In-process complex machining condition monitoring based on deep forest and process information fusion

Abnormal machining condition causes losses of quality for finished part. A machining condition monitoring system is considerably vital in the intelligent manufacturing process. Existing machining condition monitoring methods usually detect only one single abnormal condition under the same machining process, which is unrealistic and impractical for real complicated machining process. In this paper, a novel hybrid condition monitoring approach for multiple abnormal conditions' detection of complicated machining process by using deep forest and multi-process information fusion is proposed. First, various process data are obtained from a triaxial accelerometer and a sound sensor mounted on the spindle of CNC. Then, the time domain, frequency domain, and time-frequency domain features extracted from the multiple sensory signals are simultaneously optimized to select a subset with key features by the lasso technique. Furthermore, deep forest is utilized as a condition classifier by using the selected features. Finally, cutting experiments are designed and conducted, and the results show that the proposed method can effectively detect the multiple abnormal conditions under the different machining parameters.