Numerical weather models, specifically calibrated for short wave solar radiation forecast, are used in combination with statistical learning and neural networks algorithms, to offer the best irradiance forecast at every specific site.
Meteorological numerical models optimised for solar radiation forecast
Dedicated physical modules for solar radiation calibration
Statistical learning methods and neural networks
Data analysis and forecast for solar tracking plants
Algorithms trained with satellite data and site-specific data
Direct and diffuse short wave solar radiation components
Numerical weather models, specifically calibrated for short wave solar radiation forecast, are used in combination with statistical learning and neural networks algorithms, to offer the best irradiance forecast at every specific site.
Satellite irradiance data and historical plant energy generation make it possible to train the statistical algorithms and offer the best energy production forecast.
A proprietary algorithm allows us to efficiently separate direct and diffuse radiation, which are needed for plants with tilted panels and solar tracking facilities.
Satellite irradiance data and historical plant energy generation make it possible to train the statistical algorithms and offer the best energy production forecast.
For the single solar power plant IDEAM Srl provides on site forecasts of:
e.g. temperature, humidity, precipitation, wind, pressure
if requested with
detailed components:
• direct and diffuse irradiance
• near-infrared, visible and ultraviolet irradiance
• incident radiation on a tilted solar panel
• incident radiation on a tilted solar tracker
hourly or sub-hourly time steps
from 24-48-72 hours ahead up to 2 weeks
bi-daily updates (or more) with mobile or fixed temporal horizons
• confidence intervals such as 1 0 and 90 percentiles for irradiance and other fields
Specifically for solar irradiance, IDEAM Srl has developed a proprietary algorithm (MOSRH) that recalculates cloud cover and weighs appropriately the direct and diffuse components of radiation. The resulting forecasts have significantly smaller errors in situations with only partial cloud cover.
For grid operators and TSO’s IDEAM Srl can integrate the meteorological day ahead forecast with power production forecasts for grid stability, risk analysis and decision making purposes.
fundamentals parameters included in the analysis are:
• estimated instantaneous power and energy production
• hourly energy production forecasts
• daily estimated energy production forecasts
• mean seasonal and annual production values of the solar field
• ramp rate
Statistical techniques and Artificial Neural Networks are used to improve the forecast accuracy.
I f radiation and power measurements from the plant are available, the correction algorithms can be trained using the site specific data, in order to offer the best forecast performance. Otherwise, algorithms are trained using satellite data and data from other sites, and so an improvement from the basic model forecast is possible in every location, even without site specific data.
Share this page on social networks