Metaheuristic Method of Evolutionary Optimization Using Immune Approaches

Abstract

The article examines a metaheuristic method of evolutionary optimization based on the operating principles of artificial immune systems and aimed at solving constrained multi-criteria optimization problems. It is shown that contemporary optimization tasks are characterized by a high-dimensional search space, numerous local extrema, and the need to reconcile multiple criteria, which complicates the use of classical optimization techniques. The rationale for employing nature-inspired approaches — particularly evolutionary methods grounded in artificial immune system algorithms — is provided. These methods enable adaptive global search through mechanisms of cloning, mutation and selection. An evolutionary optimization approach based on the artificial immune system algorithm is proposed, with its operators and algorithmic framework formally defined. To evaluate the method’s effectiveness, computational experiments were conducted on benchmark problems using both binary and real-valued encodings. In the XdivK problem with binary encoding, the artificial immune system algorithm demonstrated a higher convergence rate than the genetic algorithm, attributable to its intensive local search within promising solution regions. In the optimisation of a bi-modal function with real-valued encoding, the artificial immune system algorithm consistently achieved the global optimum without becoming trapped in local maxima. It was established that, for small population sizes, the artificial immune system algorithm ensures faster convergence, whereas the genetic algorithm exhibits better scalability. The results confirm the effectiveness of the proposed approach for solving complex optimisation problems and highlight the potential for further development of immune-based methods within intelligent computational systems.