摘要
电力系统负荷预测实质是对电力市场需求的预测,短期电力负荷预测是电力部门的重要工作之一。目前主要的负荷预测方法有传统预测、灰色预测、混沌理论预测、智能技术预测、优选组合预测等,其中智能预测中最典型的就是人工神经网络。人工神经网络是一个极其复杂的非线性动力学系统。它的自学习功能对预测有着重要的意义,能通过学习历史负荷数据来反映出输入变量和输出变量之间的非线性关系。由于很多因素都会对电力负荷造成影响,因此可以把神经网络算法引用到负荷预测中,提高电力负荷的预测精度。基于宁夏电网短期电力负荷预测的实际需求,提出了一种基于Attention机制优化CNN-GRU混合神经网络的短期负荷预测技术。该技术通过引入Attention机制对CNN-GRU模型进行改进,有效提升了预测精度和可解释性。在宁夏电网实际数据集上进行的仿真实验表明所提出的模型具有较高的预测准确性和可靠性。
The essence of power system load prediction is to predict the demand of power market,and short-term power load prediction is one of the important tasks of power sector.At present,the main load prediction methods include conventional prediction,grey prediction,chaos theory prediction,intelligent technology prediction,optimal combination prediction,etc.Among them,the most typical intelligent prediction method is artificial neural network.Artificial neural networks form highly complex nonlinear dynamic systems.Their self-learning capability is of great significance in prediction,as they can learn from historical load data to capture the nonlinear relationships between input and output variables.Due to the various factors affecting electricity load,employing neural network algorithms in load prediction can significantly enhance prediction accuracy.This paper tries to address the practical demand for short-term load prediction in the Ningxia power grid and proposes a short-term load prediction technique based on an attention-enhanced CNN-GRU hybrid neural network.This technology improves the CNN-GRU model by introducing an attention mechanism,effectively enhancing prediction accuracy and interpretability.Simulation experiments were conducted on the actual dataset of Ningxia Power Grid,and the results showed that the model proposed in this paper has high prediction accuracy and reliability.
作者
刘会
岳东明
苗光尧
王乐乐
王国彬
朱慧娴
LIU Hui;YUE Dongming;MIAO Guangyao;WANG Lele;WANG Guobin;ZHU Huixian(State Grid Ningxia Marketing Service Center-State Grid Ningxia Metrology Center,Yinchuan 750002,China;Beijing Tsingsoft Technology Co.,Ltd.,Beijing 100000,China)
出处
《电工技术》
2024年第9期20-23,共4页
Electric Engineering
关键词
短期负荷预测
卷积神经网络
门控循环单元
注意力机制
short-term power prediction
convolutional neural network
gated recurrent unit
attention mechanism
