Application of a multi-stage MQL system for tool life prediction in Inconel-718 machining

Minimum quantity lubrication (MQL) is an important cutting technique in modern times. Cutting fluids primarily serve lubrication and cooling functions. Different oil-to-water ratios can improve cutting temperature and lubricating ability, both of which are crucial factors affecting tool life and machining quality. Therefore, this study developed a device which can adjust the minimum quantity lubrication oil-to-water ratio in real time to enhance tool life. In the first phase of the study, cutting vibration and cutting temperature values were measured using a three-axis accelerometer and a self-developed wireless temperature measurement tool holder, and the chip surface color was captured and analyzed using a charge coupled device (CCD) industrial camera to investigate the auxiliary effect of different oil-to-water ratios on machining. In the second phase, two different neural networks, a back propagation neural network (BPNN) and a general regression neural network (GRNN), were used to model and predict tool wear. The machining vibration, temperature, and chip chromaticity were used as input, and multi-feature fusion was employed to enhance prediction accuracy. Finally, the mean absolute percentage error (MAPE) of BPNN and GRNN in tool wear prediction was 3.94% and 3.77%, respectively. The prediction error percentages are within 10%, representing highly accurate prediction. The experimental results show that when the oil-to-water ratio is 75%:25%, the tool life can be improved by 223%, and an appropriate water content has a significantly inhibitory effect on cutting temperature and vibration.