Annual Load Sequence Forecasting Method Considering the Influence of Meteorological Elements

Annual load forecasting plays a key role in the medium- and long-term planning of power systems. Since it is difficult to fully characterise the influence of meteorological factors on the annual-scale load series by traditional methods, this paper proposes an annual load series forecasting method considering the influence of meteorological factors. Firstly, the correlation between meteorological elements and the annual-scale load series is analysed; secondly, the annual daily load series is classified using the SOM clustering method, and the state transfer matrix of the daily load series is calculated using the Markov chain method; then, the GRU network prediction model is constructed for meteorologically sensitive loads, and the probability distribution model of the load characteristic index is constructed for non-meteorologically sensitive loads, and the MCMC method is used to sample the Finally, the validation is based on the annual load data of the actual power grid and multi-model comparison, the results show the effectiveness of this paper’s method in capturing the annual load change pattern.