An advanced hybrid deep learning model for accurate energy load prediction in smart building

Energy Exploration & Exploitation, Journal Year: 2024, Volume and Issue: 42(6), P. 2241 - 2269

Published: Aug. 27, 2024

In smart cities, sustainable development depends on energy load prediction since it directs utilities in effectively planning, distributing and generating energy. This work presents a novel hybrid deep learning model including components of the Improved-convolutional neural network (CNN), bidirectional long short-term memory (Bi-LSTM), Graph (GNN), Transformer Fusion Layer architectures for precise forecasting. Better feature extraction results from Improved-CNN's dilated convolution residual block accommodation wide receptive fields reduced vanishing gradient problem. By capturing temporal links both directions, Bi-LSTM networks help to better grasp complicated use patterns. improve predictive capacities across linked systems by characterizing spatial relationships between energy-consuming units cities. Emphasizing critical trends guarantee reliable forecasts, transformer models attention methods manage long-term dependencies consumption data. Combining CNN, Bi-LSTM, GNN component predictions synthesizes numerous data representations increase accuracy. With Root Mean Square Error 5.7532 Wh, Absolute Percentage 3.5001%, 6.7532 Wh R 2 0.9701, fared than other ‘Electric Power Consumption’ Kaggle dataset. develops realistic that helps informed decision-making enhances efficiency techniques, promoting forecasting

Language: Английский

---

A Novel Hybrid Prediction Model of Air Quality Index Based on Variational Modal Decomposition and CEEMDAN-SE-GRU

Process Safety and Environmental Protection, Journal Year: 2024, Volume and Issue: 191, P. 2572 - 2588

Published: Oct. 9, 2024

Language: Английский

---

CRAformer: a cross-residual attention transformer for solar irradiation multistep forecasting

Energy, Journal Year: 2025, Volume and Issue: unknown, P. 135214 - 135214

Published: Feb. 1, 2025

Language: Английский

---

A noise resilient multi-step ahead deep learning forecasting technique for solar energy centered generation of green hydrogen

International Journal of Hydrogen Energy, Journal Year: 2024, Volume and Issue: 90, P. 666 - 679

Published: Oct. 8, 2024

Language: Английский

---

An effective multi-time series model of RC column backbone curve identification

Case Studies in Construction Materials, Journal Year: 2024, Volume and Issue: 20, P. e03183 - e03183

Published: April 21, 2024

Accurate identification of the backbone curves reinforced concrete (RC) columns is key to engineering design and strengthening renovation. In view problems high cost, long time, low accuracy, large dispersion calculation results discontinuous stiffness changes existing curve methods, such as experimental method, finite element simulation method semi-theoretical semi-empirical it proposed transform problem into a multi-time series prediction problem. By introducing attention mechanism combining with bidirectional short-term memory (BiLSTM), model (BC-ABiLSTM) established considering relationship between front back points curves. Compared models for BiLSTM (BC-BiLSTM), (BC-LSTM), multilayer perceptron (BC-MLP), performance BC-ABiLSTM better, mean absolute error (MAE), percentage (MAPE), root square (RMSE), R2 on testing set are 12.492kN, 10.595%, 20.838kN 0.9924, respectively, which provides new accurate, efficient cost-effective RC column under various cyclic loading levels.

Language: Английский

---

Explainable AI and optimized solar power generation forecasting model based on environmental conditions

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(10), P. e0308002 - e0308002

Published: Oct. 2, 2024

This paper proposes a model called X-LSTM-EO, which integrates explainable artificial intelligence (XAI), long short-term memory (LSTM), and equilibrium optimizer (EO) to reliably forecast solar power generation. The LSTM component forecasts generation rates based on environmental conditions, while the EO optimizes model’s hyper-parameters through training. XAI-based Local Interpretable Model-independent Explanation (LIME) is adapted identify critical factors that influence accuracy of in smart systems. effectiveness proposed X-LSTM-EO evaluated use five metrics; R-squared (R 2 ), root mean square error (RMSE), coefficient variation (COV), absolute (MAE), efficiency (EC). gains values 0.99, 0.46, 0.35, 0.229, 0.95, for R , RMSE, COV, MAE, EC respectively. results this improve performance original conventional LSTM, where improvement rate is; 148%, 21%, 27%, 20%, 134% compared with other machine learning algorithm such as Decision tree (DT), Linear regression (LR) Gradient Boosting. It was shown worked better than DT LR when were compared. Additionally, PSO employed instead validate outcomes, further demonstrated efficacy optimizer. experimental simulations demonstrate can accurately estimate PV response abrupt changes patterns. Moreover, might assist optimizing operations photovoltaic units. implemented utilizing TensorFlow Keras within Google Collab environment.

Language: Английский

---

Transfer learning and source domain restructuring-based BiLSTM approach for building energy consumption prediction

International Journal of Green Energy, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 15

Published: Nov. 2, 2024

Currently, building energy consumption prediction typically relies on vast amounts of historical data. However, for newly constructed buildings, the scarcity data leads to reduced accuracy. To address this challenge, paper proposes a novel approach that integrates transfer learning with source domain reconstruction-based BiLSTM model prediction. In first stage, both and target domains are clustered into profile types using k-means. For each type in domain, most similar profiles identified Maximum Mean Discrepancy Dynamic Time Warping. The is then reconstructed by combining these based their proportions domain. Subsequently, feature extraction method EMD-CWT-Conv introduced. Empirical Mode Decomposition applied decompose filter Continuous Wavelet Transform employed extract distinctive frequency-domain time-domain features from Final predictions made fine-tuning. Experiments grocery shop school show proposed reduces Absolute Percentage Error at least 13.19% 17.67%, respectively.

Language: Английский

---

Artificial intelligence modeling for power system planning

Electrical Engineering, Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 11, 2024

Language: Английский

---

Soft sensing of NOx emission from waste incineration process based on data de-noising and bidirectional long short-term memory neural networks

Waste Management & Research The Journal for a Sustainable Circular Economy, Journal Year: 2024, Volume and Issue: unknown

Published: July 30, 2024

Continuous emission monitoring system is commonly employed to monitor NOx emissions in municipal solid waste incineration (MSWI) processes. However, it still encounters the challenges of regular maintenance and measurement lag. These issues significantly impact accurate stable control emissions. Therefore, developing a soft sensor complement hardware becomes imperative. Considering data noise, dynamic nonlinearity, time series characteristics volatility MSWI process, this article introduces model for prediction utilizing complete ensemble empirical mode decomposition adaptive noise (CEEMDAN)-wavelet threshold (WT) method bidirectional long short-term memory (Bi-LSTM). Firstly, original signal decomposed into group intrinsic functions (IMFs) using CEEMDAN. Subsequently, WT processes high-frequency IMFs that are noise-dominant. Then, all reconstructed obtain denoized signal. Finally, Bi-LSTM predict Compared conventional modelling approaches, proposed demonstrates best predictive performance. The mean absolute percentage error, root-mean-squared error average on test set 3.75%, 5.34 mg m −3 4.34 , respectively. provides new sensing It holds significant practical value precise reference research key process parameters.

Language: Английский

---

Multimodel Ensemble Forecast of Global Horizontal Irradiance at PV Stations Based on Dynamic Variable Weight

Journal of Tropical Meteorology, Journal Year: 2024, Volume and Issue: 30(3), P. 327 - 336

Published: Aug. 14, 2024

Language: Английский

---