Analytical Target Cascading Based On The Quadratic Exterior Penalty Function And Kriging Model

The complex engineering product usually involves multiple disciplines which arecoupled strongly. There exists the computational complexity and the complexity of theinformation organization, which causes that Multidisciplinary Design Optimization (MDO)emerges as an enabling methodology for the design of complex engineering products. Basedon MDO, the complex engineering product is decomposed into several subproblems that canbe solved more easily. Then, the decomposed subproblems are coordinated by somecoordination strategy. The design optimization of the original problem is done through theoverall decomposition-coordination process. MDO method (so-called MDO strategy)completes the problem design optimization based on the decomposition-coordination. Inrecent years, a few MDO methods have been put forward and applied. Among them,Analytical Target Cascading (ATC) method is a relatively new MDO method bydecomposition and collaboration. In this paper, because of complicated coordination processand high computational cost of pre-existing coordination strategy in ATC, a new coordinationstrategy based on Quadratic Exterior Penalty Function (QEPF) is proposed. Moreover, theapproximate model can be used to replace complex, implicit unknown function or theexpensive simulation analysis to significantly reduce the computational expense. It plays animportant role and has good application in Multidisciplinary Design Optimization. So theapproximate model is employed to improve the optimization efficiency and reduce theenormous computational cost. The main works are as follows:Firstly, ATC method based on quadratic exterior penalty function is proposed. On thebasis of the comparative analysis of four MDO methods, Analytical Target Cascading that hasthe comprehensive performance is deeply studied. Due to the large computation of theexisting coordination strategies in ATC, the coordination strategy based on quadratic exteriorpenalty function is employed, whose coordinated process is simpler to make up for theshortage. And, the geometric programming problem is devoted to verify the usability andeffectiveness of ATC-QEPF method.Secondly, the approximate model technology is introduced into ATC-QEPF method.ATC-QEPF method based on Kriging model is put forward after the comparison of approximation models. This method not only make a replacement of simulation analysismodel to avoid calling simulation model, but also can avoid the effect of numerical noiseduring the iterative process of the design optimization. And the method is well confirmed bythe gear speed reducer problem, which proves the method can reduces the computational costand greatly improve the design optimization efficiency under the condition of guaranteeingthe accuracy.Finally, based on the aforementioned research, the proposed KATC-QEPF method isapplied to the conceptual design optimization of a aircraft. The ideal design results is obtainedusing the method for the general design of the aircraft, which demonstrate the KATC-QEPFmethod is an efficient and feasible MDO method that can complete multidisciplinary designoptimization of the complex engineering product system more rapidly and accurately.