Abstract: In order to realize the intelligent distribution function of natural gas distribution stations in the construction of intelligent pipeline networks, load forecasting should be carried out for users, especially the urban gas users who are greatly affected by the environmental factors, so as to provide a basis for the distribution branch to forecast and adjust the distribution control strategy with the artificial intelligence algorithm. An intelligent forecasting model based on a multilayered BP neural network was established, load forecasting was conducted for a gas user in West-to-East Gas Pipeline Network, and the results of comparison, with the weather parameters such as temperature, wind speed and direction, and the historical load value as the input, indicated that the simulation results with the forecasting method had a relative error of no more than ±8% with the actual load value, capable of accurately forecasting the 72 h short-term load of the urban gas users. In addition, the factors affecting the accuracy of load forecasting were further studied, and it was found that the change of temperature at about 1 ℃ would lead to the change of demand of about 5%-6% in winter. When the temperature exceeds 14 ℃, the user load will decrease linearly with the rise of temperature, and it will not be affected when the temperature exceeds 18 ℃. Variations in wind speed and rainfall, on the other hand, have less than 5% impact on user's demand. Accurate short-term user load forecasting is the basis for the efficient, reasonable and economic operation of the natural gas supply system by the pipeline or pipeline network companies, which can effectively ensure the economy and reasonability of the gas supply scheme, as well as the safety and efficiency of operation and scheduling.