Abstract: Aircraft preliminary configuration design is a complicated engineering design. Although the fuzzy set theory and an expert system have been applied to it, both of them have their own difficulties, i.e., the specification of membership function for a fuzzy set, and the knowledge acquisition bottleneck and the low inference efficiency in the expert system. In this paper, artificial neural networks (ANNs) are applied to the intelligent CAD of aircraft preliminary configuration designs. Parametric neural network and WAFS (Weight, Activation Function and Structure) learning algorithm are proposed to overcome the slow learning speed and hard determined network size problems in the back propagation neural network. Membership function representation by ANN and parallel reasoning methods based on ANN are studied. Some examples in aircraft preliminary configuration designs are presented to demonstrate the effectiveness of the proposed ANN methods.

Key words: ar tificial neur al networ k, aircr aft preliminar y configur ation, intelligent CAD