Application No./Patent No.: 2017107191446 Application date: August 21, 2017.
Inventor/Designer: Zheng Beirong, Xue Wei.
The invention discloses a method for predicting tool wear amount in milling machine machining process based on power detection, which mainly comprises a data acquisition step, a feature extraction step, a feature post-processing step and a prediction model step. Collect data through power signal collection and wear measurement, then extract the power effective value features and calculate the correlation to get the selected sensitive features, and then carry out order-preserving regression and exponential smoothing to realize feature post-processing. Finally, train samples, make SBL model prediction, and verify them after obtaining verification samples. According to the invention, accurate and efficient tool state self-identification and automatic early warning can be realized, the intelligent level of a machine tool is improved, the cost is effectively saved, and the efficiency is improved.
