基于PSO-SVM的发动机故障诊断研究  

Research on Automobile Engine Fault Diagnosis Based on PSO-SVM

作　　者：赵曦[1] 李颖[1] 徐江[2] 黄翰[3] 

机构地区：[1]广东科学技术职业学院计算机工程技术学院,广东珠海519090 [2]东南大学机械工程学院,江苏南京211189 [3]华南理工大学软件学院,广东广州510006

出　　处：《计算机仿真》2014年第3期171-174,共4页Computer Simulation

基　　金：国家自然科学基金(61003066;61370102);广东省自然科学基金项目(S2011040002890;S2012010010613)

摘　　要：研究发动机故障准确诊断的方法。在发动机发生故障时,利用传统算法对故障进行诊断,需要将发动机故障参数与发动机所有部件运行状态参数逐个进行对比,从而对故障部件进行诊断,存在较强的滞后性。为了避免上述传统算法的弊端,提出了一种基于PSO-SVM的发动机故障诊断方法。利用粒子群方法,对所有的发动机故障信号进行指定空间内的搜索,从而获取最优粒子,为发动机故障诊断提供依据。利用支持向量机方法,实现发动机故障信号的分类,从而完成发动机故障诊断。实验结果表明,利用本文算法进行发动机故障诊断,能够极大地提高诊断的准确性,从而满足实际生产、生活的需求,取得了令人满意的效果。In this paper, a method of engine accurate fault diagnosis was studied. When the engine is failure, the use of traditional algorithm for fault diagnosis needs to compare the engine fault parameters with the running status pa- rameters of every component one by one. Thus, it has strong lag in component fault diagnosis. In order to avoid the disadvantages of the traditional algorithm, this paper proposed an engine fault diagnosis method based on PSO - SVM. Using the method of particle swarm, all of the engine fault signals were searched within a specified space to ob- tain the optimal particle and provide the basis for engine fault diagnosis. Then using support vector machine （SVM） method, the classification of engine fault signal was achieved to complete the engine fault diagnosis. The experimental results show that the algorithm presented in this paper for engine fault diagnosis can greatly improve the accuracy, so as to meet the practical requirements of producing and living and get the favorable results.

关 键 词：发动机 故障检测 粒子群 支持向量机 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] U464[自动化与计算机技术—控制科学与工程]