Abstract: The accurate setting of negative pressure in gas extraction is the key to ensure the efficient extraction of gas extraction system. How to determine the optimal negative pressure at each position in the extraction system is a technical difficulty that needs to be solved urgently. Therefore, in order to regulate the negative pressure of gas extraction system reasonably and accurately, four prediction algorithms are compared and analyzed, and the most excellent prediction algorithm is selected and improved according to its own shortcomings. Based on the improved algorithm model, an intelligent decision-making and regulation model of negative pressure in the borehole is constructed. The solution model of gas extraction pipeline network is constructed and the solution optimization of pipeline network solution model is realized. Based on the improved particle swarm optimization algorithm, the intelligent optimization decision-making and regulation model of pipeline network is constructed. An experimental study is carried out in the field to verify the reliability of the intelligent decision-making and regulation model. The results show that (Long Short-Term Memory, LSTM) has the highest matching degree with gas extraction data features among the four prediction algorithms, and the improved (Convolutional Neural Network- Long Short-Term Memory, CNN-LSTM) model can effectively improve the problem of LSTM over-reliance on time series. The intelligent decision-making control and regulation model of negative pressure in the borehole based on CNN-LSTM model can realize the accurate prediction and regulate of negative pressure in the borehole. The improved particle swarm optimization algorithm can effectively avoid the problem of falling into local optimum and obtain the optimal solution of iterative solution. At the same time, the intelligent optimization decision-making and regulation model of pipeline network based on the improved particle swarm optimization algorithm can reasonably allocate the negative pressure of gas extraction pipeline network. In the field test, the gas flow of 1^# and 2^# test boreholes increase by 0.014 m³/min and 0.013 m³/min respectively after intelligent regulate, and the gas concentration increase by 11.91% and 10.03% respectively. After the intelligent regulation in the pipe network, the gas flow rate increase by 1.23 m³/min and the gas concentration increase by 2.87%. It indicates that the intelligent decision-making and regulation model has high reliability. The research results are of great significance to improve the extraction efficiency of gas extraction system and ensure the safety of mine production.