农机化研究2024，Vol.46Issue(3)：246-251,6.
基于随机森林算法拖拉机齿轮箱故障诊断研究
Research on Tractor Gearbox Fault Diagnosis Based on Random Forest Algorithm
摘要
Abstract
Accurate detection of mechanical components failure is an important basic condition to reduce maintenance costs,production losses and extend the service life of the machine,in the agricultural production environment,due to the poor operating environment and more interference signals,so the sensitivity of the traditional fault sensor is reduced,can not be timely judgment and processing of the fault,resulting in frequent failure of agricultural machinery work.In this study,an agricultural tractor is selected as the research target for intelligent fault diagnosis of the tractor auxiliary gear-box.The vibration signals of healthy and faulty pinions were collected under three different speed conditions(600、1350 and 2000r/min),based on discrete wavelet transform(DWT)as signal processing,by correlation feature selection(CFS)method was used for feature selection,random forest(RF)and multilayer perceptron(MLP)neural network was used for data classification,the results of the study showed that the RF fault prediction accuracy was 86.25%at 600RPM without using feature selection and 92.5%at 600r/min using the best 6 features of CFS through the corresponding values of the trained RF.At 1350r/min,the RF fault prediction accuracy was 95.0%.关键词
拖拉机/齿轮/故障分析/小波变换/随机森林Key words
tractor/gear/fault analysis/wavelet transform/random forest分类
农业科技引用本文复制引用
姚鹏飞,涂亚楠,王瑞红..基于随机森林算法拖拉机齿轮箱故障诊断研究[J].农机化研究,2024,46(3):246-251,6.基金项目
河南省高校国家级大学生创新创业训练计划项目(202013498009S) （202013498009S）
