In the rapidly evolving world of electric vehicles (EVs), predicting the charging demand power accurately is crucial to prevent overcharging and optimize energy use. A recent study published in ‘Shuju Caiji Yu Chuli’ (translated to ‘Data Collection and Processing’) tackles this very issue, offering a novel approach that could have significant implications for the maritime sector, where EVs are increasingly being considered for shore-to-ship power solutions and onboard applications.
The research, led by SONG Zongren from the Logistics Engineering College at Shanghai Maritime University, introduces a method that combines reinforcement learning (RL) and a variable weight combination model to predict EV charging demand power more accurately. The study addresses the challenges posed by the complexity of battery pack physical models and the limitations of single prediction models.
SONG Zongren and his team improved the traditional grey wolf optimization (GWO) algorithm by incorporating chaos mapping and elite reverse learning strategies. This enhancement optimizes the parameters in the least square support vector machine (LSSVM) algorithm using a dynamic weight strategy based on reinforcement learning. The researchers then used a variable weight combination method to distribute weights reasonably between the extreme learning machine prediction model and the improved LSSVM prediction model. This approach aims to overcome the shortcomings of relying on a single prediction model.
The results are promising. When tested with actual EV charging data, the new method showed a prediction accuracy improvement of 4.75%, 3.84%, and 0.38% compared to three traditional methods. “The prediction accuracy of the new method is significantly higher, which is crucial for preventing overcharging and optimizing energy use,” SONG Zongren noted.
For the maritime sector, the implications are substantial. As shipping companies and port authorities increasingly adopt EVs for various applications, accurate prediction of charging demand power can lead to more efficient energy management and cost savings. This is particularly relevant for shore-to-ship power solutions, where optimizing energy use can reduce operational costs and environmental impact.
Moreover, the maritime industry is exploring the use of EVs for onboard applications, such as auxiliary power units and electric propulsion systems. Accurate prediction of charging demand power can enhance the reliability and efficiency of these systems, contributing to the overall sustainability and cost-effectiveness of maritime operations.
The study highlights the potential of combining machine learning techniques with traditional optimization algorithms to solve complex problems in the EV sector. As the maritime industry continues to embrace electrification, such innovative approaches can pave the way for more efficient and sustainable operations.
In summary, the research by SONG Zongren and his team offers a promising solution for accurately predicting EV charging demand power, with significant commercial impacts and opportunities for the maritime sector. The findings, published in ‘Shuju Caiji Yu Chuli’, underscore the importance of leveraging advanced machine learning techniques to address the challenges of EV integration in various industries, including maritime.