Research On Tool Wear State Of CNC Lathe Based On Improved EEMD

Under the current industrial system,the manufacturing industry is still an important reference indicator for a country's comprehensive competitiveness.With the increase in manufacturing costs and the requirements for high-quality and high-precision processed products,the manufacturing industry has undergone a technological revolution from traditional manufacturing to intelligent manufacturing.The tool plays a vital role in the processing of products.The wear state of the tool affects the precision and quality of the processed products.If the wear state of the tool is not identified clearly in the process of machining,the machining materials will not be fully utilized or the accuracy and quality of the product will be seriously affected,resulting in irreparable losses.In this paper,the vibration signal is used as the monitoring signal to identify the different state of tool wear in the process of machining.The main work of this paper is as follows:Firstly,based on the monitoring signal type selected in the experiment,a signal acquisition system is built by using sensors,constant current adapters,data acquisition cards and other equipment to collect the vibration signals of different tool wear states.In view of the fact that the acquisition quality of vibration signals is related to processing parameters,the influence of sensor installation position and spindle speed on the signal acquisition quality is analyzed,through the analysis of the experimental signal,the final installation position of the sensor and the spindle speed are determined.Secondly,in view of the shortcomings of ensemble empirical mode decomposition(EEMD),such as more false components and low decomposition efficiency,the model of optimizing EEMD by bacterial foraging algorithm(BFA)is proposed.The simulation signal and experimental signal are decomposed and compared by using the model before and after optimization.The advantages of the optimization model are verified from the number of components,reconstruction error and decomposition efficiency.Finally,using the ability of hidden Markov model(HMM)to build statistical model based on time series,the model of tool wear state recognition based on HMM is established;the model of BFA-EEMD combined with HMM and the model of EEMD combined with HMM are used for training and recognition comparison,and the results show that the model of BFA-EEMD combined with HMM is more efficient in the training process,and it is more effective for tool wear state recognition.The results show that the recognition accuracy of wear state is higher,and the effectiveness of the proposed method is verified.