Damage quantification and localization algorithms for indirect SHM of bridges
نویسندگان
چکیده
This paper presents algorithms for diagnosing the severity and location of damage in a laboratory bridge model. We use signal processing and machine learning approaches to analyze the vibration responses collected both directly from the bridge model and indirectly from a vehicle passing over the model. Features are selected using principal component analysis (PCA), and a regression is performed using the kernel regression method. Various “damage” severities and positions are simulated on a laboratory bridge model by placing additional mass on the bridge. We perform two experiments; one to measure our ability to detect damage severity (i.e. size of the mass), and a second to measure our ability to detect damage location (i.e. position of the mass). In the first experiment, we vary the magnitude of the mass while keeping its location constant. In the second experiment, we vary the location of the mass while keeping its magnitude constant. In both cases, we use a portion of our data to train the algorithm, and another portion to test its validity. We report the accuracy of correctly quantifying the nature of the mass from the test data as a mean square error (MSE). state of the structure for an infinite number of mass locations and sizes within our training set. To build the regression, we apply principal component analysis to the acceleration signals, and train the kernel regression model by the collected data. This model determines the size and the location of the damage proxy using the MSE as the evaluation score. 2 EXPERIMENTAL SETUP AND PROTOCOL 2.1 Experimental setup (Cerda et al. 2013; Wang et al. 2013) A general view of the laboratory model used in this project is shown in Figure 1, and schematic of the setup is shown in Figure 2. The model consists of a vehicle that is pulled across the rails by a cable system. The vehicle starts on ‘Ramp 1’, accelerates up to a constant speed, crosses the middle section, the “bridge,” then decelerates on ‘Ramp 2.’ The vehicle has wired accelerometers so there is a cable rail above to ensure these wires do not interfere with its motion. The “bridge” is an aluminum plate, 2438 mm (8 feet) long, with two angle beams acting as girders and two rails to guide the vehicle. The vehicle model, as shown in Figure 3, has an independent suspension system. Both the vehicle and the bridge were instrumented with commercial accelerometers. On the vehicle, two sensors are on the sprung portion of the vehicle (‘front chassis sensor’ and ‘rear chassis sensor’), and two sensors are on the unsprung portion of the vehicle, rigidly attached to the wheels (‘front wheel sensor’ and ‘rear wheel sensor’), as shown in Figure 3. One sensor was placed underneath the bridge deck at midspan (‘bridge sensor’). Figure 1. The general view of the laboratory setup. The motors governing the movement of the vehicle and the data-acquisition systems are both controlled by National Instrument’s PXI system running LabView . By using a single system, we can spatially align the time series data from different runs using the vehicle’s position. More details about the experimental setup can be found in (Cerda et al. 2013). Figure 2. The illustration of the laboratory model. Figure 3. Details of the vehicle.
منابع مشابه
TWO-STAGE METHOD FOR DAMAGE LOCALIZATION AND QUANTIFICATION IN HIGH-RISE SHEAR FRAMES BASED ON THE FIRST MODE SHAPE SLOPE
In this paper, a two-stage method for damage detection and estimation in tall shear frames is presented. This method is based on the first mode shape of a shear frame. We demonstrate that the first mode shape slope is very sensitive to the story stiffness. Thus, at the first stage, by using the grey system theory on the first mode shape slope, damage locations are identified in shear frames. Da...
متن کاملEFFICIENCY EVALUATION OF PROPOSED OBJECTIVE FUNCTIONS IN STRUCTURAL DAMAGE DETECTION BASED ON MODAL STRAIN ENERGY AND FLEXIBILITY APPROACHES
Civil infrastructures such as bridges and buildings are prone to damage as a result of natural disasters. To understand damages induced by these events, the structure needs to be monitored. The field of engineering focusing on the process of evaluating the location and the intensity of the damage to the structure is called Structural Health Monitoring (SHM). Early damage prognosis in structures...
متن کاملDeployment of a Smart Structural Health Monitoring System for Long-Span Arch Bridges: A Review and a Case Study
Structural health monitoring (SHM) technology for surveillance and evaluation of existing and newly built long-span bridges has been widely developed, and the significance of the technique has been recognized by many administrative authorities. The paper reviews the recent progress of the SHM technology that has been applied to long-span bridges. The deployment of a SHM system is introduced. Su...
متن کاملA New Two-Stage Method for Damage Identification in Linear-Shaped Structures Via Grey System Theory and Optimization Algorithm
The main objective of this paper is concentrated on presenting a new two-stage method for damage localization and quantification in the linear-shaped structures. A linear-shaped structure is defined as a structure in which all elements are arranged only on a straight line. At the first stage, by employing Grey System Theory (GST) and diagonal members of the Generalized Flexibility Matrix (GFM),...
متن کاملReal-time damage detection of bridges using adaptive time-frequency analysis and ANN
Although traditional signal-based structural health monitoring algorithms have been successfully employed for small structures, their application for large and complex bridges has been challenging due to non-stationary signal characteristics with a high level of noise. In this paper, a promising damage detection algorithm is proposed by incorporation of adaptive signal processing and Artificial...
متن کاملLocalization Boyan algorithm to detect forest fires from MODIS sensor images
Of phenomena which much damage and irreparable import to forests and natural resources is the fire that each year, more than 100 fires occur in Iran and thousands of hectares of trees and plants eliminates. Given that fire risk is high in most parts of the world, full and continuous monitoring on this natural phenomenon, is essential. Use remote sensing is a way to identify and manage fire. Ahe...
متن کامل