State-Space Models for Online Post-Covid Electricity Load Forecasting Competition

Vilmarest, Joseph de; Goude, Yannig

doi:10.1109/oajpe.2022.3141883

Cited by 14 publications

(11 citation statements)

References 29 publications

(30 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Dynamic. The natural approach aims at maximizing the likelihood; in that vein, we apply an iterative greedy procedure implemented in the R package viking [21], that was applied on electricity load forecasting by [6], [7]. It yields a sparse matrix Q assumed diagonal.…”

Section: B Adaptation Of Generalized Additive Modelmentioning

confidence: 99%

“…Adaptation is especially crucial to forecast wind and solar power; this paradigm has been successfully applied to online variable selection [4] and online forecast reconciliation [5]. State-space representations have well captured the recent changes of pattern due to the coronavirus crisis [6], [7]. The majority of recent works have focused on point forecasting, whereas [8] consider adaptive probabilistic load forecasting based on hidden Markov models and is found to perform well for a range of loads; however, this approach relies on Gaussian predictive distributions and we believe it yields a relatively poor calibration (quantile bias).…”

Section: Introductionmentioning

confidence: 99%

“…This is a non-adaptive model fitted to several years of training data. We consider a state-space representation to obtain an adaptive variant, relying on the Kalman filter for the inference as in [6], [7]. Second, we focus on forecasting the quantiles of the net-demand.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Adaptive Probabilistic Forecasting of Electricity (Net-)Load

Vilmarest¹,

Browell²,

Fasiolo³

et al. 2023

Preprint

View full text Add to dashboard Cite

We focus on electricity load forecasting under three important specificities. First, our setting is adaptive; we use models taking into account the most recent observations available, yielding a forecasting strategy able to automatically respond to regime changes. Second, we consider probabilistic rather than point forecasting; indeed, uncertainty quantification is required to operate electricity systems efficiently and reliably. Third, we consider both conventional load (consumption only) and netload (consumption less embedded generation). Our methodology relies on the Kalman filter, previously used successfully for adaptive point load forecasting. The probabilistic forecasts are obtained by quantile regressions on the residuals of the point forecasting model. We achieve adaptive quantile regressions using the online gradient descent; we avoid the choice of the gradient step size considering multiple learning rates and aggregation of experts. We apply the method to two data sets: the regional net-load in Great Britain and the demand of seven large cities in the United States. Adaptive procedures improve forecast performance substantially in both use cases and for both point and probabilistic forecasting.

show abstract

Section: B Adaptation Of Generalized Additive Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Probabilistic Forecasting of Electricity (Net-)Load

Vilmarest¹,

Browell²,

Fasiolo³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…More recently, a state-space approach has been investigated to adapt GAMs over time 16 ; this online model allows GAMs to adapt to new variabilities, such as the COVID-19 pandemic, to improve forecasts 17 , 18 . Finally, the variety of models motivates predicting the demand with a combination of forecasters, yielding a better final prediction than any individual model; that is the aggregation of experts 19 , which has also shown good results on load forecasting 20 .…”

Section: Introductionmentioning

confidence: 99%

Frugal day-ahead forecasting of multiple local electricity loads by aggregating adaptive models

Lambert,

Hamrouche,

de Vilmarest

2023

Sci Rep

Self Cite

View full text Add to dashboard Cite

This paper focuses on day-ahead electricity load forecasting for substations of the distribution network in France; therefore, the corresponding problem lies between the instability of a single consumption and the stability of a countrywide total demand. Moreover, this problem requires to forecast the loads of over one thousand substations; consequently, it belongs to the field of multiple time series forecasting. To that end, the paper applies an adaptive methodology that provided excellent results at a national scale; the idea is to combine generalized additive models with state-space representations. However, extending this methodology to the prediction of over a thousand time series raises a computational issue. It is solved by developing a frugal variant that reduces the number of estimated parameters: forecasting models are estimated only for a few time series and transfer learning is achieved by relying on aggregation of experts. This approach yields a reduction of computational needs and their associated emissions. Several variants are built, corresponding to different levels of parameter transfer, to find the best trade-off between accuracy and frugality. The selected method achieves competitive results compared to individual models. Finally, the paper highlights the interpretability of the models, which is important for operational applications.

show abstract

“…The mathematical technique of singular value decomposition was employed in [13], and Nedellee et al [14] proposed a station selection method based on minimizing a V-fold cross-validation criterion. Some of those studies are devoted to modeling the effect of COVID pandemic [15]. Recently, the work [11] presents a new method for selection and combination of weather stations based on Wavelet Squared Coherence that seeks to improve the performance of forecast models for the electric load.…”

Section: Introductionmentioning

confidence: 99%

Optimal Selection of Weather Stations for Electric Load Forecasting

Caro¹,

Ruiz²,

Nouhitehrani³

2023

IEEE Access

View full text Add to dashboard Cite

For the proper planning and operation of any electric power system, it is essential to have a reliable software tool that allows for accurate short-term forecasting of electric power consumption. Temperature is one of the most important drivers of electric energy consumption and selecting the appropriate weather station or combination of weather stations is crucial for improving the forecasting accuracy. In this study, we propose an algorithm to determine the optimal selection of weather stations based on (i) an initial reduction in problem complexity, which drastically decreases the computational cost, and (ii) a posterior incremental search algorithm. The proposed methodology, which identifies the optimal number of weather locations to be used and their specific geographical locations, is exhaustively tested using three insightful case studies using real data from the Spanish mainland electric power grid and two different weather forecast databases, denoting a significant reduction in forecasting errors. The developed method is currently in use by the Spanish transmission system operator (Red Eléctrica de España, REE) to make hourly forecasts.INDEX TERMS Short-term electric load forecasting, time series, meteorological variables, seasonal Reg-ARIMA models.

show abstract

State-Space Models for Online Post-Covid Electricity Load Forecasting Competition

Cited by 14 publications

References 29 publications

Adaptive Probabilistic Forecasting of Electricity (Net-)Load

Adaptive Probabilistic Forecasting of Electricity (Net-)Load

Frugal day-ahead forecasting of multiple local electricity loads by aggregating adaptive models

Optimal Selection of Weather Stations for Electric Load Forecasting

Contact Info

Product

Resources

About