基于随机森林方法的地铁车门故障诊断

针对现有地铁车门故障诊断方法存在的诊断速度慢以及大量故障检修数据未得到合理利用等问题,提出一种基于信息增益率的随机森林故障诊断方法.该方法将地铁车门历史故障数据集转化成决策表,通过Bootstrap重抽样,建立多棵基于信息增益率的决策树,形成随机森林故障诊断模型,实现地铁车门故障的快速诊断.且随着故障数据的增加,其故障诊断模型可以自动更新完善.通过地铁车门实际故障数据,验证了该方法的有效性.同时,通过对随机森林模型中决策树的数目讨论分析,确定了该方法模型的最优设计结构.

Subway Door Fault Diagnosis Based on Random Forest Method

In order to solve the problems of the existing subway door fault diagnosis methods such as slow diagnosis and the failure of reasonable utilization of a large number of troubleshooting data,a random forest fault diagnosis method based on information gain ratio is proposed.This method historical of fault data set subway doors is transformed into decision table,and multiple decision trees based on the information gain ratio is built through Bootstrap re-sampling to form a random forest fault diagnosis model,which can realize the rapid diagnosis of subway doors faults.With the increase of fault data,the fault diagnosis model can be automatically updated and perfected.The effectiveness of the method is verified by the actual fault data of subway door.At the same time,the optimal structure of the model is determined by discussing and alalyzing the number of decision tree in the random forest fault diagnosis model.