基于归一化RBFNN的执行器非线性模型及其特性补偿

A Nonlinear Model of Actuator Based on Normalized RBF Neural Networks and Its Performance Compensation

执行器的动态非线性特性是影响控制系统控制效果的一个重要因素,采用归一化径向基函数神经网络NRBFNN,通过竞争学习算法RPCL确定RBF中心,用递归最小二乘法估计网络的输出权值,建立了执行器动态非线性特性模型及其逆模型,通过信号补偿方式来改善执行器的动态特性,仿真结果表明了该方法的有效性.

The dynamic nonlinearity of actuator is an important element affecting the performance of a control system. With Normalized Radial Basis Functions Neural Network （NRBFNN）, competitive learning algorithm RPCL for RBF centers, and recursive least squares algorithm for net output weights, the model and inverse model of actuator nonlinearity were established in this paper. The dynamic performance of the actuator has been improved by using signal compensation with its inverse model, Simulation results have proved the validity of this method.