A laptop screen showing a power forecast graph with annotations about electric car usage and solar panel generation

Forecasting Methodology

Our machine learning models form the core of our product. They constantly ingest, manage, and analyze time-series data from various sources, including weather, grids, ISO, and meters.

Trusted by leaders in power & utility markets

Advancing the energy industry

Employing cutting-edge ML methodologies, advanced statistical techniques, and sophisticated mathematical models, we analyze the data for complex relationships between historical and real-time information.

Our AI algorithms continuously learn and update with new data, allowing the models to evolve dynamically. As a result, our models adapt to ever-changing market dynamics and can identify patterns in large datasets that previously went unnoticed.

Frost-covered transmission tower under clear blue sky representing energy infrastructure resilience

Short-term
forecast model

Amperon's "short-term" forecasts consist of 15-day forward-looking predictions for demand, price, and renewable asset generation. Our models run every hour as new data becomes available. See our product pages to learn more about our modeling capabilities and use cases.

Solar panel array installation in desert environment with tracking mounts

Long-term
forecast model

Amperon's "long-term" forecasts consist of demand predictions up to 5 years out. Our models run every week as new data becomes available. Amperon uses a combination of our short-term forecast methodologies and simulation techniques to generate hundreds of scenarios for demand.

Amperon AI Engine methodology diagram showing inputs from grid data, weather, distribution, on-prem data, enterprise API, and carbon intensity flowing to outputs including grid forecast, portfolio demand, meter demand, and asset renewables, with access points through platform UI, API, and third-party integrations

Platform inputs & outputs

Amperon integrates market and proprietary inputs based on the scope and nature of customer needs. Typical inputs include:

Grid Data: ISO/RTO data on territory operating conditions and historic load.

Distribution Data: AMI/AMR SCADA data that provide near-time load data from substations and other grid measurement points.

Weather Data: 40k weather points from 4 different vendors. See weather page for more information.

On-Premises Data: Smart meter, on-premises EMS, and BTM data.

Enterprise Data APIs: CIS, ETRM, and EDI/Billing data.

Grid Carbon Intensity Data: WattTime’s localized grid carbon intensity data

The Amperon system is fully API enabled, allowing programmatic access to forecasts and automated loading of data. System outputs, usually in CSV files, are designed to be highly flexible for end users to work with. Specific standard output requirements can be accommodated.

Forecasting Methodology

Advancing the energy industry

Platform inputs & outputs

More data points. More models. More accurate forecasts.