摘要
该文提出了高精度ＲＢＰ－模糊推理复合学习系统．系统主要由基于鲁棒估计的鲁棒ＢＰ学习环节和基于混合合成推理的模糊推理环节构成．该学习系统的主要特点是可由鲁棒ＢＰ算法和ｍｉｎ－ｍａｘ，ｍａｘ－ｍｉｎ模糊推理算法简单地实现．最后通过在目标跟踪问题中应用结果，表示了该算法的高精度和鲁棒性．
A new accurately robust BP-fuzzy reasoning system for learning combination is proposed. This learning system is mainly constructed with a robust BP network with fuzzy reasoning which replaced robust estimation and mixed fuzzy reasoning.The main feature of this learning system is a simple algorithm constructed from the following three parts: RBP learning algorithm, max-min fuzzy reasoning and min-max fuzzy reasoning. This learning system is applied to a target tracking problem.The results of test show that this tracking system is more accurate and more robust.
出处
《自动化学报》
EI
CSCD
北大核心
1995年第4期392-399,共8页
Acta Automatica Sinica
关键词
神经网络
模糊推理
学习系统
BP network
fuzzy reasoning
robust estimator
target tracking system.
