Competitive-Cooperative Multi-Task Optimization Algorithm with Historical Success Archive

Multi-task optimization plays a pivotal role in various scientific and engineering disciplines. Within this field, Competitive Multi-Task Optimization (CMTOP) has garnered significant attention as a specialized branch aimed at solving multiple optimization tasks concurrently under competitive settings. However, existing CMTOP algorithms suffer from inherent shortcomings, such as inefficient resource allocation and undirected randomization in auxiliary task selection, which considerably limit their overall performance. To overcome these challenges, this article proposes a novel Competitive Multi-Task Historical Success Archive Synergy Algorithm with a Success-Memory Framework (CMHSA-SMF). The algorithm incorporates three key innovations: (i) an adaptive task affinity weighting mechanism with dynamic thresholds that guides auxiliary task selection by capitalizing on distinct phase-dependent demands during early and late stages of optimization; (ii) a historical parameter-aware adaptive resource allocation strategy that dynamically distributes computational resources based on iterative performance feedback and success memory; and (iii) a collaborative co-evolutionary dynamic coupling strategy, which integrates adaptive random mating probability control and a dynamic cooperative evolutionary operator to enhance elite diversity, strengthen inter-task communication, and promote cross-task knowledge transfer for accelerated convergence. Extensive evaluations on widely-used benchmark suites, including C2TOP, C4TOP, and CCPLX, along with real-world applications in sensor coverage optimization and photovoltaic model parameter extraction, demonstrate that CMHSA-SMF significantly outperforms seven state-of-the-art peers (DE, SHADE, MFEA, MFDE, DEORA, MTSRA, and MTEA-HKTS) in terms of solution quality, convergence speed, and robustness.