【作者】 康忠健； 罗霖；

【Author】 Zhongjian Kang;Lin Luo;College of New Energy,China University of Petroleum;

【机构】 中国石油大学(华东)新能源学院；

【摘要】 配电网线损率是电力企业的重要考核指标之一,对电力企业的经济效益有着重大影响,所以对线损异常情况进行精确诊断成为了电力企业迫切需要解决的问题。本文提出一种基于长短期记忆神经网络的配电网线损预测模型,首先,通过粗时间粒度的线损预测,将本文预测模型与BP神经网络、径向基神经网络进行对比试验,验证了本文预测模型的准确性优势。其次,使用本文预测模型在细时间粒度上进行线损预测。最后,利用某市电力公司所管辖的配电网采集数据进行仿真试验,试验结果表明,在大数据集情况下,本文所提基于长短期记忆网络的预测模型准确率高于其他神经网络算法。更多还原

【Abstract】 The line loss rate of the power grid is one of the urgent assessment indexes of power enterprises and has a significant impact on the economic benefits of power enterprises, precise diagnosis of abnormal line loss conditions has become an urgent issue for power enterprises. In this paper, the forecasting method of power grid line loss based on long short-term memory was studied. Firstly, the forecasting model in this paper is compared with BP neural network and radial basis function neural network through coarse time granularity line loss forecasting, and the accuracy advantage of this forecasting model was verified. Secondly, the forecasting model in this paper is used to perform line loss forecast at fine time granularity. Finally, the simulation test was carried out with the data collected from the power grid managed by a municipal power enterprise. The experimental results showed that in the case of large data sets, the veracity of the forecasting model based on the LSTM network proposed in this paper was better than other algorithms.更多还原