Application Research Of Immune Clone Selection Algorithm

In the field of Artificial Immune System, many methods adapted the theory of clone selection, or embedded the mechanism in the process of realization. But the application of clone selection principle is not perfect. Many characteristics are only applied by metaphor. They are not realized in practice. In the paper, clone selection algorithm is improved. It is applied in function optimization, PID control, Job shopping, clustering. The main work is as follows:Based on the concentration regulation in the process of antibody cloning, which is explained by immune network theory, it presents different evolution strategies at the beginning and at the end of evolution. Immune network clone optimization(INCOA) is presented. It can increase the diversity of antibody population at the end of evolution. Thus it can avoid premature. It is tested by non-linear equation sets and multi-peaks function. The test results show the validity of the algorithm.Simple clone selection algorithm(SCSA) is presented to optimize the parameters of PID control system. Through the experiments, including pendulum, one-order delay function model, six-order function model, and the real time control of pendulum, the results show that it is simple and has less computing complexity. It is stable for searching the solution of object function. Its convergence speed is clearly higher than that of basic clone selection algorithm(BCSA).SCSA is used to solve the problem of optimizing the parameters of PID. Its whole performance is better than those of GA and BCSA.The niching theory is used to design the self-adaptive clone heuristic algorithm. Self -adaptive clone heuristic algorithm(SACHA) and niching self-adaptive clone heuristic algorithm(NSACHA) are presented to the problem of Job shop. It simulates the self-adaptation and co-evolution among immune system. It expresses to be the competition to the amount of antibodies among different population, different antibodies. It improves the cover rate to the problem space and enhance the global searching ability. NSACHA can get the global optimization solution and local optimization solution at the same time. So it has real meaning in practice. At last, for the problem of clustering, antibody memory clone clustering algorithm(AMCCA) is proposed. It uses clone selection, clone memory to produce memory cells and suppresses the worse antibodies. Thus, it can reach the aim of suppressing data. The test results show that random search AMCCA can find better solution than traditional algorithms. Its performance is close to the other typical clustering algorithm. So it can be used to cluster.The analysis results show that clone selection algorithm based on biology clone selection principle has equal or better performance with other typical methods. To some extent, it has real application value in practice.