A Study on Evaluating the Suitability of Machine Learning for Safety of Enhancing Safety of Gas Facilities

Abstract

Safety management method must be considered as all energy industries because of their own inherent risks. The safety management for gas facilities should be considered more carefully because those facilities have been operated by various kinds of gas and noxious materials. So, deriving risk analysis is very important for preventing and corresponding accidents by means of a specific analysis method. Recently, machine leaning methods are tried to use in this gas safety domain in order to analyze risk and accident sign. Data mining algorithms in machine learning have been used only without the reliability check. So, most attempts to apply gas facilities are unavailing until now. This paper preferentially aims to check the feasibility of machine learning analysis in order to apply a safety of gas facility. The feasibility study must be generally selected target gas facility, collected what kinds of risk factor, and then considered the appropriated machine learning method. Therefore, This research check two kinds of the feasibility using Bayesian and decision tree in real data. As a result, these machining learning algorithms are able to use in gas domain for enhancing safety of facilities.