A multi-objective scheduling method for hybrid integrated energy systems via Q-learning-based multi-population dung beetle optimizers

A hybrid integrated energy system (HIES) optimal scheduling model is developed by considering energy storage (ES), solar thermal (ST), wind turbine (WT), demand response (DR), and carbon capture and storage (CCS) to minimize the economic cost and pollutant gas emission; A multi-population dung beetle optimizer based on Q-learning (MODBO-QL) is designed to solve the HIES optimal scheduling model. In MODBO-QL, three subpopulations are constructed first: Global search subpopulation (GSS), Local search subpopulation (LSS), and Balanced search subpopulation (BSS). Second, the search and evolution strategies corresponding to the population characteristics are given. The GSS and LSS perform global search and local search, respectively. At the same time, the BSS is responsible for randomly exploring or exploiting the solution to increase the diversity of the population. Finally, the number of populations is dynamically and adaptively adjusted using Q-learning to enhance information exchange between different sub-populations and improve solution performance. An optimal dispatch experiment was conducted using three typical seasonal load data and renewable energy forecast data in a region of China. The experimental results show that compared with the original system, the three typical quarters’ average economic costs of the dispatch results obtained by the method in this paper are reduced by 8.7%, 20.99%, and 3.97%, and carbon emissions are reduced by 617.35t, 284t, and 19.5t, respectively; MODBO-QL performs better compared to MOPSO, MOWOA, and MODBO.