| With the rapid development of manufacturing industry, CNC machine tool is playing an increasing important role in the manufacturing industry. At the same time, the probabilities of failure are also greatly rising, which results in massive losses. Therefore, it will be rather significant and urgent to have a method of health status evaluation of CNC machine tools.This paper mainly focuses on a new method of health evaluation and the application to the operational health evaluation in machining process. And this integrated method is based on the spindle current, the extraction of spindle vibration signal processing. Specific work is done as follows:First of all, the wavelet packet analysis is used to decompose the spindle current and spindle vibration signal in multi-levels. After that, we combine data mining algorithms such as the fisher discriminant analysis, principal component analysis and the improved K-means algorithm and so on to extract basic characteristic with the energy value of the signal at the bottom as the basic characteristics.Then, we carry on the health evaluation model training of the extracted feature information with the method of logistic regression and improve logistic regression algorithm with the method of iterative averaging. Ultimately a multi-state regression model is generated.Finally, the experiment is done on XHK715 machining centers to train and validate the multi-group experiments in different configurations. The experimental results confirm the effectiveness of the health model and reveal that the more the training state is, the more accurate the model is and thus the more accurate the health assessment results will be.In this paper, both theoretical derivation and experimental verification has proved the validity and effectiveness of health assessment methods. This method can also be applied to other parts of the machine tool as well as other similar machine tools to have the health assessment. |
