Research On Remaining Useful Life Prediction Of CNC Machine Tools Based On Deep Learning

In the context of the era of big data,intelligent manufacturing has become the inevitable direction of the development of the manufacturing industry.As an significant part in CNC machining,the problem of tool wear is one of the main problems in CNC machining.In the milling process of CNC machine,tool wear degradation is inevitable.Once the tool failure occurs,the workpiece surface quality will not meet the requirements.This leads to low efficiency of the machine,and even causes damage to the machine tool when the tool wear is serious.In order to solve this problem,this paper proposed CNC machine tool remaining useful life prediction method.The main research contents of this paper are as follows:(1)Research on data processing and feature engineering method of CNC machining.First of all,based on the analysis of the original data,the removal of non-processed data,abnormal empty files,files containing shutdown fragments,and trend items are processed.Next,feature indexes are extracted from three aspects: time domain,frequency domain and time frequency domain.Then,select the optimal feature set according to monotonicity and trend evaluation criteria.Finally,kernel principal component analysis algorithm is used for feature dimension-reduction processing.(2)Research on remaining useful life prediction method based on time series data regression.Direct prediction strategy is used to study the method and process of tool residual life prediction.According to the time series characteristics of CNC machining data,the remaining useful life label of cutting tool is constructed manually.Then,the attention mechanism is integrated into the long short-term memory network,and the mapping model between features and label is established.(3)Research on remaining service life prediction method based on degradation trend prediction.Indirect prediction strategy is used to study the method and process of tool residual life prediction.Based on the dynamic time warping algorithm,a health indicator which can represent the tool wear state is constructed.Then the long short-term memory network model is used to predict the trend of health indicator.According to the predicted degradation trajectory,the time from the failure threshold is calculated,and the remaining useful life of CNC machine tool is predicted.According to the above research contents,the remaining useful life prediction model of CNC machine tool is completed.An example is analyzed by using the real machining tool data.The feasibility of the proposed method is verified.