A PNN self-learning tool breakage detection system in end milling operations

With the advance of technology over the years, computer numerical control (CNC) has been utilized in end milling operations in many industries such as the automotive and aerospace industry. As a result, the need for end milling operations has increased, and the enhancement of CNC end milling technology has also become an issue for automation industry. There have been a considerable number of researches on the capability of CNC machines to detect the tool condition. A traditional tool detection system lacks the ability of self-learning. Once the decision-making system has been built, it cannot be modified. If error detection occurs during the detection process, the system cannot be adjusted. To overcome these shortcomings, a probabilistic neural network (PNN) approach for decision-making analysis of a tool breakage detection system is proposed in this study. The fast learning characteristic of a PNN is utilized to develop a real-time high accurate self-learning tool breakage detection system. Once an error occurs during the machining process, the new error data set is sent back to the PNN decision-making model to re-train the network structure, and a new self-learning tool breakage detection system is reconstructed. Through a self-learning process, the result shows the system can 100% monitor the tool condition. The detection capability of this adjustable tool detection system is enhanced as sampling data increases and eventually the goal of a smart CNC machine is achieved. (C) 2015 Elsevier B.V. All rights reserved.