Evolutionary Algorithm And Its Application In Aircraft Conceptual Design

Aircraft design is a complex system engineering which synthesizing many different disciplines. During the process of preliminary conceptual design, the suitable optimization model is formulated in terms of the design objectives. Aircraft conceptual design is performed on the base of effective engineering numerical methods, optimization tools and Decision-Making (DM). And the satisfied solutions can be obtained, which are prepared for detail designs.The robust and effective optimization algorithms are of great importance to aircraft conceptual design. Conventional optimization methods are implemented with great efficiency, nevertheless they extremely rely on the mathematical model of the design object, and that the search with a single point in the design space can not ensure that the global optimal solutions are found. Compared with the conventional optimization methods, Evolutionary Algorithm (EA) is more attractive. It has been applied in many fields because of its independence on the optimization model. The search in EA is based on population, which provides the information sharing mechanism that improves the efficiency of the computation. Therefore, EA is adaptable to aircraft conceptual design.A framework of aircraft conceptual design was built, which comprise engineering numerical methods, optimization tools and DM methods.In order to build a more effective optimization method, a branch of the EA, Particle Swarm Optimization (PSO), was investigated. The principle and convergence characteristics were analyzed. PSO was compared with Genetic Algorithm (GA) in optimization operations and searching techniques. It concluded that the information sharing mechanism made all the individuals tend to converge to the current global best solution quickly in the intelligent evolution. Therefore, PSO is more efficient. In order to improve the convergence characteristics of the algorithm and keep the diversity of the population, several improved PSO algorithms were built. The algorithms were used to solve the optimization of several numerical functions, truss structural design and structural design of the wing. PSO proved to be a robust and effective optimization tool and suited for large-scale optimizations in engineering.In multi-objective optimization, a robust and well distributed noninferior set is expected, which can help the designers to understand the project and make decisions. PSO was developed to solve multi-objective optimization. In order to improve the searching efficiency and keep the diversity of noninferior set, a new algorithm, Niche Based Simulated Annealing - Multi-objective Particle Swarm Optimization was built. It was applied to the multi-objective optimization of numerical functions and wing shape design of the Reusable Launch Vehicle (RLV).The engineering design is associated with the experiences and preferences of designers. In Physical Programming (PP), the objective functions are mapped to preference functions in terms of designers' expectations, and then they are integrated into an aggregate preference function. The tradeoff solution is obtained which can express designers' preferences correctly. PP provides a method of multi-objective decision-making. PP can generate the well distributed noninferior set of multi-objective optimization by tweaking the preferences of designers. A new Interactive Physical Programming (IPP) was presented. The preferences were adjusted in terms of DM matrix, and design the process was related to designers' intentions. PP method was employed to solve the multi-objective DM problem of numerical functions, civilian aircraft conceptual design and air-breathing hypersonic vehicle fore body / inlet configuration design.In the framework of aircraft conceptual design, the previous EA and MD methods were applied to aerodynamic configuration optimization of aircraft. A new wing/body configuration of a glide hypersonic vehicle was presented, which was generated by conic lofting design method. The geometry of the vehicle was parameterized in terms of conic control points and shape parameters, which improved the efficiency of configuration design. The revised Newtonian Theory was applied to estimate the hypersonic aerodynamic characteristics. And then the trajectory of the vehicle was simulated and optimized aimed to maximizing the flight range. The aerodynamic configuration was optimized with respect to lift-to-drag ratio and volumetric efficiency given the constraint of longitudinal stability. The Particle Swarm Optimization, Multi-objective Particle Swarm Optimization and Physical Programming were used to perform the optimization. And the preliminary conceptual design of the vehicle was achieved.