Proceedings of the

The 33rd European Safety and Reliability Conference (ESREL 2023)
3 – 8 September 2023, Southampton, UK

A Knowledge Graph Method for Risk Factor Analysis of Underground Gas Storage

Mingyuan Wua, Jinqiu Hub, Xiaowen Fanc and Laibin Zhangd

College of Safety and Ocean Engineering, China University of Petroleum (Beijing), China. Key Laboratory of Oil and Gas Safety and Emergency Technology, Ministry of Emergency Management, China.


In recent years, the data from underground gas storage stations have become more complex and scaled up. This paper proposes a knowledge graph method for risk factors analysis to use textual information such as production reports during the operation period of gas storage and underground gas storage. The technique extracts relationships from textual data of the gas storage operation period, identifies risk factors using a Bi-directional Long-Short Term Memory network and Conditional Random Field algorithm (Bi-LSTM-CRF), finds the connections among them, and builds a knowledge graph of risk factors based on the extraction results using Neo4j graph database. In addition, this paper compares Bi-LSTM-CRF with other models, and its accuracy, recall, and F1 value metrics are improved by 3.6%, 2.9%, and 3.2%, respectively. The results show that the Bi-LSTM-CRF risk identification method has the highest accuracy rate of 94.3% and the best results in unstructured text extraction from gas storage reservoirs. This paper proposes that the risk factor analysis method based on a knowledge graph can characterize the relationship between risk factors and effectively improve underground gas storage sites' risk management capability.

Keywords: Risk factor analysis, Knowledge graph, Underground gas storage, Relationship extraction, Bi-directional Long-Short Term Memory network Conditional Random Field algorithm (Bi-LSTM-CRF).

Download PDF