Improvement on regulating definition of antibody density of immune algorithm

摘要

A new heuristic optimization algorithm, which has the advantage of keeping the diversity of the solution, immune algorithm (IA), has been developed quickly in recent years. However, the calculation of antibody density by entropy in original IA has some shortcomings, i.e. the complex calculation and constants determined by experience will slower the speed of convergence. For an advancement of a performance, we modify the definition of antibody compared to the original immune algorithm, bring up a improved IA (IIA) based on the vector distance, proved its convergence and contrast the search results of optimization of numerical function with standard genetic algorithm (SGA) and original IA. Experimental results show that the IIA can effectively preserve diversity than SGA in population and it has faster speed than original IA in convergence.