Font Size: a A A

Research On Artificial Immune System For Solving Function Optimization Problems

Posted on:2006-10-10Degree:MasterType:Thesis
Country:ChinaCandidate:Q WangFull Text:PDF
GTID:2168360152970668Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Many practical engineering problems can be abstracted to corresponding function optimization problems. Until now there are many heuristic algorithms to solve function optimization problems. Compared to conventional algorithms, the merit of heuristic algorithms is their better global search ability, which avoids converging to local optimization early. And traditional Genetic Algorithm is a classical algorithm for function optimization problems. Immune system is a distributed self-adapted complex system. Artificial Immune System is an idea corresponding to Biology Immune System. People find useful strategy to solve engineering and scientific problems. How to design effective algorithm for optimization by simulating biology immune activity is a meaningful subject. Clonal Selection Principle is very important in Artificial Immune System. Inspired by the Clonal Selection Principle, the Immune Algorithm can solve function optimization well.The work we do in this thesis is as followed:(1) The principle of biology immune system and the common immune algorithm for optimization at present are introduced. The characters and the differences of the immune algorithm and genetic algorithm are summarized. ECJ which we use to implement GA and do numerical experiment is introduced in this thesis(2)A clonal selection algorithm based on clone selection principle to solve numerical function optimization is proposed. There are mainly two differences between the algorithm and the present clonal selection algorithm, the way of coding and mutation. Real number coding is adopted. A kind of Gaussian mutation is adopted, which make the individuals searched towards the best fitness. The algorithm has no fixed mutation probability and can find best solutions without controlling so it can quicken the speed of convergence.(3)In order to test the performance of the algorithm, several object functions are optimized including single-modal and multi-modal function. And we compare the results of the immune algorithm to that of a Genetic algorithm. The numerical experiment results demonstrate the immune algorithm can find the best solutionswith fast convergence speed, which have best results than the Genetic algorithm.(4) Through numerical experiment, we did a research on the two important parameters of the immune algorithm proposed in this thesis. And we studied how the two parameters affect the performance of the algorithm.
Keywords/Search Tags:genetic algorithm, artificial immune system, immune algorithm, clonal selection, function optimization
PDF Full Text Request
Related items