An Enhancement Method Based on Long Short-Term Memory Neural Network for Short-Term Natural Gas Consumption Forecasting

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