| At Present, great progress has been made on structural optimization technology, and various optimization algorithms can be applied to more and more optimization problems Although great progress has been made on structural optimization, the optimization of the rigid framed structural cross-section parameters is still a conundrum. Most of the researches are about the optimization problems of the truss frame structure, so only rod units need to be considered, and the cross-sectional area is the only parameter of the mechanical characteristics of the truss frame cross-section. Engineering software applied to rigid frame structure is relatively scarce, because the cross-sectional mechanical characteristic parameters of the beam unit are relatively complex, which belongs to multiobjective optimization problems. At present, researches on multivariable complex structure optimization are few; former optimization methods aims at the complex structure, and the multiple irrelevant mechanical characteristic variables of the beam unit cross-section are transformed into cross-sectional geometrical variables, so as to aim at weight reduction, rigidity enhancement; weighed multiobjective optimization models are set up and optimized. However, the structural shape obtained by the optimization mode usually can not be realized in engineering practices for other reasons. Therefore, the thesis adopts the mechanical characteristics of mathematical model direct coupled beam cross-section put forward by topic group for optimization, and during engineering practice optimization results can be referred to; according to the requirements, structural shapes are designed. Before optimization, the range of the mechanical characteristic parameters is defined, thereby enabling the optimized results to be realized in engineering.The genetic algorithm is an evolution algorithm with relatively strong global searching capability, and is widely applied to structural optimization. Therefore, the optimization on the cross-sectional mechanical characteristic parameters of the rigid frame structure by the genetic algorithm is of great significance. The thesis mainly discusses the following work. 1. Mathematical model establishment. The mathematical model comprises design variables, restrictions and objective functions. In the original mathematical model, the design variables have a bending inertia moment, a twisting inertia moment and a sectional area on the cross-sectional characteristic parameters to be optimized for the beam units; the restrictions are the upper and lower limit restrictions and the displacement restriction of the loading point in the direction of the load; the objective function is weight minimum and rigidity maximum, and the inverse measure of strain energy represents the rigidity; therefore, the objective function can be expressed as weight minimum and strain energy minimum. In the model, the variation of the inertia moment has no effect on the objective of weight, and only has relation with the strain energy. Therefore, under the action of the objective of strain energy minimum, the inertia moments are respectively the upper limit of each restriction interval, and such an optimization result means nothing to the rigid frame optimization. In order to solve this problem, the thesis adopts the new mathematical model put forward by the topic group. The restrictions are the upper and lower limit restrictions of weight minimum, strain energy minimum and the design variables, and the displacement restrictions of the loading point in the direction of the load; the objective functions are the bending inertia moment, the twisting inertia moment and the sectional area.2. Elitist genetic algorithm optimization. The Genetic algorithm solves the optimization problem of a single objective, while the parameter optimization of the rigid frame structure belongs to the optimization problem of multiple objectives with restrictions. In view of the situation, the thesis adopts weighting factor technique to convert the multiple objective problems into a single objective problem. A penalty function method is adopted, and the restriction conditions are merged to the objective function.In the fundamental genetic algorithm, there are many deficiencies; although the more and more elite individuals will be generated with the evolutionary process of colonies, individual with the best adaptability in the current colony may be damaged because of the randomness of genetic manipulation such as crossing, mutation, etc., thereby affecting the operating efficiency and the convergency of the genetic algorithm. Evolution manipulation is added after the genetic manipulation in the Elitist genetic algorithm, and the individual with best adaptability can be kept until the next generation of the colonies. The implementation of the strategy can ensure that the most elite individual can not be damaged by the inheritance operations such as crossing, mutation, etc, furthermore, the global convergency of the genetic algorithm can be ensured.3. The optimization on the mechanical characteristic parameters of a simple rigid frame cross-section by the elitist genetic algorithm. More optimized results are calculated by an algorithm example of the simple rigid frame structure rigid frame structure, so as to prove that the mathematical model and the inheritance optimistic algorithm put forward by the topic group are effective in the optimization design of the rigid frame structure.4. Analysis on inheritance operators. During the operation of the genetic algorithm procedures, the crossing probability and the mutation probability need to be selected in advance, and the two parameters have great influence on the performance of the genetic algorithm. As for a concrete problem, weighing the fitness of the parameter settings is based on the convergency conditions of multiple operations and the quality of solutions for judgment. In the thesis, repeated operations and comparative researches are carried out to select the crossing probability and the mutation probability which have the best effect on the parameter optimization of the cross-sectional mechanical characteristics of the rigid frame structure.5. The optimization of the concept vehicle frame structure by the elitist genetic algorithm. According to the mathematical model put forward by the topic group, calculation is carried out by adopting the crossing probability and the mutation probability obtained by previous comparison. The optimization analysis of the statics and the dynamics is carried out on the rigid frame structure of the concept vehicle with preferable effects. Therefore, the topic group provides a practical method for solving the problems of the design method in the multiobjective optimization of engineering.6. Adaptive genetic algorithm and niche genetic algorithm. From the comparative analysis of the adaptive genetic algorithm, we can see that the self-adapting design of the genetic algorithm can improve the deficiency that the iterative initial speed of convergency is too high, and the diversity of the species can be maintained to a certain degree, thereby increasing the probability of finding the globally optimal solution. The genetic algorithm based on the niche technology is another improved method of the genetic algorithms. By comparative calculative analysis, we have the conclusion that the niche genetic algorithm obviously excels fundamental genetic algorithm in the aspects of global convergency reliability and convergency speed, and the evolution speed of the niche genetic algorithm is comparatively stable; even when the nearly global optimal solution is found, the niche genetic algorithm can also maintain relatively high diversity of species, and the phenomenon of premature is inhibited so as to provide potential momenta for further evolution. However, in evolution calculation, the magnitudes of the adaptability need to be compared repeatedly in the niche genetic algorithm, so calculation is increased virtually, and calculation time is increased. |
PDF Full Text Request