Study On Intelligent Diagnosis Technology Of Tool Wear Based On Bispectrum, Fractal And Neural Network

In milling process, the quality of products is affected by cutter wear condition. To survey and define tool wear exactly is an important subject that needs to be solved urgently both inside and outside the nation in automatic production at present. Tool condition monitoring is a key technique in automatic and unmanned machining process. Therefore, the prediction of tool wear and failure is very urgent.In accordance with the non-stationarity of the acceleration and AE signal due to the progressive wear of the cutter, a method for extracting the cutter wearing feature by bispectrum analysis is presented. The characteristics of the bispectra of different wear stages are analyzed. Experimental results show that the method is effective and suitable for milling cutter condition monitoring. The bispectra can be used to identify the non linearity of a system. It is very useful to find out that the bisoectra and their contours for the similar working conditions have some similarity while for different working conditions exist some different characteristics. The method proposed in the paper provides a useful tool for condition monitoring of working machinery.In this paper, we set up an experiment system of the milling tool wear monitoring and collect a variety of fault date using vibratory sensor. On the basis of the theory of fractal geometry the paper presents a new method of describing signals complexity using fractal dimension according. From the point of view of engineering application, it introduces the algorithm of box dimension. It also researches into the changing rules of the box dimension about the vibration signals in the whole course of tool wear through experiments. 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. The result of the experiments showed that the monitoring can identify the different tool wear states more correctly under different cutting conditions.By analyzing and dealing with the vibration signal and AE signal based on bispectrum analysis and fractal dimension, comparing their strongpoints and weakpoints , Intelligent Diagnostic System is achieved by the model identify function of neural network. We mostly apply Elman network of ANN to show the cutter'states.