Cutting tool life management in turning process: a new approach based on a stochastic wear process and the Cox model

In machining processes, a significant cost contribution is related to the cutting tool insert. A precocious replacement leads to lesser profitability of the cutting tool while a late replacement tends to produce more scraps due to advanced wear of the cutting tool insert. To optimize the replacement times, there is a need to develop an integrated monitoring framework to assess the wear of the cutting tool for different cutting conditions during the machining process and to predict the remaining useful life. The aim of this paper is to propose a complementary approach based on a gamma process to model the stochastic behaviour of the flank wear evolution and a Cox proportional hazard model to consider different cutting conditions. Experimental data measured in turning is used to fit a piecewise stationary gamma process on 29 cutting tool inserts using identical cutting conditions. Another set of data is used to fit a Taylor law to take into account different cutting speeds. The piecewise stationary gamma process is then adapted to simulate random flank wear paths for a defined cutting speed range. Using this model, several cutting tool lifetimes are simulated and used to feed a Cox proportional hazard model. The fitting procedure relies on a learning phase and a control phase to ensure the accuracy of the model. The results of both models are then discussed, and the robustness of the Cox Proportional Hazards Model to noise in the data is assessed.