Vibration Control Of Active Mass Damper Based On Adaptive Dynamic Programming

In order to cope with the huge damage caused by earthquakes and typhoons to buildings,structural vibration control technology has always been a hot topic in the field of structural engineering.As an active vibration control technology,the active mass damper(AMD)outputs active control force according to a certain control strategy through an actuator installed at the top of the structure to reduce the vibration of the structure in earthquakes and strong winds.The core of this technology is how to design a control strategy to output optimal control.Due to the uncertainty of structural parameters of complex civil engineering and the randomness of seismic action,traditional control algorithms based on accurate models are difficult to play a good role in damping control.As a data-driven control algorithm,adaptive dynamic programming algorithm(ADP)uses the function approximation structure to fit the Hamilton-Jacobi-Bellman equation(HJB)solution,and collects system state variable data for offline iteration or online optimization.Finally,an approximate optimal control strategy is obtained.The algorithm can control without relying on the model parameters of the system.ADP has the ability to learn and optimize,can solve the problem of “dimension disaster”,and has certain advantages in solving the problem of strong nonlinear and complex system control.In this paper,the robust controller based on ADP algorithm is studied systematically and applied to AMD system to control the structure with unknown parameters under random disturbance.The main work includes the following aspects.(1)The Lagrange method is used to establish the AMD system dynamics equation.The system state variables and output variables are selected to establish the system state space expression.The basic principle of ADP algorithm is studied,and a robust controller based on ADP algorithm is designed and applied to AMD system.(2)The AMD system vibration response simulation data is collected in MATLAB/SIMULINK.The system cost function and the optimal controller are approximated by the evaluation neural network and the execution neural network respectively.The offline neural network weights are obtained by offline iteration,and the neural network approximation is performed optimally.The approximate optimal controller is applied to the active control device of the AMD system to simulate and analyze the controlled structure under earthquake.(3)On the small vibration table of Quanser,the AMD system control experiment is carried out based on ADP robust algorithm.The test results and simulation results are compared to verify the effectiveness of the AMD system based on the data-driven ADP algorithm on structural vibration reduction control.