State Monitoring And Early Warning Of Engine Transmission Body Casing And Gear Processing Equipment

The domestic CNC machining tool is difficult to meet the processing requirements of casing and gear of aeroengine driver.Improving the intelligence of equipment is an effective way to solve this problem.To improve equipment utilization and reduce failure rate,the study of remote monitoring technology and early warning technology were carried out.The typical CNC machining tools of casing and gear production line and interfaces of servo driver and CAN module are introduced.By studying the signal definition in softPLC and the parameter of PLC and CNC kernel,the device communications program based on Socket protocols was developped.Then the confirmatory experiment was carried out.Referencing the requirements Analysis of remote monitoring system,the browser/server(B/S)architecture and the development mode based on the Frontend-Backend Separation were adopted.The Web interface was designed based on React,including login and registration functions,single/multiple device monitoring functions,production line monitoring functions and information recording or querying functions.Based on Flask framework and multi-thread technology,the server-side program was developed.The functions were realized including collecting the state information of CNC machining tools and operating PostgreSQL database implemented by using ORM framework.The test optimization of the monitoring system was carried out.The server response time was in milliseconds.The Web response speed increased 13.3 times.And by using Locust tool the function of multi-user operating was tested efficiently.The practicability of the monitoring system was proved.The key parameters to reflect the CNC equipment state were analyzed Further.The early warning values of basic motion parameters were determined.Combining with the experimental data,the warning methods of vibration signal,noise signal,motor temperature and follow-up error were studied.The method of multi-source parameters early warnings based on machine learning algorithm was proposed inventively.According to data set extracted from equipment state information,the early warning models were built respectively using the decision tree algorithm,the KNN algorithm,the multilayer neural network algorithm,and the support vector machine(SVM)algorithm.Best of all,the accuracy of health warning model trained by SVM with RBF kernel can reach 100%.Finally the function of automatically generating and pushing the early warning report was reallized.