Research On The Prediction Technology Of NC Machining Working Hours Based On Big Data Analysis

Manufacturing enterprises need to use NC machining working hours to determine product delivery cycle,make production plan,calculate cost and analyze the capacity and load of NC machining equipment.Therefore,the accuracy of NC machining working hours plays an important role in the field management of manufacturing enterprises.With the development of intelligent manufacturing and Internet of things,manufacturing enterprises can collect and acquire massive data with typical big data characteristics(manufacturing big data)generated in the manufacturing process of parts from CNC machining equipment,and make full use of these data can better predict NC machining hours,thus bringing new possibilities for improving the accuracy of working hours.Based on the analysis of the influencing factors of working hours,this paper puts forward and explores the prediction method of NC working hours based on big data analysis,so as to improve the accuracy of NC working hours prediction and further promote the scientific and refined quota management.Firstly,the direct influencing factors of working hours include the type of machining,the amount of material removal and cutting parameters.According to the influencing factors,the main source and classification of manufacturing big data are determined,and the overall framework and technical framework of working hours prediction based on big data analysis technology are proposed.Secondly,In view of the problems existing in the collected manufacturing big data,define the rules of process coding,formulate the cleaning strategy of manufacturing big data,code and clean the manufacturing big data,and the pre-processed manufacturing big data was stored based on Hadoop Distributed File System(HDFS)and structured database.Then,the structure of working hours prediction model based on back propagation(BP)neural network.In view of the problems of falling into local minimum and slow convergence speed in some regions in the training process of working hours prediction model,Levenberg Marquardt(LM)algorithm is used to optimize the working hours prediction model.Through case analysis,it is proved that LM-BP working hours prediction model has better prediction performance.Aiming at the problem that the training efficiency of working hours prediction model is low when the data volume of manufacturing big data is large,a parallel design scheme of LM-BP algorithm based on MapReduce framework is proposed.Through case analysis,it is proved that the parallel LM-BP working hours prediction model consumes less training time and improves the training efficiency of the working hours prediction model.Finally,the prototype system for CNC machining working hours prediction is designed and developed,and the application case at a workshop has initially proved the feasibility and effectiveness of the method.