EDformer family: End-to-end multi-task load forecasting frameworks for day-ahead economic dispatch

Applied Energy, Journal Year: 2025, Volume and Issue: 383, P. 125319 - 125319

Published: Jan. 15, 2025

Language: Английский

---

An energy consumption prediction approach in smart cities by CNN-LSTM network improved with game theory and Namib Beetle Optimization (NBO) algorithm

The Journal of Supercomputing, Journal Year: 2025, Volume and Issue: 81(2)

Published: Jan. 20, 2025

Language: Английский

---

PilotCareTrans Net: an EEG data-driven transformer for pilot health monitoring

Frontiers in Human Neuroscience, Journal Year: 2025, Volume and Issue: 19

Published: Jan. 29, 2025

Introduction In high-stakes environments such as aviation, monitoring cognitive, and mental health is crucial, with electroencephalogram (EEG) data emerging a keytool for this purpose. However traditional methods like linear models Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) architectures often struggle to capture the complex, non-linear temporal dependencies in EEG signals. These approaches typically fail integrate multi-scale features effectively, resulting suboptimal intervention decisions, especially dynamic, high-pressure pilot training. Methods To overcome these challenges, study introduces PilotCareTrans Net, novel Transformer-based model designed decision-making aviation students. The incorporates dynamic attention mechanisms, convolutional layers, feature integration, enabling it intricate dynamics more effectively. Net was evaluated on multiple public datasets, including MODA, STEW, SJTUEmotion EEG, Sleep-EDF, where outperformed state-of-the-art key metrics. Results discussion experimental results demonstrate model's ability not only enhance prediction accuracy but also reduce computational complexity, making suitable real-time applications resource-constrained settings. findings indicate that holds significant potential improving cognitive strategies thereby contributing enhanced safety performance critical environments.

Language: Английский

---

Enhancing Energy Consumption in Automotive Component Manufacturing: A Hybrid Autoregressive Integrated Moving Average–Long Short-Term Memory Prediction Model

Sustainability, Journal Year: 2025, Volume and Issue: 17(4), P. 1586 - 1586

Published: Feb. 14, 2025

The automotive industry faces continuing challenges with regard to advancing sustainability and reducing energy consumption vehicle emissions. South Africa accounts for half of the total CO2 emissions in is world’s 12th-largest emitter. In this study, we aimed develop a model combining autoregressive integrated moving averages (ARIMAs) long short-term memory (LSTM) determine best fit prediction using lowest root mean square error configuration enhance component manufacturing. ARIMA dissects time-series data into components level, trend, seasonality, while automatic function refines parameters. Simultaneously, utilizing historical data, LSTM uses specific algorithms predict future electricity generation carbon component’s manufacturing sector. According our results, predicted variables’ interdependence revealed an enhancement intensity body part products equal 29%, cumulative savings 7.22%, increase efficiency 16.25%. Our model’s predictive fitness holds significant potential allowing manufacturers make informed economic technical decisions toward development low-carbon products. Critically, improved activities critical lowering consumption, greenhouse gas emissions, sustainable transportation, production costs.

Language: Английский

---

Statistical evaluation of a diversified surface solar irradiation data repository and forecasting using a recurrent neural network-hybrid model: A case study in Bhutan

Renewable Energy, Journal Year: 2025, Volume and Issue: unknown, P. 122706 - 122706

Published: March 1, 2025

Language: Английский

---

Leveraging IoT-enabled machine learning techniques to enhance electric vehicle battery state-of-health prediction

Journal of Energy Storage, Journal Year: 2025, Volume and Issue: 120, P. 116409 - 116409

Published: April 3, 2025

Language: Английский

---

Drift Detection Methods on Machine Learning Systems: a Discussion over Discrete Live Data

Published: May 7, 2025

Context: Machine learning has become an essential tool for addressing complex problems in information systems, encompassing industrial, commercial, and residential applications. Problem: systems without frequent retraining are prone to data concept drift, compromising predictive accuracy. This issue is particularly critical scenarios where infeasible due high computational costs or unavailability. Solution: study evaluates the performance of drift detection methods discrete time series with controlled changes mean standard deviation using synthetic Gaussian signals. IS Theory: The General Systems Theory underpins by emphasizing how interplay between adaptive contributes maintaining stability efficiency dynamic environments. Method: Experiments were conducted variations mean, deviation, both parameters simultaneously order obtain qualitative patterns detectors behaviors. ADWIN, KSWIN, Page-Hinkley tested under this scenario. Summary Results: findings reveal that ADWIN exhibited greater precision robustness, while KSWIN showed excessive sensitivity, leading a number false positives. Contributions Field: research offers comprehensive analysis detectors’ performance, specifically involving providing useful reference designing resilient machine learning-based forecasting systems. Impacts on advances development can adapt environments characterized shifts direct applications industrial contexts energy management.

Language: Английский

---

Prediction of Surrounding Rock Deformation in a Highway Tunnel Using an LSTM‐RF Hybrid Model

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

Published: Jan. 1, 2025

Accurate tunnel deformation prediction is critical for mitigating construction risks and ensuring stability. This study introduces a novel hybrid model integrating long short‐term memory (LSTM) networks random forest (RF) to enhance the precision of predictions during construction. Bayesian optimization was utilized fine‐tune parameters, optimal performance. Validated with multidepth data from Yangjiashan highway in China, demonstrates remarkable adaptability complex geological conditions. The results show that LSTM‐RF achieves mean square error (MSE) 0.0025, root‐mean‐square (RMSE) 0.0052, coefficient determination ( R 2 ) 0.9810, outperforming individual models other frameworks predicting trends. By effectively capturing temporal dependencies modeling nonlinear residuals, provides robust reliable solution improving safety efficiency tunneling projects. These findings emphasize potential approaches geotechnical engineering, particularly predictive maintenance infrastructure monitoring.

Language: Английский

---

EleKAN: Temporal Kolmogorov-Arnold Networks for Price and Demand Forecasting Framework in Smart Cities

Lecture notes in electrical engineering, Journal Year: 2025, Volume and Issue: unknown, P. 182 - 192

Published: Jan. 1, 2025

Language: Английский

---

Exploring the relationship between tourism development and environmental pollution using an LSTM-based time series model

Frontiers in Environmental Science, Journal Year: 2025, Volume and Issue: 13

Published: May 15, 2025

With the rapid development of tourism, understanding its relationship with environmental pollution has become a critical issue. Traditional research methods often struggle to effectively capture complex time series data and nonlinear associations, limiting their ability accurately analyze predict interactions between tourism changes. In response these challenges, this introduces modeling framework leveraging LSTM-Attention-Random Forest (LARF). The LSTM model captures temporal dynamics in data, Attention mechanism enhances focus on steps, Random improves prediction accuracy by relationships through ensemble learning. Experimental results demonstrate that LARF significantly outperforms traditional generalization across multiple datasets, an average improvement 18.2% MSE 16.5% MAPE compared baseline models like LSTM, GRU, Forest. Specifically, achieves 30.0 Global Tourism Data 35.0 China City Air Quality Data, highlighting robustness reliability. Furthermore, provides innovative insights for pollutant risk quantification management, offering actionable recommendations sustainable governance. This study contributes not only advancing methodologies analyzing systems but also offers versatile can be applied other predictive decision support future.

Language: Английский

---