Design And Implementation Of CNC Equipment Fault Diagnosis System Based On Edge Intelligence

Numerical control equipment is an important foundation of manufacturing modernization and has become an indispensable part of enterprise production activities.Once CNC equipment breaks down,it will cause huge losses to enterprises and even casualties.Therefore,the realization of fault diagnosis of NC equipment is an inevitable choice to ensure industrial production.Motor is the core component of NC equipment.Through the fault diagnosis of motor,the state of NC equipment can be monitored in real time.Traditional fault diagnosis methods rely on artificial feature extraction,and require researchers to have a priori knowledge,which is timeconsuming and laborious.In order to improve the accuracy and intelligence of fault diagnosis system,deep learning is applied to the field of fault diagnosis,and the advantages of neural network in feature learning are fully used to process the collected signals.In addition,with the rapid development of the Internet of things,massive data has brought great pressure to the server.Edge computing can effectively reduce data processing delay.Therefore,edge computing is combined with intelligent fault diagnosis method.Through edge computing,the computing center is placed on the edge side to reduce the pressure of the server,so as to realize edge intelligence,so that the motor can get the result of fault diagnosis without connecting to the cloud.Firstly,the structure and common faults of the motor are introduced,and the common motor condition monitoring methods such as vibration signal monitoring,electronic current monitoring and temperature signal monitoring are analyzed.Finally,the vibration signal is used for motor fault diagnosis and analysis.In view of the fact that the traditional convolution neural network can not fully extract the multi-scale information in the original data because it uses the convolution kernel with the same size,the multi-scale convolution method proposed in this paper can well solve this problem.Considering that the original data is the timing signal of the motor,the multi-scale convolution network mainly extracts the local spatial features and has insufficient ability to extract the timing information between the data.In this paper,a multi-scale time convolution neural network model is proposed,which inputs the fused features of multi-scale convolution into the time convolution network to extract the information of time series.Finally,the superiority of the proposed method is verified by experiments.Finally,the fault diagnosis system based on edge intelligence is realized.In the hardware part,Siemens nanobox PC is used as the edge device,and the data acquisition card of Ni usb-6289 is selected to collect the motor vibration signal.The software part adopts Java language to develop each module to realize the functions of user login,fault early warning and so on.