Abstract
Although a considerable number of studies have been made on the trouble shooting based on simple rules to control the manufacturing systems, little attention has been given to the point of trouble handling based on the higher strategy level. The purpose of this paper is to propose a new system, a strategy-based trouble handling system, that consists of a filter module, the trouble prediction system, and the trouble handling proposal system. First the filter module calculates several characteristic quantities with sensing data concerned with each trouble, then distinguishes them into three levels (L, M, H) by comparing the calculated quantities with standard ones. The trouble prediction system, which is constructed of association mode of the neural network algorithm, converts output of each level through the filter module into figure patterns. The system predicts the corresponding troubles from the figure patterns even if they are vague. The trouble handling proposal system suggests the handling instruction considering the situation of the whole system, which is used through the back propagation of the neural network algorithm. The evaluation tests were carried out in a NC drill machine. The results showed that the developed system could judge the troubles clearly and suggest handling instruction as exactly as mature engineers did.
