Research On Nonlinear Compensation For Sensors And Actuators Using Neural Networks

A method for sensors and actuators nonlinear compensation, which is based on neural networks dead-zone inverse approximation, is presented. Theory analysis and experiment results show that the method can solve the negative effects caused by nonlinear sectors in control systems, and convert the system into a generalized linear system. This article also provides a method of dead-zone compensation of actuator based on second-order neural net, which realized the dynamic compensation for unknown nonlinear dead-zone. This method doesn't require any hypothesis of nonlinear and constraint of dead-zone, and realized the compensation strategy composed of estimator and compensator using neural networks, which supplies a new solution for nonlinear compensation. Moreover, BP and RBF neural networks are used to research the adaptive inverse control for nonlinear systems.