A hybrid memetic enhanced particle swarm optimisation algorithm for cost and emission dispatch of smart micro-grids with renewable generation and demand response

The growing integration of stochastic renewable generation and demand response (DR) has increased the complexity of day-ahead economic and environmental dispatch in smart micro-grids. Although particle swarm optimisation (PSO)-based methods provide effective global exploration, they may stagnate before sufficiently refining cost-sensitive hourly power allocations. The main novelty of this study is the integration of adaptive population-based PSO exploration with a domain-aware memetic local refinement mechanism that exploits the marginal-cost structure of hourly economic dispatch. HM-EPSO integrates opposition-based initialisation, adaptive multi-operator mutation, a differential-evolution-based escape mechanism, and a greedy memetic local search that reallocates hourly generation according to marginal-cost information. The local refinement primarily targets economic dispatch, while the adaptive population-based framework is retained for emission minimisation.HM-EPSO is evaluated over 20 independent runs on a 24-hour stochastic smart micro-grid comprising seven energy sources with 40% DR participation and is compared with PSO-mutation, EPSO-M, SA-EPSO, and published MOPSO results. For cost minimisation, HM-EPSO achieves a mean cost of $1370.29 and a best cost of $1271.61. These values are respectively 13.55% and 19.77% lower than the published best MOPSO cost of $1585, while HM-EPSO significantly outperforms all internally evaluated baselines. Ablation analysis identifies the memetic local search as the principal contributor, with its removal increasing the mean best cost by 10.38%. For emission minimisation, HM-EPSO achieves a mean of 1845.53 kg and a best of 1757.00 kg, respectively 2.20% and 6.89% below the published best MOPSO value of 1887 kg, while remaining statistically comparable to the enhanced PSO variants. Overall, the results demonstrate that domain-aware memetic refinement substantially strengthens economic dispatch while preserving competitive emission performance, highlighting the objective-dependent benefits of hybrid search in smart micro-grid scheduling.