Yield prediction for wind and solar plants

Accurate Intraday, DayAhead and Automated Performance Analysis

Accurate power and yield predictions for wind and photovoltaic systems

Our performance and yield forecasts for wind and solar systems are produced using a learning algorithm. The forecast provides an accurate insight into the expected energy production of your wind or solar farm – whether for the current day, the next day or long-term periods. Our yield forecasts provide you with accurate data tailored to the needs of your wind or solar farm.

Creation of our yield forecast

Machine learning models form the basis of our forecasts and are based on current weather data and historical production values. Meteorological factors such as air pressure, temperature, solar radiation, wind speed and direction are included, as well as micro factors (e.g. topography, park effects, icing). Regular re-training of the ML models ensures continuously optimised forecasts, which are available for individual turbines and solar modules as well as for entire generation parks as intraday, day-ahead and automated performance analysis.

4cast forecast model - wind energy forecasting process with machine learning: weather and customer data are fed into power forecasting and event identification models to create accurate energy forecasts.

Visualization of the wind energy forecast: precise yield forecasts for optimum wind energy production.

Accurate yield forecasting pays off

An accurate day-ahead forecast reduces the cost of balancing power, enables reliable feed-in contracts and optimises revenue planning for plant operators.

Your benefits with our yield prediction

Optimal calculation of energy production and revenue
Increase revenue by incorporating relevant market data
Better data to ensure grid stability
Avoid penalties with accurate feed-in reports
Identify ideal maintenance windows
Optimized system operation

Day-Ahead Forecasts

Our DayAhead forecasts are based on gradient boosting models trained on historical weather and production data. We provide these forecasts every morning for the next few days. They start at midnight the following day, have a 15-minute time resolution, and can cover up to four days. If you order forecasts for more than 24 hours in a row, you will receive multiple forecasts for the same period. In this case, we recommend that you always use the most recent forecast. If you wish, we can overwrite outdated forecasts if the agreed delivery method allows.

Our automated performance analysis is a valuable tool in the planning and development of wind and solar farms. It provides an accurate assessment of the future performance of a site.

An arrow pointing from the 'DayAhead Yield Prediction' on the left side to a date labeled '21.09.' on the right side.

A blue circle frames the illustration. At the top is an intraday forecast, below is a calendar page with the date 21.09.

Intraday Forecasts

Our intraday forecasts use a neural network trained on historical weather and production data. In addition, live data is continuously fed into the system to produce continuously updated intraday forecasts.

In order for your live data to be included in the forecasts, a binding data transfer method must be agreed upon. All common transmission methods can be used. If you wish to change the transmission method, please give us one week’s notice so that we can adjust the system accordingly. If desired, the connection can also be encrypted to ensure secure transmission.

Modules for complete flexibility in yield forecasting

Forecast modules

Basic module

The power curve of your wind energy plant is adjusted according to the air density and is combined with the weather forecast to predict yields. We also have the right solution for your solar plant.

Machine learning module

The machine learning model is trained using historical production data and weather predictions to apply the learned patterns to current weather predictions and thus produce yield forecasts.

Intraday module

We use live production data to optimize yield forecasts for short time windows, as used for trading on the electricity exchange market.

Optimization modules

Event module

We monitor the output of your wind or solar plants to identify influencing factors and refine the forecasts.

Retraining module

The machine learning model is retrained using new production data.

FAQ

How does your machine learning work?

We use machine learning to identify patterns and correlations in various datasets. After training the model, these patterns are applied to the unknown datasets to produce the forecasts. We’d be happy to provide more details.

Where do you get your weather data?

Our production forecasts are based on numerical weather predictions (NWPs) from a range of providers. We consider and process a combination of these NWPs on a case-by-case basis.

Can a contract be enhanced during an existing term?

We’re your partner for any concern or request you have. Please reach out to us and we’ll be sure to find a solution together.

Can newly built turbines be integrated into an existing contract?

Yes, existing contracts can be expanded.

How do I communicate with your team?

You’ll be assigned a dedicated contact person who will be on hand to answer all your questions.

Our experts are here for you