不确定非线性系统的智能自修复控制

The intelligent self-repairing control scheme for nonlinear system in presence of structure fault is studied. Some new intelligent fault feature extraction and pattern recognition techniques are proposed, such as improved discrete Fourier transform algorithm and fast fuzzy clustering algorithm. A new idea of composite fault detection based on wavelet neural network and self-organizing fuzzy CMAC neural network is presented, and new fault identification method based on hierarchic residual sorting neural network is presented. Some new intelligent fault diagnosis techniques for nonlinear system are proposed, which include radial basis function neural network observer, support vector machine, wavelet singularity detection principle, multiple step prediction neural network, neural network observer and multiple neural network. Neural network model following adaptive control, fuzzy multiple model matching pseudo-inverse reconfiguration control and decentralized neural network inverse system reconfigurable control techniques are proposed to apply to intelligent self-repairing control. The main contribution of this researching is that some results about dynamical performance, robustness and tracking performance of self-repairing system are obtained.

研究不确定非线性系统结构故障自修复方法：提出预测及残差神经网络故障诊断及小波神经网络组合故障模式识别新方法，可对多故障实时准确定位；提出广义模型跟随神经网络直接自修复控制及模糊多模型匹配伪逆自修复新方法，对不确定非线性系统实施重构，可确保系统鲁棒稳定性与跟踪性；建立结构故障可识别性与可修复性等框理论框架并应用于工程实践。