Neural basis expansion analysis with exogenous variables: Forecasting electricity prices with NBEATSx

International Journal of Forecasting, Journal Year: 2022, Volume and Issue: 39(2), P. 884 - 900

Published: May 5, 2022

We extend neural basis expansion analysis (NBEATS) to incorporate exogenous factors. The resulting method, called NBEATSx, improves on a well-performing deep learning model, extending its capabilities by including variables and allowing it integrate multiple sources of useful information. To showcase the utility NBEATSx we conduct comprehensive study application electricity price forecasting tasks across broad range years markets. observe state-of-the-art performance, significantly improving forecast accuracy nearly 20% over original NBEATS up 5% other well-established statistical machine methods specialized for these tasks. Additionally, proposed network has an interpretable configuration that can structurally decompose time series, visualizing relative impact trend seasonal components revealing modeled processes' interactions with assist related work, made code available in dedicated repository.

Language: Английский

---

Electricity price forecasting with high penetration of renewable energy using attention-based LSTM network trained by crisscross optimization

Energy, Journal Year: 2022, Volume and Issue: 254, P. 124212 - 124212

Published: May 10, 2022

Language: Английский

---

An ARIMA-LSTM model for predicting volatile agricultural price series with random forest technique

Applied Soft Computing, Journal Year: 2023, Volume and Issue: 149, P. 110939 - 110939

Published: Oct. 20, 2023

Language: Английский

---

Harnessing eXplainable artificial intelligence for feature selection in time series energy forecasting: A comparative analysis of Grad-CAM and SHAP

Applied Energy, Journal Year: 2023, Volume and Issue: 353, P. 122079 - 122079

Published: Oct. 17, 2023

This study investigates the efficacy of Explainable Artificial Intelligence (XAI) methods, specifically Gradient-weighted Class Activation Mapping (Grad-CAM) and Shapley Additive Explanations (SHAP), in feature selection process for national demand forecasting. Utilising a multi-headed Convolutional Neural Network (CNN), both XAI methods exhibit capabilities enhancing forecasting accuracy model efficiency by identifying eliminating irrelevant features. Comparative analysis revealed Grad-CAM's exceptional computational high-dimensional applications SHAP's superior ability revealing features that degrade forecast accuracy. However, limitations are found with Grad-CAM including decrease stability, SHAP inaccurately ranking significant Future research should focus on refining these to overcome further probe into other methods' applicability within time-series domain. underscores potential improving load forecasting, which can contribute significantly development more interpretative, accurate efficient models.

Language: Английский

---

Review of virtual power plant operations: Resource coordination and multidimensional interaction

Applied Energy, Journal Year: 2023, Volume and Issue: 357, P. 122284 - 122284

Published: Dec. 12, 2023

Virtual power plants (VPPs) have become an important technological means for large-scale distributed energy resources to participate in the operation of systems and electricity markets. However, VPPs is challenged by stochastic resource characteristics, complex control features, heterogeneous information structures, strategic game behaviors among stakeholders. To clarify key problems solutions these challenges, this article describes coordination multidimensional interaction mechanism, it elaborates overall decision-making process VPPs. It also discusses different specific operational stages that should attach importance from three separate perspectives: energy, market. From each perspective, every section first analyzes motivation decision-making, then complexity problem models, summarizes modeling methods solving techniques, thus completing a comprehensive review VPP operation. Furthermore, adopts interdisciplinary approach, utilizing literature technical statistics capture multifaceted contributions operations. delves into evolving trends technology, analyzed coupling cyber-physical-social perspective. Finally, future trajectory research issues deliberated.

Language: Английский

---

Evolutionary game-theoretical approaches for long-term strategic bidding among diverse stakeholders in large-scale and local power markets: Basic concept, modelling review, and future vision

International Journal of Electrical Power & Energy Systems, Journal Year: 2025, Volume and Issue: 166, P. 110589 - 110589

Published: March 9, 2025

Language: Английский

---

Deep learning models for price forecasting of financial time series: A review of recent advancements: 2020–2022

Wiley Interdisciplinary Reviews Data Mining and Knowledge Discovery, Journal Year: 2023, Volume and Issue: 14(1)

Published: Sept. 28, 2023

Abstract Accurately predicting the prices of financial time series is essential and challenging for sector. Owing to recent advancements in deep learning techniques, models are gradually replacing traditional statistical machine as first choice price forecasting tasks. This shift model selection has led a notable rise research related applying forecasting, resulting rapid accumulation new knowledge. Therefore, we conducted literature review relevant studies over past 3 years with view aiding researchers practitioners field. delves deeply into learning‐based models, presenting information on architectures, practical applications, their respective advantages disadvantages. In particular, detailed provided advanced such Transformers, generative adversarial networks (GANs), graph neural (GNNs), quantum (DQNNs). The present contribution also includes potential directions future research, examining effectiveness complex structures extending from point prediction interval using scrutinizing reliability validity decomposition ensembles, exploring influence data volume performance. article categorized under: Technologies > Prediction Artificial Intelligence

Language: Английский

---

Two-step deep learning framework with error compensation technique for short-term, half-hourly electricity price forecasting

Applied Energy, Journal Year: 2023, Volume and Issue: 353, P. 122059 - 122059

Published: Oct. 18, 2023

Prediction of electricity price is crucial for national markets supporting sale prices, bidding strategies, dispatch, control and market volatility management. High volatility, non-stationarity multi-seasonality prices make it significantly challenging to estimate its future trend, especially over near real-time forecast horizons. An error compensation strategy that integrates Long Short-Term Memory (LSTM) network, Convolution Neural Network (CNN) the Variational Mode Decomposition (VMD) algorithm proposed predict half-hourly step prices. A prediction model incorporating VMD CLSTM first used obtain an initial prediction. To improve predictive accuracy, a novel framework, which built using Random Forest Regression (RF) algorithm, also used. The VMD-CLSTM-VMD-ERCRF evaluated from Queensland, Australia. results reveal highly accurate performance all datasets considered, including winter, autumn, spring, summer, yearly predictions. As compared with without (i.e., VMD-CLSTM model), outperforms benchmark models. For predictions, average Legates McCabe Index seen increase by 15.97%, 16.31%, 20.23%, 10.24%, 14.03%, respectively, relative According tests performed on independent datasets, can be practical stratagem useful short-term, forecasting. Therefore research outcomes demonstrate framework effective decision-support tool improving accuracy price. It could value energy companies, policymakers operators develop their insight analysis, distribution optimization strategies.

Language: Английский

---

Understanding electricity prices beyond the merit order principle using explainable AI

Energy and AI, Journal Year: 2023, Volume and Issue: 13, P. 100250 - 100250

Published: March 1, 2023

Electricity prices in liberalized markets are determined by the supply and demand for electric power, which turn driven various external influences that vary strongly time. In perfect competition, merit order principle describes dispatchable power plants enter market of their marginal costs to meet residual load, i.e. difference load renewable generation. Various models based on this when attempting predict electricity prices, yet is fraught with assumptions simplifications thus limited accurately predicting prices. article, we present an explainable machine learning model German day-ahead foregoes aforementioned principle. Our designed ex-post analysis builds features. Using SHapley Additive exPlanation (SHAP) values disentangle role different features quantify importance from empiric data, therein circumvent limitations inherent We show wind solar generation central driving as expected, wherein affects more than Similarly, fuel also highly affect do so a nontrivial manner. Moreover, large ramps correlated high due flexibility nuclear lignite plants. Overall, offer influence main drivers Germany, taking us step beyond explaining relation each other.

Language: Английский

---

Selection of contributing factors for predicting landslide susceptibility using machine learning and deep learning models

Stochastic Environmental Research and Risk Assessment, Journal Year: 2023, Volume and Issue: unknown

Published: Sept. 13, 2023

Language: Английский

---