Wear Prediction of Woodworking Cutting Tools based on History Data

Abstract The tool lifespan is among the most contributing factors in machining economics. Accurate lifetime prediction maximizes the utilization of each tool and the machine tool. In order to predict the remaining tool life accurately the past has to be captured continuously. In this paper the capturing of tool usage data and its analysis is explained. With the captured data tool operation circumstances can be calculated. These are mainly responsible for tool wear. This circumstances data is captured by machine tool integration and combined with production planning data. The generated context of the machine tool information of the covered cutting path is combined with the material used. The combination results in specific tool wear. This cutting circumstances are logged in the tool history data. The cutting conditions are described with the material used, feed rate per tooth, cutting width, cutting speed, cutting path and number of revolutions. A model has been developed which describes the steps in detail. A forecast algorithm uses the history data to predict the occurring tool wear and end of life of the cutting tool. Input for the calculation are the cutting path, cutting circumstances, tool type and material type. With this information it is possible to forecast the remaining tool life by using a specifically tailored machine learning algorithm. The prediction will become more accurate after each learning cycle. In this paper the algorithm is explained and the generated characteristic diagram is displayed for each tool and its tool life.