Leveraging Meter Load Data to Build a Bottom-up Spatial Load Forecasting

(DER), and it informs outage mitigation strategies. Load forecasting is a complex task that requires information on operational, climate and economic trends. This includes data on historical load transfer (when available), EV penetration, population changes, construction projects, weather patterns and economic trends. Many utilities simply rely on past peak data to project future load through linear regression. However, this does not capture well underlying trends, nor does it help build a more targeted approach to infrastructure investments and outage mitigation strategies.

We developed a bottom-up spatial forecast model that predicts future load using meter data at the lowest level. The predictions are then aggregated at a geographic level (neighbourhood, city) and an electrical connectivity level (isolation region, protective region, circuit, full network). Using meter-level data helps studying shifts in consumer behaviour. It also helps identifying electrical regions with future load limitations thanks to connectivity data (asset-toasset mapping to identify protective and isolation regions of a network).

In this paper we will present:

- The power of load data analysis through two examples:

o Load transfer identification at the circuit level (through time series decomposition and identification of change points in de-seasonalized load data) o Inference of EV charger penetration through graph signal processing of time series, using meter-level load data combined with neighbourhood economic data

- The gain in accuracy of a bottom-up spatial load forecast compared to a simple linear regression model based on peak data.

- The benefit of the bottom-up spatial load forecasting (leveraging machine learning models) approach in planning infrastructure investments, leveraging GIS tools to map the most vulnerable regions of the network at different stages in the future (1year, five years, 10 years).

By predicting future electricity needs with greater precision, utilities can better plan for the transmission of distribution of electricity and minimize the risk of blackouts and overloading.