摘要
电力负荷具有不确定性、随机性及波动性的特点,难以对其实现精准预测。为此,提出了一种基于改进型自适应白噪声完备集成经验模态分解算法和经改进型鲸鱼算法优化的双向长短期记忆网络预测模型。首先对选取的澳大利亚某电网负荷数据进行预处理;其次利用经验模态分解方法将负荷数据分解为一系列子序列;然后利用改进的鲸鱼算法对双向长短期记忆网络进行超参数寻优;最后将分解后得到的各分量数据输入到优化模型中进行预测。结果表明,所提算法实现了电力负荷的精准预测,得到了比其他单一基准模型和多数组合模型更好的预测效果,具有一定的适用性与应用价值。
Power load has the characteristics of uncertainty,randomness and volatility,which makes it difficult to achieve accurate prediction.Therefore,a bi-directional long short-term memory network prediction model based on improved whale optimization complete ensemble EMD with adaptive noise algorithm and improved whale algorithm optimization was proposed.Firstly,the load data of a selected Australian power grid was preprocessed;secondly,the empirical mode decomposition method was used to decompose the load data into a series of subsequence;then,the improved whale algorithm was used to optimize the parameters of the bi-directional long short-term memory network;finally,the decomposed data of each component was input into the optimization model for prediction.The results show that the proposed algorithm achieves accurate prediction of power load,achieving better prediction performance than other single benchmark models and most combination models,and has certain applicability and application value.
作者
焦家俊
刘田园
Jiao Jiajun;Liu Tianyuan(School of Electrical and Control Engineering,Liaoning Technical University,Fuxin Liaoning 125100,China;College of International Education,Hebei University of Technology,Tianjin 300401,China)
出处
《电气自动化》
2024年第2期36-39,共4页
Electrical Automation
关键词
电力负荷
负荷预测
经验模态分解算法
改进型鲸鱼算法
双向长短期记忆网络
electric load
load forecasting
empirical modal decomposition algorithm
improved whale optimization algorithm
bi-directional long short-term memory network
作者简介
通信作者:焦家俊(2002-),男,安徽黄山人,主要研究方向为现代人工智能在电力系统方面的应用;刘田园(2002-),女,山西太原人,主要研究方向为现代人工智能在电力系统方面的应用。
