Regional Grid High-Temperature Warning Based on Long Short-Term Memory Networks

During hot summers, high temperatures significantly increase power demand, especially due to air conditioning usage, posing immense pressure and challenges to the power grid and potentially leading to severe consequences like power supply shortages, voltage instability, and widespread blackouts. This paper introduces a high-temperature warning model for regional power grids utilizing Long Short-Term Memory (LSTM) networks. By processing and analyzing weather observation data from Hangzhou during the summer of 2023, LSTM, CNN, and ARIMA models were constructed and their temperature prediction performance was compared. The findings indicate that the LSTM surpasses the CNN and ARIMA in performance across six metrics. Notably, the LSTM model also excels in air conditioning load forecasting, demonstrating its potential and accuracy in practical applications.