Real-time adaptive fault detection and isolation on a high-speed multi-injection molding machine

This paper discusses a data driven fault detection and isolation problem. An adaptive diagnostic module that allows real-time control of all parts produced by an injection molding machine is presented. The approach is based on a structural pattern recognition with a multi-signature system taking into account the qualitative and quantitative aspects of the sensor signals. Detection of unknown defects is also considered. Experimental results obtained on the production machine are illustrated in the last section.