Improved artificial immune algorithm based on dynamic shrinkage strategy

In order to overcome the shortcomings of artificial immune algorithm in the process of function optimization, such as low convergence accuracy and weak local search ability, an improved artificial immune algorithm is proposed by introducing a variety of methods and designing a dynamic shrinkage strategy. Firstly, Logistic-Cubic cascade chaotic mapping is introduced to initialize the algorithm, so that the antibodies are evenly distributed in the whole solution space to enhance the global search ability in the early stage of the algorithm. Secondly, when the antibody population is updated, the mutated clonal memory cells are used to replace the poor antibodies in the population to avoid the premature stagnation of the algorithm, and the dynamic weight is added to the calculation of antibody expectation value to make the algorithm adapt to more possible situations. Then, designing a dynamic shrinkage strategy with the Pearl curve model to improve the algorithm's convergence speed and accuracy. Seven typical test functions are selected as the experimental objects, and five different optimization algorithms are used as the comparison algorithm to optimize the experimental functions. The experimental results show the effectiveness and superiority of the improved artificial immune algorithm on the function optimization problem.