Application And Research Of Immune Genetic Algorithm In Computer Aided Design

Computer aided design (CAD), which has nearly half a century's history till now, plays a significant role in the design industry. The evaluation of a scheme is always the most important and complicated step in the design. The quality of the scheme is crucial to whether the project will be success or not in computer aided design projects. The quality of a design constrained by a variety of objective criteria can't be judged by a single and subjective way of evaluation. The evaluation of the scheme has always been a complicated and difficult task, especially in the computer aided landscape design.In order to get an accurate evaluation, a design scheme should be evaluated not only from the practical, economic and other factors but also from the current specific criteria at the same time. It will be great valuable to study on comprehensive and intelligent evaluation method for the overall design in application. Support vector machine (SVM) have good generalization ability of the parameters aiming at small samples of training set in evaluation. The SVM is introduced to the evaluation by making use of the better generalization ability to the small sample training set which reaches a relatively high rate of correct classification in the testing sample set.Genetic Algorithm(GA) is a collateral stochastic searching algorithm,which adopts natural selection and natural genetic mechanism. At first, it was designed and applied in the field of numerical optimization and combinatorial optimization. For many years,genetic algorithm has obtained great attention and research. It has been widely used in computer science, artificial intelligence, information technology, engineering practice and many other fields. However, researchers find that due to a variety of problems, GA causes "premature convergence", which affects the global search for the optimal solution. And then the Immune Genetic Algorithm (IGA) is introduced.This thesis develops mainly around the intelligent approach to computer aided landscape design. Based on IGA's and SVM's theories and CAD's features, we propose the SVM algorithms optimized by IGA,and apply them in the computer aided landscape design .Based on the objective analysis of the practical problems in the relative field, main points and the innovation of the thesis are as follows.1 Based on in-depth analysis of impact on the SVM's parameters of the classification capacity,this thesis proposes parameters selection methods by SVM instead of the previous way of determining the parameters——cross-validation method or the experience of scholars. The optimal parameter can be searched automatically in a limited scope by grid algorithm which ensures the order of the search. It gets better effect than the previous methods.2 Based on the theory of the IGA and SVM, the thesis proposes another method of parameters selection in IGA-SVM. IGA is proposed in order to solve the problem of "premature convergence" in GA. It can solve the problem of choosing the optimal parameters in SVM by its global optimization performance.3 Through the consulting of a great variety of design materials and the integration of the decisive evaluation elements of landscape design, the SVM is introduced into the landscape's overall layout evaluation model in the system of computer aided landscape design. By the practices of the samples data, an intelligent landscape's overall layout evaluation method is realized. IGA can optimize the parameters of the model and kernel in the SVM, and it promotes the rate of correct classification.