Research On Precision Workpiece Classification Algorithm Based On Support Vector Machine

With the development and progress of manufacturing industry,intelligent manufacturing technology is more and more popularized.As the work piece,Precision work piece belongs to non-standard products always be used in special occasions,which has higher requirements for intelligent manufacturing technology.In the process of precision piece machining,the information of the work piece can be obtained automatically by reading the radio frequency chip in the work piece tray(the chip stores the relevant information of the work piece),which updates the traditional manual information input method and improves the intelligence of precision machining.However,in the heat treatment stage of the work piece,the work piece is separated from the tray,which causes the information association between the work piece and the tray to fail and affects the subsequent intelligent processing.Aiming at the failure of the correlation between the work piece and its information,the image processing technology is used to realize the matching of the work piece before and after heat treatment,so as to realize the correlation between the work piece and its information and improve the intellectualization of precision work piece processing.The main research of this paper are as follows(1)Aiming at the problem that single feature cannot accurately describe the precision work piece,when the feature of the precision work piece was studied,the fusion feature was proposed to describe the precision work piece.Firstly,according to the characteristics of precision work piece,the fusion features are obtained by SIFT feature and HOG feature.Secondly,in order to reduce the data volume of the fusion feature,KPCA dimension reduction algorithm was used to reduce the dimension of SIFT feature and HOG feature respectively.Finally,tandem fusion was carried out to obtain the fusion feature which can well describe the precision work piece.The experimental results show that the classification accuracy of fusion features is 86.7%.Compared with a single feature,the fusion feature proposed in this paper can describe the precision work piece more effectively.(2)In order to solve the problem of small batch and multi-variety production of precision work piece,An adaptive Levy Flight Particle Swarm Optimization Algorithm(ALPSO)was proposed to identify the corresponding work piece before and after heat treatment.Firstly,the image of precision work piece was collected and the fusion features are extracted.Secondly,the improved Levy flight PSO algorithm was used to optimize the SVM to solve the problem of the traditional SVM’s hyper parameter selection,so as to improve the classification ability and robustness of the SVM classifier.Finally,the precision work piece images are divided into training set and test set,the improved ALPSO-SVM classification algorithm was trained and verified.Experimental results show that the classification accuracy reaches 95.6%.Compared with the common classification algorithm,the ALPSO-SVM classification algorithm proposed in this paper can effectively classify the precision work piece with higher classification accuracy and stability.