Self-tuning of Teachless Process Monitoring Systems with Multi-criteria Monitoring Strategy in Series Production

Abstract

Modern monitoring systems in machine tools are able to detect process errors promptly. Still, the application of monitoring systems is restricted by the complexity of parameterization for save monitoring. In most cases, only specially trained personnel can handle this job at multi-spindle machines or turn-mill centers. The aim of the research project “Proceed” is to figure out in which extent a self-parameterization and independent optimization of monitoring systems in industrial series production can be realized. Therefore, the complete parameterization of the processing chain, consisting of the choice of signal sources, character extraction, the monitoring- and decision making strategy, shall be automated. This paper deals with the self-parameterization of a multi-criteria monitoring system based on a genetic algorithm.