Research On Tool Wear Monitoring System Based On Multi-sensor Information Fusion

In the context of informatization,the continuous integration of a new generation of information technology and industry has accelerated the process of automation,networking and intelligence of mechanical processing and manufacturing equipment.CNC machine tools are the main equipment for manufacturing high-end parts in the machining industry,and their intelligence determines the efficiency and quality of machining.The machining accuracy of CNC machine tools is the core indicator of machine tool quality,and its accuracy depends on the quality of machine tool manufacturing,control algorithms and machining tools.CNC machine tools are direct participants in product processing,and the degree of tool wear is the core element to ensure product quality in the machining process.Therefore,the development of a tool life management system for machine tool processing tools is an urgent problem to be solved in the intelligentization of machine tool equipment.This paper takes the signals collected by the sensors of the CNC milling machine during the machining process as the research object,uses feature engineering and deep learning algorithms to perform data mining on the collected signals,uses a method based on multi-sensor fusion features to monitor tool wear values,and builds Tool wear value monitoring and life management system based on deep learning model.The main research contents of this paper are as follows:1.First,collect the vibration sensor,force sensor,and acoustic emission sensor signals through the CNC machine tool workbench,and then perform the time domain,frequency domain and wavelet packet coefficient feature extraction on the collected signals.Among them,the time domain,frequency domain and frequency domain characteristics of the vibration signal,force sensor signal and acoustic emission signal are collected respectively,and the feature vector of each sensor is constructed to represent the current wear state of the CNC milling machine tool.2.Secondly,in order to be able to accurately identify the tool wear status,considering the noise environment in practical applications,the adaptive denoising ability of the deep residual shrinkage network is used to classify and recognize the tool wear status to increase the robustness of the model.In this experiment,the multi-sensor combination feature is directly used as the training data,and the wear state of the tool is divided into three stages: the initial stage,the middle stage and the later stage,which are divided into three categories as the training label data.By adding noises with different signal-to-noise ratios to the signal,and comparing different deep neural network models,the deep residual shrinkage network’s ability to identify the degree of tool wear in a strong noise environment is verified.3.Then,use the tool wear value as the label data to establish a deep residual network,and use the data collected by each sensor as a training sample in turn to construct a deep residual network model.Through comparative analysis,explore the optimal tool wear value monitoring sensor Types,provide a reference for sensor selection for the intelligent transformation of traditional CNC machine tools.Further,based on the deep neural network method,the features of multiple sensors are merged to construct a deep learning model,and the monitoring effect of fused multi-sensor features compared with a single sensor feature in tool wear value monitoring modeling is explored.4.Finally,save each deep neural network model,use python,Qt5 and other development tools to develop a tool wear value real-time monitoring and life management system based on the deep learning model.Experiments show that the monitoring model and adaptive denoising classification recognition model used in this article have more accurate monitoring effects and higher recognition accuracy.At the same time,it can be migrated to tool life management software for application.