Application of fuzzy set theory and back-propagation neural networks in progressive die design

Abstract Fuzzy set theory is introduced to process the design details of the uncertainty portion of die design and assist the designer in transforming design items with fuzziness into definite and reasonable design attributes. For design parameters in die design that possess intermediate features, fuzzy cluster analysis is used to obtain design attributes. For theoretical or empirical formulas possessing uncertainty coefficients ranges or preference design parameters, the fuzzy weighted average method is adopted to obtain the feature parameters that conform to die design requirements. For the design of uncertain operation stations, such as the installation of an idle station, the linguistic fuzziness or linear membership function is matched with the network recall value and learning pattern through neural network learning, and then the designer can decide whether it is necessary to install such an operation station. Finally, this study establishes an expert system prototype to combine the uncertainty problems in three kinds of die design and help the designer obtain a definite design strategy while faced with uncertain design items.