| Tool wear is one of the most important factors affecting machining accuracy.The wear process contains complex physical and chemical changes.Therefore,how to effectively monitor the condition of the tool is of great significance to ensure normal production.Aiming at the above problems,this article collected vibration signals at different positions in the machining area,extracted signal characteristics that can characterize different wear conditions and merges them,and finally conducts wear state recognition research.The main research contents of this paper are as follows:(1)Set up a tool wear monitoring test platform,on this basis,performed turning tests and collected vibration signals,used single factor analysis to study the impact of the three elements of cutting on tool wear rate and workpiece surface roughness.(2)Used the wavelet packet decomposition method which based on the cost function and the multi-resolution SVD package to process the vibration signal and extract the ratio of energy to the total energy.The comparison results show that: the Multi-resolution SVD package is better.(3)Analyzed the ability of three feature fusion methods to reflect the different wear states of the tool,the results shows that: the second way of fusion is the best.In addition,the effect of feature fusion on the accuracy of tool wear status recognition is compared and analyzed.The results show that the recognition accuracy is significantly improved after feature fusion.(4)Researched and analyzed the Bagging(Guided aggregation algorithm)and the LVQ(Learning Vector Quantization)network,constructed and trained the Bagging-LVQ neural network model for tool wear status recognition.Used Bagging-LVQ and traditional LVQ neural network to compare,all tests in this paper are carried out to monitor and identify.The results show that the accuracy of the Bagging-LVQ neural network in this paper is higher than that of the traditional LVQ neural network. |
PDF Full Text Request