Adaptive Multi-objective Particle Swarm Optimization based on Signal-to-Noise Ratio Distance

In Multi-Objective Particle Swarm Optimization (MOPSO), an appropriate search strategy is crucial for achieving satisfactory algorithm performance. However, traditional algorithms often fail to adapt search strategies based on particle states, leading to insufficient convergence and diversity when solving complex problems. To address this, we propose an evolutionary state estimation strategy based on SNR (Signal-toNoise Ratio) distance to evaluate particle states, enabling adaptive search strategies for particles in different states. We introduce an improved Logistic map to dynamically adjust particle acceleration coefficients. Additionally, a population stagnation detection method using Hausdorff distance is proposed; when the population is trapped in local optima, a genetic mutation algorithm is applied to evolve the external archive, injecting new vitality and effectively guiding the algorithm to escape local optima. These innovations significantly enhance the global optimization performance of the algorithm in complex multi-objective scenarios. Simulation experiments were conducted on 12 standard multiobjective benchmark problems (ZDT and DTLZ series) comparing AMOPSO-SNRd with eight state-of-the-art multiobjective optimization algorithms. The results demonstrate that AMOPSO-SNRd outperforms most competing algorithms in terms of convergence and diversity metrics.