Method And Application Of Winged Missile Oriented Multidisciplinary Robust Design Optimization

The thesis is financially supported by the National Doctoral Disciplinary Science &Research Foundation under Grant No.20010487024, the theory and application research ofbionic evolvement, and the research work of winged missile (WM) conceptual designproposed by the third academy of China Aerospace Science and Industry Corporation(CASIC). According to the requirement of multidisciplinary robust design optimization(MRDO), some key techniques, such as decomposition and collaboration strategy, robustoptimization design (ROD), uncertainty analysis, the integration of multidisciplinary designoptimization (MDO) and ROD, and so on, have been addressed. According to the analysis of function and performance of WM, the mathematical modelsin the fields of simultaneous dynamic and structural optimization, simultaneous aerodynamicand concealment optimization, and simultaneous flight and control optimization have beenformed. Some general sources of uncertainties in design are analyzed and the theory ofMRDO is put forward. Based on this, a general mathematical MRDO model of WM isestablished through robust adjustment of the optimization design model of WM. To solve the optimization design problem of WM system, a decomposition approachbased on artificial neural network (ANN) and genetic algorithm (GA) is proposed forpartitioning the large-scale missile system design problem into smaller, more tractablesubsystems. The optimal partitioning scheme are implemented and evaluated on a gearreducer problem and a WM problem, respectively. It proves that the decomposition strategybased on artificial intelligence is a potential and efficient method. In the MDO of WM, there are two challenges: 1) information exchange among coupledsubproblems; 2) complexity of system analysis. The use of ANN-based response surface (RS)approximations in MDO is proposed. The approach provides some mechanism forcoordinating the overall design problem and progressing toward improved designs. Theapproximate optimization is a two-level optimization architecture, with the discipline levelonly satisfies the local constraints and the system level ensures that all of the disciplinedesigns agree on a single value for those linking/coupled variables. It proves theoretically thatthe optimum of this built model is identical with that of the original problem, and a numericalexample and a simultaneous WM and engine design are adopted as two examples to verify theeffectiveness of this approach. Based on the theory of three designs, a robust optimization method based on the idea oforthogonal experiment is proposed. It is applied to the robust optimization problem of slidingshoe in the ball mill. Based on the use of response surface approximations in MDO, aconcurrent subsystem robust design optimization (CSRDO) is proposed. It provides systemlevel with RS approximations substituted for subsystem information, improves the IIcollaborative efficiency. The collaborative optimization preserves the autonomy of eachindividual discipline, which makes computing in a concurrent processing environment to bepossible. Furthermore, this approach incorporates a robust optimization formulation, using alinearization approach, and generates global robust designs in an iterative fashion. As shownpractically, compared to other approaches, the proposed techniques can obtain global feasiblerobust optimum. Using object-oriented language and database technique, a WM-based MRDO platform isdeveloped to completely achieve the above ideas and methods. This platform is successfullyapplied to a X-type WM system and a Horomill system. It proves general, feasible andscientific.