Forecasting
Demand, generation, price, and weather forecasting
The prediction layer of Meridia — machine learning forecasts that every other module consumes. Predict facility demand at hourly to annual horizons, forecast solar and wind generation from weather models, anticipate wholesale market prices for procurement and dispatch decisions, and estimate grid carbon intensity for emissions-aware scheduling. Every forecast includes confidence intervals and automatic accuracy tracking so you know exactly how much to trust the prediction.
Key capabilities
Load & demand forecasting
ML-based demand forecasting at facility, site, and portfolio level — from 15-minute ahead to 3 years out. Weather-adjusted models account for temperature, humidity, solar gain, and wind chill. Separate base load, weather-sensitive load, and production-driven load components. Automatic model retraining as consumption patterns evolve with seasonal changes, occupancy shifts, and equipment upgrades.
Solar & wind generation forecasting
Forecast renewable generation output from numerical weather prediction models, satellite imagery, and on-site measurements. Intra-day nowcasting for dispatch decisions, day-ahead forecasts for market bidding, and seasonal projections for PPA profile matching. Probabilistic forecasts with P10/P50/P90 bands so storage and trading modules can price uncertainty.
Price forecasting
Wholesale electricity and gas price forecasts across day-ahead, intraday, and balancing markets. Feature-engineered models incorporating demand forecasts, renewable generation forecasts, fuel prices, interconnector flows, and planned outages. Forecast carbon allowance prices (EU ETS, UK ETS) for emissions cost planning. Confidence intervals and scenario fans for procurement risk assessment.
Peak demand prediction
Predict demand charge peaks before they happen — not after. Identify the days, hours, and facilities most likely to set new peak demand records based on weather forecasts, production schedules, and historical patterns. Trigger pre-emptive alerts to shift loads, pre-cool buildings, or dispatch battery storage before the peak materializes.
Grid carbon intensity forecasting
Forecast the carbon intensity of grid electricity at hourly resolution — enabling emissions-aware scheduling. Shift flexible loads and storage dispatch to low-carbon windows. Calculate the marginal vs. average emission factor for each hour to accurately attribute avoided emissions from demand response and behind-the-meter generation.
Forecast accuracy & model governance
Continuous backtesting of every forecast against actuals. Track MAPE, RMSE, and bias metrics by forecast horizon, facility, and model version. Automatic anomaly detection flags when model accuracy degrades — triggering retraining or alerting the energy team. Full model versioning and audit trail for regulatory and ISO 50001 reporting.
How it connects
Forecasting is the prediction layer that every operational module consumes. Procurement uses demand and price forecasts for contract timing and risk modeling. Battery Storage uses price and generation forecasts for dispatch optimization. Demand Response uses load and price forecasts for flexibility activation. Renewable Assets uses generation forecasts for performance monitoring. Carbon & ESG uses grid carbon intensity forecasts for emissions-aware scheduling. All forecasts are built on the live metered data from Facilities & Metering.