A Study on Process of Pressure Pattern based Clustering for Gas Regulator System

For the convenience of consumers’ usage of the city gas, the maintenance of the pressure of the city gas is crucial in Korea as it is widely utilized throughout the country. Gas regulators reduce high-pressure gas to a usable pressure. This study proposes a clustering approach for the pressure patterns of gas regulator systems and aims to derive the characteristics of these clusters by linking them with regional factors. This helps to identify groups of pressure regulators with similar environmental conditions and pressure patterns, which can then be used as a foundation for establishing common maintenance strategies. For the clustering approach, k-means clustering was employed. DTW (Dynamic Time Warping) was used as distance measures. The Elbow method and Silhouette Coefficient were used to evaluate the models’ performance. Additionally, to identify the regional characteristics of the gas regulators, public data and the Kakao API were used to consider variables such as residential facilities and transportation which then enabled the development of a suitable clustering approach. A total of 10 clusters were derived, and the characteristics of each cluster were summarized.