SHIPENGINEERING船舶工程Vol.45No.12023总第45卷,2023年第1期—116—基于随机森林算法的船舶电站故障诊断陈冠宇,杨鹏,陈宁(江苏科技大学能源与动力学院,江苏镇江212003)摘要:针对船舶电站故障诊断中常用的BP神经网络算法存在的收敛速度慢和诊断准确率不高等问题,提出一种基于随机森林算法的船舶电站诊断模型。在Simulink软件中搭建船舶电站故障模型,通过在Simulink中仿真得到船舶电站故障数据,分析基于随机森林算法的船舶电站故障诊断原理。在MATLAB软件中分别建立基于随机森林算法和BP神经网络算法的船舶电站故障诊断模型,并对二者的故障诊断结果进行对比分析。结果表明,基于随机森林算法的诊断模型相比基于BP神经网络的诊断模型,能显著提高船舶电站故障诊断的效率和准确率。关键词:随机森林算法;船舶电站;故障诊断;Simulink软件中图分类号:U665.12文献标志码:A【DOI】10.13788/j.cnki.cbgc.2023.01.18FaultDiagnosisofShipPowerStationBasedonRandomForestAlgorithmCHENGuanyu,YANGPeng,CHENNing(SchoolofEnergyandPower,JiangsuUniversityofScienceandTechnology,Zhenjiang212003,Jiangsu,China)Abstract:InordertosolvetheproblemsofslowconvergenceandlowdiagnosticaccuracyofBPneuralnetworkalgorithminfaultdiagnosisofmarinepowerstation,adiagnosismodelbasedonrandomforestalgorithmisproposed.ThefaultmodelofshippowerstationisbuiltinSimulinksoftware,andthefaultdataofshippowerstationareobtainedbysimulationinSimulink,andthefaultdiagnosisprincipleofshippowerstationbasedonrandomforestalgorithmisanalyzed.ThefaultdiagnosismodelsofshippowerstationbasedonrandomforestalgorithmandBPneuralnetworkalgorithmareestablishedinMATLABsoftware,andthefaultdiagnosisresultsofthealgorithmsarecomparedandanalyzed.TheresultsshowthatthefaultdiagnosismodelbasedonrandomforestalgorithmcansignificantlyimprovetheefficiencyandaccuracyoffaultdiagnosisofshippowerstationcomparedwiththemodelbasedonBPneuralnetwork.Keywords:randomforestalgorithm;marinepowerstation;faultdiagnosis;Simulinksoftware0引言船舶电站主要用来为船舶动力系统和照明系统等提供稳定的电力,近年来随着船舶的负载容量不断增加,负载类型不断增多,其复杂程度越来越高,船舶对其稳定运行提出了更高的要求。对船舶电站进行...