Knowledge representation in machine tool supervision systems

A deeply coupled numeric-symbolic machine tool supervision system for tool breakage detection is developed. The supervision system hierarchically integrates real-time signal-processing algorithms in a knowledge-based processing environment where rules and objects coexist. A numeric-symbolic model that incorporates physical models and empirical knowledge is developed. An application to tool damage detection in milling is described, with emphasis on the integration of symbolic and numeric processing. The decision strategy for this problem is described, and a validation study is presented