Feature-Driven Microgrid Economic Dispatch via Solar-Induced Contingent Balancing Satisficing

• Published in: IEEE transactions on sustainable energy pp. 1 - 21
• Main Authors: Dong, Hanjiang, Wu, Lubin, Li, Shenglin, Liang, Zipeng, Chung, Chi-Yung, Zhu, Jizhong
• Format: Journal Article
• Language: English
• Published: IEEE 2025
• Subjects: Costs; Economic dispatch; Fuels; Heating systems; Indexes; k-means clustering; microgrid; Microgrids; Optimization; Pipelines; Production; Random variables; robust satisficing optimization; soft dynamic time warping; solar power; Uncertainty
• ISSN: 1949-3029; 1949-3037
• DOI: 10.1109/TSTE.2025.3569485

For the multiple energy microgrid (MEMG), this paper builds an adaptive Wasserstein distance-based ambiguity set instead of characterizing its size for solving the economic dispatch (ED) problem considering uncertain solar outputs. With the ambiguity set, an electrothermal balancing satisficing (EBS) framework is proposed by defining a solar-induced contingent electrothermal balancing risk measure (SCEB-RM) to interpret the risk level for maintaining electrothermal balance under solar uncertainty. For comprehensive contingent cost targets, a multi-objective EBS is extended to refine the solar impact on electric and thermal energy resources, respectively, and provide the corresponding out-of-sample cost bounds. Using the realized solar data, a feature-driven EBS model is formulated by incorporating soft dynamic time warping-based k-means clustering into the ambiguity. A tractable counterpart is obtained by developing a solution approach on bilateral responsive affine resource adaptation, where satisfactory targets are specified by comparing with sample robust optimization as the baseline, serving as an instance. Using real-world datasets, combinations of IEEE test systems and thermal networks validate that EBS eliminates the risky condition of uncertainty while revealing the connection between targets and solutions. It is shown that cluster-wise information makes EBS withstand greater uncertainty as effectively as the state-of-the-art in most cases and sometimes better.