Doubly time-dependent Hawkes process and applications in failure sequence analysis

Lu-ning Zhang, Jian-wei Liu () and Xin Zuo
Additional contact information
Lu-ning Zhang: China University of Petroleum
Jian-wei Liu: China University of Petroleum
Xin Zuo: China University of Petroleum

Computational Statistics, 2023, vol. 38, issue 2, No 21, 1057-1093

Abstract: Abstract Since the Hawkes process is proposed in 1971, it has become increasingly widely applied in the field of event sequence analysis, such as social network analysis, electronic medical record analysis, click recommendation, financial analysis, and so on. Similar to the idea of electronic medical record analysis, we hope that the time-dependent Hawkes process can be used to analyze the failure of the compressor station system in the process of oil and gas gathering and transportation. However, at present, the existing Hawkes process model that has been proposed cannot meet our demands well. Most of the existing Hawkes process research so far assumes that the Hawkes process is time-independent, and its background intensity and trigger pattern will not change with time. In addition, recently some researchers put forward some Hawkes process models, while the trigger pattern is related to time, but the background intensity remains constant over time, or background intensity changes with time, and the trigger pattern remains unchanged, while we intend to analyze in between failure trigger pattern change and the trend of the background intensity changes over time. Therefore, we come up with a new doubly time-dependent Hawkes process model and its corresponding effective parameter learning method based on our requirements. We change the constant background intensity to time dependent background intensity, which obeys the Weibull distribution. Since background intensity and trigger pattern between events for the new proposed Hawkes process are all time dependent, we call it as the doubly time-dependent Hawkes process (DTDHP). To verify DTDHP, we carried out verification experiments in several artificial and real-world datasets and put forward some suggestions for the practical production of compressor stations.

Keywords: Time-dependent Hawkes process; Failure analysis; Background intensity; Infection function; Trigger kernel; Weibull distribution (search for similar items in EconPapers)
Date: 2023
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s00180-022-01269-6 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:compst:v:38:y:2023:i:2:d:10.1007_s00180-022-01269-6

Ordering information: This journal article can be ordered from
http://www.springer.com/statistics/journal/180/PS2

DOI: 10.1007/s00180-022-01269-6

Access Statistics for this article

Computational Statistics is currently edited by Wataru Sakamoto, Ricardo Cao and Jürgen Symanzik

More articles in Computational Statistics from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().