船舶短期电力负荷预测面临负荷序列非线性强、多波动、易受航行工况与环境因素耦合干扰等难题。BP神经网络强大的非线性拟合能力与自适应学习机制，可从复杂的时序负荷数据中自动提取特征，有效逼近和预测此类动态多变的负荷曲线。因此，研究基于BP神经网络的船舶短期电力负荷预测方法。通过拉格朗日插值法对原始船舶多源时序数据进行缺失数据填补，运用改进F-score特征选择算法由填补后多源时序数据中筛选出最优特征子集，以此为输入构建三层结构的BP神经网络模型，得到船舶短期电力负荷预测结果。结果表明，该方法的缺失数据插值填补与最优特征子集筛选效果均较为显著，以此为基础，对海上航行与靠泊装卸两种船舶工况下的24h电力负荷预测误差始终低于0.03，且在负荷波动较大处仍保持较高预测精度，预测性能稳定可靠。
Short term power load forecasting for ships faces challenges such as strong non-linear load sequences, multiple fluctuations, and susceptibility to coupling interference from navigation conditions and environmental factors. The powerful nonlinear fitting ability and adaptive learning mechanism of BP neural network can automatically extract features from complex time-series load data, effectively approximating and predicting such dynamic and variable load curves. Therefore, research is conducted on a short-term power load forecasting method for ships based on BP neural network. By using Lagrange interpolation method to fill in missing data in multi-source time-series data of the original ship, an improved F-score feature selection algorithm is applied to select the optimal feature subset from the filled multi-source time-series data. Based on this, a three-layer BP neural network model is constructed as input to obtain the short-term power load prediction results of the ship. The results showed that the missing data interpolation and optimal feature subset screening effects of this method were significant. Based on this, the 24-hour power load prediction error for both maritime navigation and berthing loading and unloading ship conditions remained below 0.03, and high prediction accuracy was maintained in areas with large load fluctuations, indicating stable and reliable prediction performance.
2026,48(4): 84-88 收稿日期：2025-8-28
DOI：10.3404/j.issn.1672-7649.2026.04.013
分类号：U665；TP183
作者简介：叶明壕(2000-),男,硕士,研究方向为人工智能、神经网络
参考文献：
[1] 陈建勇, 陈亚杰, 高海波, 等. 新型绿色船舶电力推进系统关键技术及应用分析[J]. 船海工程, 2023, 52(6): 1-7
CHEN J Y, CHEN Y J, GAO H B, et al. Analysis of the key technologies and application for new green ship propulsion systems[J]. Ship & Ocean Engineering, 2023, 52(6): 1-7
[2] 王谦, 高海波, 左文. 基于概率稀疏自注意力的船舶短期电力负荷预测[J]. 大连海事大学学报, 2024, 50(1): 134-142
WANG Q, GAO H B, ZUO W. Ships short-term power load prediction based on ProbSparse self-attention[J]. Journal of Dalian Maritime University, 2024, 50(1): 134-142
[3] PANG S, ZOU L, ZHANG L, et al. A hybrid TCN-BiLSTM short-term load forecasting model for ship electric propulsion systems combined with multi-step feature processing[J]. Ocean Engineering, 2025: 1-11
[4] 姜俊道, 邹亮, 刘星斗, 等. 结合负荷多步预测的氢能船舶双层能量调度策略[J]. 电力系统自动化, 2025, 49(13): 166-176
JIANG J D, ZOU L, LIU X D, et al. Bi-level energy scheduling strategy for hydrogen-powered ships combined with multi-step load forecasting[J]. Automation of Electric Power Systems, 2025, 49(13): 166-176
[5] 周寅正, 陈俐. 基于模型预测控制的双机组混合动力船舶能量管理研究[J]. 中国舰船研究, 2024, 19(z1): 74-83
ZHOU Y Z, CHEN L. Study on energy management of dual-diesel generator sets hybrid power ships based on model predictive control[J]. Chinese Journal of Ship Research, 2024, 19(z1): 74-83
[6] 洪远远, 施伟锋. 基于飞轮储能技术的船舶区域配电系统冲击负荷供能策略[J]. 船舶工程, 2023, 45(1): 103-109
HONG Y Y, SHI W F. Impulse load energy supply strategy of ship regional distribution power system based on flywheel energy storage technology[J]. Ship Engineering, 2023, 45(1): 103-109
