基于平衡学习的CMAC神经网络非线性滑模容错控制  

Fault-tolerant control of nonlinear system based on balanced learning CMAC neural network

作　　者：朱大奇[1] 孔敏[2] 

机构地区：[1]上海海事大学信息工程学院,上海浦东200135 [2]江南大学智能控制研究所,江苏无锡214122

出　　处：《控制理论与应用》2008年第1期81-86,共6页Control Theory & Applications

基　　金：国家自然科学基金资助项目(50775136);国家863计划课题资助项目(2006AA09Z210);上海市教委科研资助项日(06FZ038)

摘　　要：以一改进的信度分配CMAC(cerebellar model articulation controllers)神经网络为在线故障诊断的手段,将变结构滑模控制技术引入容错控制器设计之中,提出一种动态非线性系统主动容错控制方法.在常规CMAC学习算法中,误差被平均地分配给所有被激活的存储单元,不管各存储单元存储数据(权值)的可信程度.改进的CMAC中,利用激活单元先前学习次数作为可信度,其误差校正值与激活单元先前学习次数的-p次方成比例,从而提高神经网络的在线学习速度和精度;在此基础上利用滑模控制算法进行容错控制律的在线重构,实现动态非线性系统在线故障诊断与容错控制的集成.分析了系统的稳定性,仿真结果表明改进故障学习算法及容错控制的有效性.A fault-tolerant control scheme of dynamic nonlinear system based on improved CMAC neural network and sliding model control technique is presented in this paper. In the conventional CMAC learning scheme, the correcting amounts of errors are equally distributed into all addressed hypercubes, regardless of the credibility of those hypercubes, The proposed improved learning approach employs the number of learnings of addressed hypercubes as the credibility of learned value. The correcting amounts of errors are inversely proportional to the pth power of the number of learnings of addressed hypercubes. With this idea, the fault learning speed can be improved. Based on the improved CMAC learning approach and the effective control law of reconfiguration strategy using sliding model control technique, a combination of on-line diagnosis and fault-tolerant control for a dynamical nonlinear system is realized. The system stability and performance are also analyzed. Finally, the numerical simulation demonstrates the effectiveness of the improved CMAC algorithm and the proposed fault-tolerant controller.

关 键 词：信度分配 平衡学习 故障诊断 容错控制 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]