Demand Time Series Prediction of Stacked Long Short-Term Memory Electric Vehicle Charging Stations Based on Fused Attention Mechanism

Energies, Journal Year: 2024, Volume and Issue: 17(9), P. 2041 - 2041

Published: April 25, 2024

The layout and configuration of urban infrastructure are essential for the orderly operation healthy development cities. With promotion popularization new energy vehicles, modeling prediction charging pile usage allocation have garnered significant attention from governments enterprises. Short-term demand forecasting piles is crucial their efficient operation. However, existing models lack a discussion on appropriate time window, resulting in limitations station-level predictions. Recognizing temporal nature occupancy, this paper proposes novel stacked-LSTM model called attention-SLSTM that integrates an mechanism to predict electric vehicles at station level over next few hours. To evaluate its performance, compares it with several methods. experimental results demonstrate outperforms both LSTM models. Deep learning methods generally outperform traditional series In test set, MAE 1.6860, RMSE 2.5040, MAPE 9.7680%. Compared model, reduced by 4.7%and 5%, respectively; while value decreases 1.3%, making superior overall. Furthermore, subsequent experiments compare performance among different stations, which confirms exhibits excellent predictive capabilities within six-step (2 h) window.

Language: Английский

---

Renewable and Sustainable Energy—Current State and Prospects

Energies, Journal Year: 2025, Volume and Issue: 18(4), P. 797 - 797

Published: Feb. 8, 2025

The energy transition is seen as a fundamental engine of economic development and factor in improving the quality life [...]

Language: Английский

---

Multi-Step Ageing Prediction of NMC Lithium-Ion Batteries Based on Temperature Characteristics

Batteries, Journal Year: 2024, Volume and Issue: 10(11), P. 384 - 384

Published: Oct. 31, 2024

The performance of lithium-ion batteries depends strongly on their ageing state; therefore, the monitoring and prediction battery state health (SoH) is necessary for an optimized secured functioning systems. This paper evaluates compares three artificial neural network architectures multi-step cells: Recurrent Neural Network (RNN), Gated Unit (GRU) Long short-term memory (LSTM). These models use features extracted from cell’s temperature to predict capacity. are experimental measurements surface selected based Spearman correlation analysis. results were evaluated compared considering different percentages training dataset: 60%, 70%, 80%. Training testing data generated experimentally accelerated cycling tests. During these experiments, four Nickel Manganese Cobalt/Graphite (NMC) cells cycled under a controlled environment dynamic current profile Worldwide Harmonized Light Vehicles Test Cycles.

Language: Английский

---

EXPANDING THE POSSIBILITIES OF USING ENERGY STORAGE AND GENERATION SYSTEMS WITH RENEWABLE SOURCES FOR ENERGY SUPPLY OF RAILWAY INFRASTRUCTURE FACILITIES

Actual problems of improving of current legislation of Ukraine, Journal Year: 2023, Volume and Issue: 11(189), P. 132 - 147

Published: Nov. 14, 2023

The article analyzes the options for implementing renewable energy systems and storage on railway infrastructure rolling stock. It is noted that sector of Ukraine, as well some European countries, not diversified by sources, use sources (RES) limited. possibility using hybrid with photovoltaic batteries a wind generator to meet electricity needs facility considered. concept regulated crossing operating in combination electric vehicles (electric bicycle tire) proposed, which will ensure balancing power supply system, excess RES generation additional needed. In accordance proposed concept, possible scenarios functioning load were identified. consumption estimated typical schedule determined.

Language: Английский

---