A fuzzy expert system for buried pipeline corrosion assessment based on DCVG measurements

摘要

Despite protective measures such as external coating and cathodic protection, regular in-line inspection is essential to ensure buried steel pipelines integrity against corrosion. To inspect pipe sections considered unpiggable, TIGF - which operates a 5000 km-long gas network in southwest France -conducts DC-Voltage Gradient (DCVG) surveys, which locate coating defects using surface electrical measurements. Every year, TIGF detects about 7000 coating defects and collects for each of them a set of 26 features allowing for corrosion-deterioration risk assessment. However, identifying the most critical cases requiring to be excavated is a complex task. In this work, we introduce a decision support tool, based on a Fuzzy Expert System (FES) named ExpressIF, to evaluate coating defects criticality. FES are artificial intelligence tools designed to reproduce human reasoning based on an inference engine. The proposed system supports excavation decisions by evaluating 7 different types of deterioration risk together with the severity level of a potential pipeline failure. It embeds more than 300 rules gathered from experts and fulfils operational needs through a dedicated interface. For each decision made, references to the relevant rules are given, which therefore provides an efficient framework for improving selection rules from excavation feedbacks. This tool is now in operation and has fully contributed to the definition of the excavation campaign planned by TIGF for 2016.