Hybrid modeling with data enhanced driven learning algorithm for smart generation control in multi-area integrated energy systems with high proportion renewable energy

Expert Systems with Applications, Год журнала: 2024, Номер 261, С. 125530 - 125530

Опубликована: Окт. 9, 2024

Язык: Английский

---

Multi-Scale Building Load Forecasting Without Relying on Weather Forecast Data: A Temporal Convolutional Network, Long Short-Term Memory Network, and Self-Attention Mechanism Approach

Buildings, Год журнала: 2025, Номер 15(2), С. 298 - 298

Опубликована: Янв. 20, 2025

Accurate load forecasting is of vital importance for improving the energy utilization efficiency and economic profitability intelligent buildings. However, restricted in popularization application conventional techniques due to great difficulty obtaining numerical weather prediction data at hourly level requirement conduct predictions on multiple time scales. Under condition lacking meteorological forecast data, this paper proposes utilize a temporal convolutional network (TCN) extract coupled spatial features among multivariate loads. The reconstructed are then input into long short-term memory (LSTM) neural achieve extraction features. Subsequently, self-attention mechanism employed strengthen model’s ability feature information. Finally, carried out through fully connected network, multi-time scale model building loads based TCN–LSTM–self-attention constructed. Taking hospital as an example, predicts cooling, heating, electrical next 1 h, day, week. experimental results show that scales, proposed more accurate than LSTM, CNN-LSTM, TCN-LSTM models. Especially task predicting 1-week scale, achieves improvements 16.58%, 6.77%, 3.87%, respectively, RMSE indicator compared with model.

Язык: Английский

---

Online decoupling feature framework for optimal probabilistic load forecasting in concept drift environments

Applied Energy, Год журнала: 2025, Номер 392, С. 125952 - 125952

Опубликована: Апрель 25, 2025

Язык: Английский

---

A review of deep learning methods for multi-energy load joint forecasting in integrated energy systems

Journal of Physics Conference Series, Год журнала: 2025, Номер 3001(1), С. 012017 - 012017

Опубликована: Апрель 1, 2025

Abstract To accommodate the large-scale integration of renewable energy, and enhance utilization efficiency multiple energy types, such as electricity, gas, cooling, heat, Integrated Energy System (IES) has emerged in recent years. The forecasting loads, is a key challenge guiding operational strategies IES, development deep learning (DL) technology, with its advantages accuracy, provides an effective solution. This review first explains uniqueness challenges IES multi-load forecasting, which involves predicting load time series while accounting for temporal characteristics each their interdependencies. It then summarizes traditional methods analyses DL-based methods, focusing on aspects capability dealing characteristics, coupling, multi-task learning, privacy protection. Finally, future trends DL are discussed.

Язык: Английский

---

Deep reinforcement learning-based multi-objective optimization for electricity-gas-heat integrated energy systems

Expert Systems with Applications, Год журнала: 2024, Номер unknown, С. 125558 - 125558

Опубликована: Окт. 1, 2024

Язык: Английский

---

Two-stage dual-attention spatiotemporal joint network model for multi-energy load prediction of integrated energy system

Sustainable Energy Technologies and Assessments, Год журнала: 2024, Номер 72, С. 104085 - 104085

Опубликована: Ноя. 18, 2024

Язык: Английский

---