基于SSA-VMD-BiLSTM模型的充电站负荷预测方法

扫码查看

原文链接

国家科技期刊平台
NETL
NSTL
万方数据

中文摘要：随着电动汽车的普及,维护电网安全稳定运行的压力越来越大,为制定实施高效的需求响应策略,充电站的短期负荷预测尤为重要.然而针对充电站电力负荷变化不稳定、影响因素多的问题,现有方法缺乏对负荷数据的噪声分离和平滑分解以及对分解后的负荷数据进行针对性分析,为进一步提高充电站负荷预测精度,提出一种基于麻雀搜索算法(sparrow search algorithm,SSA)优化变分模态分解(variational mode decomposition,VMD)算法结合双向长短期记忆(bidirectional long short-term Memory,BiLSTM)神经网络的短期负荷预测方法.首先利用SSA算法优化 VMD的参数,再通过 VMD 将不平稳的负荷数据分解成噪声集中的非周期性主分量和多个平滑的周期性分量;鉴于 2 种分量数据前后依赖性的不同,对多个周期性的分量直接基于历史数据结合BiLSTM模型方法进行负荷预测;对噪声集中的非周期性的主分量,考虑其负荷变化的不确定性,分析主要外部原因,基于特征因素数据结合 BiLSTM模型方法进行预测.最后通过结果重构的方式得到综合预测结果.通过算例分析,考虑误差评估参数将所提方法与其他模型方法进行对比,验证所提方法具有更高的精确度、可靠性.

外文标题：Load Prediction Method of Charging Station Based on SSA-VMD-BiLSTM Model

外文摘要：With the popularity of electric vehicles,the pressure on the safe and stable operation of the power grid is increasing.In order to develop and implement an efficient demand response strategy,short-term load prediction of charging stations is particularly important.However,in view of the unstable changes of the power load of charging stations and the many influencing factors,the existing methods lack noise separation and smooth decomposition of the load data and targeted analysis of the decomposed load data.In order to further improve the forecasting accuracy of charging station load,this paper presents a short-term load prediction method based on sparrow search algorithm(SSA)optimization variational mode decomposition(VMD)algorithm combined with bidirectional long short-term memory neural network(BiLSTM).Firstly,the SSA algorithm is used to optimize VMD parameters,and then the unsteady load data is decomposed into aperiodic principal component in noise concentration and multiple smooth periodic components through VMD.In view of the different dependence of the two component data before and after,multiple periodic components are directly based on historical data and combined with BiLSTM model method for load prediction.For the aperiodic principal component in the noise concentration,the paper considers the uncertainty of load change,analyzes the main external causes and makes the prediction based on the characteristic factor data and BiLSTM model method.Finally,the comprehensive prediction results are obtained by means of result reconstruction.The proposed method is compared with other models by example analysis,considering the error evaluation parameters,to verify it has higher accuracy and reliability.

外文关键词：

charging station loadshort-term load forecastingvariational mode decomposition(VMD)sparrow search algorithmbidirectional long and short memory neural network

作者：

林彦旭、高辉

展开 >

作者单位：

南京邮电大学自动化学院/人工智能学院,江苏南京 210003

关键词：

充电站负荷短期负荷预测变分模态分解麻雀搜索算法双向长短期记忆神经网络

基金：

国家自然科学基金项目国家电网公司总部科技项目

项目编号：

520771075400-202416211A-1-1-ZN

出版年：

2024

DOI：

10.3969/j.issn.1007-290X.2024.06.006

广东电力

广东电网公司电力科学研究院，广东省电机工程学会

广东电力

CSTPCD北大核心

影响因子：0.527

ISSN：1007-290X

年,卷(期)：2024.37(6)