在复杂海洋环境中，船舶能源管理系统对既能提供高精度又能实时响应的预测模型的需求越来越大。与传统预测方法相比，由于强大的学习能力和泛化能力，人工神经网络算法在处理船舶电力负荷预测等非线性数据方面具有潜在优势，但每种神经网络智能算法都有其独特的优势、局限性和应用合理性。采用RBF、BP、Elman和LSTM这4种类型的智能方法来预测船舶在恶劣海况下的短期电力负荷。实验结果表明，RBF网络预测模型的平均相对误差、均方根误差等评价指标优于其他神经网络，RBF神经网络在收敛速度、预测精度和泛化能力方面表现最优，是预测船舶电力负荷的有效工具，为船舶电力系统的实时优化调度和能效管理提供了参考。
As electric propulsion systems in maritime vessels continue to advance rapidly, the proportion of electric propulsion load can now exceed 70% of the total installed capacity, significantly escalating the demands placed on energy management systems within power stations. In the challenging maritime environment, there is an increasing demand for forecasting models that offer both high accuracy and rapid response times. Compared to traditional forecasting methods, artificial neural networks excel in handling nonlinear data, particularly in the context of power load predictions. Owing to their robust learning capabilities and generalization strengths, artificial neural networks have emerged as a focal point in the field of load forecasting. Each intelligent algorithm presents distinct advantages, limitations, and specific contexts for rational application. Four types of intelligent methods, namely RBF, BP, Elman and LSTM, are adopted to predict the short-term power load of ships under adverse sea conditions. Experimental results demonstrate that the evaluation metrics of the RBF neural network predictive model surpass those of the other networks. The RBF neural network model is not only simple to establish and quick to converge, but also exhibits low error rates, making it an effective tool for forecasting power loads, thus providing a solid foundation for optimal energy scheduling and enhancing the efficiency of the ship's power network.
2026,48(1): 174-181 收稿日期：2025-3-31
DOI：10.3404/j.issn.1672-7649.2026.01.025
分类号：U665.1
基金项目：广西自然科学基金资助项目（2025JJH160005）；广西高校中青年教师科研基础能力提升项目（2024KY0442）；钦州市科技计划项目（20233219）
作者简介：蒙宁佳(1989-),男,硕士,讲师,研究方向为船舶电力负荷预测、智能优化算法
参考文献：
[1] ZHOU B W, WANG H, XIE Y C, et al. Regional short-term load forecasting method based on power load characteristics of different industries[J]. Sustainable Energy, Grids and Networks, 2024, 38: 1-13.
[2] CHEN Y M, SUN B Z, XIE X W, et al. Short-term forecasting for ship fuel consumption based on deep learning[J]. Ocean Engineering, 2024, 301: 1-11.
[3] XU Y, LI J, DONG Y, et al. Power load combination forecasting system based on longitudinal data selection[J]. Engineering Applications of Artificial Intelligence, 2024, 130: 1-17.
[4] AHMAD O A. Effective RNN-based forecasting methodology design for improving short-term power load forecasts: application to large-scale power-grid time series[J]. Journal of Computational Science, 2023, 68: 1-13.
[5] ZHANG Y, LIU L F, YUAN F M, et al. Multifactor and multiscale method for power load forecasting[J]. Knowledge-Based Systems, 2023, 268: 1-9.
[6] ZHENG X Q, RAN X Y, CAI M X. Short-term load forecasting of power system based on neural network intelligent algorithm[J]. IEEE Access, 2024, 1: 1-8.
[7] FENG Q M, QIAN S P. Research on power load forecasting model of economic development zone based on neural network[J]. Energy Reports, 2021, 7: 1447-1452.
[8] 严文博, 黄云辉, 熊斌宇, 等. 基于神经网络的船舶综合电力系统负荷组合预测研究[J]. 舰船科学技术, 2024, 46(7): 112-120.
YAN W B, HUANG Y H, XIONG B Y, et al. Research on load combination forecasting of ship integrated power system based on neural network[J]. Ship Science and Technology, 2024, 46(7): 112-120.
[9] 刘银波, 吕越. RBF神经网络在船舶电力负荷预测中的应用[J]. 舰船科学技术, 2021, 43(18): 121-123.
LIU Y B, LV Y. Application of RBF neural network in ship power load forecasting[J]. Ship Science and Technology, 2021, 43(18): 121-123.
[10] 蒙宁佳, 梅家豪. 船舶相关智能算法研究综述[J]. 珠江水运, 2025(4): 72-75.
MENG N J, MEI J H. Research review on ship-related intelligent algorithms[J]. Waterage of Pearl River, 2025(4): 72-75.
[11] HAN X S, SHI Y, TONG R J, et al. Research on short-term load forecasting of power system based on IWOA-KELM[J]. Energy Reports, 2023(9): 238-246.
[12] SHENG Z Y, AN Z Y, WANG H W, et al. Residual LSTM based short-term load forecasting[J]. Applied Soft Computing, 2023, 144: 1-14.
[13] CHEN H H, ZHU M Y, HU X, et al. Research on short-term load forecasting of new-type power system based on GCN-LSTM considering multiple influencing factors[J]. Energy Reports, 2023(9): 1022-1031.
[14] HAN L J, WANG X H, YU Y, et al. Power Load Forecast Based on CS-LSTM Neural Network[J]. Mathematics, 2024(12): 1-16.
[15] FAN C D, NIE S H, XIAO L Y, et al. A multi-stage ensemble model for power load forecasting based on decomposition, error factors, and multi-objective optimi-zation algorithm[J]. International Journal of Electrical Power & Energy Systems, 2024, 155: 1-19.
[16] RASOUL J, MAHMOUD R D, SAYED F G. SAC-ConvLSTM: A novel spatio-temporal deep learning-based approach for a short term power load forecasting[J]. Expert Systems with Applications, 2024, 237: 1-14.
[17] 赵亮. 电力推进船舶的能量管理与负荷预测研究[D]. 武汉: 武汉理工大学, 2016.
