Tool Wear Condition Monitoring Technology Based On Fractal Theory

Tool wear in the cutting process will reduce the surface quality of the workpiece and affect the machining accuracy.Excessive wear is likely to damage workpieces and machine tools and even threaten personal safety.However,tool condition monitoring technology can monitor tool wear states and change tool in time,which is helpful to improve processing quality,improve production efficiency and reduce production cost.This paper presents a method based on fractal theory,binary gray wolf optimization algorithm and support vector machine by collecting cutting force signals and vibration signals for tool condition monitoring in turning and milling process.The specific research contents are as follows:Firstly,this paper introduces the research status of tool wear condition monitoring and fractal theory,and expounds the form and process of tool wear.Cutting force signal and vibration signal data are obtained by designing tool wear turning experiment and using public milling data set.The original signal is denoised by wavelet threshold,which can eliminate the influence of noise.Then,fractal theory is used to analyze the relationship between cutting force signals and vibration signals and tool wear in the cutting process,and fractal features are extracted.The results show that the signals in cutting process have self similarity and scale-free.The box dimension and correlation dimension of some components of cutting signal have high linear correlation with tool wear,which is helpful to identify tool wear state.In view of the limitation of single fractal,multifractal method is introduced to study cutting process signal from different measurement aspects.The multifractal characteristics and long-range correlation of signals were discussed by multifractal detrended fluctuation analysis method.The trend of multifractal characteristic parameters with tool wear was analyzed after getting the multifractal characteristic parameters.The multifractal characteristics influencing mechanism of cutting process signals are revealed by rearrangement and substitution of signal sequences.The results show that the signals of cutting process have long-range correlation,and some multifractal characteristic parameters are highly sensitive to tool wear.The multifractal characteristics of cutting process signal are mostly caused by the long-range correlation of small and large fluctuations of data.Finally,support vector machine(SVM),particle swarm optimization(PSO)and binary gray wolf optimization algorithm(BGWO)are combined to classify the tool wear state.BGWO method is used for feature selection and particle swarm optimization algorithm is used to optimize the parameters of support vector machine.The tool wear state is ultimately recognized by the trained support vector machine model.The results show that the tool wear monitoring model based on BGWO-PSO-SVM method and fractal theory features can get 93.3%and 98.7%recognition accuracy in the turning experiment and milling experiment respectively and BGWO has higher recognition accuracy compared with other common feature selection methods.