Disturbance mechanism and correction model of extreme weather events on regional power grid electricity consumption

The incidence of extreme weather has been increasing year by year, posing a huge challenge to the power balance of the new electrical power systems (EPS), and it is difficult to guarantee the safe and stable operation of EPS. The frequent occurrence of extreme weather has led to an annual increase in temperature sensitive load electricity consumption. As a high-quality control resource for virtual power plants, temperature sensitive loads urgently need to analyze the impact of meteorological changes on such loads. In response to the characteristics of low data volume and strong randomness in load forecasting under extreme weather scenarios, this paper explores the disturbance mechanism of extreme weather events on regional power grid electricity consumption, and combines convolutional neural networks (CNN) with long short-term memory networks (LSTM) to design a combined forecasting method based on variational mode decomposition (VMD) and enhanced CNN-LSTM. Firstly, this method decomposes the load sequence using VMD to obtain sub sequences of different periods, and then uses CNN-LSTM to predict the daily maximum load during extreme weather periods. Simulation experiments show that the proposed method can effectively solve the issue of regional power grid load forecasting under extreme weather conditions, with high precision.