Complex numbers simplify solar forecasting

The total power received at ground level from the sun, or the global solar irradiance, is the main parameter influencing the generation of photovoltaic power. Anticipating that energy production makes it easier to balance the electrical grid, decreases cost, and increases the energy share of clean solar power.

Voyant et al. developed an innovative and simple forecasting technique to predict short-term solar irradiance using complex numbers and linear predictive methodology. They tested the model on experimental times series data gathered on Corsica and found good agreement.

Previous methods for solar forecasting include numerical weather prediction models, statistical or artificial intelligence methods, and sky imagery. Bettering predictions is an overall goal, but the team believes those improvements should not come at the cost of simplicity and robustness.

“The approach based on complex number generation and least squares optimization is technically simple to implement, the predictive results are good for the horizons studied, and there is the possibility to extend the deterministic mode to the probabilistic one,” said author Cyril Voyant. “The use of complex (or even hypercomplex) numbers could greatly improve predictive strategies in many areas.”

The real part of the irradiance data corresponds to the measurement, while the imaginary part captures the volatility. By including both, the model achieves better accuracy than classical versions and requires very little input data.

In the future, the researchers plan to modify the imaginary part to include new variables. They also aim to apply the method to other intermittent sources of energy such as wind.

Source: “Complex-valued time series based solar irradiance forecast,” by Cyril Voyant, Philippe Lauret, Gilles Notton, Jean-Laurent Duchaud, Luis Garcia Gutierrez, and Ghjuvan Antone Faggianelli, Journal of Renewable and Sustainable Energy (2022). The article can be accessed at https://doi.org/10.1063/5.0128131 .