Research On Reentry Trajectory Optimization Method Based On Multi-objective For Hypersonic Vehicles

Hypersonic vehicles have large advantages in flight speed, defense penetration and flight range, as well as the good prospects in military and civilian aspects. Therefore, hypersonic technology has been developed rapidly in recent years, and become the focus of research gradually. The trajectory optimization technique as a key of hypersonic technology has also been concerned widely. However, the complex environment of reentry process and the dramatic changes in atmosphere bring about many challenges in optimization process. Therefore, a more effective algorithm is very necessary to reentry trajectory optimization. Aiming at the reentry trajectory optimization problem for hypersonic vehicles, this dissertation expands the following research in three areas:For reentry trajectory optimization problem with multiple objectives and priority, the direct collocation method is adopted to transform motion equations into algebraic constraints, and the varying domain multi-objective optimization algorithm is also introduced to establish the optimization model. Firstly, the original problem is transformed into a nonlinear multi-objective programming with priorities using direct collocation approach. Then, the objectives are fuzzified into fuzzy goals, and the constant tolerance of each objective is substituted by the varying domain. According to the principle that the objective with higher priority has higher satisfactory degree, the priority order is modeled as the order constraints of the varying domain. The corresponding two-side, single-side and hybrid-side varying domain models are presented for different fuzzy relations respectively. By regulating the parameter, optimal reentry trajectory satisfying priority requirement can be achieved. The effectiveness and advantage of the proposed method are verified by simulation.For the mesh refining problem in direct collocation method, the concepts of density function and error estimation are introduced to divide meshes effectively, thereby increasing computing speed and solution accuracy. Firstly, the density function and its role in mesh refinement are discussed, and the error estimating methods are also presented to determine the accuracy of solution. Then, based on the theories above, four adaptive mesh refinement algorithms and their detailed description are presented. The differences in performance and application of these algorithms are analyzed in simulation.Considering the real-time trajectory generating problem with known and unknown target point, two effective algorithms are proposed. When target point is known, the 3-DOF motion model is divided into longitude model and latitude model, and solved respectively. For vertical trajectory control variable is designed as a piecewise linear function to reduce dimension, and the bank-reversal strategy is used to control lateral trajectory. To the problems with unknown target point, an enhanced optimization algorithm based on Gauss pseudospectral method is presented. The methods of computing terminal constraints and path constraints are changed to reduce the nonlinearity of optimization model, then computational efficiency can be enhanced consequently. The final simulations demonstrate the effectiveness and feasibility of the two algorithms.