Learning Models in Dynamic Manufacturing

Abstract

In a dynamic environment, continuous update of the operational knowledge is neces sary. A computer learning technology is needed to capture and update that knowledge. When the relevant knowledge of the production process and equipment is available and remains relatively stable, the expert system technology can be applied to develop an intelligent support system for monitoring, diagnosis, and recovery processes. However, many existent learning algorithms are in applicable to this case because of their slow learning rates. To improve the performance, appropriate incorporation of a human expert is required. This paper proposes such a computer learning model and describes how an expert should play the role in assisting the computer. Simulation studies using data from an actual paper making factory are carried out to compare performances of two learning strategies employing different levels of the expert's assistance.