Research On State Prediction And Fault Warning System Of High Grade CNC Machine Tools Based On Machining Process Data

As the "Made in China 2025" strategy is put forward,the crucial task is to realize the "intelligent","high efficiency" and "digital" transformation of CNC machine tools in traditional manufacturing industry.Meanwhile,it's an urgent problem to avoid or reduce the machine failure time and improve machine utilization.Traditional manufacturing processes discard critical process data,which cause in serious data waste.While the process real-time data is the directly response to machine state.Therefore,this paper firstly studies the data acquisition method of high-grade CNC machine tools.Then,analyzes the state predict methods and study the fault reasoning model about weighted fuzzy Petri Net based on processes data.Finally,design and develop the fault warning system,which provides a new solutions for digital transformation of CNC factory.The main works of this paper are summarized as follows:1.Research the data acquisition method of Siemens 840 Dpl and HZ-8 CNC system.Study the hardware and software structure of CNC machine tools and compare the characteristics of each acquisition methods.Then develope the acquisited software of two CNC system based on OPC specification and third party interface.Then clearfy the storage area and format of NC data.Finally,complete the validation tests in a CNC factory and the correctness of software is proved.2.Research on machine state prediction model based on multiple matching of multi-dimensional time series.Proposed the state model and measurement model of machine tool based on the characteristic of process data.Then emphatically establish the multi-matching model by using sliding window.With the proposed ?-coupling similarity index,the maximum similarity set?optimal sliding time w and prediction time L is determined.Next,the DBSCAN clustering algorithm is adopted to achieve the exactly match of machine state.Finally,the simulations for the collected parameters are carried out which verify the effectively and superiority of proposed modeling algorithm.3.Research on warning model based on weighted fuzzy fault Petri Net(WFPN).The WFPN model is proposed to describe the machine fault triggering process based on the traditional Petri net.Furthermore,by using the passing matrix and state equation the fault reasoning algorithm is proposed.Then this paper presents simlary BP algorithm,fish optimization algorithm and fault rules self-learning algorithm which had solved the difficult in getting model parameters and fault rules.Next,the WFPN model is established for the key components of the machine tool and the fault warning process is completed based on the results of the previous chapter.Finally,the simulation results had proved the effectiveness of the proposed optimization algorithm.4.Develope the state prediction and fault warning system.Firstly,the overall design of system architecture is proposed according to the condition of factory.Then the fault warning client is developed by using C# and Matlab mixed programming.The web client of MVC.Net framework is completed though HTML/JavaScript.Next,develope and design the Oracle database of whole system though ADO.Net.Meanwhile,the part of communication module is developed by WCF service.Finally,simply complete the functional test and performance test of whole system.