摘要
基于MATLAB模糊逻辑工具箱 ,利用维数缩减技术和模糊减法聚类法 ,对表征电液伺服阀工作状态的实测样本数据进行聚类分析 ,实现合理的特征空间划分和寻找适当的规则数目 ,从而实现了自适应神经—模糊推理系统 (即 :ANFIS)的结构辨识。在此基础上 ,利用BP算法与最小二乘法相结合的混合算法 ,实现ANFIS的参数辨识 ,建立了适用于电液伺服阀的故障模式识别的ANFIS ,从而有效地解决了电液伺服阀故障的多样性和不确定性的难题 。
s:Based on MATLAB Fuzzy Logical Toolbox, this paper clusters the practical sampled data which reflects the work states of electrohydraulic servo valve, gets the rational feature space partition and rule number by using dimensional decrement and fuzzy subtractive clustering algorithm. Consequently,the ANFIS initial structure is identified . On the basis of that, the parameters of the ANFIS is identified by the hybrid algorithm which combines BP algorithm with LSE and the applied ANFIS is created for the fault pattern recognition of electrohydraulic servo valve. Thereby,this effectively solves the problem of its fault variety and uncertainty and realizes the intelligent recognition for the fault patterns of the valve.
出处
《机床与液压》
北大核心
2003年第3期308-309,113,共3页
Machine Tool & Hydraulics
关键词
模糊减法聚类
自适应神经-模糊推理系统
电液伺服阀
故障诊断
模式识别
液压伺服系统
Fuzzy subtractive clustering
Adaptive network-based fuzzy inference system
Electrohydraulic servo valve
Fault diagnosis
Pattern recognition
