Loss Sensitivity Calculation and Analysis Based on Graph Computing

Loss sensitivity in electricity systems refers to the responsiveness of power losses in electrical transmission and distribution networks to various influencing factors. This concept is critical for optimizing the efficiency and reliability of power delivery from generation sources to end consumers. Especially in the electricity market trading process, network losses, although only a small part of the total cost, have a significant impact on each specific transaction and grid user. Therefore, in-depth understanding and analysis of the impact of generation transactions on the grid network loss can provide a basis for the power purchase plan for the grid to provide reference for the grid network loss management, which is of great significance. The current mainstream calculation and analysis method for network loss sensitivity is the marginal loss coefficient method, which determines the network loss share of each bus by analyzing the impact of bus injected power on the total network loss changes, but the nature of its serial calculation cannot properly solve the balance between real-time analysis of network loss optimization and accuracy, and with the increase in system size, this problem has gradually become a major obstacle to the realization of real- time analysis. This paper presents a novel parallel computing method based on graph computing for determining the sensitivity of system network loss and node-injected power. Graph computing is a parallel computing method that stores, computes, and analyzes the information in a system by constructing a topological graph model and based on it. The method alleviates the computational burden associated with calculating network loss sensitivity as the scale of the power system expands by leveraging the parallel computing capabilities of graph databases. Furthermore, it facilitates the graphical calculation and analysis of network loss sensitivity through the utilization of graph storage.