| The system design of flight vehicle involving many disciplines coupled with each other, computation complexity and communication complexity makes the overall design process extremely complex. In order to reach the highest potential benefit of integrated design and improve the design quality, the Multidisciplinary Design Optimization (MDO) method has gained much attention. Followed the multi-disciplines essence of the flight vehicle design process, the MDO method adopting the valid design/analysis/optimization process to integrate the knowledge of various disciplines and carrying out the organization and management in the design process of whole system, the overall optimal solution of the system can be obtained by coherently exploits the synergism of mutually interacting phenomena. MDO has stimulated the rapidly interest in aerospace realm all over the world, been applied to the overall design process of many kinds of flight vehicles successfully, and has produced the huge benefit nowdays.Combined with practical demands of the MDO research of flight vehicles in our country, four key technologies including the sensitivity analysis, the approximation strategies, the search strategies and the optimization process are studied in the dissertation, by which a suit of integrated system design method has been formed. The method is applied successfully to the design process of the solid strategic ballistic missile system and the hypersonic vehicle system in the dissertation.The sensitivity analysis technology is studied systematically. In order to solve the sensitivity analysis problem of discrete variables, the generalized sensitivity concept is presented, which fits continual/discrete variables well. The dissertation implements the generalized sensitivity analysis of continual variables by both the automatic differentiation method and the complex variables method, as well as the improved orthogonal design method and the improved neural network method of the generalized sensitivity analysis for discrete variables. In the above foundation, the software package of the generalized sensitivity analysis (GSASP) is developed successfully, which has the excellent robustness and broad applicability. Finally, the system sensitivity analysis method is studied on the foundation of disciplinary generalized sensitivity.The MDO approximation strategies based on generalized sensitivity—MTPACA (i.e. modified two-point approximation based cumulative approximation strategy) is studied. According to coupling degree between the design variables, the approximation functions are divided into three types, i.e. capable-of-separated, weak-coupling and strong-coupling. To the former two types, incomplete three-point approximation strategy and complete two-point approximation strategy are selected, and which are improved by application of the generalized sensitivity analysis. To the third type, considering the merits of the high precision of global approximation strategy and small scale computation of mid-range approximation strategy, MTPACA strategy is studied, which integrates the Rasmussen cumulative approximation strategy and complete two-point approximation strategy. So the evaluation costs can be reduced with the guarantee of approximation accuracy, while it also can be easily implemented in projects.MDO search strategy based on the generalized sensitivity is studied. For optimization question of the continual variables, the generalized reduced gradient method based on the complex variables is presented by the introduction of generalized sensitivity to calculate the sensitivity of design variables. For optimization question of the mixed discrete-continual variables, the generalized sensitivity computation is adopted to calculate the sensitivity of continual variables and the discrete variables directly. Combining with checking-point technique, the mixed variables direct search strategy based on the generalized sensitivity is presented.The Concurrent Subspace Optimization (CSSO) process is studied. It is systematically analyzed of merits and disadvantages of the standard CSSO process, improved CSSO process and response surface based CSSO process, presenting the improvement steps. Integrating with the aforementioned research results such as generalized sensitivity analysis, approximation strategies and search strategies, the move limit strategy is improved, based on which the Generalized Sensitivity Based Concurrent Subspace Optimization (i.e. GSBCSO) is presented. Compared with the existing CSSO process, the GSBCSO process has the characteristic of broad applicability, small computation scale etc.The system design method presented in the dissertation is applied to the weak-coupling questions such as the overall design of solid strategic ballistic missile. Firstly, mathematic states of design and optimization issue of the missile in GSBCSO process are established. Then, the GSBCSO process is carried out to optimize and solve the question. Finally, the comparison is made between the results obtained by the GSBCSO process and that of Multidisciplinary Discipline Feasible (MDF) process. It is showed that the GSBCSO process can complete the MDO problem of complicated engineering system effectively. The efficiency of optimization was improved remarkably. Feasibility and validity has been proved in this case.The method presented in the dissertation is also applied to the strong-coupling questions such as the overall design of hypersonic vehicle. The hypersonic vehicle being a typical strong-coupling system, the GSBCSO process is suitable to be applied to optimization and solution of its system performance. The computational results indicate that the GSBCSO process will reduce numbers of system analysis and calculation complexity as well as improving the system performance of the hypersonic vehicle to a certain degree comparing with that of MDF process, which lay the foundation for system design research of the hypersonic vehicle. The dissertation conducts thorough research on several key technologies of the MDO method of flight vehicles, forms the achievement that can be applied directly in projects. Applied in the flight vehicle design practice, the achievement takes the good effects. The dissertation provides a series of preferable methods, which will impel the MDO theoretical research and promote the combination of MDO theory and application in projects, having great theoretical and application value.
|
PDF Full Text Request