Capturing Spatio-Temporal Dependencies in the Probabilistic Forecasting of Distribution Locational Marginal Prices

Toubeau, Jean‐François; Morstyn, Thomas; Bottieau, Jérémie; Zheng, Kedi; Apostolopoulou, Dimitra; Grève, Zacharie De; Wang, Yi; Vallée, François

doi:10.1109/tsg.2020.3047863

Cited by 23 publications

(11 citation statements)

References 44 publications

(46 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Probabilistic estimation was used in power system and energy discipline with great success, i.e. load forecasting [18][19][20][21][22], locational marginal prices forecasting [23], and wind forecasting tasks [24]. Probabilistic estimation utilizes a variety of approaches such as quantile regression (QR), quantile GBRT (Q-GBRT), regression neural network (QRNN) methods to estimate the results in the forms of quantiles prediction intervals (PIs), etc.…”

Section: Motivationmentioning

confidence: 99%

Multi-Quantile Recurrent Neural Network for Feeder-Level Probabilistic Energy Disaggregation Considering Roof-Top Solar Energy

Zhang¹,

Watkins²,

Kuenzel³

2021

Preprint

View full text Add to dashboard Cite

The purpose of feeder-level energy disaggregation is to decouple the net load measured at the feeder-head into various components. This technology is vital for power system utilities since increased visibility of controllable loads enables the realization of demand-side management strategies. However, energy disaggregation at the feeder level is difficult to realize since the high-penetration of embedded generation masks the actual demand and different loads are highly aggregated. In this paper, the solar energy at the grid supply point is separated from the net load at first via either an unsupervised upscaling method or the supervised gradient boosting regression tree (GBRT) method. To deal with the uncertainty of the load components, the probabilistic energy disaggregation models based on multi-quantile recurrent neural network model (multi-quantile long short-term memory (MQ-LSTM) model and multi-quantile gated recurrent unit (MQ-GRU) model) are proposed to disaggregate the demand load into controlled loads (TCLs), non-thermostatically controlled loads (non-TCLs), and non-controllable loads. A variety of relevant information, including feeder measurements, meteorological measurements, calendar information, is adopted as the input features of the model. Instead of providing point prediction, the probabilistic model estimates the conditional quantiles and provides prediction intervals. A comprehensive case study is implemented to compare the proposed model with other state-of-the-art models (multi-quantile convolutional neural network (MQ-CNN), quantile gradient boosting regression tree (Q-GBRT), Quantile Light gradient boosting machine (Q-LGB)) from training time, reliability, sharpness, and overall performance aspects. The result shows that the MQ-LSTM can estimate reliable and sharp Prediction Intervals for target load components. And it shows the best performance among all algorithms with the shortest training time. Finally, a transfer learning algorithm is proposed to overcome the difficulty to obtain enough training data, and the model is pre-trained via synthetic data generated from a public database and then tested on the local dataset. The result confirms that the proposed energy disaggregation model is transferable and can be applied to other feeders easily. <br>

show abstract

Section: Motivationmentioning

confidence: 99%

Multi-Quantile Recurrent Neural Network for Feeder-Level Probabilistic Energy Disaggregation Considering Roof-Top Solar Energy

Zhang¹,

Watkins²,

Kuenzel³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…These values are typically estimated using a grid search approach, which is embedded within a cross-validation scheme. 40…”

Section: Machine-learning Modelsmentioning

confidence: 99%

Automatic risk adjustment for profit maximization in renewable dominated short‐term electricity markets

Bottieau

Bruninx

Sanjab

et al. 2021

Int Trans Electr Energ Syst

Self Cite

View full text Add to dashboard Cite

show abstract

“…In other words, the activation function is the non-linear transformation applied to the data that allows the NARNET to model the non-linear properties of the data. Some of the most common activation functions used in forecast models include the Logistic Sigmoid function, Tangent Hyperbolic function and Rectifier Linear function [8,14]. NARNETs may have a single layer or have layers connected in series or in parallel [1,5].…”

Section: Introductionmentioning

confidence: 99%

“…Many efforts have been channelled to develop forecast models with high accuracy, although very few studies report the statistical significant difference between the analysed models. In other words, many authors often do not explain if the reported performance could be attributed to randomness in the data used [2,4,6,10,14,15,[19][20][21][22][23][24][25][26][27][28][29][30].…”

Section: Introductionmentioning

confidence: 99%

Statistical evaluation of wind speed forecast models for microgrid distributed control

2022

View full text Add to dashboard Cite

With the increasing needs to decarbonise existing energy systems, there is an effort to integrate small-scale distributed generation sources, such as wind generators, with the electric demand in circuits known as microgrids. The operation of distributed variable renewable resources is subject to an optimum operating regime, ahead of real-time, which relies on output forecast. However, many wind speed forecast models are designed for centralised controllers, which are vulnerable to control failures. A suitable wind forecast model for a distributed control system is, therefore, required for optimal and reliable use of renewable generation. This paper presents a comparison of wind speed forecast models suited for distributed control, evaluating them in terms of the statistical significant difference in accuracy and computational resource requirements. This is essential since computational resources are limited in distributed control schemes. The data used in this paper is the historical wind speed of the Auchencorth Moss Atmospheric Observatory from 2016 to the end of 2019. Two forecast model types based on Auto-regression and Artificial Neural Network (ANN) are compared using the Diebold-Mariano test. Results show that ANN models with parallel hidden layers have the highest accuracy with statistical significant difference, while remaining suitable for microgrid distributed control.

show abstract

Capturing Spatio-Temporal Dependencies in the Probabilistic Forecasting of Distribution Locational Marginal Prices

Cited by 23 publications

References 44 publications

Multi-Quantile Recurrent Neural Network for Feeder-Level Probabilistic Energy Disaggregation Considering Roof-Top Solar Energy

Multi-Quantile Recurrent Neural Network for Feeder-Level Probabilistic Energy Disaggregation Considering Roof-Top Solar Energy

Automatic risk adjustment for profit maximization in renewable dominated short‐term electricity markets

Statistical evaluation of wind speed forecast models for microgrid distributed control

Contact Info

Product

Resources

About