Automatic Dynamic Graph Calculation for the Fault Location System in a Grid with a Large Number of Unreliable Renewable Energy Sources

RES.

The proposed method performs an automatic dynamic calculation of a state graph for the integrated tasks of fault location and supply restoration in networks with intermittent RES. The algorithm consists of several key stages.

1. Initial Data & State Analysis: Gathering network parameters, recording the status of switching devices and equipment under maintenance, and identifying available controllable elements.

2. Formation of a State Graph: At each step, only remote controlled switchgear is considered. For each permissible switching action, a scenario is built, the new network operating mode is assessed, and an integral penalty function is calculated. This function incorporates following criteria: branch and source overloads, power losses, and overall system reliability and stability. 3. Dynamic Programming & Optimal Path Search: The algorithm employs an

"informed search" method, utilizing a priori information to exclude inefficient scenarios and avoid a full combinatorial search. At each step, scenarios with the minimal value of an integral penalty function are selected, gradually converging on an optimal restoration sequence until all consumers are restored or switching options are exhausted.

4. Post-Restoration Network Optimization: After determining the optimal switching sequence, a further optimization of device parameters (e.g., transformer tap changers, capacitor banks, energy storage systems) is performed to minimize losses and stabilize voltages.

The application of this method yields significant benefits: increased restoration speed by minimizing switching operations, enhanced resilience to RES intermittency, reduced energy losses, and full automation of the process, which is crucial for large, distributed networks.

Future research may focus on real-time system integration, predictive forecast of weak links of power grid with incorporated RES, and the application of machine learning to handle uncertainties, further improving the algorithm's accuracy and speed in complex dynamic grids.