A short term load forecasting method based on sensitive meteorological factor extraction and BiGRU-Attention network

The new power system is the key to achieving low-carbon energy transformation. Accurate short-term load forecasting is crucial for ensuring the safe and stable operation of emerging power systems and electricity market transactions. With the continuous development of social demand for cooling/heating, the coupling relationship between meteorological factors represented by temperature and load has become increasingly prominent. In this paper, a short-term load forecasting method considering the effect of meteorological factors is proposed. Firstly, based on the maximum correlation and minimum redundancy algorithm, this paper analyzes the correlation between meteorological factors and power load in time series and selects a set of meteorological factors that have a strong correlation with the trend of electricity load changes. Then, a power load prediction model based on BiGRU-Attention is constructed to extract the coupling characteristics between meteorological factors and power load and to realize high-precision short-term power load forecasting. Finally, a case study was conducted using real data from a certain region in China. The results show that compared with traditional methods, the proposed model reduces the error by 2.12%, verifying its effectiveness.