Font Size: a A A

Multi-objective Optimization Method And Its Engineering Applications

Posted on:2006-03-17Degree:DoctorType:Dissertation
Country:ChinaCandidate:W G AnFull Text:PDF
GTID:1112360212967732Subject:Aircraft design
Abstract/Summary:PDF Full Text Request
The multi-objective optimization design of aircraft is very necessary. But the shortcomings of existent multi-objective optimization algorithms and the difficulty in their engineering applying, always block the application of multi-objective optimization algorithms in aviation and astronautics field. So, the research on this paper focuses on multi-objective optimization algorithms and their application. The main contents of this paper are described as follows:1. Existent multi-objective optimization algorithms are categorized and compared with each other. The research shows that multiple objective particle swarm optimization is better than other optimization algorithms.2. In order to overcome the weak local searching abilities of multiple objective particle swarm optimization (MOPSO) and particle swarm optimization (PSO), simplex method-multiple objective particle swarm optimization (SM-MOPSO) and simplex method-particle swarm optimization (SM-PSO) are proposed. Validated by using standard test functions, SM-MOPSO and SM-PSO inherit all the merits of particle swarm optimization, overcome their shortcomings, and could attain good quality Pareto solution or Pareto set.3. Evolutionary algorithm is very inefficient when it is used to solve large and complex multi-objective engineering optimization problem, because the large number of evolutions is needed in the evolutionary algorithm and time-consuming high-fidelity analysis have to be done the same times as the number of population in each generation. In order to cut down the computational expense, the single objective model management framework and the multi-objective model management framework are proposed. By using the model management framework, accurate approximation models of the entire searching space can be constructed, and evolutionary algorithm would not only avoid many time-consuming high-fidelity analyses but also attain good optimization results efficiently. In this paper, we give two engineering examples: (1) aerodynamic layout optimization design of cosmonaut transportation vehicle, (2) the wing structure optimization design of a certain unmanned aviation vehicle. By using the framework proposed in this paper, evolutionary algorithm not only attains good optimization results but also makes high...
Keywords/Search Tags:multi-objective optimization algorithm, model management framework, parallel compute, multi-objective decision-making method, interactive algorithm
PDF Full Text Request
Related items