Strategic Two-Stage Diagonal Quadratic Approximation Method for Economic Dispatch with Energy Storage

Solving large-scale stochastic dynamic economic dispatch problems involving energy storage over multiple time periods is inherently challenging. Decomposing the optimization problem across time scales to achieve parallel optimization is an effective strategy for enhancing computational efficiency. However, the introduction of energy storage complicates this task significantly due to its non-convex characteristics. To address this challenge, this paper presents a novel strategic two-stage Diagonal Quadratic Approximation Method (DQAM) that transforms the original problem into a two-stage structure amenable to parallel solving. In this structure, each sub-problem focuses on a single time period, effectively handling the mixed-integer and strong temporal coupling characteristics of energy storage. This extension broadens the applicability of the original DQAM to non-convex problems. Numerical results under various conditions demonstrate the effectiveness and performance of the proposed two-stage DQAM, achieving significantly improved solution efficiency while maintaining good solution accuracy. Leveraging computational processor and memory resources, the proposed method can achieve a maximum enhancement of 36.86 times in computational efficiency on large-scale systems with multiple time periods. Note to Practitioners - With the increasing scale of power systems and the growing penetration of renewable energy, the accuracy and real-time requirements for economic dispatch solutions have been continuously enhanced. Solving large-scale optimization problems over multiple time periods in power grids has become more challenging, especially with the additional computational burden introduced by the deployment of energy storage to mitigate renewable energy fluctuations. The configuration of energy storage generates a large number of binary decision variables, which results in the original problem becoming a large-scale mixed-integer programming model which is difficult to solve. This paper proposes a strategic two-stage DQAM to solve the stochastic dynamic economic dispatch problem with energy storage, utilizing a two-stage structure to determine an approximate energy storage optimization schedule through an optimization approximation function. Subsequently, the equivalent dynamic economic dispatch problem is transformed into a sizable form with smaller and more easily solvable sub-problems. By relaxing the temporal coupling constraint, the computational efficiency is significantly improved. Test results on systems of various scales and time periods demonstrate that the two-stage DQAM reduces computation time in all scenarios.