Use of redundant sensory information for fault isolation in manufacturing cells

The competition and high precision requirements in manufacturing processes have forced companies to develop and implement automated manufacturing cells on the factory floor. However, due to the high precision requirements of these cells, small changes in the cell components, such as bearing wear, clamp wear, and sensor loss of sensitivity, can degrade their performance such that they fail to function properly. Furthermore, on-line direct measurement of the condition of cell components is very difficult. This paper presents on-line and off-line algorithms that use redundant sensory information to determine the condition of these cell components without directly measuring them. The effectiveness of these algorithms has also been tested and confirmed on the factory floor. This work suggests that for higher productivity, high precision manufacturing cells should have self-diagnosis capability to predict which parts need repair or maintenance