Study On Monitoring Method Of Tool Wear Of Multi-variety And Small Batch Production Lines

Due to large variety and small batch as well as cross parallel processing of workpieces,complex production process and environment of the production line in the aerospace enterprise,the information and emergencies become multi-source heterogeneous,complex and uncertain which result in low production efficiency,poor stability and other issues easily.As a result,the production line safe operation and the product quality is definitely influenced.Thus,to achieve effective monitoring of tool wear is crucial to ensure the safe operation of machine tools and the stability of product quality in aerospace complex component production line.Thus,the machine tools of a production line in an aerospace enterprise is taken as the research object.It makes a deeply analyzes of the machine tool wear problem in the production line in an aerospace enterprise,and then conducts a systematically studies the of machine tool wear monitoring technology.The specific research contents are as follows:(1)The problem of small size of sample and under-sampling with wear data set caused by flexible mixed flow machining and multi-process cross-parallel in multi-varieties and small batch production lines.To solve this problem,a method of machine tool wear data based on compressed sensing and noise processing is proposed.First,the compressed sensing is introduced to reconstruct the data to improve the sparsity and balance the sample class.Then,the random Gaussian noise is added to increase the number of training samples to cope with the problem of under-sampling and insufficient samples,in this way,the data reconstruction is achieved and the number of samples of machine tools wear in the production line of aerospace enterprises is increased.(2)The issue of difficulty and low accuracy of tool state recognition caused by multi-source heterogeneous data and small sample size of machine tool wear data in multi-variety and small batch aerospace enterprise production line.To address this issue,a tool state recognition method for multi-variety and small batch aerospace enterprise production line based on multi-source heterogeneous data fusion is presented.Firstly,the improved stack sparse self-encoder network based on Dropout method is used to solve the problems of low accuracy and over-fitting of deep learning network.Then,the tool state recognition model for small sample multi-source heterogeneous data is constructed which combined with the improved D-S evidence theory.Finally,the fusion of recognition results is realized,and the accuracy of tool state recognition is improved.(3)The problem of poor tool wear monitoring effect result from the uncertainty of quality and quantity of machining objects in a aerospace complex component production line.According to this problem,a tool wear critical state recognition method based on deep learning network and deep convolutional neural network is performed to realize the recognition of tool wear critical state.Moreover,the tool wear state monitoring model based on deep convolutional neural network is formulated,in this way,the intelligent monitoring of tool wear state of multi-varieties and small-batch aerospace complex structure production lines is realized.