Supply-reliability based method of intellectual optimization on preventive maintenance strategy for natural gas pipeline system

Maintenance management of devices is fundamental to ensure the supply reliability of natural gas pipeline system. A descriptive model integrating the state transition with the degeneration process of units was established. Then a method for calculating the maximum gas supply capacity of the pipeline system under scenarios of unit failure was developed. The maintenance strategy optimization model with a high supply reliability and a low maintenance cost was established, and a deep reinforcement learning-based method of maintenance optimization for the pipeline system was presented. The results show that compared with the scheduled maintenance, the proposed preventive maintenance strategy can improve the supply reliability and reduce the system maintenance costs. The average supply reliability increases from 99.41% to 99.54%, and the maintenance cost is reduced by 9.38%. © 2023 University of Petroleum, China. All rights reserved.