An approach for maintaining pipeline infrastructure based on graphical images is disclosed. The approach receives a plurality of parameters in a pipeline infrastructure by using a plurality of sensors. The approach generates a digital twin of the determined one or more susceptible and/or vulnerable points/areas/joints in the pipeline infrastructure. The approach simulates the determined susceptible areas/points/joints in the pipeline infrastructure for determining one or more damages and/or defects. The approach predicts one or more proactive maintenance actions based on the determined damage and/or defect in the pipeline infrastructure to prevent the users from being near the susceptible area/joints/portions and prioritizing the proactive maintenance actions based on the level and extent of damage and/or defect.