Fast Tool Breakage Monitoring: A Decision Level Fusion-Based Approach Using Vibration and Sound Signal

Abstract

Abstract. In the process of high-speed machining for difficult-to-cut material, rapid tool wear on one hand and tool breakage on the other hand are obstacles for unmanned machining. Sudden tool breakage is difficult to predict in advance compared to the slower tool wear process, but a rapid detection of tool breakage can save the machine tool and parts from further damage. In automated machining systems, tool breakage monitoring and correction in the shortest time is of paramount importance. However, few studies have considered a fast-time reaction in tool breakage monitoring to avoid catastrophic shank fractures. This technical brief develops a millisecond time response tool breakage monitoring approach based on the sensor fusion decision. The fast response is achieved by combining (a) sensitive indicator construction and (b) fusion alarm decision. An abrupt change point identification model for the amplitude fluctuation in monitoring indicators is introduced. Experimental results show that the new monitoring approach based on both vibration and sound signals greatly reduces false alarms using dual checks, and response time is only one-tenth of traditional spectrum-based methods. The shortest window size is discussed for tool breakage monitoring, and the monitoring response time is that no more than one rotation of the milling tool is revealed. The validation experiments conducted with two cases of tool breakage data under different cutting parameters and tool diameters demonstrate that this method exhibits strong adaptability to varying cutting conditions.