A Hydraulic Transmission System Based on Double Fuzzy Recurrent Neural Network Sliding Mode Control

Aiming at the nonlinear and time-delay problems caused by flexible load in hydraulic servo drive control system, a double fuzzy recurrent neural network sliding mode controller (DFRNNSMC) is proposed in this paper. The fuzzy recurrent neural network (FRNN) is an improved neural network which combines fuzzy system and recursive structure. Because it adds a recursive layer, it can realize dynamic mapping. Therefore, compared with RBF neural network, it can improve the accuracy of approaching the nonlinear sliding mode equivalent control term, and improve the robustness of the system. For the problem of time delay, a fuzzy controller is designed to adaptively adjust the flow gain of the valve core opening to reduce the response time. The method can improve the tracking ability of the hydraulic transmission system to a certain extent. The stability of the system is verified by Lyapunov theorem and the digital simulation is carried out by using Simulink. By establishing the simulation model of the hydraulic transmission system, it is verified that the proposed DFRNNSMC has faster convergence speed and better robustness. The dynamic tracking experiment is carried out on the hydraulic test bench, and compared with the traditional control method. Through the comparison of the dynamic tracking accuracy, the feasibility of this method and the superiority of its control performance are verified.