An Adaptive Prediction Approach to Machining Operation. A Concept of Adaptive Prediction and Application to Evaluation of Tool Failure.

Adaptive prediction approach is proposed for autonomous operation planning system, which can be installed on each machine tool. This approach enables us to predict machining troubles more accurately by adaptation of characteristics for analysis. This approach is applied to evaluation of tool failure in interrupted cutting. In adaptive process of tool failure, fracture probability distribution is modified by Johnson's ranking approach, which takes data of unbroken tools into account. Then fracture characteristics of tool are obtained by minimizing prediction error using conjugate direction method. In prediction process, tool failure is evaluated using updated characteristics obtained in adaptive process. Fracture characteristics finally obtained agree sufficiently with experimental result. In order to plan machining operation after adaptive prediction, neural network is constructed for rapid prediction of tool failure.