Research On Milling Chatter Detection Based On Fourier Decomposition Method And Rbf Neural Network

Milling is a commonly used processing method.When the cutting parameters are not properly selected,chattering occurs easily,which seriously affects the surface quality and processing efficiency of the workpiece.If the milling state can be detected in order to take timely measures to suppress chatter,it can avoid causing greater losses.In response to this problem,this article conducted the following research from the aspects of experimental design,signal analysis,feature recognition and software development:Based on the milling dynamics model,by solving the dynamic equations,the stability lobe diagram was drawn.Based on this,milling experiments with different machining parameters were designed,vibration acceleration signals were collected,and data sample sets for different milling states were established.The vibration signal characteristics under the flutter state were analyzed.Considering the non-stationarity of the milling vibration signal,a new time-frequency analysis theory is introduced,namely Fourier decomposition method(FDM).The decomposition results of the simulation signal show that,FDM can effectively suppress modal aliasing and over decomposition phenomenon compared with EMD.The milling vibration signals are extracted in the time domain,frequency domain and time-frequency domain respectively.In view of the poor accuracy of the detection method using a single feature to set the threshold,a milling chatter detection method based on radial basis neural network is proposed.The Pearson correlation coefficient method was combined with the test results of the algorithm model to filter the feature quantity,and the threedimensional feature vector was input to the radial basis neural network for training and testing.The test results show that the method can detect the milling state and has a high Classification accuracy.Designed and developed the milling chatter detection software,which realized the functions of data acquisition,waveform display,milling state monitoring and feedback adjustment of processing parameters.