4cast is dedicated to generate power production forecasts at the highest level of precision imaginable.
Especially wind power is notoriously hard to predict. Regional small-scale effects, like vegetation or neighboring plants affect the power production on a large scale. This makes forecasts based solely on weather data unreliable as their geographical resolution is too coarse.
The solution is using your historical production data. All those effects are hidden in it! Our algorithms can read your data like an open book and infer the regional peculiarities of the site. Complemented by weather data from a multitude of sources, we apply machine learning technology to your production data to learn an optimal model which suits your specific needs. Our specialty is learning analytical models using symbolic regression, nevertheless our algorithms also feature all widely used deep learning techniques. The result is a forecast of unprecedented precision.
Whether your needs are intra-day high precision forecasts tuned to minimizing financial loss in energy trading or short-term predictions most accurate in estimating the power output to fulfill regulatory requirements, choose 4cast.
The 4cast project is partially funded by the Ministry for Labour, Social Affairs, Health, Women and Family (MASGF) of the Federal State of Brandenburg from resources by the European Social Fund (ESF).
Our basic model relies on the manufacturer's power curves. A coarse-grid weather prediction is mapped to the turbine spot.
The standard model is based on the manufacturer's power curves, your production data and a coarse-grid weather prediction. Our learning algorithm produces an improved parametrized power curve such that some local conditions are now included.
Based on your production data and a whole bunch of weather predictions, we let our machines learn an improved power curve. We choose the best model according to your specification. By increasing complexity, our learners include generalized linear regression, artificial neural networks, and symbolic regression.
Together with you we define customized targets which are fed into the learning algorithms to produce optimal forecast given your individual requirements.
An example is the generation of a forecast that maximizes your profit – this may differ from a best forecast of the production alone!