A FUZZY-NEURAL SYSTEM INCORPORATING UNEQUALLY IMPORTANT EXPERT OPINIONS FOR SEMICONDUCTOR YIELD FORECASTING

Abstract

Yield forecasting is a very important task to a semiconductor manufacturing factory. To enhance both the precision and accuracy of semiconductor yield forecasting, a fuzzy-neural system incorporating unequally important expert opinions is constructed in this study. In the proposed methodology, multiple experts construct their own fuzzy yield learning models from various viewpoints to predict the yield of a product. Besides, these expert opinions can also be considered unequally important. To aggregate these fuzzy yield forecasts, a two-step aggregation mechanism is applied. At the first step, fuzzy intersection is applied to aggregate the fuzzy yield forecasts into a polygon-shaped fuzzy yield forecast, in order to improve the precision of yield forecasting. After that, a back propagation network is constructed to defuzzify the polygon-shaped fuzzy yield forecast and to generate a representative/crisp value, so as to enhance the accuracy. For evaluating the effectiveness of the proposed methodology and to make comparison with some existing approaches, all approaches were applied to the practical data of three products in a real semiconductor manufacturing factory. According to experimental results, the proposed methodology improved both the precision and accuracy of semiconductor yield forecasting by 48% and 38%, respectively.