Intelligent Prediction And Control Of Grinding Dimensions Based On Statistical Learning

Grinding is an important process of high precision parts.Grinding accuracy directly affects the quality and performance of parts.Therefore,it is necessary to study how to improve the grinding accuracy.Measurement and control in the process of machining is one of the key elements to improve the accuracy of machining.Active Measurement technology is a real-time online measurement of workpiece size and monitoring of the processing state in the process of machining.It is widely used in modern grinding process.Grinding Active Measuring Instrument is the research result of active measurement technology.Through real-time monitoring of grinding process data,it guides grinding machine to change grinding parameters(grinding wheel speed,grinding wheel feed speed,etc.)and achieves complete closed-loop feedback control in the process of grinding.At present,the grinding active gauge produced in China can not realize the function of prediction in grinding,and the adjustment of grinding parameters lags behind the grinding process.If the change trend of workpiece size can be predicted in the process of machining,the active gauge can take corresponding measures in advance and feed back the processing information in time.The active gauge can guide the machine tool to change the grinding parameters and complementary adjustment values,so as to improve the quality and intelligence of grinding process.In this paper,on the premise of meeting the actual needs of engineering,the active measurement technology of grinding,the method of statistical learning prediction and the construction method and optimization method of prediction model are discussed and analyzed theoretically.So this paper studies the intelligent prediction and control method of grinding dimension based on statistical learning,and carries out the applied research on the size prediction model in practical engineering.In this paper,intelligent discontinuous surface treatment based on support vector machine,compensation value prediction based on grey support vector machine and remote grinding monitoring and alarm system are studied.The main research contents and results are as follows:(1)Analysis of grinding dimension error based on active measurement mode.The sources and causes of dimension errors in measurement and processing are analyzed.The corresponding measures are put forward according to the analysis results,which provides the basis for the prediction and control of grinding dimension.(2)Research on grinding dimension prediction and control method.Aiming at the problem that the adjustment lag of grinding parameters affects the accuracy of grinding dimension in the grinding mode of active measuring instrument and grinding machine,a prediction and control method of grinding dimension based on statistical learning theory is proposed.In this paper,grinding parameters are adjusted online according to the trend of grinding size change.Then the goal of improving the dimensional accuracy of grinding workpiece is realized.(3)Research and optimization of grinding dimension prediction model.By analyzing the factors affecting the grinding dimension accuracy,a prediction model of grinding dimension based on grey relational support vector machine(GRSVM)is proposed on the basis of statistical theory.The model combines the advantages of grey relational system and support vector machine.The complexity of modeling is reduced by screening the input of the model.The training set of the prediction model is optimized by convex hull algorithm and KKT condition.This provides a theoretical basis for the construction of grinding dimension prediction model based on online incremental learning.The hybrid function theory and cross validation parameter optimization method are applied to the grinding dimension prediction model,which further improves the prediction accuracy of the prediction model.The grinding parameters are adjusted according to the trend of predicted size,which improves the dimensional accuracy of grinding.(4)Applied research and experimental analysis of grinding dimension prediction and control.Based on the existing laboratory experiments,the validity and feasibility of the intelligent discontinuous surface processing method based on support vector machine,prediction of grinding compensation value based on grey support vector machine and remote grinding alarm system based on configuration software are verified.Experiments show that the research enriches the function ofthe prediction module of the active gauge system.It also improves the precision of grinding and promotes the intelligent degree of grinding.In this paper,a method of predicting and controlling grinding dimension based on statistical learning is proposed to solve the problem of affecting the accuracy of grinding dimension in active measurement grinding mode.The prediction model is optimized by studying the influence factors of grinding size and statistical learning theory,so that the prediction accuracy of the optimized prediction model is higher.The feasibility of the prediction and control method of grinding dimension is verified by the application research and experiment of grinding dimension prediction and control.It is proved that this method can effectively improve the intelligent level of grinding and the processing accuracy of products,and has the value of popularization and application.