Short-Term Thermal Load Probabilistic Forecasting Considering Electrical-Thermal Coupling Characteristics

With the improvement of the consumption potential of new energy assisted by thermal system, the coupling and tightness of power and heat supply are continuously strengthened. Higher requirements are put forward for probability prediction of thermal load. In the existing research, point prediction models are usually established according to the relevant characteristics of thermal system, which is difficult to reflect the influence of the uncertainty of thermal load and the electrical-thermal coupling characteristics on the thermal load prediction result. In view of this, a probabilistic forecasting method of short-term thermal load considering electrical-thermal coupling characteristics is proposed. Firstly, the collected data is preprocessed, and the timescale is matched. Secondly, Pearson coefficient is used to analyze the electrical-thermal coupling characteristics, and the feature set considering electrical-thermal coupling is constructed. Finally, Gaussian process regression is used to establish a probabilistic forecasting model of short-term thermal load.