复杂工况下基于RBF神经网络的全电推进船舶负荷预测  

Research on ship load forecasting of electric propulsion based on rbf neural network under complex working conditions

作　　者：钱宇轩 俞鑫 柴婷逸 严文博 黄云辉[2] 熊松 QIAN Yu-xuan;YU Xin;CHAI Ting-yi;YAN Wen-bo;HUANG Yun-hui;XIONG Song(Changzhou Power Supply Branch of State Grid Jiangsu Electric Power Co.,Ltd.,Changzhou 213000,China;Wuhan University of Technology,Wuhan 430000,China)

机构地区：[1]国网江苏省电力有限公司常州供电分公司,江苏常州213000 [2]武汉理工大学自动化学院,湖北武汉430000

出　　处：《舰船科学技术》2023年第17期97-101,共5页Ship Science and Technology

基　　金：国网江苏省电力有限公司研发服务资助项目(6410402000N6);国家自然科学基金资助项目(52177221)。

摘　　要：全电力推进船舶的复杂工况使其负荷情况难以预测,无法确立精确的数学模型刻画,而应用RBF神经网络建立船舶电力系统负荷预测模型具有可靠性和准确性。通过对全电力推进船舶负荷特点的分析和对RBF神经网络负荷预测的基本原理研究,提出一种基于RBF神经网络的全电力推进船舶的负荷预测方法,选取合理的历史负荷数据,将其归一化处理后输入至RBF神经网络预测模型,再将模型输出反归一化后得到负荷预测结果。在Matlab/Simulink中对某全电力推进船舶在恶劣复杂工况下实际短期运行的负荷情况进行预测,预测准确率高达96.4%。预测结果表明,该方法实现了对复杂工况下全电力推进船舶短期负荷的精准预测,模型拟合程度很高。It is difficult to predict the load of all-electric propulsion ship because of its complex working conditions,and it is difficult to establish an accurate mathematical model.Based on the analysis of the characteristics of the load of the all-electric propulsion ship and the study of the basic principle of the load forecasting by RBF neural network,a load forecasting method of the all-electric propulsion ship based on RBF neural network is proposed,the reasonable historical load data are selected and input into RBF neural network forecasting model after normalization,and then the output of the model is normalized to get the load forecasting result.In Matlab/Simulink,the short-term load of an all-electric propulsion ship under severe and complex conditions is predicted,and the accuracy of the prediction is as high as 96.4%.The prediction results show that the method can accurately predict the short-term load of full-electric propulsion ship,and the fitting degree of the model is very high.

关 键 词：复杂工况 全电力推进船舶 RBF神经网络 负荷预测 

分 类 号：U661[交通运输工程—船舶及航道工程]