PROBABILISTIC NET LOAD FORECASTING BASED ON TCN AND GAUSSIAN PROCESS-ENABLED RESIDUAL MODELING LEARNING APPROACH
Net load forecasting is of great importance for grid operation and management with high penetration of renewable energy.This paper proposes a net load prediction method based on temporal convolutional network(TCN)and Gaussian process(GP),which can provide accurate point prediction and probabilistic prediction results.First,TCN is used to extract the variation pattern of net load from a large amount of historical data,and the excellent time series modeling capability of TCN can discover the complex mapping relationship between the input and output of the net load forecasting task.Then,a composite kernel function is designed for the Gaussian process to correct the prediction residuals of TCN,which can further improve the accuracy of point prediction based on TCN prediction,and also quantify the uncertainty of net load prediction by using the uncertainty quantification ability of Gaussian process.Finally,experiments are conducted on real net load datasets,and the effectiveness of the proposed method is illustrated by comparing it with a large number of state-of-the-art models.
prediction modelsphotovoltaic outputprobability density functionresidual neural networkstemporal convolutional network(TCN)
NETL
NSTL
万方数据
