基于分位数损失的短期概率负荷预测

Short-term Probabilistic Load Forecasting Based on Quantile Loss

为提高电力负荷预测精度,并针对用电不规律导致的负荷数据波动性大的问题,提出一种分位数损失引导的门控卷积神经网络模型。首先,使用卷积神经网络提取出历史数据中更多的重要特征;其次,在模型训练过程中采用分位数损失指导训练,然后添加门控单元使得关键特征更加明显;最后,模型输出每个分位点的预测值。实验结果表明,相对于多种概率负荷预测模型,所提的分位数引导的门控卷积神经网络模型在概率负荷预测方面具有更好的预测精度,同时具有良好的实际应用前景。

In order to improve the accuracy of power load forecasting,and solve the problem of large fluctuation of load data caused by irregular electricity consumption,a quantile loss-guided gated convolutional neural network model is proposed.Firstly,the convolutional neural network is used to extract more important features from historical data.Secondly,the quantile loss is used to guide the training of the model.Then,the gated unit is added to make the key features more obvious.Finally,the model outputs the predicted value of each quantile.The experiments show that compared with other probabilistic load forecasting models,the proposed quantile-guided gated convolutional neural network model has better forecasting accuracy,and good application prospects in probabilistic load forecasting.