Power supply quality prediction method based on LSTM and self-attention mechanism DOI
Yan Yang,

Yu Chang

Journal of Computational Methods in Sciences and Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: April 26, 2025

Existing LSTM-based power quality (PQ) prediction models primarily rely on historical information, which limits their ability to fully capture contextual dependencies. Furthermore, these process inputs sequentially without accounting for the varying importance of different time steps, leading significant inaccuracies. To address limitations, this study proposes an enhanced PQ model that integrates Bidirectional Long Short-Term Memory (BiLSTM) with a Self-Attention (SA) mechanism. The BiLSTM module is introduced both forward and backward temporal dependencies, enabling more comprehensive long-term patterns in series data. SA mechanism dynamically adjusts steps through weighted summation, enhancing model’s focus critical features improving its capacity nonlinear relationships. from layer are then mapped connected generate final outputs. Experiments were conducted using data Nanchang as primary dataset, additional datasets Nanjing, Wuhan, Changsha, Beijing used generalization testing. results demonstrate BiLSTM-SA outperforms traditional LSTM across all metrics, achieving mean absolute error (MAE) 0.09 voltage deviation, 0.05 improvement over single-layer LSTM. Notably, maintains robust performance complex supply scenarios, generalized MAE only 0.2 Beijing. These findings highlight effectiveness combining reducing errors ensuring stability quality, offering advancement methodologies.

Language: Английский

Enhanced multi-energy load forecasting via multi-task learning and GRU-attention networks in integrated energy systems DOI Creative Commons
Swee‐Huay Heng

Electrical Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 4, 2025

Language: Английский

Citations

0
Short-Term Residential Load Forecasting Based on the Fusion of Customer Load Uncertainty Feature Extraction and Meteorological Factors DOI Open Access
Wenzhi Cao,

H. Liu,

Xiangzhi Zhang

et al.

Sustainability, Journal Year: 2025, Volume and Issue: 17(3), P. 1033 - 1033

Published: Jan. 27, 2025

With the proliferation of distributed energy resources, advanced metering infrastructure, and communication technologies, grid is transforming into a flexible, intelligent, collaborative system. Short-term electric load forecasting for individual residential customers playing an increasingly important role in operation planning future grid. Predicting electrical households more challenging with higher uncertainty volatility at household level compared to total feeder regional levels. The previous research results show that accuracy using machine learning single deep model far from adequate there still room improvement.

Language: Английский

Citations

0
Predicting Oil Temperature in Electrical Transformers Using Neural Hierarchical Interpolation DOI Creative Commons
Abdeltif Boujamza, Saâd Lissane Elhaq

Journal of Engineering, Journal Year: 2025, Volume and Issue: 2025(1)

Published: Jan. 1, 2025

Effective electricity consumption planning is critical for power distribution. Ensuring the distribution network aligns with expected demand fluctuations a challenging task influenced by various time‐related and seasonal variables. This study focuses on improving transformer oil temperature forecasting, an indicator of health, using neural hierarchical interpolation time series (NHITS) model. The NHITS model’s architecture designed to handle long‐term forecasting efficiently, making it ideal capturing extended trends in temperature. It incorporates multirate signal sampling via MaxPool layers merge predictions across different scales. proposed methodology involves two key phases: data preparation model development. In phase, (ETT) datasets are used, normalized standard scaler, essential features such as external load selected. During development trained its hyperparameters optimized optimal performance. evaluates performance under conditions, including comparison multivariate univariate series, effects short horizons, impact temporal resolution. was validated ETT dataset, our results were benchmarked against previous that employed same dataset used Informer indicate outperforms model, showing average decrease 51.37% mean squared error (MSE) 37.83% absolute (MAE). These findings highlight ability capture both short‐term characteristics data, promising solution temperatures.

Language: Английский

Citations

0
Cloud-based estimation of lithium-ion battery life for electric vehicles using equivalent circuit model and recurrent neural network DOI
Ziqing Chen, Jianguo Chen,

Zhicheng Zhu

et al.

Journal of Energy Storage, Journal Year: 2025, Volume and Issue: 114, P. 115718 - 115718

Published: Feb. 10, 2025

Language: Английский

Citations

0
Bidirectional Long Short-Term Memory (BiLSTM) Neural Networks with Conjoint Fingerprints: Application in Predicting Skin-Sensitizing Agents in Natural Compounds DOI Creative Commons

Huynh Anh Duy,

Tarapong Srisongkram

Journal of Chemical Information and Modeling, Journal Year: 2025, Volume and Issue: unknown

Published: March 3, 2025

Skin sensitization, or allergic contact dermatitis, represents a critical end point in toxicity assessment, with profound implications for drug safety and regulatory decision-making. This study aims to develop robust deep-learning-based quantitative structure-activity relationship framework accurately predicting skin sensitization toxicity, particularly the context of natural-product-derived compounds. To achieve this, we explored advanced recurrent neural network architectures, including long short-term memory (LSTM), bidirectional LSTM (BiLSTM), gated unit (GRU), GRU, model intricate structure-toxicity relationships inherent molecular We aim optimize improve predictive performance by training cohort 55 models diverse set fingerprints. Notably, BiLSTM model, which integrates SMILES tokens RDKit fingerprints, achieved superior performance, underscoring its capability effectively capture key determinants sensitization. An extensive applicability domain analysis coupled an in-depth evaluation feature importance provided new insights into attributes that influence propensity. further evaluated using natural product data set, where it demonstrated exceptional generalization capabilities. The accuracy 86.5%, Matthews correlation coefficient 75.2%, sensitivity 100%, area under curve 88%, specificity 75%, F1-score 88.8%. Remarkably, categorized products discriminating sensitizing from non-sensitizing agents across various subcategories. These results underscore potential BiLSTM-based as powerful silico tools modern discovery efforts assessments, especially field products.

Language: Английский

Citations

0
Energy Consumption Prediction of Cold Storage Based on LSTM with Parameter Optimization DOI
Yabo Wang,

Junhao Chen,

Bo Cao

et al.

International Journal of Refrigeration, Journal Year: 2025, Volume and Issue: unknown

Published: March 1, 2025

Language: Английский

Citations

0
Feature Engineering for Short Term Residential Load Forecasting Using RNN-Based Neural Networks DOI
Nosirbek N. Abdurazakov, Р. Алиев,

A. Mirzaalimov

et al.

Advances in Science, Technology & Innovation/Advances in science, technology & innovation, Journal Year: 2025, Volume and Issue: unknown, P. 83 - 90

Published: Jan. 1, 2025

Language: Английский

Citations

0
Temperature field and curing degree prediction of large composite blades based on coupled finite element analysis and machine learning DOI Creative Commons

Yue Wu,

Zhong Cao, Chen Liu

et al.

Polymer Composites, Journal Year: 2025, Volume and Issue: unknown

Published: March 27, 2025

Abstract Since large composite blades are variable curvature and thickness components with dimensions, the temperature field analysis will produce a inhomogeneous field, which requires lot of time. In this paper, new method combining finite element machine learning is proposed. By constructing numerical model blade curing using zoned heating to optimize gradient in tongue groove region, maximum reduced by 74.18% degree 21.987% compared conventional profile curing. A long short‐term memory(LSTM) neural network was used predict variations, Grey Wolf algorithm parameters high prediction accuracy. The instructive for online monitoring control process customized hot press tanks. Highlights improves temperature‐field balance. tandem LSTM constructed as an agent model. Enabling be connected cure. Optimizing grey wolf algorithm.

Language: Английский

Citations

0
Optimization and prediction of mechanical properties of TPU-Based wrist hand orthosis using Bayesian and machine learning models DOI
Kaplan Kaplan, Osman Ülkir, Fatma KUNCAN

et al.

Measurement, Journal Year: 2025, Volume and Issue: unknown, P. 117405 - 117405

Published: March 1, 2025

Language: Английский

Citations

0
A Load Classification Method Based on Hybrid Clustering of Continuous–Discrete Electricity Consumption Characteristics DOI Open Access
Jing Li, Yarong Ma, Hao Li

et al.

Processes, Journal Year: 2025, Volume and Issue: 13(4), P. 1208 - 1208

Published: April 16, 2025

There are numerous quantities and types of electrical loads, their characteristics have similarities differences. To adapt to the development trend refined management scheduling on load side, it is necessary explore electricity consumption patterns loads classify them. However, classification performance affected by data redundancy, complexity feature selection, diversity power behavior. imperative based characteristics. Firstly, a statistical analysis load-side data, monthly each throughout year extracted reflect continuous load. By calculating annual rate, maximum utilization hours, rated capacity then using Gaussian Mixture Model (GMM) for clustering analysis, discrete obtained. Then, K-prototypes model, method proposed hybrid setting weight between characteristics, optimal number categories can be determined through elbow method. Finally, 86 industrial electricity-consuming enterprises in region Northwest China as experimental subjects, results demonstrate that this study outperforms K-means, GMM, Gower.

Language: Английский

Citations

0