Abstract: Booster station is an important part of gathering and transportation system in gas fields, so its energy efficiency evaluation and prediction is of great practical significance to improving the operation efficiency of the gathering and transportation system. In this paper, 25 booster stations in the gathering and transportation system of one certain gas field were taken as the examples. The RS-BP Neural Network model was established based on the advantages of Rough Set (RS) theory and BP Neural Network. Then, 8 evaluation indexes were selected, and the data were pretreated by virtue of the RS attribute reduction approach. Thus, 20 booster stations (1^#-20^#) were evaluated comprehensively. Finally, in order to verify the accuracy of the evaluation results, the energy efficiency of 21^#-25^# booster stations was predicted 4 times in the software MATLAB and compared with the prediction results without reduction. It is indicated that compared with the pure BP Neural Network, RS-BP Neural Network has shorter training time and smaller error, so it is significantly valuable in actual application.