Chapter 8 A Machine Learning Framework for Automated Functionality Monitoring of Movable Bridges Masoud Malekzadeh and F. Necati Catbas Abstract Functionality of movable bridge highly depends on the performance of the mechanical components including gearbox and motor. Therefore, on-going maintenance of these components are extremely important for uninterrupted operation of movable bridges. Unfortunately, there have been only a few studies on monitoring of mechanical components of movable bridges. As a result, in this study, a statistical framework is proposed for continuous maintenance monitoring of the mechanical components. The efficiency of this framework is verified using long-term data that has been collected from both gearbox and motor of a movable bridge. In the first step, critical features are extracted from massive amount of Structural Health Monitoring (SHM) data. Next, these critical features are analyzed using Moving Principal Component Analysis (MPCA) and a condition-sensitive index is calculated. In order to study the efficiency of this framework, critical maintenance issues have been extracted from the maintenance reports prepared by the maintenance personnel and compared against the calculated condition index. It has been shown that there is a strong correlation between the critical maintenance actions, reported individually by maintenance personnel, and the condition index calculated by proposed framework and SHM data. The framework is tested for the gearbox. Keywords Structural health monitoring • Machine learning • Big data • Movable bridge • Automated condition monitoring 8.1 Introduction Obtaining reliable and timely assessments of bridge condition, performance and safety throughout increasingly longer service lives represents a considerable challenge for bridge owners, engineers and the Federal Highway Administration (FHWA). The ability to quantitatively characterize existing bridges may lead to more cost-effective and efficient maintenance management decisions, and more robust evaluations of structural safety. Presently, the long-term performance and condition of most bridges are evaluated on the basis of biennial visual inspection data. Visual inspection data are inherently qualitative and are subject to other important limitations that can hamper their effectiveness for assessing bridge performance and safety. A study by the FHWA on the reliability of visual inspection [1] revealed many of the uncertainties associated with this assessment approach. Structural Health Monitoring (SHM) is an emerging approach that has gained significant attention because it promises to enable more quantitative, reliable and timely assessments of bridge condition and performance than are possible using only visual inspection data [2, 3]. The health status of structures is under threat from several sources including environmental effects, overload, design and construction issues. Proximity to waterways, mechanical system failure and fatigue due to the stress fluctuations during the operation make the situation even worse for movable bridges. As a consequence, there is an unavoidable progressively demand for monitoring the behavior of such a structures over time [2] . Operational monitoring of the most critical machinery components of movable bridges will provide information about any possible issues associated with mechanical and electrical parts of these structures. Subsequently, the maintenance plan can be rescheduled and maximized in light of such information to enhance the service life of the structure. Furthermore, the design code for movable bridges can be revised and promoted based upon the data captured from the monitoring system. M. Malekzadeh Metal Fatigue Solutions, 7251 West Lake Mead Boulevard Suite 300, Las Vegas, NV 89128, USA e-mail: m.malekzadeh@knights.ucf.edu F.N. Catbas ( ) Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, FL 32816, USA e-mail: catbas@ucf.edu © The Society for Experimental Mechanics, Inc. 2016 S. Pakzad, C. Juan (eds.), Dynamics of Civil Structures, Volume 2, Conference Proceedings of the Society for Experimental Mechanics Series, DOI 10.1007/978-3-319-29751-4_8 57
RkJQdWJsaXNoZXIy MTMzNzEzMQ==