Internal model control of valve‐controlled asymmetric cylinder system based on sectional neural network inverse system

Abstract

In order to improve the control performance of the valve-controlled asymmetric cylinder system with negative overlap, it is proposed that the complex nonlinear system is transformed into a linear system based on the neural network inverse system. The structure of the inverse system is been constructed by the state model of the system, and optimised by adding state parameter feedback. So the pseudo-linear system is reduced to second order. An improved back-propagation neural network based on a genetic algorithm is used to solve the inverse model. Aiming at the problem of negative overlap of servo valves and laminar and turbulent flow, the sectional inverse system switched by reference speed is established to improve the accuracy of the inverse system model. For pseudo-linear systems, an internal model controller is designed to improve the control performance of the composite system. The proportion–integration–differentiation and the internal model control (IMC) of the sectional inverse system are compared with that of AMESIM and Simulink joint platform. The results show that the system controlled by the IMC of the sectional inverse system has better response performance, eliminates the asymmetric characteristics in extending and retracting motion response and is less affected by load fluctuation.