基于神经网络模型的船舶电网短期电力负荷预测  

Short-term load forecasting for marine electric network based on neural network models

作　　者：张宇涵 高海波[1] 商蕾[1] 林治国[1] 陈亚杰[2] ZHANG Yuhan;GAO Haibo;SHANG Lei;LIN Zhiguo;CHEN Yajie(School of Energy and Power Engineering,Wuhan University of Technology,Wuhan 430063,China;No.711 Research Institute,China Shipbuilding Industry Corporation,Shanghai 201101,China)

机构地区：[1]武汉理工大学能源与动力工程学院,湖北武汉430063 [2]中国船舶重工集团公司第七一一研究所,上海201101

出　　处：《应用科技》2021年第5期12-15,22,共5页Applied Science and Technology

基　　金：国家自然科学基金重点项目(U1709215)国家自然科学基金项目(51579200);中央高校基本科研业务费项目(2018III053GX)。

摘　　要：恶劣海况时电力推进船舶的电网负荷波动较大,发电机组会频繁投入或退出电网,准确的电力负荷预测将有助于优化能量管理策略,保障电力系统的安全性,并提升用电效率。人工神经网络拥有很强的学习能力和泛化能力,能够有效的进行短期电力负荷预测。通过对反向传播(BP)、径向基神经网络(RBF)、Elman共3种不同的网络模型进行原理阐述、数据处理、模型建立及参数调整后,再对其在短期电力负荷预测的表现进行比较,RBF神经网络的预测效果及各项评价指标最优,且其模型建立最简便,因此相较于另外两种网络更适合进行短期电力负荷预测。The electric power load of an electric propulsion ship fluctuates greatly under rough sea,and the generator sets will be frequently put into or out of the power grid.Accurate power load forecasting will be conductive to optimizing power management system,so as to ensure the security of power system and improve power efficiency.Artificial neural network has strong learning ability and generalization ability,and can effectively carry out short-term power load forecasting.After principle elaboration,data processing,modelling and parameter adjustment of three different network models—back propagation(BP),radial basis function(RBF)and Elman,their performance in short-term power load forecasting is compared.It is shown that RBF neural network has the best forecasting effect and the highest score of various evaluation indexes,and that its model is the easiest to build,for which it is more suitable for short-term power load forecasting than the other two networks.

关 键 词：短期电力负荷预测 BP神经网络 RBF神经网络 ELMAN神经网络 船舶电网 计算机应用 人工智能 机器学习 

分 类 号：TM715[电气工程—电力系统及自动化]