Dirichlet Sampled Capacity and Loss Estimation for LV Distribution Networks With Partial Observability

With low voltage (LV) distribution networks increasingly being re-purposed beyond their original design specifications to accommodate carbon technologies, the ability accurately calculate actual spare capacity is critical. Traditionally, within Great Britain (GB) power system, there has been limited monitoring of LV networks, making this difficult. This paper proposes a method for estimating unmonitored using demand data from customer Smart Meters. In particular, proposed infers existing network capacity, as well losses, across scenarios where only number customers have Meters installed. Typical daily load profiles with are learned Dirichlet sampled Gaussian mixture model (GMM). Learned then applied all unmetered estimate parameters. Method accuracy assessed by comparing estimations simulated, fully observed, models. The also compared benchmark models establishing unobserved profiles. Overall, results in show that outperforms terms assessing substation headroom, particularly 10–50%