Research Of Turning Tool Wear Recognition Method Based On Force Signal

In the cutting process,the tool wear or breakage will directly affect the quality of processing parts,severe tool breakage will result in scrap parts,the interruption of cutting process and machine damage,this will cause huge economic losses;Therefore,real-time accurate grasp of the tool wear and breakage is significant to improve product quality,improve productivity,and save processing costs.The paper do research on tool wear recognition method based on the cutting force signal.The main research contents are as follows:(1)Research of turning tool wear feature extraction method based on cutting force signal.The signal processing methods such as the time domain analysis,frequency domain analysis and time-frequency analysis were used to extract the features of force signal and 30 features sensitive to tool wear were selected by the correlation coefficient,and 33 features in addition to the cutting parameters were fused;After data processing,a total of 600 sample data of 33 dimension was gotten,the sample data was divided into training set and test set,and the 33 dimension feature vector was reduced to 7 dimensional by KPCA,the feature optimization technique greatly improved the speed and accuracy of subsequent pattern recognition.(2)Research on the recognition method of tool wear state.For the recognition of tool wear state,the recognition model such as C4.5 decision tree,pruning decision tree and random forest were constructed,after the model performance tested,the recognition accuracy of three models were 95%,96.67% and 99.17%.(3)Research on the recognition method of tool wear.For the recognition of tool wear,GA_BP,BP_AdaBoost and GA_SVR were constructed by improving conventional pattern recognition method such as BP and SVM,and the average absolute error of all model was below 0.0127 mm,root mean square error was less than 0.0170 mm.