基于深度学习的矿井电力短期负荷预测方法

Mine power short-term load forecasting method based on deep learning

短期电力负荷预测能准确评估出煤矿的整体电力负荷变化情况,保证煤矿供电系统的安全与可靠运行。由于煤矿电力负荷预测受多种因素影响,难以实现精确预测,文章针对此问题,基于深度学习理论,提出了一种卷积神经网络(CNN)和双向门控循环单元(BiGRU)相结合的矿井电力短期负荷预测方法,并用于煤矿的实际电力负荷预测中。首先,构建了煤矿电力负荷预测的混合学习模型;然后,给出了数据处理方法,设计了模型评判指标,搭建了仿真平台并进行了多种算法的分析与对比;最后,基于组态软件开发了电力监控与预测系统,并应用于煤矿实际监控中。经现场试验表明,设计的方法可以实现对矿井短期电力负荷的准确预测,为煤矿电力系统的安全运行提供准确的决策支撑。

Short-term power load forecasting can accurately evaluate the change of the whole power load in coal mine,and ensure the safe and reliable operation of the power supply system in coal mine.Due to the influence of many factors,it is difficult to achieve accurate prediction of coal mine electric load prediction.To solve this problem,this paper proposes a short-term mine electric load prediction method based on deep learning theory,which combines convolutional neural network(CNN)and bidirectional gated cycle unit(BiGRU),and applies it to the actual coal mine electric load prediction.Firstly,the hybrid learning model of coal mine power load forecasting is constructed.Then,the data processing method is given,the model evaluation index is designed,the simulation platform is built,and a variety of algorithms are analyzed and compared.Finally,based on the configuration software,the power monitoring and forecasting system is developed and applied to the actual monitoring of coal mine.The field test shows that the proposed method can accurately predict the short-term power load of the mine and provide accurate decision support for the safe operation of the coal mine power system.