Research On Wear Monitoring Technology Of Milling Cutter Based On M-SDSAE Network

With the constant development of machining technology,people put forward higher requirements for the machining accuracy of parts in aerospace,automobile,precision instruments and other high-tech fields.In the process of machining,tool wear is one of the main factors that affect the machining accuracy of parts,so the research on tool wear condition monitoring has high practicab ility and social-economic value.The selecting principles of blunt tool criterions and sensor are determined by analyzing the tool wear form,wear process and the influence of tool wear on various signals.Choosing the appropriate hardware and planning the function and structure of the software for the requirement of tool wear monitoring system.Then the data acquisition module,signal processing module and tool wear monitoring module are designed and implemented.In order to solve the problem that the original data of sensor can't directly reflect the change of tool wear state in the process of machining.We use statistical theory and signal processing technology to preprocess cutting signals and extract features from cutting signals.Then build a test platform to obtain the real and effective multi-sensor signal in the process of machining.By analyzing the time domain,frequency domain and time-frequency domain of the signal,we can extract the signal features which can reflect the change of tool wear stat e.In order to make up for the defect of traditional tool wear monitoring model.In this paper,deep learning knowledge is used to solve this problem in the field of tool wear monitoring.So we used SDSAE network to Monitor the wear state of milling cutter.Compared with SDSAE network and BP neural network,SDSAE network has better performance in tool wear monitoring.It is concluded that the SDSAE network has superiority in tool wear monitoring field.By using the information fusion method in machine learn ing,a tool wear monitoring method that M-SDSAE multi-mode stack noise reduction self coding network is proposed.In this method,multi-sensor information is fused in the feature layer.Combined with the previous algorithm,develop and integrate of each module of tool wear condition monitoring system in the MATLAB 2018 a and VC++6.0 development environment.This paper presents a method of information fusion,it is M-SDSAE neural network.The M-SDSAE neural network can fuse information in feature level.In matlab2018 a and VC++6.0 development environment,we combine with the previous algorithm to develop and effectively integrate the modules of the tool wear condition monitoring system.