Recognition Of Tools Wear Of Machine Tool And Quality Analysis Of Workpieces With Spindle Track In A Digital Factory

In recent years,the"Internet+"technology that promotes the upgrading of traditional industries has received enthusiastic attention.In this context,digital factories have emerged.As a brand-new organization of production and manufacturing,the digital factory supervises,controls,simulates,evaluates and optimizes the entire production process by collecting and processing data throughout the entire processing and manufacturing life cycle.Integrating computer simulation,big data processing,artificial intelligence and modern digital man-ufacturing technology,it has the characteristics of digitization,integration,virtualization,dynamics,distribution,rigor,synergy,and complementarity.The digital factory strengthens the connection between product design and product manufacturing so that the factory's pro-duction efficiency is improved,the supply is balanced,and the management is intelligent.As the key equipment for product processing in the manufacturing industry,the digitization of CNC machine tools and the monitoring of the entire process of product processing and manufacturing are the first technologies that need to be paid attention to and studied.How to analyze and predict the working status of the machine tool through the big data obtained by the digital factory,and then analyze and predict the quality of the product and improvement measures to improve the yield of the product,is the primary demand for the intelligentization of the digital factory in this field.This article focuses on the above issues,based on the machine tool data obtained by a digital factory and public dataset,and carries out the following researches.(1)Based on the machine tool processing data provided by the manufacturing plant,collect the relevant data of the entire life cycle of the manufacturing process components for analysis,and study the methods of obtaining the station production status,machine tool efficiency,machine tool alarm analysis,etc.This paper studies the methods of obtaining the station production situation,machine tool efficiency and machine tool alarm analysis in a digital factory,and realizes the real-time monitoring of NC machine tool processing status through the statistical analysis of power signal and the analysis of tools and workpieces moving speed,so as to provide the monitoring,analysis and service for the digital factory under the background of industrial Internet.(2)Tool wear is a key factor in the quality of CNC machine tools.Aiming at the problem of tool wear identification and classification by high-dimensional data,an algorithm(Lasso Feature Importance&Extreme Learning Machine,LFI-ELM)that combines Lasso feature importance and extreme learning machine algorithm is proposed.From the perspective of feature importance,Used the Lasso feature importance analysis method,selected from 232data features to obtain the minimum X-axis output power,the maximum spindle acceleration error,the maximum spindle output voltage,the Y-axis position error crest factor,and the minimum Y-axis feedback current.The tool wear is predicted and analyzed.The LFI-ELM algorithm is further applied to the field of tool wear identification.The accuracy of tool wear recognition based on the traditional overrun learning machine algorithm is only 60.87%,while the accuracy of tool wear recognition based on LFI-ELM is as high as 97.39%.(3)Based on the machining data of machine tools,it is proposed to apply the random forest algorithm to the quality prediction of the machining workpieces and the prediction of the spindle trajectory.The quality of the processed workpieces is analyzed from the per-spective of geometric error,and the modules length of error vector in plane is used as one of the evaluation indicators.The quality of the manufactured workpieces is studied,and on this basis,the quality prediction of the processed workpieces is further realized through the prediction of the spindle trajectory,which provides a referential resolution for the managers,which is convenient for the managers to make timely adjustments to the CNC machine tools,thereby reducing the waste caused by The loss of raw materials.(4)In product manufacturing,the moving speed of the workpiece is an important pro-cess parameter.This paper proposes a method for predicting the moving speed of a workpiece based on the Pearson Correlation Coefficient weighted k-nearest neighbor algorithm(PCC-KNN),which calculates the Pearson Correlation Coefficient between the electrical signals of the motor and the moving speed of the workpieces and use it as the weight to weight the Euclidean distance.This algorithm improves the performance of the k-nearest neighbor al-gorithm,with RM SE reduced by 1.6581 and 0.9256,and R~2Score increased by 0.3285 and0.1793.When the conclusions of RM SE and R~2Score are inconsistent,comprehensively consider the root mean square error RM SE and the coefficient of determination R~2Score two evaluation methods,and propose a hybrid evaluation method R~2S?-RM SE is used as the evaluation method of the two comprehensive considerations.Using R~2S?-RM SE as the evaluation method,the prediction result of the workpiece moving speed based on PCC-KNN is 0.0972 and 0.1599 higher than that of the traditional k-nearest neighbor algorithm.