Monitoring and control of machining processes using neural networks

In this dissertation, a study on the acoustic emission (AE) from orthogonal metal cutting process is done based on Kannatey-Asibu and Dornfeld's work (61) and is further extended to both oblique and three dimensional single point metal cutting processes. The study is particularly concentrated on providing a theoretical background for the proportionality between the energy content of AE and the power dissipated in metal cutting process, and on deriving relations for theoretical RMS AE level. For orthogonal cutting, a new expression for the RMS AE level which is not a function of the chip-tool contact length, and the length of sliding friction on the rake face, is developed from Kannatey-Asibu and Dornfeld's model (61), and is experimentally evaluated. In the experimental evaluation of reasonably good correlation with the theory for the variations in cutting speed, or depth of cut was observed.; In this study, an on-line tool wear detection system for turning operations is developed, and experimentally evaluated. The results of experimental evaluation show that the system works well over a wide range of cutting conditions, and the ability of the system to detect tool wear is improved due to the generalization, fault-tolerant and self-organizing properties of the neural network. It is also demonstrated that the performance and the reliability of tool wear detection can be significantly improved by fusing information from force and AE sensors using the neural network structure.; In this dissertation, the feasibility of using an adaptive resonance network (ART2) with unsupervised learning capability for tool wear detection in turning operations is investigated. The experimental results show that tool wear can be effectively detected with or without minimum prior training using the self-organization property of the ART2 network.; The final topic of this dissertation is control of machining process using the multilayered perceptron type neural networks. The results of the experimental evaluation manifest that the input/output relationship of the machining process can be effectively simulated with the neural networks and the adaptive optimal control system based on the neural network works reasonably well. (Abstract shortened with permission of author.)...