Experimental evaluation, modeling and sensitivity analysis of temperature and cutting force in bone micro-milling using support vector regression and EFAST methods

In orthopedic surgeries, examining the effect of machining conditions are very crucial due to the possibility of damage to the bone tissue, and the occurrence of thermal necrosis. Considering that a comprehensive and methodical study based on the design of experiment method, artificial intelligence modeling and sensitivity analysis can complement the previous investigations, so we found this motivation to explore the behavior of the two important variables of cutting force and temperature in the cortical bone micro-milling operation. Accordingly, in this paper, the effect of micro-milling conditions, including feed rate, tool rotational speed, tool diameter, and cutting depth on cutting force and temperature have been investigated simultaneously. For this purpose, based on the design of experiment technique, the set of experimental tests was performed on fresh cortical bone. Next, a machine learning technique known as support vector regression (SVR) is used to model and predict the force and temperature in the machining procedure. Subsequently, based on the attained model, a big dataset is prepared, and by utilizing the extended Fourier amplitude sensitivity test (EFAST), the effectiveness of the output variables on the input parameters was determined. The SVR predictor achieved the root mean square error of (0.329, 2.133), the mean absolute error of (0.277, 1.820), the mean absolute percentage error of (9.266, 5.353), and the correlation factor of (0.971, 0.904) for (force, temperature). Also, it is realized that, feed rate, rotational speed, cutting depth, and tool diameter, affect the (force, temperature) by (46%, 8.9%), (34%, 9.2%), (15%, 1.5%), and (4%, 80.3%), respectively. By using the attained model of bone micro-milling operation, the surgeons can predict the temperature and force under different machining parameters before the surgery, consequently choosing the optimized machining conditions to reduce the temperature and cutting force.