摘要
针对非线性系统提出了一种基于动态数据驱动的LS-SVM多模型预测主动容错控制方法,使系统对已知故障容错的同时,利用动态数据驱动对模型的补充,实现了对未知故障的主动容错.该方法首先基于LS-SVM建立系统正常或已知故障模式的动态模型库,实际运行时依据系统对性能容忍度指标和模型失配度指标的实时计算分析,判断系统所处的运行模式,当系统发生已知故障时,直接调用动态模型库中已有的模型,并采用经局部线性化近似预测控制算法计算控制律;当系统发生未知故障时,则选用模型库中最接近当前运行模式的模型进行故障过程的过渡容错控制,并以动态数据5步补充循环算法,快速建立该未知故障的LS-SVM模型,进而利用新模型实现系统对未知故障的主动容错.并以一非线性系统仿真实例验证了所述方法的可行性和有效性.
For the non-linear systems, the predictive active fault-tolerant control of LS-SVM multiple models driven by dynamic data is presented,which achieves fault-tolerant confrol for the system with un- known faults. For this purpose LS-SVM is used to establish dynamic models to compose their basic bank. During the operation of the system,it judges different modes based on two indexes. If known faults occur, it attains the controllerts parameters on the basis of the known models in the model bank. On the contrary, if unknown faults occur,it attains the parameters after a new model is established. Simulations are given to demonstrate the validity of the proposed approach at the end.
出处
《甘肃科学学报》
2009年第3期88-93,共6页
Journal of Gansu Sciences
基金
甘肃省自然科学基金项目(3ZS051-A25-032)
甘肃省教育厅高等学校研究生导师科研项目(050301)
兰州理工大学特色学术梯队基金项目(0950)
关键词
主动容错控制
数据驱动
多模型
预测控制
active fault-tolerant control
driven by data
multiple models
predictive control
