Research On The Key Technologies And Applications Of Uncetainty-based Multidisciplinary Design Optimization For Flight Vehicles

The design of flight vehicles involves multiple disciplines and computational expensive simulation models, and also relates to many kinds of uncertainties. To design the flight vehicles with high performence, meanwhile enhance the robustness and reliability becames a critical problem in recent year. Uncertainty-Based Multidisciplinary Desgin Optimization(UMDO) handles the optimal design problem of coupled multidisciplinary system under uncertainty. The design quality of flight vehicles can be greatly enhanced by using UMDO. The design process of flight vehicles is a hierarchical process. In this paper, the UMDO methods for hierarchical system under aleatory uncertainty and mixed uncertainty is studied, including four key technologies, which are approximation modeling, uncertainty analysis, hierarchical multilevel optimization under uncertainty, and UMDO procedure. A compelte set of UMDO methods for hierarchical system is established and applied in the aeroelasitc tailoring of composite wing, the system design of small satellite, and programming of on-orbit refuling mission.The part of UMDO theoretical research is structured as follows.Firstly, the approximate method for implicit limit state function is discussed. Based on Least Squares Support Vector Machines(LSSVM), a Weighted Least Squares Support Vector Regression(WLSSVR) method is proposed for the implicit function approximation. To improve the global approximation quality of implicit function, the sequential modeling method is studied. A sequential modeling method based on the Potential Greatest Gradient Point(PGGP) is proposed. Considering the reliability analysis problem is widely exist in the engineering application, a sequntianl modeling method based on the Potential Most Probable Point(PMPP) is developed for efficient reliability analysis. The test results validate the efficacy of the aforementioned methods.Secondly, the UMDO method for hierarchical system under aleatory uncertainty is studied. Two kinds of probabilistic analysis method are proposed, which are AIO(All in one) probabilistic analysis and hierarchically probabilistic analysis. Based on these methods, the Sequential Probabilistic Analytical Target Cascading(SPATC) approach is established. SPATC approach is applied in the problem with approximate Guassian interrelated responses, non-Guassian interrelated responses and high dimension. The test results validate the advantages of SPATC approach on efficacy and accuracy.Finally, the UMDO method for hierarchical system under mixed uncertainty is studied. Considering the design variables include mix uncertainty, the uncertainty propagation method, optimization method and UMDO procedure are developed. Based on RIA(Reliability Index Approach) and PMA(Performance Measurement Approach), two mixed uncertainty analysis and propagation methods are derived, which are RIA-MUP(Mixed Uncertainty Propagation) and PMA-MUP. RIA-MUP and PMA-MUP can be applied to the uncertainty analysis and propagation problem with uncertainty integration variables. Using PMA-MUP to transform the believable constraint to deterministic constraint, the optimization problem under mixed uncertainty can be figured out by sequentially solve deterministic optimization and mixed uncertainty analysis. It provides a way to solve the optimization problem under mixed uncertainty. A Mixed Uncertainty based Analytical Target Cascading(MUATC) approach is established. MUATC is an integration of the aforementioned methods and SPATC. The effectiveness of MUATC has been validated by some test examples.The part of UMDO application research is organized as follows.Firstly, the aeroelastic tailoring problem of composite wing is discussed. The mathematical model of the aeroelastic tailoring problem is set up. This model is then hierarchical decomposed into several subproblems. ATC method is used to orgranize and slove the problem after decomposition. The results show that the ATC method has a good performance on this application. Considering there are aleatory and epistemic uncertainties existing in the model, SPATC and MUATC are used to solve the UMDO problems respectively, and both achieve the optimal schemes which satisfy the reliablility or believable constraints.Secondly, the system design of a small satellite is studied. The disciplinary models in the system design are established. According to the hierarchical property of small satellite system design, the system design model is decomposed. ATC method is applied to solve the decomposed model and obtained good optimization results. SPATC and MUATC are used to solve the UMDO problems of the preceding small satellite system design problem under aleatory uncertainty and mixed uncertainty respectively, and achieved satisfying optimization results.Finally, the programming of on-orbit refuling mission is studied. The refueling mission for geosynchronous satellites is chosen as the study object. The mathematical model is establish and decomposed into four subproblems, which are, the problem of solving the times that servicer visits depot, optimal location problem, optimal path planning problem. The mission programming problem is also hierarchical decomposed. Twenty geosynchronous satellites are taken as the future refueling objects. Considering “one to one” and “one to multi” refueling stragety, the mission programming problem is sovled. The results show that “one to multi” refueling stragety is much better than “one to one” refueling stragety. Based on these results, the programming of “one to multi” refueling mission under aleatory uncertainty and mixed uncertainty are also dicussed. SPATC and MUATC are applied to these problems and obtained the optimal schemes which satisfy reliablility or believable constraints.To sum up, the key issues in the UMDO theory are studied according to the characteristics of flight vehicles design in this paper. A complete set of UMDO methods is constituted. The proposed UMDO methods are applied in the component level design problem, system level design problem and top level design problem. The research work in this paper laid a good foundation to enhance UMDO theories and the design quality of flight vehicles...