Research On The Remaining Life Prediction Method Of Bearings In Wooden Door Production Equipment Based On Deep Learnin

Rolling bearing is an important component which is used in wood gate processing equipments and the main cause of equipment failure.To reduce the possibility of the failure and help enterprises to make repairing schedule,hence,it is very important to find a method of predicting rolling bearing remaining useful life which can be used in wood gate industry.The research contents of the article are as follows:Firstly,the bearing vibration signals contain much of noise.This article used a denoising method based on variational modal decomposition.To find the best parameters of variational modal composition,simulated annealing algorithm was used to solve this problem.The results of the experiential signal and simulate signal showed that the denoising method can effectively eliminate noise.Secondly,because of the sensitivity of different characteristic indicators to different bearing fault types and bearing fault degrees are inconsistent.I extracted12-dimensional characteristic indicators in three domains including time domain,frequency domain and wavelet packet domain.To avoid the error caused by different working condition,normalization was applied to all characteristic indicators.Thirdly,this thesis researched an attention-based LSTM model,the attention mechanism helped the model to process different characteristic indicators.The experiential results showed that the method has good predict performance.Finally,to put all the algorithms in used,a software system of rolling bearing remaining useful life prediction was developed,the system can realize mechanical system data acquisition and real-time remaining useful life prediction.This proposed rolling bearing remaining useful life prediction helps enterprises to realize the intelligence upkeep of processing equipment.This research gives some reference value to processing equipment monitoring and repairing.