Cloud Energy Storage Online Optimal Scheduling Strategy Based on Approximate Dynamic Programming

Aiming at the optimal operation of cloud energy storage, an online optimal scheduling strategy based on approximate dynamic programming is proposed. Fully considering the stochasticity of renewable energy power generation, load and electricity price, the non-linearity of energy storage, and the constraints of AC power flow, voltage and line capacity, a stochastic non-convex nonlinear programming model for cloud energy storage online operation is established. In order to solve this complex problem accurately and efficiently, it is transformed into a Markov decision process to realize the decoupling of the original problem and reduce the calculation scale of each sub-problem. In addition, the offline training of approximation value function is combined with online decision-making. It can reduce the impact of inaccurate forecast information on the online operation of cloud energy storage, and avoid the iterative process of approximate dynamic programming in the online decision-making process, so as to quickly obtain a near-optimal cloud energy storage online optimization scheduling strategy. Finally, case studies on various test system verifies the effectiveness of the proposed method.