A computing transfer demand response system for coordinating data centers and power grids
Description: Patent No.: ZL 2022 1 1002763.0
Introduction： The present invention discloses a computing transfer demand response system for coordinating data centers and power grids. This system is innovative and economical, and can play a role in alleviating the pressure on the power system, exploring potential resources, achieving the optimal allocation of response resources. I. Innovation and Practicality 1. The operation of the coordinated power grid and communication network is coordinated, and the constraints and characteristics of both the communication network and the power grid of data centers with different nodes are proposed. By using the method of computing task transfer, the load is transferred to nodes with distributed generation or energy storage, enabling the data center to participate in the power grid demand response plan and playing a positive role in peak shaving and valley filling for the power system during peak electricity consumption periods. 2.Based on the declaration calculation module, the optimization algorithm reconstruction module is constructed. It is used to receive the mathematical models of the objective function module and the declaration constraint module, and the improved and efficient Lagrange reconstruction multiplier method is adopted to solve the optimal value. The key point lies in that the improved and efficient Lagrange reconstruction multiplier method reconstructs the inequality constraints in the index constraint function into the objective function in the form of penalty terms. The calculation speed is faster than that of the typical genetic algorithm and has a higher global optimal value. While maintaining the accuracy of the optimal solution, it greatly shortens the calculation time, thereby playing a role in reducing energy consumption for the data center. 3. Based on this system, an intelligent quotation module is constructed. It is used to receive the temporal virtual output boundary output by the declaration calculation module, the historical market clearing result - the temporal winning quantity, and the execution result output by the historical execution calculation module, including the temporal output capacity and the execution resource combination. The key point lies in continuously using historical data and the existing data through machine learning methods to find the optimal declared electricity price for data Center A to participate in the current power market demand response, maximizing the benefits, and outputting the sequential reported electricity prices for the current participation in the power market demand response. 4. Based on this system, the execution calculation module is constructed. It is used to receive the time series output capacity obtained by data center cluster B from data center A. Through the internal optimization calculation module, the optimal execution result of data center cluster B is output, including the time series output capacity and the combination of execution resources. The key point lies in implementing the virtual output winning scalar of data center A in the electricity market while constructing a mathematical model with the goal of minimizing the power generation cost of data center cluster B, so that data center managers can obtain the optimal resource combination and the highest revenue.
Organisation: Guangdong Power Grid Corporation Zhaoqing Power Supply Bureau
Innovator(s): Ruixin Tang,Yu Lei, Jiayun Huang, Hongguang Ning, Yanfang Wu, Zhihao Liang,Yeheng Jing, Guangdong Power Grid Corporation Zhaoqing Power Supply Bureau
Category: Energy
Country: China