Active Disturbances Rejection Control Based On PSOBP Neural Network Combined Inverse Of Two-Motor Variable Frequency Speed-Regulation System

Multi-motor variable frequency speed-regulation system has been widely used in the field of modern industrial automation control. The problems of control accuracy and running cost have also followed. How to realize the high precision control and no-pressure-sensor operation of such a complex multivariable system under disturbed environment has become the focus of the research on power transmission control and identification. Two-motor variable frequency speed-regulation system is chosen as the research object, a new control strategy that BP neural network based on particle swarm optimization(PSOBPNN) combined inverse active disturbance rejection control(ADRC) has been proposed in this paper. The research on tracking and decoupling effect, system robustness and identification effect of tension under disturbance is studied.Firstly, the mathematical model of two-motor variable frequency speed-regulation system under vector control is deduced. Then the right and left reversibility of the system model are analyzed in order to obtain the system expression of the corresponding right and left inverse of the system. PSOBP network is used to approximate the analytical expressions of the right and left inverse systems to build the PSOBP network right inverse system and PSOBP neural network based on the tracking differentiator(TD-PSOBPNN) left inverse system. These two systems are contacted before and after the original two-motor system. As follow, the PSOBPNN combined inverse system has been constructed to realize the tracking and decoupling of speed and tension, real-time measurement of tension under disturbance.Secondly, the basic principle and structure of ADRC are briefly introduced, then the ADRC controllers are designed separately for the equivalent pseudo linear subsystems in order to restrain the influence of load mutation in the control process of two-motor system.Thirdly, the simulation modeling of two-motor system platform is settled on Matlab/Simulink to build the model of two-motor variable frequency speed-regulation system.The comparative analysis is given between TD-PSOBPNN left inverse identification and traditional BPNN left inverse identification by simulation; and then the simulation verification of ADRC-PSOBPNN combined inverse control strategy is accomplished next.Finally, based on S7-300 PLC master control experiment platform, the comparative experimental verification partly between the new control strategy proposed in this paper and PID control is given.The simulation and experimental results show that, the convergence accuracy and speed of neural network training are enhanced significantly by the new control strategy, and the tension intwo-motor system can be real-time identified precisely under the disturbance conditions.Meanwhile, the fast tracking and decoupling control of variables in two-motor system can be implemented efficiently and the ability of anti-interference is improved obviously.