混沌耗散离散粒子群算法及其在故障诊断中的应用  

Chaotic dissipative particle swarm optimization and its application to fault diagnosis

作　　者：王灵[1] 俞金寿[1] 

机构地区：[1]华东理工大学自动化研究所,上海200237

出　　处：《控制与决策》2007年第10期1197-1200,共4页Control and Decision

基　　金：教育部博士点专项基金项目(20030251003)

摘　　要：针对大型化工过程生产系统的高维度数据及其噪声严重影响故障诊断的性能,采用基于故障特征选择和支持向量机(SVM)的故障诊断方法.为了确保在线故障诊断的实时性和准确性,提出一种新型的混沌耗散离散粒子群(CDDPSO)算法,用于故障诊断中特征变量的搜索.仿真结果表明,CDDPSO算法能有效地搜索到全局最优解,而基于故障特征选择的故障诊断方法具有良好的故障诊断性能.Considering the high dimensionality of data and the noises in large-scaled chemical process industry system seriously spoil the fault diagnosing performances, a fault diagnosis method based on fault feature selection and support vector machines is proposed. To ensure the real-time capability and correct rate of diagnosing online, a novel chaotic dissipative discrete particle swarm optimization （CDDPSO） is developed to search fault feature variables for diagnosis. Simulation results show that CDDPSO finds the global optima more effectively and the proposed fault diagnosis method possesses better diagnosing performance.

关 键 词：故障诊断 粒子群优化算法 特征选择 支持向量机 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]