Development of Robust DER Contribution Estimation Model Using Machine Learning Under Incomplete Data

Demand for electricity is soaring internationally due to the development of artificial intelligence technologies and the continued spread of data centers. Renewable energy is actively being integrated to address potential stability issues in power distribution systems caused by the surge in demand. For stable operation, renewable energy-based power supply requires the ability to accurately estimate the influence and contribution of distributed generation(DG) to overall power output. This paper proposes a machine learning-based methodology applicable in industrial environments with limited measurement infrastructure. The proposed power tracing model for DG addresses missing data through machine learning techniques. To verify the proposed methodology, the performance and accuracy of the DG contribution estimation model were evaluated using a reduced test system based on branch points within the distribution network.