Minute-level ultra-short-term power load forecasting based on time series data features

Applied Energy, Journal Year: 2024, Volume and Issue: 372, P. 123801 - 123801

Published: July 3, 2024

Electricity is fundamental to the development of national economies and societies, reliant on accurate power load forecasting for its stable supply. Ultra-short-term analyzes historical data predict changes within next hour. This crucial achieving efficient dispatching, improving emergency management, ensuring operation system. However, with increasingly widespread application renewable energy, inherent intermittency exacerbates complexity randomness loads, posing a challenge models accurately capture features. In addressing this challenge, study presents novel method feature extraction from time series data, aimed at enhancing accuracy forecasting. By analyzing trend, periodicities, randomness, it simplifies complex into several features, significantly reducing noise-induced errors identification understanding Moreover, applies five prevalent deep learning models. Experimental results show that using reduces mean absolute percentage error by an average 54.6905%, 42.6654%, 51.3868% datasets three different substations in China. These not only affirm method's efficacy but also provide new technical foundations reliable functioning future systems.

Language: Английский

---

InfoCAVB-MemoryFormer: Forecasting of wind and photovoltaic power through the interaction of data reconstruction and data augmentation

Applied Energy, Journal Year: 2024, Volume and Issue: 371, P. 123745 - 123745

Published: June 20, 2024

Language: Английский

---

Feature fusion temporal convolution: Wind power forecasting with light hyperparameter optimization

Energy Reports, Journal Year: 2025, Volume and Issue: 13, P. 2468 - 2481

Published: Feb. 8, 2025

Language: Английский

---

A hybrid error correction method based on EEMD and ConvLSTM for offshore wind power forecasting

Ocean Engineering, Journal Year: 2025, Volume and Issue: 325, P. 120773 - 120773

Published: March 3, 2025

Language: Английский

---

FTLNet: federated deep learning model for multi-horizon wind power forecasting

Discover Internet of Things, Journal Year: 2025, Volume and Issue: 5(1)

Published: March 10, 2025

Language: Английский

---

A Novel Hybrid Deep Learning Model for Day-Ahead Wind Power Interval Forecasting

Sustainability, Journal Year: 2025, Volume and Issue: 17(7), P. 3239 - 3239

Published: April 5, 2025

Accurate interval forecasting of wind power is crucial for ensuring the safe, stable, and cost-effective operation grids. In this paper, we propose a hybrid deep learning model day-ahead forecasting. The begins by utilizing Gaussian mixture (GMM) to cluster daily data with similar distribution patterns. To optimize input features, feature selection (FS) method applied remove irrelevant data. empirical wavelet transform (EWT) then employed decompose both numerical weather prediction (NWP) into frequency components, effectively isolating high-frequency components that capture inherent randomness volatility A convolutional neural network (CNN) used extract spatial correlations meteorological while bidirectional gated recurrent unit (BiGRU) captures temporal dependencies within sequence. further enhance accuracy, multi-head self-attention mechanism (MHSAM) incorporated assign greater weight most influential elements. This leads development based on GMM-FS-EWT-CNN-BiGRU-MHSAM. proposed validated through comparison benchmark demonstrates superior performance. Furthermore, forecasts generated using NPKDE shows new achieves higher accuracy.

Language: Английский

---

Short-term wind power forecasting using integrated boosting approach

Frontiers in Energy Research, Journal Year: 2024, Volume and Issue: 12

Published: May 30, 2024

Rapidly increasing global energy demand and environmental concerns have shifted the attention of policymakers toward large-scale integration renewable resources (RERs). Wind is a type RERs with vast potential no pollution associated it. The sustainable development goals: affordable clean energy, climate action, industry, innovation infrastructure, can be achieved by integrating wind into existing power systems. However, will bring instability challenges due to its intermittent nature. Mitigating these necessitates implementation effective forecasting models. Therefore, we proposed novel integrated approach, Boost-LR, for hour-ahead forecasting. Boost-LR multilevel technique consisting non-parametric models, extreme gradient boosting (XgBoost), categorical (CatBoost), random forest (RF), parametric linear regression (LR). first layer uses algorithms that process data according their tree architectures pass intermediary forecast LR which deployed in two processes forecasts one models provide final predicted power. To demonstrate generalizability robustness study, performance compared individual CatBoost, XgBoost, RF, deep learning networks: long short-term memory (LSTM) gated recurrent unit (GRU), Transformer Informer using root mean square error (RMSE), (MSE), absolute (MAE) normalized (NRMSE). Findings effectiveness as superior improvement MAE recorded 31.42%, 32.14%, 27.55% datasets Bruska, Jelinak, Inland farm, respectively CatBoost revealed second-best performing model. Moreover, study also reports literature comparison further validates

Language: Английский

---

Short-Term Prediction of Photovoltaic Power Based on DBSCAN-SVM Data Cleaning and PSO-LSTM Model

Energy Engineering, Journal Year: 2024, Volume and Issue: 121(10), P. 3019 - 3035

Published: Jan. 1, 2024

Accurate short-term photovoltaic (PV) power prediction helps to improve the economic efficiency of stations and is great significance arrangement grid scheduling plans.In order accuracy PV further, this paper proposes a data cleaning method combining density clustering support vector machine.It constructs model based on particle swarm optimization (PSO) optimized Long Short-Term Memory (LSTM) network.Firstly, input features are determined using Pearson's correlation coefficient.The feature information clustered density-based spatial applications with noise (DBSCAN), then, in each cluster cleaned machines (SVM).Secondly, PSO used optimize hyperparameters LSTM network obtain optimal structure.Finally, different models established, generation results obtained.The show that methods effective PSO-LSTM DBSCAN-SVM outperforms existing typical methods, especially under non-sunny days, effectively improves prediction.

Language: Английский

---

Enhancing Wind Power Forecasting Accuracy with Hybrid Deep Learning and Teaching-Learning-Based Optimization

Cleaner Energy Systems, Journal Year: 2024, Volume and Issue: 9, P. 100139 - 100139

Published: Aug. 17, 2024

Language: Английский

---

Ultra-short-term wind power forecasting based on long short-term memory network with modified honey badger algorithm

Energy Reports, Journal Year: 2024, Volume and Issue: 12, P. 3548 - 3565

Published: Sept. 25, 2024

Language: Английский

---