Analysis Of Bearing Fault Diagnosis Based On Industrial Internet Of Things Cloud Platform

The regular maintenance mode of operation and maintenance of traditional industrial equipment has caused problems such as untimely maintenance of equipment and high cost of operation.The emergence of equipment fault diagnosis based on artificial intelligence algorithms provides a new solution for equipment operation and maintenance.At present,most of the existing research on equipment fault diagnosis stays at the theoretical level of the algorithm,without combining specific industrial equipment operation and maintenance scenarios,and the actual effect of the diagnosis model needs to be verified.Data-driven fault diagnosis is the mainstream direction in the current field of fault diagnosis,but it requires higher sensor data quality and performance of fault diagnosis algorithms.Integrating fault diagnosis into the Internet of Things system to directly obtain equipment monitoring data for training prediction of the fault diagnosis model can greatly improve the accuracy of the model.In this paper,combined with actual industrial Internet of Things application scenarios,industrial data acquisition technology and convolutional neural network fault diagnosis algorithm are studied.Further,we develop a set of industrial Internet of Things cloud platform system to monitor,analyze and store data collected by various industrial sensors and integrate a fault diagnosis model,which is based on convolutional neural network,into the industrial Internet of Things cloud platform system to complete real-time monitoring of equipment status and operation status estimation,so that it is possible to give recommendations about operation and maintenance.This paper mainly works in the following three aspects.1.Research related technologies of the Industrial Internet of Things,select and debug industrial sensor technology,industrial information collection technology,and industrial Internet of Things data transmission protocols applied in the Industrial Internet of Things cloud platform system,and realize a complete data link for device status data collection and upload.2.Design and develop an industrial Internet of Things cloud platform system.The system selects B / S(Browser / Server)architecture,and uses JAVA programming language and a series of development components to develop an industrial Internet of Things cloud platform.The system uses the message middleware Kafka cluster to enhance the platform's data communication capabilities,and finally deploys it on the server,where the fault diagnosis system is implemented as a submodule of the system based on the convolutional neural network algorithm.Through a series of test work on the platform,it is verified that the various functions of the platform are complete and the performance meets the practical demands.3.Aiming at the difficulty of feature extraction of various sensor data in industrial scenes,this paper chooses a convolutional neural network with strong capabilities in feature learning to build a fault diagnosis model and compare the performance of CNN fault diagnosis model with LSTM algorithm model,DNN model and decision tree by using the bearing data set of Case Western Reserve University.Considering comprehensively the prediction accuracy and operation time of the model,the experimental results prove that the fault diagnosis models of convolutional neural networks are superior to several comparison algorithms.