基於多種群量子粒子群的粗糙集屬性約簡演算法在故障診斷中的應用

Applications in Fault Diagnosis Based on Quantum-Behaved Particle Swarm Rough Set Attribute Reduction by a Multi-Swarm Algorithm

針對故障診斷過程中的特徵選擇問題，提出一種基於多種群量子粒子群約簡演算法；該方法引入多種群分群的改進策略，利用兩層結構分別實現雙向搜索，保證群體進化快速收斂於最小約簡，選出真正代表故障的特徵變數；通過對田納西－－伊斯曼 （Tennessee-Eastman, TE） 公司的故障診斷應用，並與主元分析法等實驗對比，結果表明，基於MQPSO的粗糙集屬性約簡演算法應用於故障診斷具有非常實際的意義，能夠最大程度上提高故障診斷的正確率。

Considering the feature selection problem in a fault diagnosis process,the author proposes a method based on quantum-behaved particle swarmattribute reduction by a multi-swarm algorithm. This method introduced amulti-swarm clustering strategy, using a two-tier structure to realize abidirectional search, to ensure fast convergence to the minimum populationreduction,and select feature variables that truly represent the characteristicsof the fault. Through the application of Tennessee-Eastman's faultdiagnostic and other experiment using principal component analysis, theresults show that a rough set attribute reduction algorithm based onMQPSO is practical and can improve the accuracy of fault diagnosis to thegreatest extent.