ANALYZING DATABASES IN THE TAIWAN BRIDGE MANAGEMENT SYSTEM USING BIG DATA APPROACHES

Abstract

This paper describes some preliminary results of this research. Before conducting a comprehensive big data analysis of the whole system of Taiwan Bridge Management System (TBMS), a pilot test was performed to explore possible outcomes. A total number of 34,099 records each having 305 fields of 24 bridges in a specific region were analyzed using the famous R software. The resulting correlations between inventory fields and deterioration of bridge components are categorized in three types: explicable, inexplicable, and irrelevant. For those inexplicable correlations, interviews with experts are found necessary in order to explore useful or meaningful maintenance information. Two more similar tests are also being planned for two groups of bridges that are nearby sea or solely in the mountainous area. In addition to the correlation between a specific deterioration and all existing inventory fields, other maintenance information such as frequently maintained components as well as their maintenance frequencies and costs, and other crucial relationships among fields in the TBMS are highly anticipated in this research in order to establish beneficial suggestions to the bridge maintenance strategies in Taiwan.