A multi-objective power-computing collaborative optimization strategy for data centers considering wind and solar uncertainty

The rapid expansion of data centers is an inevitable trend in the intelligent era. To address the challenges of high energy consumption and carbon emissions, a multi-objective optimization model for power-computing collaboration in data centers is proposed, with uncertainty in renewable energy output considered. An improved k-means algorithm is applied to reduce annual forecast scenarios of wind and solar power output. A set of typical representative scenarios is extracted to mitigate the impact of renewable energy uncertainty on system operation. Based on the flexibility of computing loads and energy storage systems, a collaborative optimization model is constructed. The objectives minimize the daily total cost of data center operation and the curtailment rate of wind and solar energy. The optimization model is solved under the optimal scenario of renewable energy output using the non-dominated sorting genetic algorithm-Ⅱ (NSGA-Ⅱ) with an elitism selection strategy. The results demonstrate that joint scheduling of wind, solar, and storage systems, together with flexible computing loads can effectively reduce daily operating costs, and improve renewable energy utilization, and maintain user satisfaction.