Study On Tool-Wear Recognition Based On Fractal Theory

In milling process, surface quality and dimensional accuracy of the workpiece is affected by cutting-tool wear condition. To survey and define cutting-tool wear exactly, it is an important problem in automation production to be urgently solved at present. Cutting-tool wear is a very complex process. Various factors have a direct or indirect effect on cutting-tool wear, resulting in uncertainty. So it is difficult for experimental data and result to have good stability. Further studies have shown that cutting-tool wear is non-linear, random and dissipated. There are a large number of non-stationary signals in the fields of condition monitoring and fault diagnosis for cutting-tool. Researching and developing effective engineering methods for processing non-stationary signals are necessary for promoting sustained development of fault diagnosis technology.Vibration analysis is a very important means for condition monitoring and fault diagnosis. This paper aims at the research on the methods of signal processing and pattern recognition. Therefore, firstly, an experimental platform was set up for the failure diagnosis of cutting-tool, on which we have done a lot of experiments. Then the vibration signals of cutting-tool in different wear conditions were collected. Wavelet noise reduction is used to pretreat the vibration signals. The method of fractal recognition can diagnosticate the wear degree quantificationally.According to the non-stationarity of vibration signals, we introduce the de-noising method based on wavelet decomposition and reconfiguration. The influence on denoising effect is compared by using different mother wavelet and wavelet threshold rules. For the test data of this thesis, the denoising effect adopting the mother wavelet 'db4' and the default threshold rule is the best.This paper introduces the algorithm of box dimension and information dimension from the point of view of engineering application. The influence of noise on the computational precision of box dimensition is discussed.The algorithm of correlation dimension (G-P algorithm) is interpreted and the method to choose parameters is given. Time-delay and Minimum Embedding Dimension have obtained by using mutual information method and Cao's method respectively. It avoids choosing the parameters blindly and confirms the precision of correlation. The validity of algorithm is validated by sin signal, Lorenz signal, FBM signal simulation, and we also use it to cutting-tool condition monitoring. With the increase of cutting-tool wear, the wear between cutting-tool and workpiece gets sharp and the waveform of vibration signal changes irregularly, then the fractal dimension increases gradually. Though box dimension and information dimension have small changes, the trend is clear. Correlation dimension has great changes. Correlation dimension of the new cutting-tool is minimum, broken-down cutting tool's correlation dimension is visibly increasing. The result of the experiments shows the changing rules of the fractal dimension about the vibration signals in the whole course of tool wear. The variance of the fractal dimension, which reflects the geometric characters of the vibration signals, has the same tendency as that of the flank wear. The tool wear monitoring can be realized effectively by using the fractal dimensions as the feature of the vibration signal.Simulation and practical application shows that fractal recognition provides a simple way to describe the normal and abnormal state of the vibration signal. It can identify the different tool wear states quantificationally. So, it is more intuitionistic than traditional qualitative pattern recognition methods.