Odor recognition of thermal decomposition products of electric cables using odor sensing arrays

Yuanchang Liu, Shintaro Furuno, Sosuke Akagawa, Rui Yatabe, Takeshi Onodera, Nobuyuki Fujiwara, Hidekazu Takeda, Seiichi Uchida, Kiyoshi Toko

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

An odor sensing system with chemosensitive resistors was used to identify the gases generated from overheated cables to prevent fire. Three different electric cables for a distribution cabi-net were used. The cables had an insulation layer made of polyvinyl chloride (PVC) or cross-linked polyethylene (XLPE). The heat resistance of the cables was tested by differential thermal and ther-mogravimetric analyses. The thermal decomposition products of the cables were investigated by gas chromatography-mass spectrometry (GC-MS). For the odor sensing system, two types of 16-channel array were used to detect the generated gases. One contains high-polarity GC stationary phase materials and the other contains GC stationary phase materials of high to low polarity. The system could distinguish among three cable samples at 270 °C with an accuracy of about 75% through both arrays trained with machine learning. Furthermore, the system could achieve a recall rate of 90% and a precision rate of 70% when the abnormal temperature was set above the cables’ allowable conductor temperature at 130 °C. The odor sensing system could effectively detect the abnormal heating of the cables before the occurrence of fire. Therefore, it is helpful for fire prediction and detection systems in factories and substations.

Original languageEnglish
Article number261
JournalChemosensors
Volume9
Issue number9
DOIs
Publication statusPublished - Sep 2021

All Science Journal Classification (ASJC) codes

  • Analytical Chemistry
  • Physical and Theoretical Chemistry

Fingerprint

Dive into the research topics of 'Odor recognition of thermal decomposition products of electric cables using odor sensing arrays'. Together they form a unique fingerprint.

Cite this