As energy management and grid optimisation become increasingly data-driven, the need for accurate and granular load visibility grows. However, the deployment of measurement devices on low-voltage distribution grids requires a significant financial investment, often unsustainable at large scales.
To address this challenge, CEZ Distribuce developed a Multi-Layer Perceptron model that leverages artificial intelligence and machine learning to redistribute transformer-level load data across individual feeders, eliminating the need for extensive hardware installations by inferring localised consumption patterns. The model allowed for a significant increase in the accuracy of load estimation compared to previous methods, and CEZ Distribuce is currently focusing on scaling this solution across its entire distribution network.
