Study On Optimization Design For The Reinforced Concrete Frame-shearwall Structures Based On Immune Genetic Algorithm

The artificial immune system simulates some basic concepts and mechanism about information processing of the biological immune system. It comprises two branches-- immune clonal selection algorithm and immune evolution algorithm. Immune clonal selection algorithm simulates the body’s own immune system. Compared with standard genetic algorithm, the algorithm retains more local information in the process of searching, because it pays more attention to the mutation operator. Immune evolution algorithm simulates the artificially acquired immunity, adds the immune operator in the process of standard evolutionary algorithm. The immune operator can avoid the degradation or invalid individual appearing and improve the search direction for the characteristic information and prior knowledge of actual problems are used to adjustment of population evolution partially. Through the study of bionic algorithm, it is found that the most suitable method for optimization design on layout of shearwalls is the immune genetic algorithm.The main research contents and achievements of this paper include:(1) Through deducing the anti lateral and torsional stiffness matrix of the frame shear structure and analyzing the sensitivity of different optimization variables to structural stiffness, the conclusion that the shear wall length and distance from the centroid distance is the main factors affecting the structure torsional is gotten.(2) The optimal design model of reinforced concrete frame shear structure is established. The optimization objective of frame shear structure is the cost of beams, columns and walls. According to the sensitivity analysis of the optimized variables to the structural stiffness, the idea that the optimization variables should be divided into two levels is put forward: the layout of the wall should be determined firstly and then optimize the section size of members. The first level variables are the length of shear wall, the shear wall distance from the centroid, and the concrete grade of beams, columns and walls. The second level variables are the section size of beams, columns and walls. Corresponding to the optimization variable, the constraints are divided into two levels, too. The first level includes the whole specifications of the structure, and the second level includes the bearing capacity and the detailing requirements of the members. The first level variables are optimized by the immune algorithm and the second level variables are optimized by the grid search method. Both the immune genetic algorithm and grid search method are suitable for discrete variable, which can make the structure optimization results fit the modulus.(3) Three types of vaccines are designed for the immune genetic algorithm: the vaccine for wall horizon layout, the vaccine for concrete grades and the vaccine for structure whole constraints. The vaccines for wall horizon layout and for concrete grades introduced the design habits and the prior knowledge into the algorithm. They make all individuals feasible. They make up the blindness of the standard genetic algorithm. The vaccine for structure whole constraints optimizes the search direction, improves the overall fitness of the population, increase the efficiency of the algorithm.(4) The software SWOD for optimizing the reinforced concrete frame-shear wall structures is made. The software uses VC++ programming language and object-oriented programming method. The software can import the data file STRUSTRU.SAT and LOAD.SAT of PKPM to ensure the accuracy of the input data, which make optimizing complex structures be possible. The internal force analysis and structure design in the optimization process are completed by SWOD. The operation of SWOD is simple because most parameters are set according to the codes and only a few parameters should set by users. SWOD has high flexibility by using polymorphism. If SWOD needs to be changed, the modified workload for the old codes is small.(5) SWOD uses the CSC matrix storage strategy and the LDL solver based on super node, which is suitable for solving large sparse matrix. CSC matrix storage strategy significantly reduces the demand of storage space. the structure of dynamic analysis and static analysis are completed by the super node LDL solver, so that optimizing the large frame-shear wall structures becomes possible.(6) Five examples are analyzed and compared at the last. Example one proves the results of internal force analysis and structure design calculated by SWOD are correct. Example two proves that the grid search method is effectiveness. Example three displays the difference between three design patterns. Example four and five show the optimization effect of multi-floor frame-shear wall structure and high-floor frame-shear wall structure. SWOD can provid feasible preliminary design scheme to the designers, because some design habits introduce to the optimization process, which makes optimized structures meet the design specifications.