一种混合智能故障诊断方法及柴油机故障诊断

提出了一种模糊聚类、粗糙集理论与神经网络集成的混合智能故障诊断方法。引入聚类有效性函数和点分布密度函数。对模糊c-均值聚类算法进行改进,形成了自适应模糊聚类算法并依据该算法将连续的故障特征值离散化。应用粗糙集理论处理离散化的故障诊断数据。采用基于信息熵的方法,约简冗余的故障特征。依据约简结果构建神经网络,采用遗传算法优化网络的权值和阈值。将该方法用于柴油机气门故障诊断,并与普通神经网络进行对比。结果表明,该方法提高了故障诊断的正确率。

A Hybrid Intelligent Fault Diagnosis Method and Fault Diagnosis for Diesel Engine

A hybrid intelligent fault diagnosis method integrating fuzzy clustering,rough sets theory and artificial neural network was proposed.Adaptive fuzzy clustering algorithm was formed by the introduction of cluster validity index and distribution density function of data point to improve fuzzy c-means fuzzy clustering algorithm.And continuous values of fault feature were discretized according the algorithm.Rough sets theory was employed to deal with the fault diagnosis data and redundant features were reduced by the information entropy-based method.Neural networks were established on the basis of reduction,and the weights and biases of which were optimized by genetic algorithm.Applying the method to the fault diagnosis of diesel engine valve and comparing with general neural network,the results indicate that the presented method improves the accuracy of fault diagnosis.