基于小波分解的非线性系统多模型自适应控制

提出了一种改进的基于小波分解的非线性系统辨识算法,利用小波函数的逼近能力在线辨识被控对象的非线性项.针对基于小波分解的辨识算法缺乏预测能力,提出了根据线性鲁棒自适应控制器提供的当前控制信息预测未来的非线性项值新方法,并结合多模型方法,根据所定义的切换指标自动切换到当前最优控制器.仿真结果表明,改进的基于小波分解的辨识算法能够有效逼近非线性系统,基于小波分解的非线性系统多模型自适应控制方法改善了系统性能,随着系统运行跟踪误差明显减小,说明了该方法的有效性和可行性.

Multiple Models Adaptive Control of Nonlinear System Based on Wavelet Decomposition

A nonlinear system identification algorithm based on improved wavelet decomposition is proposed which uses the approximation ability of wavelet function to identify nonlinear terms. Since the identification algorithm based on wavelet decomposition lacks of prediction ability, a new method is proposed which relies on the current control information provided by linear robust adaptive controller to predict the nonlinear term, and uses multiple models to build two adaptive models and switch to the optimum controller according to switching function. The results of two simulation examples show that the identified system based on improved identification algorithm is close to the real nonlinear system, and the control approach can improve the system performance greatly. The obviously reduced error illustrates the effectiveness of this method.