Tehnički glasnik, Vol. 18 No. 1, 2024.
Izvorni znanstveni članak
https://doi.org/10.31803/tg-20230221152352
AI-based Electric Fire Detection State Judgment Data Set Construction
Hee-Chul Kim
; Gwangju University, 277 Hyodeok-ro, Nan-gu, Gwangju, Korea, 61743
Sažetak
In this paper, we create a virtuous cycle ecosystem of AI data for judging electric fire status. Data collection reflects feedback on inspection results such as collection of electric fire status judgment data through cloud sourcing and purification, processing, inspection and data disclosure of the collected data. It is necessary to determine the cause of the damage through fire forensics in order to confirm the property damage caused by the fire. The damage investigation so far is based on the experience of the investigator, and it is difficult to conduct a sufficient investigation and analysis of multiple fires. Accordingly, by building a data set for AI learning for the cause analysis of electric fires, The AI composition that can overcome the subjective and unprofessionalism of the forensic of electric fires is made. Therefore, we study the reliability and system development feasibility of digital conversion of fire detection report and data for AI learning.
Ključne riječi
1st dragon mark; 2nd dragon mark; cause of fire; electric fire; electrical fire causes; fire detection; melt marks
Hrčak ID:
313794
URI
Datum izdavanja:
15.2.2024.
Posjeta: 196 *