Research On Tool Wear Prediction Based On Cointegration Modeling

During the machining process, the cutting tool gradually wears out and the blunt tool will inevitably affect the surface quality and decrease the processing efficiency. Therefore, it is essential to monitor the tool wear condition. The artificial intelligent (AI) method and statistics method are the most frequently used methods for tool wear monitoring. However, the training process of the AI based methods usually need plentiful samples and the process for obtaining these samples is very costly and time consuming, while the multiple linear regression (MLR) method often leads to the spurious regression to make the predicted result inaccurate.In this paper, the cointegration method is applied to construct the model between the signal features and tool wear so as to predict the tool wear. The cointegration theory and modeling process are firstly introduced in this paper and the difference between the cointegration method and MLR method is investigated by case analysis. Side milling experiments of TC4 were then carried out to testify the effectiveness of the proposed model and the cutting force, vibration and acoustic emission signals are analyzed. Features correlated with the tool wear are selected for model construction. Finally, the cointegration model is constructed to reflect the relationship of tool wear and signal features by a series procedures such as unit root test and Johansen test. The cointegration model is found to predict the tool wear accurately and distinguish the existence of spurious regression by comparison with MLP method. In addition, the general regression neural network and wavelet neural network are also applied for tool wear prediction and are found to be able to provide satisfied result as well. But its ability to predict samples out of range of the modeling samples is poorer than the cointegration method.The research in this paper demonstrates that the cointegration model can predict the tool wear accurately and efficiently. This application has positive significance for improving the surface quality and the machining efficiency.