A New Active Vibration Isolation Control Method Based on CMAC Neural Network

摘要

In terms of the nonlinearity in active vibration isolation system and the deficiencies of the generally used control methods, a new adaptive neuro-control method was proposed. In the new control structure, a CMAC (Cerebellar Model Articulation Controller) network was adopted as controller, and an improved simultaneous perturbation algorithm, which was proven a stochastic gradient approximation method, was proposed to update controller weights without modeling the secondary path. The performance of the control method was verified by numerical simulation, and simulation results showed that the proposed method has preferable isolation performance for system with nonlinearity when compared with Filtered LMS method.