Session: OAC-01-01 Safety, Reliability, and Risk Management

Paper Number: 121299

121299 - Knowledge Graph-Based Decision Model for Gas Pipeline Emergency Response 

Abstract： Facing the problems of complexity and variability of oil and gas pipeline accidents and difficulty in giving emergency decision points efficiently, an emergency decision-making method for gas pipeline accidents based on knowledge graph was proposed. First of all, based on the evolutionary characteristics of oil and gas pipeline accidents, we built the model layer of oil and gas pipeline accident emergency knowledge map and construct the ontology model and conceptual framework of emergency knowledge map. Then, the bidirectional long and short-term memory neural network-conditional random field algorithm model was used to extract entities and relationships from 343 oil and gas pipeline accident investigation reports to populate the data in the schema layer. Then, the establishment of entity nodes and relationships was realized through the Neo4j graph database, and the oil and gas pipeline emergency response domain knowledge graph was constructed. Finally, the recommendation and ranking of emergency decision points were realized by clustering the nodes and relations of oil and gas pipeline emergency response tasks, which improves the accuracy of the emergency decision process. The method was applied to the Songyuan pipeline explosion accident in Jilin Province through example analysis. The research results show that the method realizes the rapid generation of the points of the emergency response process for newly occurred accidents. Compared with the accident investigation report, the recommended emergency response points supplement the original emergency response measures with countermeasures during the accident evolution process, combustible gas concentration monitoring, and emergency response upgrading, which further improves the emergency decision-making process. This method not only effectively solves the contradiction that the current massive data in the field of oil and gas pipeline emergency response cannot provide decision support for emergency response, but also provides a new way for emergency response decision-making in oil and gas pipeline accidents.

Highlight：

1. 343 investigation reports of  gas accidents were analyzed to provide a reference for scenario evolution and emergency decision-making of gas accidents.

2. Established a knowledge map for emergency decision-making of gas pipeline accidents, which provides a reference for the transformation from data to information to knowledge in the field of emergency response.

3. A knowledge mapping-based fast decision generation method for gas pipeline accidents is proposed, which provides auxiliary decision support for gas accidents.

4. Through example analysis, the knowledge graph-based emergency decision generation method for gas pipeline accidents achieves the rapid generation of the main points of the emergency disposal process and its emergency decision plan for new accidents, which solves the contradiction that the current massive data in the field of pipeline emergency but cannot provide decision support for emergency.

Presenting Author: 徐 厚佳 北京中国石油大学

Presenting Author Biography: Houjia Xu was born on December 27, 1996 in Chizhou City, Anhui Province, China. He is pursuing a master's degree in safety engineering from 2019 to 2022, and a PhD degree in safety science and engineering from 2022 to present at China University of Petroleum (Beijing). Dr. Xu is mainly engaged in oil and gas equipment failure and integrity management. Up to now, Dr. Xu has received 3 provincial and ministerial awards, including the First Prize of Science and Technology of China Occupational Safety and Health Association, and has published 5 papers and patents.

Authors:

Xu Houjia China University of Petroleum Beijing
Shuai Jian China University of Petroleum Beijing