基于深度学习的城市短期电力负荷预测方法

Urban Short-term Electricity Load Forecasting Method Based on Deep Learning

城市短期电力负荷预测是城市电网系统中的关键技术,其直接影响电网的动态平衡,而电力平衡又是社会生活和工业运行的基础,对人们的正常生活和工业生产至关重要.然而,短期电力负荷序列的复杂性使得传统单一模型的预测精度不尽如人意.因此,基于BWO-CNN-BiGRU-AT模型,提出了一种创新的短期电力负荷预测方法.通过结合卷积神经网络、双向门控循环单元和Bahdanau Attention机制,并运用白鲸优化算法进行优化,深入挖掘了电力负荷数据中的复杂关联,从而显著提升了预测性能.实验结果表明,该方法在预测准确性方面表现优异,相较于其他对比模型有更高的有效性.

Short-term electricity load forecasting is a key technology affecting the dynamic balance of the urban electricity grid system,and electricity balance as the foundation of social life and industrial operations plays an important role in ensuring normal living and industrial production for people.However,the complexity of short-term electricity load series makes the prediction accuracy of traditional single model unsatisfactory.Based on the BWO-CNN-BiGRU-AT model,we propose an innovative short-term electricity load forecasting method.Researchers explore the complex correlation in electricity load data with convolutional neural network,bidirectional gated recurrent units,Bahdanau Attention mechanism,and beluga whale optimization algorithm,thus significantly improve the prediction performance.The experimental results show that the method performs better in prediction accuracy in comparison with other models(such as BP,GRU,BiGRU,and BiGRU-AT).