Short-term prediction of natural gas load based on WPD-SCA-ELM model

With increasing natural gas consumption, it is of great significance to accurately predict the daily consumption load of natural gas in the future for the rational allocation of natural gas resources. To solve this problem, a natural gas load prediction model based on the WPD-SCA-ELM model was established based on the idea of “decomposing-prediction-reconstruction”. The wavelet basis function and decomposition layers affecting the wavelet packet decomposition were optimized, and the factors affecting the daily load were selected, and the temperature factor hysteresis was corrected by a translation operation. Finally, the algorithm is compared with other models. The results show that the daily load data in the heating period is not normally distributed and has great fluctuation. The Fk4-order two-layer decomposition can better reflect the variation trends and daily load characteristics. The correlation coefficients of daily maximum temperature and daily minimum temperature are larger than average temperature, and the correlation between temperature and daily load can be improved by translating and sliding the temperature. The MAPE, RMSE and DS of WPD-SCA-ELM model are 0.59, 7321 and 0.920, respectively. Compared with other models, the evaluation index is the best, which proves that the model is useful.