Accelerating Mixed-Integer Optimization in Energy Storage Systems: A Novel Approach Using Heuristics, Interior Point Methods, and Clustering

In this study, we propose a novel approach to accelerate online mixed-integer optimization using heuristics, the interior point method, and an integrated clustering and sorting technique. Our method constructs a Cardinality cutting plane that leverages linear programming (LP) solutions without the CROSSOVER step, forming a feasible region for cutting plane segmentation. It carefully configures parameters to ensure feasibility and utilizes 0-1 variables derived from clustering energy storage device information. Our method seamlessly integrates into any Mixed-Integer Programming (MIP) solver. Empirical results show that our algorithm outperforms traditional methods in rapidly exploring feasible regions and significantly accelerates lower bound improvements, enhancing optimization efficiency for energy system scheduling.