Immune Clonal Selection Algorithm Improved And Its Application Research

AIS (Artificial Immune System) based on the functions, principles, basiccharacteristics of biological immune system and other related theories is a computingsystem set up to resolve various complicated problems, and represents a new directionof intelligent computation research following the Artificial neural network, andevolutionary computation. This thesis aims to research and explore the evolutionarylearning mechanism in the biological immune system, and design an effectiveartificial immune algorithm so as to overcome the combinatorial optimizationproblem and machine character recognition problem.A kind of immune clustering colonel selection algorithm for resolving large-scaleTSP is put forward in this thesis, which combines together the advantages of bothclustering algorithm and immune colonel selection algorithm. The algorithm makesuse of the k-means dynamic clustering method for neighborhood partition of TSP, andof immune colonel selection algorithm for the precise optimization of TSP subdividedwithin the group, and then of neighborhood connection according to K method forgravity to rapidly get massive global connections TSP and get a satisfactory solution.On the other hand, a dynamic chaotic adaptive immune colonel selection algorithm isintroduced in this thesis. It takes the advantages of the ergodicity and randomness ofnonlinear chaotic to initialize antibody and of the global optimization ability ofimmune dynamic colonel selection algorithm to optimize the problems, making fulluse of the respective merits of the immune operator and chaotic operators.In the testing of TSP, the results show that the algorithm can effectively improvethe precision, lower time complexity, and be superior to other intelligent algorithms interms of high accuracy and fast convergence speed. Finally, by applying the chaoticadaptive dynamic immune colonel selection algorithm to machine characterrecognition, the experimental results show that colonel selection algorithm achievedthe parallel and serial process of character recognition, and that the algorithmdiscussed in this thesis has intelligent machine learning ability.