Identification Of Tool Wear State Of Gear Hobbing Machine Based On Vibration Signal Analysis

Tool condition monitoring is one of the core technologies urgently needed to be overcome in the intelligent development of CNC machine tools.Hobbing is currently the most widely used gear forming technology.Its core component hob is expensive and the manufacturing process is complicated.Excessive use will aggravate the hob wear and lead to reduced gear accuracy.The hob is scrapped and cannot be sharpened,which greatly affects the company.Production efficiency.Therefore,the monitoring of the wear state of the hob is of great significance for saving processing costs,ensuring processing quality,and improving gear production efficiency.To solve this problem,this paper collects the vibration signal of the hob spindle in the z direction and denoises it;extracts the characteristic value of the hob wear energy in the vibration signal based on Hilbert Yellow Transform(HHT);uses the IPSO algorithm to optimize the RBF neural network and build IPSO-RBF neural network model;finally realize the recognition of hob wear status.The main contents of the thesis research are as follows:Firstly,the mechanism of hob wear is introduced and the vibration signal of the hob spindle in the z direction is collected.The reasons and forms of hob wear and the method of formulating hob grinding standards are expounded;the characteristics of various signals and their advantages and disadvantages for recognizing the state of hob wear are analyzed;The relationship of hob wear proves that it can effectively reflect the wear state of the hob;an experimental scheme for collecting vibration signals is designed,and the signal collection work of its full life cycle is completed.Secondly,the noise reduction method of the hob spindle's z-direction vibration signal is designed.Based on the orthogonal experiment method,the parameters combination of wavelet threshold denoising are selected to obtain a more representative wavelet threshold parameter combination;according to the signal denoising effect evaluation method,the denoising effect of each combination is evaluated to obtain the optimal denoising parameters and use them for wavelet threshold denoising;draw and analyze the signal spectrum before and after denoising to verify that the method can effectively remove the high frequency noise of the vibration signal and retain its effective low frequency components.Then,the energy features that can effectively characterize the wear of the hob are extracted from the vibration signal.Aiming at the limitation of the traditional timefrequency domain method for analyzing non-stationary vibration signals,the HHT transform method with strong adaptability is used to perform EMD decomposition and Hilbert spectrum analysis on the vibration signals;the calculation screening has a strong correlation with the denoised vibration signals The energy of the IMF component and the marginal spectrum energy are used as the energy characteristic values that characterize hob wear;comparing the transformation trends of the eigenvalues at different wear stages of the hob,they are verified as effective as feature vectors.Finally,an IPSO-RBF neural network model for identifying the wear state of the hob was constructed.An improved strategy for standard particle swarm optimization(PSO)that is easy to fall into local optimization is proposed,and the improved PSO algorithm is used to optimize the RBF neural network to obtain a recognition model;the full life cycle vibration signal energy characteristic value of the input hob is verified that the model can be effectively recognized Hob wear state;design comparison experiments to verify that the IPSO-RBF neural network model is far superior to other recognition models in recognition rate and recognition time.