Expert Systems with Applications, Год журнала: 2024, Номер unknown, С. 126193 - 126193
Опубликована: Дек. 1, 2024
Язык: Английский
Expert Systems with Applications, Год журнала: 2024, Номер unknown, С. 126193 - 126193
Опубликована: Дек. 1, 2024
Язык: Английский
Expert Systems with Applications, Год журнала: 2024, Номер 259, С. 125376 - 125376
Опубликована: Сен. 11, 2024
Язык: Английский
Процитировано
10Energies, Год журнала: 2024, Номер 17(16), С. 4026 - 4026
Опубликована: Авг. 14, 2024
Accurate and reliable PV power probabilistic-forecasting results can help grid operators market participants better understand cope with energy volatility uncertainty improve the efficiency of dispatch operation, which plays an important role in application scenarios such as trading, risk management, scheduling. In this paper, innovative deep learning quantile regression ultra-short-term power-forecasting method is proposed. This employs a two-branch architecture to forecast conditional power; one branch QR-based stacked conventional convolutional neural network (QR_CNN), other temporal (QR_TCN). The CNN used focus on short-term local dependencies sequences, TCN learn long-term constraints between multi-feature data. These two branches extract different features from input data prior knowledge. By jointly training branches, model able probability distribution obtain discrete forecasts ultra-short term. Then, based these forecasts, kernel density estimation estimate function. proposed innovatively ways priori knowledge injection: constructing differential sequence historical feature provide more information about ultrashort-term dynamics and, at same time, dividing it, together all features, into sets inputs that contain according demand forecasting task; dual-branching designed deeply match corresponding branching computational mechanisms. injection methods effective for performance understandability model. point forecasting, interval probabilistic comprehensively evaluated through case real plant. experimental show performs well task outperforms state-of-the-art models field combined QR. paper technical support scheduling, management time scale system.
Язык: Английский
Процитировано
4Energy Conversion and Management, Год журнала: 2024, Номер 323, С. 119155 - 119155
Опубликована: Ноя. 18, 2024
Язык: Английский
Процитировано
3Sustainable Energy Technologies and Assessments, Год журнала: 2025, Номер 75, С. 104242 - 104242
Опубликована: Фев. 25, 2025
Язык: Английский
Процитировано
0Mathematics, Год журнала: 2025, Номер 13(5), С. 782 - 782
Опубликована: Фев. 26, 2025
Considering the influence of high-speed obstacle avoidance trajectory in optimization design stage intelligent bus aerodynamic shape. A collaborative method aiming at structure and control system for rollover stability is proposed to reduce interference lateral load caused by large side area on driving improve safety process. At conceptual stage, a multidisciplinary co-design frame aerodynamics/dynamics/control built, an adaptive Gaussian Process Regression approximate modeling establish model high-precision high-efficiency evaluation index with as objective resistance crosswind constraints. Taking objectives, integrated static structural parameters dynamic buses carried out. The results show that MDO can obtain shape vehicle body low sensitivity safe stable trajectory. Compared initial trajectory, peak transfer rate during process decreases 33.91%, which significantly reduces risk rollover. traditional serial method, has obvious advantages further performance buses.
Язык: Английский
Процитировано
0Renewable Energy, Год журнала: 2025, Номер unknown, С. 122945 - 122945
Опубликована: Март 1, 2025
Язык: Английский
Процитировано
0Sustainability, Год журнала: 2025, Номер 17(7), С. 3239 - 3239
Опубликована: Апрель 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.
Язык: Английский
Процитировано
0Lecture notes in electrical engineering, Год журнала: 2025, Номер unknown, С. 226 - 240
Опубликована: Янв. 1, 2025
Язык: Английский
Процитировано
0Energy, Год журнала: 2025, Номер unknown, С. 136309 - 136309
Опубликована: Апрель 1, 2025
Язык: Английский
Процитировано
0Expert Systems with Applications, Год журнала: 2025, Номер unknown, С. 127932 - 127932
Опубликована: Апрель 1, 2025
Язык: Английский
Процитировано
0