Forecasting daily natural gas load accurately is difficult because it is affected by various factors. A large number of redundant factors existing in the original dataset will increase computational complexity and decrease the accuracy of forecasting models. This study aims to provide accurate forecasting of natural gas load using a deep learning (DL)-based hybrid model, which combines principal component correlation analysis (PCCA) and (LSTM) network. PCCA is an improved principal component analysis (PCA) and is first proposed here in this paper. Considering the correlation between components in the eigenspace, PCCA can not only extract the components that affect natural gas load but also remove the redundant components. LSTM is a famous DL network, and it was used to predict daily natural gas load in our work. The proposed model was validated by using recent natural gas load data from Xi’an (China) and Athens (Greece). Additionally, 14 weather factors were introduced into the input dataset of the forecasting model. The results showed that PCCA–LSTM demonstrated better performance compared with LSTM, PCA–LSTM, back propagation neural network (BPNN), and support vector regression (SVR). The lowest mean absolute percentage errors of PCCA–LSTM were 3.22% and 7.29% for Xi’an and Athens, respectively. On these bases, the proposed model can be regarded as an accurate and robust model for daily natural gas load forecasting.

Artificial intelligence models have been widely applied for natural gas consumption forecasting over the past decades, especially for short-term consumption forecasting. This paper proposes a three-layer neural network forecasting model that can extract key information from input factors and improve the weight optimization mechanism of long short-term memory (LSTM) neural network to effectively forecast short-term consumption. In the proposed model, a convolutional neural network (CNN) layer is adopted to extract the features among various factors affecting natural gas consumption and improve computing efficiency. The LSTM layer is able to learn and save the long-distance state through the gating mechanism and overcomes the defects of gradient disappearance and explosion in the recurrent neural network. To solve the problem of encoding input sequences as fixed-length vectors, the layer of attention (ATT) is used to optimize the assignment of weights and highlight the key sequences. Apart from the comparisons with other popular forecasting models, the performance and robustness of the proposed model are validated on datasets with different fluctuations and complexities. Compared with traditional two-layer models (CNN-LSTM and LSTM-ATT), the mean absolute range normalized errors (MARNE) of the proposed model in Athens and Spata are improved by more than 16% and 11%, respectively. In comparison with single LSTM, back propagation neural network, support vector regression, and multiple linear regression methods, the improvement in MARNE exceeds 42% in Athens. The coefficient of determination is improved by more than 25%, even in the high-complexity dataset, Spata.