Research On Tool Wear State Recognition Method Based On Feature Selection

With the advent of the era of intelligence,information technology continues to promote the development of the manufacturing industry,and traditional manufacturing is gradually transforming into "intelligent manufacturing and high-end manufacturing".With the upgrading of the manufacturing industry,the machining system is becoming more sophisticated and complex.Development,in which the tool is a key executive component in the processing system,and its wear state affects the work efficiency of the machine tool and the quality of the processed product,and even causes hidden dangers such as personal safety.Therefore,studying the tool wear state is of great significance to improving equipment efficiency,the quality of machined products and ensuring personal safety.The main research work in this paper is as follows:(1)The mechanism of tool wear and the main forms of tool wear are analyzed.The CUSUM algorithm is used to divide the stages of tool wear.Considering the characteristics of the changing stages of tool wear,the following research is conducted.(2)For the unstable vibration signal of the tool,the angle domain resampling method is used to resample the unstable vibration signal to eliminate the instability,and the multi domain feature extraction is carried out for the resampled vibration signal to obtain the tool wear feature set.(3)A feature fusion evaluation method based on Fisher Score is proposed to select tool wear features.The method uses feature monotony and robustness to build a comprehensive evaluation coefficient and combines Fisher Score algorithm to calculate the importance score of tool wear features,select features with high scores,and improve the recognition speed and accuracy of tool wear state recognition model.The classical Fisher Score algorithm and Relief F algorithm are used for comparative analysis to verify the superiority of the feature fusion evaluation algorithm based on Fisher Score.(4)A tool wear state recognition model based on CSA-TWSVM is constructed.The parameters of Twin Support Vector Machine(TWSVM)are optimized using the Crow Search Algorithm(CSA).The CSA-TWSVM model is simulated and analyzed using the UCI dataset to verify the validity and reliability of the model.The tool wear state identified by the model is compared with the real value of wear,which proves that the proposed method has high accuracy in tool wear state recognition.Finally,an application system prototype based on CSA-TWSVM tool wear state recognition model is developed and designed.