Low-Voltage Network Estimator

Network planning aide utilising smart meter data to identify the correct low-voltage phase of customers. Data driven techniques provide visualisation of the combined electrical network topology. Intuitive user interface allows for result analysis and interpretation.

Traditionally, electrical network providers do not meticulously record which voltage phase each of their customers are on. Houses are often swapped from phase to phase by power linesmen without documenting changes. Poor phase distribution across a network can lead to challenges such as inefficiencies, increased cost, potential blackouts and the inability of users to sell power back to the grid. The advent of smart meter usage has presented an opportunity for the development on new techniques to identify customer voltage phases, allowing for improved network design planning and proactive phase balancing.

A variety of statistical analysis and unsupervised machine learning techniques are used to identify customer phases based on voltage data. These techniques include spectral clustering, fuzzy c-means clustering and dynamic time warping identification. Correlation techniques are then used to generate a minimum spanning tree of the electrical network feeding out from connected transformers. Novel accuracy metrics are used to quantify the accuracy and consistency of the developed methods. Visualised results and analysis are displayed on an intuitive user interface designed with extensibility in mind.