(A) study on the monitoring system for intelligent turning operation

Abstract

According to the increase in demand for automated and unattended operation, it results that each machine tool must be used more effectively rather than whole machining system. In this thesis, the intelligent machine tool is defined by comparison with an intelligent human machinists. In order to realize the intelligent machining system, the skilled machinist``s know-how should be replaced by machine controller. One of the most important roles of skilled machinist in conventional machining operation is to monitor the machining process. The ability for detection and diagnosis of errors in a manufacturing system is necessary to build the autonomous controls for machine tools and factory systems. In this thesis, not only the signal processing methods for individual monitoring object but also the system integration using neural networks are focused to realize the in-process and real-time monitoring system in turning operation. The cutting state is classified into five groups as one normal and four abnormal cutting states(tool fracture, tool wear, long chip formation and chatter). The acoustic emission sensor and the tool dynamometer are used to detect the abnormal cutting states. In order to detect the tool fracture in turning process, the feasibility of using acoustic emission and cutting force signals for the detection of massive tool breakages as well as small fractures of cutting tool are investigated. The fracture of cutting tools is successfully detected through the analysis of these dual signals in the several types of tool fracture. The normalized disparities are defined by the use of modeling of static force component to detect the flank wear. The dynamic fluctuations of cutting force is analyzed to detect crater wear, and the increase in magnitude of dynamic fluctuations is originated from the formation of built-up edge. These relationships turn out to be proper signal features to detect crater under the various cutting conditions. The envelope signal o...