Research On Artificial Immune Algorithm Based On DNA And Its Applications

Artificial immune system (AIS) has developed based on the mechanism and function of the human immune system and related immune theory. Artificial immune algorithm is a highly parallel adaptive information learning system for solving complex optimization problems.It is a new type of intelligent algorithm after neural network, evolutionary computation, fuzzy system and a hot topic of research in life science and computer science.Based on analyzing the characteristics of the antibody population diversity and the process of antibody production, inspired by the biological expression of genetic information, this thesis has carried on deep research to the DNA based artificial immune algorithm and applications. The main contents of this thesis are as follows:(1). The antibodies of artificial immune algorithm are encoded by using DNA coding, and DNA molecular operations at the level of genes are introduced into the antibody production process, the artificial immune algorithm based on DNA is presented. In this Algorithm, the crossover operator and mutation operator are designed to enrich the antibody production process, and enhance the diversity of the antibody population. Through the contrast experimental study for the test functions with different dimension, the validity of the algorithm is confirmed.(2). To the lack of the antibody similarity criterion, the artificial immune algorithm based on code weight pseudo vector distance is proposed. The proposed algorithm endues with different weights according to the gene position. Which reflects the antibody decoding effect of the difference between high and low gene position. Abandoning the disadvantage of traditional artificial immune algorithms of measuring antibody similarity only from the candidate solution space, the value space criterion is also used in this algorithm. With some typical test functions, the optimum solutions confirm the effectiveness of improving antibody diversity. The proposed algorithm is applied to tune PID controller parameters of a servo system, and the satisfactory dynamics is reached.(3). The artificial immune algorithm with two populations is proposed based on an evaluation of diversity. In this algorithm, the original antibody population is taken as the main antibody population, and a dual antibody population is added. The later antibody population evaluates the distance within oneself and the distance between the two populations to guide the main antibody population evolution, and to improve the main antibody population diversity by the intercross strategy with it. The search capability of the proposed algorithm is promoted. With some typical test functions and the PID controller parameter tuning problem of the ABS control system, the effectiveness of the proposed algorithm is verified.