Abstract:Purpose - The purpose of this paper is building the regression model related to tool wear, and the regression model is used to identify the state of tool wear. Design/methodology/approach - In this paper, genetic programming (GP), which is originally used to solve the symbolic regression problem, is used to build the regression model related to tool wear with the strong regression ability. GP is improved in genetic operation and weighted matrix. The performance of GP is verified in the tool vibration, force and acoustic emission data provided by 2010 prognostics health management. Findings - In result, the regression model discovered by GP can identify the state of tool wear. Compared to other regression algorithms, e.g. support vector regression and polynomial regression, the identification of GP is more precise. Research limitations/implications - The regression models built in this paper can only make an assessment of the current wear state with current signals of tool. It cannot predict or estimate the tool wear after the current state. In addition, the generalization of model has some limitations. The performance of models is just proved in the signals from the same type of tools and under the same work condition, and different tools and different work conditions may have influences on the performance of models. Originality/value - In this study, the discovered regression model can identify the state of tool wear precisely, and the identification performances of model applied in other tools are also excellent. It can provide a significant information about the health of tool, so the tools can be replaced or repaired in time, and the loss caused by tool damage can be avoided.

Application of Genetic Programming in the Identification of Tool Wear