Fuzzy-neural network control for a Magnetorheological elastomer vibration isolation system

Abstract

This paper proposes the design method of a fuzzy-neural network controller (FNNC) for a MRE vibration control system with sinusoidal excitations to improve the vibration attenuation. A semi-active fuzzy controller (FC) is employed to obtain the controlled force based on the feedback signals of the relative displacement and absolute displacement of the isolation structure. Then the dynamic performance experiment of MRE isolator is carried to obtain the output forces generated by the isolator with variable amplitude and frequency displacement excitation and controlled current. Based on the test results, BPNN (back propagation neural network) with good learning capability is employed to emulate the nonlinear relationship between output force of FC, excitation displacement, velocity and controlled current applied to the coil in MRE isolator. The acceleration responses of the system with FNNC are evaluated by physical experiments, and the results show that the proposed FNNC could achieve satisfactory control effect for the sinusoidal excitation, which also outperforms conventional FC. The reason is that the neural network controller can approximate the nonlinear function relationship among the controlled force, excitation displacement, velocity and controlled current. However, the function relationship with conventional FC is only considered as the linear function.