In recent years, production automation has been the focus of this research endeavor to improve product quality and to increase productivity. Implementation of computer-based untended machining has attracted great attention in the manufacturing community. For a successful implementation of untended machining, a better understanding of the machining processes and the functions they perform is required. This necessitates the development of sensors and intelligent decision-making systems.

In this thesis work, a framework for sensor-based intelligent decision-making systems to perform on-line monitoring is proposed. Such a monitoring system interprets the detected signals from the sensors, extracts the relevant information, and decide on the appropriate control action. In this thesis, emphasis is given to applying neural networks to perform information processing, and to recognize the process abnormalities in a machining operation. For signal detection, an instrumented force transducer is designed and implemented in real time turning operation. A neural network program based on feedforward back-propagation algorithm is developed. The program is tested by the simulation data and varified by the experimental data. The superior learning and noise suppression abilities of the developed program enable high success rates for monitoring the tool wear and surface roughness under machining of advanced ceramic materials.

It is evident that the development of hardware neural networks provides the monitoring system with fast computational capabilities. The advances in sensor technology enables inexpensive sensors to be easily mounted for monitoring the machining process. All these evidences justify that neural networks will become an attractive modeling tool for use in on- line monitoring of machining processes in near future