A dynamic multi-objective optimization evolutionary strategy based on adaptive selection for wet flue gas desulfurization process

Wet flue gas desulfurization (WFGD) process is crucial for reducing SO 2 emissions in coal-fired power plants. In this process, the limestone slurry reacts with the SO 2 emissions. Meanwhile, the reaction efficiency is sped up by pumps and oxidation fan, which is positively correlated with their electrical load. Herein, three conflicting objectives need to be minimized: SO 2 emissions, electrical load, and use of limestone slurry. Meanwhile, many uncontrollable factors change over time, resulting in rapidly or random changing Pareto-optimal solution set (PS), i.e., a dynamic environment. Therefore, we formulate WFGD problem as a dynamic multi-objective optimization problem (DMOP). To solve WFGD problem, a simple yet effective algorithm, a dynamic multi-objective optimization evolutionary strategy based on adaptive selection (ASS), is proposed in this paper. When a change occurs, ASS provides diversified solutions by different proposed strategies, namely, a diverse direction prediction and a center-guided self-correcting prediction. Based on the severity of environmental change, an adaptive selection mechanism can adjust the selection probability of each strategy. ASS consists of two different prediction strategies, enabling more responsive to different changes. Comprehensive empirical studies shows that ASS achieves an excellent performance on CEC2018 DMOP benchmarks and well solve real-world WFGD problem compared to four state-of-the-art algorithms.