基于TCN-GRU模型的短期负荷预测方法

为了进一步提高短期负荷的预测精度,为电力系统的稳定运行提供更加有力的保障,文中提出了一种将时间卷积网络(TCN)和门限循环单元(GRU)相结合的短期负荷预测方法TCN-GRU。首先,将采集的训练数据划分为时序数据和非时序数据;其次,将时序数据输入到TCN模型中以提取时序特征;然后,将提取出来的时序特征与非时序数据组合起来输入到GRU模型中对模型进行训练;最后,利用训练好的模型实现对短期电力负荷的预测。基于广东省佛山市某行业真实负荷数据验证了TCN-GRU模型的负荷预测能力,并通过对比多种深度学习模型的预测效果,验证该模型具有更高精度的短期负荷预测能力。

A forecasting method for short-term load based on TCN-GRU model

In order to improve the accuracy of short-term load prediction and provide a more powerful guarantee for the sta-ble operation of the power system, a short-term load prediction method that combines a temporal convolutional network (TCN) and a gated recurrent unit (GRU) is proposed in this paper. Firstly, the training data are divided into time series data and non-time series data. Secondly, the time series data is selected as the input of the TCN model in order to extract time series features, then the integration of the time series features and non-time series data is entered into the GRU model for training the model. Finally, the trained model is used to predict short-term power load. Based on the real load data of an industry in Foshan City, Guangdong Province, the load forecasting ability of the TCN-GRU model was verified, and by comparing with the prediction effects of various deep learning models, the model was verified to have higher short-term load forecasting ability.