Multi-disciplinary Optimization Method And Its Application On Hydro-generator Design

With the development of science, the design of large mechanical productdevelope from design of part to design of system, design requirements developefrom single obejctive to multi-objective, fields involved in design develope from asingle field to multiple fields.Hydro generator, as one of large mechanical product, its design involvedmechanical, electronic, control and it is a collaborative design process, designerfrom different fields are needed to complete the design task by mutual cooperation.The design process of hydro generator requires establishing an approximate modelfor the electromagnetic fields, decoupling the Coupling Model of mechanical andelectrical fields, what’s more multi-objective needs to be optimized.Two main approximation modeling methods, Polynomial approximationresponse surface modeling techniques and approximation modeling techniquesbased on support vector machine, are applied in this article to obtain theapproximation model of implicit design model. For implicit design model with lowdimension design space and a small number of variables, this article appliedPolynomial approximation response surface modeling techniques to obtain theapproximation model, for implicit design model with high dimension design spaceand a small number of variables, this article applied approximation modelingtechniques based on support vector machine to obtain the approximation model. Inthis article, the approximation modeling of the cost of materials andelectromagnetic field were carried out.For the coupled design problems with the approximation model obtained, thisarticle applied two methods, decoupling optimization technology based onminimum variance and multidisciplinary decoupled optimization technique basedon genetic algorithms. In this article, the coupling model of hydro generator statorweight is decoupling optimized.For the multi-objective optimization problem in the design of hydro generator,this article proposed a new optimization method named multi-objectiveoptimization based on function feature set synthesis, the method is to synthesizethe multi-obejctive function into one, and then using the Single-objectiveoptimization approach to solve the problem. Algorithm validation and examplesvalidation for the proposed algorithm are given in this article.On the basis of theoretical research, this article develop a multi-discplinaryoptimization system of hydro generator. Several modules are developed to solvethe Approximate modeling, decoupling optimization and multi-objectiveoptimization problem.