP network is used to establish the predicted model that predicts the displacement and velocity response of the structure in the future moment according to the known displacement and velocity response,and the above is the design identifier based on the time-delay problem to provide the basis for active control decision.(4)In this article the proposed state observer and time-delay identifier based on PSO-BP network are applied to the classical optimal control of AMD system of a high-rise building structure and a footbridge for entering platform.The robustness of the observer is verified by investigating the control effects under various excitations,and comparison of predicted error is investigated between PSO-BP neural network state observer and the other two observers including modal filter and system state observer.The results show that the proposed observer has higher prediction accuracy and is more suitable for engineering application,and the time-delay identifier has a good effect in the response prediction of the delay time segment in a certain range.In summary,this research utilizes Particle Swarm Optimization algorithm to optimize BP neural network,using this method to solve the time-delay problem and the demand for structural full state feedback problem,which have certain significance for the engineering applications of active control technology.