Improved Grey Model And Its Application In Honing Size Prediction

Honing processing is the last process in the processing of cylinder parts.Therefore,the quality of honing directly affects the final quality and performance of the parts,and the machining accuracy becomes more and more urgent.However,the processing dimensional accuracy is one of the important indicators to measure the processing quality.The uncertainty and nonlinearity of the honing process will lead to the problem that the traditional roughing process has an oversized size.Establishing a dimensional prediction model suitable for honing processing and online correction of machining parameters for real-time monitoring of machining processes via online measurement systems have become one of the effective means to effectively ensure the dimensional accuracy of the machining and control the quality of the honing process.In this dissertation,the prediction model and optimization algorithm were established for the problem of honing processing dimensional accuracy.The main research contents include:(1)The classical GM(1,1)prediction model was taken as the research object,aiming at improving the prediction accuracy and reliability of the classical GM(1,1)prediction model by analyzing the modeling process factors affecting the classical GM(1,1)prediction model.Andaseries of improvements to the classical GM(1,1)prediction model were made for the influencing factors.A cumulative AGM(1,1)prediction model was established by using background value reconstruction and parameter cumulative estimation improvement methods.The improved prediction model was applied to the honing processing size prediction.The applicability and prediction accuracy of the model were verified by simulation experiments.On this basis,in order to solve the application of the prediction model in medium and long-term prediction,an equal-dimensional replenishment method was established to establish an equal-dimension-cumulative AGM(1,1)prediction model,which is added to the modeling process based on prediction information or real-time processing information.The prediction model could be established.The method could accurately describe the change trend of the size,make the change trend of the size more accurate,improve the fitting degree between the predicted value and the measured value,and reduce the relative error.(2)Using the combined model method,this dissertation establisheda combined model of equal-dimension-cumulative AGM(1,1)prediction model based on support vector regression(SVR)residual correction.The gray wolf optimization algorithm(GWO)was used to optimize the parameter parameters of SVR,and the SVR prediction model was established.The residual value of the prediction model of the equal-dimension-cumulative gray AGM(1,1)prediction model is used to correct the residuals,and the prediction is further improved.Precision.(3)Finally,the standard error(MSE)and the mean standard error(MAE)were used as the evaluation criteria.The simulation experiment was established and compared with the prediction results.The results show that the combined model of the equal-dimension-cumulative AGM(1,1)prediction model with SVR residual correction has higher prediction accuracy,and the combined model can significantly improve the prediction accuracy when applied to honing processing size prediction.The combined model of the equal-dimension-cumulative AGM(1,1)prediction model for SVR residual correction obtained in this paper satisfies the accuracy requirements of honing processing size prediction,and provides a theoretical basis for dynamic adjustment of honing processing parameters and on-line control of processing quality.