摘要
首次将模糊控制的原理引入时变系统的辨识中，提出了一种遗忘因子模糊自调整原理，导出了一种实现比较简便，对参数时变具有较高跟踪速度和精度的辨识方法。该方法不仅适用于故障诊断的参数估计，也适合于时变系统参数估计问题，理论分析和仿真验证了该方法的有效性。
When time-varying system identification algorithm is applied to fault diagnosis,obviously the effectiveness of such an algorithm is important.When forgetting factor RLS(recursive least squares)algorithm is used for such an algorithm,the forgetting factor should be quickly and easily adjustable to make the algorithm always effective.We apply fuzzy control to make such an adjustment quick and easy.Fig.1 shows schematically that we first fuzzify and then defuzzify.Fuzzification requires eqs.(6a)thorugh (6f).Defuzzification requires eq.(7).Between fuzzification and defuzzification there is in Fig.1 a fuzzy control table,which is given in full as Table 2.The effectiveness of our new forgetting factor RLS identification algorithm is confirmed by simulation tests. The cost functions, which should be as small as possible if effectivess is to be realized,are shown in Figs.2 and 4 as varying with time. Our fuzzy-controlled forgetting factor(curve 1 in Fig.2) gives much better results than fixed forgetting factors 0. 95 (curve 2) and 0. 99 (curve 3).Fig.4 magnifies curve I in Fig. 2. In Fig. 4 the quick jump and quick return in the neighborhood of 200 time steps show clearly that our new method can effectively follow step -function changes in the system.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
1996年第4期517-521,共5页
Journal of Northwestern Polytechnical University
基金
国家教委博士点基金
关键词
时变系统
遗忘因子
模糊控制
故障诊断
time -varying system identification,forgetting factor, fuzzy control, fault diagnosis
