考虑多因素的深度学习融合方法实现负荷预测

Load Forecasting in the Deep Learning Fusion Method Considering Multiple Factors

综合考虑电力负荷预测受多因素影响的特性,提出一种深度学习融合方法即K近邻和长短期记忆网络融合模型(KNN-LSTM)实现电力负荷预测。为了充分挖掘影响电力负荷预测的关键因素,引入了KNN算法衡量各因素与负荷数据之间的欧式距离以遴选最邻近的K个负荷影响因素,基于选取的负荷因素数据构建监督学习数据集,引入LSTM模型实现了电力负荷高精度预测。结果表明,KNN-LSTM方法可以有效地提取关键负荷因素并获得良好的预测性能。

Considering that power load forecasting is affected by many factors,a deep learning fusion method,namely K-nearest neighbor and long short term memory(KNN-LSTM),was proposed to realize power load forecasting.In order to fully mine key factors affecting power load forecasting,KNN algorithm was introduced to measure the Euclidean distance between each factor and load data to select the nearest K load-influencing factors.Based on the selected load factor data,the supervisory learning data set was constructed,and an LSTM model was introduced to achieve high-precision power load forecasting.The results showed that the proposed KNN-LSTM method could effectively extract key load factors and achieve good prediction performance.