IGA-FEM For Structure Optimization

The Immune Genetic Algorithm(IGA) based on immune principle and the IGA-FEM were studied in this thesis. The IGA-FEM was used to optimize the combustor floating-wall and a good optimizing result was obtained. The study work mainly included the following:(1) Both methods, information entropy and Euclidean distance, are developed to calculate the concentration. By simulating the immune mechanism of antibody self-adjustment and memory function in above both methods,four IGAs are conducted. The results of the optimization show that IGAs have better convergence than ESGA; Among the four IGAs, the IGA, simulating the mechanism of antibody self-adjustment and calculating concentration by information entropy, has the best convergence performance.(2) The effects of coding length, mutation rate and generation on the optimization performance of IGAs were studied. In the study, the IGAs simulated the mechanism of antibody self-adjustment and calculated concentration by information entropy and Euclidean distance respectively.(3) Based on the above IGAs (studied in the second section work), both AIGAs were put forward by introducing the crossover and mutation rate self-adaptive technique. The optimization results show that AIGAs have good performance on the global convergence, and the AIGA which based on information entropy is better.(4) The idea of combining Immune Genetic Algorithm (IGA) and Finite Element Method (FEM) was put forward, and IGA-FEM and AIGA-FEM were established for structural optimization. In the study, IGA and AIGA were based on antibody self-adjustment and information entropy. The optimization results of flat structure show that AIGA-FEM is better for structural optimization. (5) The floating-wall model of combustor was optimized by AIGA-FEM. A strategy of optimization was adopted, which was achieved by adding the variables and the complexity conditions step by step. In the optimization process, not only the geometrical restriction and stress but also the stiffness of casing, the place of middle pole and temperature were considered.