Assessing Gas Leakage Detection Performance Using Machine Learning with Diff erent Modalities

• Published in: Transactions on electrical and electronic materials pp. 653 - 664
• Main Authors: Gaurav Kumar Kasal, Vivek Pratap Singh, Saurabh Kumar Pandey
• Format: Journal Article
• Language: English
• Published: 한국전기전자재료학회 01.10.2024
• Subjects: 전기공학
• ISSN: 1229-7607; 2092-7592
• DOI: 10.1007/s42341-024-00545-0

Artifi cial intelligence technologies are reviving the entire proactive repair system for leakage issues in industries with a real-time strategy for the current era of industry 4.0. In the present work, Gas sensor and an infrared thermography is used in this work to analyze the eff ectiveness of fi nding gas leaks. The eff ectiveness of the proposed work is to extract the statistical features from gas sensor and infrared thermography to identify gas leaks. Moreover, diff erent machine learning methods are renowned for performing exceptionally well in classifi cation tasks, to obtain accurate categorization. To properly identify and classify gas leakages, four separate categories have been established on the most popular online dataset. On a diff erent gas leakage dataset, diff erent tests have been conducted to verify the methodology and evaluated our strategy against other machine learning techniques that are frequently employed in gas leakage detection. The results show how well the proposed method works for precisely identifying and categorizing gas leaks. The comparison analysis demonstrates the diff erent machine learning methods and their superiority in terms of effi ciency and classifi cation precision. KCI Citation Count: 0