Deep Learning Forecasting Model for Market Demand of Electric Vehicles DOI Creative Commons
Ahmed İhsan Şimşek, Erdinç Koç, Beste DESTİCİOĞLU

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(23), P. 10974 - 10974

Published: Nov. 26, 2024

The increasing demand for electric vehicles (EVs) requires accurate forecasting to support strategic decisions by manufacturers, policymakers, investors, and infrastructure developers. As EV adoption accelerates due environmental concerns technological advances, understanding predicting this becomes critical. In light of these considerations, study presents an innovative methodology demand. This model, called EVs-PredNet, is developed using deep learning methods such as LSTM (Long Short-Term Memory) CNNs (Convolutional Neural Networks). model comprises convolutional, activation function, max pooling, LSTM, dense layers. Experimental research has investigated four different categories vehicles: battery (BEV), hybrid (HEV), plug-in (PHEV), all (ALL). Performance measures were calculated after conducting experimental studies assess the model’s ability predict vehicle When performance (mean absolute error, root mean square squared R-Squared) EVs-PredNet machine regression are compared, proposed more effective than other methods. results demonstrate effectiveness approach in considered have significant application potential assessing vehicles. aims improve reliability future market develop relevant approaches.

Language: Английский

Resilience Evaluation and Its Spatiotemporal Analysis of China’s NEV Industry Using Enhanced GRA-CRITIC-CPM DOI

Qiong Yang,

Haibin Liu

Journal of Cleaner Production, Journal Year: 2025, Volume and Issue: unknown, P. 145360 - 145360

Published: March 1, 2025

Language: Английский

Citations

1
Forecasting of New Energy Vehicle Sales and Evaluation of Regional Development Based on BP Neural Network and EWM-TOPSIS DOI
Xieyang Wang, Zirui Xu, Y. Li

et al.

Highlights in Business Economics and Management, Journal Year: 2025, Volume and Issue: 53, P. 97 - 110

Published: March 17, 2025

This paper focuses on the new energy vehicle market, utilizing big data technology and artificial intelligence algorithms to perform statistics, analysis, forecasting in both temporal spatial dimensions. In time dimension, sales volume is forecasted by piecewise cubic Hermite interpolation, polynomial fitting, ARIMA model BP neural network model, results between different models are compared analyzed. Meanwhile, factors affecting this market analyzed using entropy weight method. development level of each province assessed TOPSIS comprehensive evaluation method, stage which provinces located classified K-means cluster analysis. The show that developing rapidly, but there still problem uneven some regions. At same time, study also found has higher credibility prediction, method EWM-TOPSIS can effectively assess city, analysis intuitively differences stage. research provide technical support theoretical for industrial China's era data.

Language: Английский

Citations

0
A study on monthly sales forecasting of new energy vehicles in urban areas using the WOA-BiGRU model DOI Creative Commons

X. Li

PLoS ONE, Journal Year: 2025, Volume and Issue: 20(4), P. e0320962 - e0320962

Published: April 21, 2025

To accurately predict the sales of new energy vehicles (NEVs) in Chinese cities and explore applicability optimization algorithms for GRU models forecasting urban NEV sales., this paper conducts a spatiotemporal analysis data. The Whale Optimization Algorithm (WOA) is then employed to optimize parameters Bidirectional Gated Recurrent Unit (BiGRU) model, thereby proposing WOA-BiGRU-based model monthly prediction NEVs. Its results are compared with those particle swarm (PSO) algorithm. research findings as follows: growth has reversed declining trend overall automobile China; Cities higher predominantly concentrated four major economic hubs--the Pearl River Delta, Yangtze Beijing-Tianjin-Hebei region, Chengdu-Chongqing. techniques such WOA can improve accuracy predicting city-level NEV. WOA-BiGRU outperforms both standalone BiGRU PSO models, achieving Mean Absolute Error (MAE) 3051.89, which 526.18 lower than 104.72 that model. This study improves NEVs, offering critical insights development industry China, deployment charging infrastructure, stabilization power grid, emission reduction transportation sector.

Language: Английский

Citations

0
Approach towards the Purification Process of FePO4 Recovered from Waste Lithium-Ion Batteries DOI Open Access
Liuyang Bai,

Guangye Liu,

Yufang Fu

et al.

Processes, Journal Year: 2024, Volume and Issue: 12(9), P. 1861 - 1861

Published: Aug. 31, 2024

The rapid development of new energy vehicles and Lithium-Ion Batteries (LIBs) has significantly mitigated urban air pollution. However, the disposal spent LIBs presents a considerable threat to environment. Recycling these waste not only addresses environmental issues but also compensates for resource shortages generates substantial economic benefits. Current recycling processes primarily focus on extraction valuable metals, often overlooking treatment residual post-extraction. This project targets iron phosphate (FePO4) derived from lithium (LFP) battery materials, proposing direct acid leaching purification process obtain high-purity phosphate. purified can then be used preparation LFP aiming establish complete regeneration cycle that recovers carbonate materials production LFP. study investigates parameters such as types concentrations, time, number cycles. results demonstrate that, after purification, levels impurity metals decrease while content increases correspondingly. Under optimized experimental conditions, dilute sulfuric rates Al, Cu, Ca, Ni reached 36.0%, 51.4%, 89.5%, 90.9%, respectively. Furthermore, hydrothermal in phosphoric achieved 87.9%, 85.8%, 98.4%, 99.1% Ni, microstructure characterization revealed significant changes phase grain morphology during acid, which are likely associated with liberation atoms lattice. These findings indicate is highly effective removing impurities recycled LIBs.

Language: Английский

Citations

2
New Energy Vehicle Development and Electricity Demand Forecasting Based on Random Forest Model DOI Creative Commons
Lin Zhou, Kun Wang, Weiwei Zhang

et al.

E3S Web of Conferences, Journal Year: 2024, Volume and Issue: 573, P. 02014 - 02014

Published: Jan. 1, 2024

With the implementation of green economy and decarbonization strategy, new energy automobile industry has developed rapidly in China, which poses challenges to balance stability power system. This paper predicts development trend China's vehicle through random forest model, analyses impact vehicles on demand. The results show that number China is expected increase significantly, accounting for a quarter total vehicles, charging piles will significantly meet industry, demand electricity load whole society are maintain rapid growth, supply grid. study provides an important reference government regulation, grid adaptation enterprise planning.

Language: Английский

Citations

0
Deep Learning Forecasting Model for Market Demand of Electric Vehicles DOI Creative Commons
Ahmed İhsan Şimşek, Erdinç Koç, Beste DESTİCİOĞLU

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(23), P. 10974 - 10974

Published: Nov. 26, 2024

The increasing demand for electric vehicles (EVs) requires accurate forecasting to support strategic decisions by manufacturers, policymakers, investors, and infrastructure developers. As EV adoption accelerates due environmental concerns technological advances, understanding predicting this becomes critical. In light of these considerations, study presents an innovative methodology demand. This model, called EVs-PredNet, is developed using deep learning methods such as LSTM (Long Short-Term Memory) CNNs (Convolutional Neural Networks). model comprises convolutional, activation function, max pooling, LSTM, dense layers. Experimental research has investigated four different categories vehicles: battery (BEV), hybrid (HEV), plug-in (PHEV), all (ALL). Performance measures were calculated after conducting experimental studies assess the model’s ability predict vehicle When performance (mean absolute error, root mean square squared R-Squared) EVs-PredNet machine regression are compared, proposed more effective than other methods. results demonstrate effectiveness approach in considered have significant application potential assessing vehicles. aims improve reliability future market develop relevant approaches.

Language: Английский

Citations

0