A Framework for Developing an Estimation Model of Damages on Bridge Elements Using Big Data Analytics

According to the Ministry of Land, Infrastructure and Transport in Korea, the number of bridges over 30 years old will be three times in 2025 than three thousands in 2015 so that proper maintenance enforcement has emphasized. The Korean government legislated on the Special Act on Safety Control for Infrastructure in 1995 and rated the condition grade of ?A? to ?E? to indicate that under the ?C? grade of infrastructure requires critical repairs. Since the lack of maintenance budget and decrease of professional inspectors, more effective and efficient monitoring solutions for bridge conditions are required. The primary purpose of this study is to develop a model to predict deficiencies of bridge elements using big data analytics and the framework is proposed in this paper. The model analyzed a dataset of Bridge Management System (BMS) developed by the Korea Institute of Civil Engineering and Building Technology (KICT) (e.g., general bridge information, structural information, and inspection and maintenance history) and found combinations of significant factors causing damages of bridge elements. Data mining algorithms (Apriori and Ripper) were applied for the pattern analysis and the partitional clustering algorithm grouped similar patterns for more efficient and fast damage prediction. Machine learning concepts using bagging and boosting were also employed for resampling and upgrading the estimation model to enhance estimation performance. The expected results showed potential to predict causes, locations, current status, and the types of bridge damages that can be used for preventive bridge maintenance planning.