数据驱动逆系统方法的多模型主动容错控制

摘要

针对非线性系统提出了一种基于数据驱动逆系统思想的多模型主动容错控制方法;该方法首先采用最小二乘支持向量机(LS-SVM)对系统正常以及各种先验故障情形建立其逆模型,并以此构造系统的基本逆模型库.实际运行时依据系统性能指标切换函数,经在线计算分析,判断系统所处的运行模式.当系统发生已知故障时,调用与之匹配的LS-SVM逆模型,使其始终与被控对象串联保持为不变的伪线性系统,并以固定的内模控制器实现非线性系统对已知故障的快速容错;当系统发生未知故障时,切换至基于数据驱动技术的无模型自适应控制器(MFAC)进行过渡容错控制,在保证系统稳定的同时利用过渡容错期间的输入输出(I/O)数据建立系统当前故障情形的逆模型并添加到模型库中,使其具有自学习能力;最后以一仿真算例验证了文中所述方法的有效性.%For nonlinear system was proposed based on data - driven of inverse system of the multi - model active fault tolerant control method. First using least squares support vector machine (LS - SVM) system normal and various failure scenarios prior to model the inverse system, and thus construct the basic inverse model library system, the actual run - time system performance based on the switch function, calculated by the online analysis of the operating mode in which to judge the system, known to occur when the system fails, call the matching inverse model of controlled object so that it always remains the same series pseudo linear system to a fixed internal model controller for nonlinear system fault quickly known fault tolerance; When an unknown fault, switch to data - driven technique based on model - free adaptive controller (MFAC), the transition fault tolerant control, to ensure system stability, fault tolerance of the unknown active fault tolerant by using the transition period while the input and output data set system is currently unknown faults inverse model and added to the model library, it has a self - learning ability. Finally, a simulation example verifies the validity of the method.