基于粒子群优化支持向量机的水面无人艇故障诊断  

Fault Diagnosis of Unmanned Surface Vessel Based on Support Vector Machine with PSO Algorithm

作　　者：廖建锋[1] 刘庭瑞 LIAO JianFeng;LIU TingRui(Henan Economy and Trade Vocational College, Zhengzhou 450000, China;College of Automation, Harbin Engineering University, Harbin 150001, China)

机构地区：[1]河南经贸职业学院,郑州450000 [2]哈尔滨工程大学自动化学院,哈尔滨150001

出　　处：《船舶工程》2018年第4期15-18,27,共5页Ship Engineering

摘　　要：通过故障诊断对水面无人艇（USV）潜在的故障进行预报、分析和判断，从而及时调整控制策略以抑制故障的继续发展，为消除故障和维修设备提供准确的技术支持。支持向量机（SVM）在解决小样本、高维度和非线性模式识别问题中有独特优势，而最佳的属性参数却很难直接得到。粒子群优化算法（PSO）具有全局搜索能力和易于实现的优势，将其应用到SVM属性参数的优化选择中。USV故障诊断实例分析结果表明，PSO—SVM的故障诊断精度高于BP-NNs、GS—SVM和GA-SVM，适用于USV故障诊断。Fault diagnosis is used to forecast, analyze and judge the potential faults concealed inside Unmanned Surface Vessel （USV）, so that the control strategy can be timely adjusted to suppress the continued development of the faults, and the accurate technical support is provided to eliminate the faults and maintain the equipments. Support Vector Machine （SVM） has unique advantages in solving pattern recognition problems featuring small sampling, high dimension and non-linear. However, it is very difficult to get appropriate SVM parameter directly. Particle swarm optimization （PSO） has the advantages of overall search and easy implementation, and it is applied to the optimization of SWM attribute parameters. The analysis results of USV fault diagnosis experiment show that the PSO-SVM method can achieve higher diagnostic accuracy than BP-NNs, GS-SVM and GA-SVM,, and it is applicable for the fault diagnosis of USV.

关 键 词：水面无人艇 故障诊断 粒子群优化 支持向量机 参数优化 

分 类 号：U661.323[交通运输工程—船舶及航道工程] TP273[交通运输工程—船舶与海洋工程]