The study of a self-adaptive evolutionary programming and using in engine CBR design system

Abstract

In this paper, a improved self-adaptive evolutionary algorithm is researched and adopted in CBR engine design system. Because attribute weight of CBR system can obviously influence the accuracy of case-search, the evolutionary programming algorithm is used to get attribute weight for adapting to changes of requirement. This paper analyses the coding method of chromosome, educes the adaptability function based on reference case base REF and testing case base TEST, develops a new self-adaptive aberrance method that can suit the changes of adaptability function and overcome the lack of conventional evolutionary programming algorithm. This paper also gives the algorithm procedure of self-adaptive evolutionary programming and ensures self-adaptive evolutionary programming algorithm possesses the furthest dispersive search ability during the solution space and the furthest accurate search ability during the local situation. Experiment results prove the validity of self-adaptive algorithm and CBR design system is used successfully in engine design process. (6 pages)