基于聚类方法和神经网络的非线性系统多模型自适应控制

针对具有参数跳变的非线性系统,联合聚类算法和神经网络提出新的多模型自适应控制方法。首先对系统的输入输出数据进行模糊聚类,然后基于递推最小二乘法建立多个固定模型。为提高系统的暂态性能,同时建立两个自适应模型,并在此基础上设计鲁棒自适应控制器。此外,为了补偿系统的非线性部分,建立非线性预测模型,并设计非线性神经网络自适应控制器。所提方法可使控制切换系统具有稳定性保证。最后,通过性能指标对控制器进行平滑切换。仿真结果表明,所提方法能够保证系统具有良好的控制性能。

Multiple model adaptive control of nonlinear systems based on clustering method and neural network

Aiming at the nonlinear system with parameter jump, a multiple model adaptive control method is proposed, which combines clustering algorithm and neural network. Firstly, the fuzzy c-means clustering algorithm is used to adaptively cluster the input and output data. Then the recursive least squares method is used to establish multiple fixed models for all kinds of data. To improve the transient performance of the system, two adaptive models are established and a robust adaptive controller is designed. In addition, in order to compensate the nonlinear part of the system, the nonlinear prediction model is established and the nonlinear neural network adaptive controller is designed. The proposed method can ensure the stability of the control switching system. Finally, the controller is smoothly switched through the performance index. The simulation results show that the proposed method can guarantee the system has a good control performance.