The coupled evaluation of the reliability and operational capability of power grid unit personnel is generally described as a constrained multi-objective problem. However, traditional GA/PSO variants often get trapped in local optima and lack good scalability. To address this, this paper proposes an adaptive weighted GA-PSO hybrid strategy, which features (i) diversity-aware inertial scheduling, (ii) stagnation-triggered elite crossover mutation, and (iii) asynchronous parallel fitness computation on CPU/GPU. The coupled model uses entropy regularization weighting and dual consistency constraints to comprehensively score reliability (e.g., SAIDI/SAIFI/MTTR) and capability. Results: On three real-world datasets (unit logs, training records, and outage slips) and 30 repeated experiments, our proposed method improves the coupling degree of the optimal solution from 0.84 (GA-PSO) to 0.89, while consuming 34% less time than previous methods and exhibiting a smaller variance (SD 0.021). Statistical tests show that it improves the significance within the 95% confidence interval (p<0.01). Therefore, we can conclude that this is an optimizer that can achieve fast convergence and be executed in parallel, as well as an easy-to-understand coupled evaluation process, which helps in the rational allocation of human resources and targeted training.
Xiao et al. (Thu,) studied this question.