Siavash Dorvash

Localized Structural Damage Detection: A Change Point Analysis

Many current damage detection techniques rely on the skill and experience of a trained inspector and also require a priori knowledge about the struc- tures properties. However, this study presents adapta- tion of several change point analysis techniques for their performance in civil engineering damage detection. Lit- erature shows different statistical approaches which are developed for detection of changes in observations for different applications including structural damage detec- tion. However, despite their importance in damage de- tection, control charts and statistical frameworks are not properly utilized in this area. On the other hand, most of the existing change point analysis techniques were originally developed for applications in the stock mar- ket or industrial engineering processes; utilizing them in structural damage detection needs adjustments and ver- ification. Therefore, in this article several change point detection methods are evaluated and adjusted for a dam- age detection scheme. The effectiveness of features from a statistics based local damage detection algorithm called Influenced Coefficient Based Damage Detection Algo- rithm (IDDA) is expanded for a more complex structural system. The statistics used in this study include the uni- variate Cumulative Sum, Exponentially Weighted Mov- ing Average (EWMA), Mean Square Error (MSE), and multivariate Mahalanobis distances, and Fisher Crite- rion. They are used to make control charts that detect and localize the damage by correlating locations of a sen- sor network with the damage features. A Modified MSE statistic, called ModMSE statistic, is introduced to re- move the sensitivity of the MSE statistic to the variance of a data set. The effectiveness of each statistic is analyzed.

Application of state-of-the-art in measurement and data analysis techniques for vibration evaluation of a tall building

Recent advancements in sensing and data acquisition technology have made monitoring of structures and infrastructure more affordable and, at the same time, more comprehensive. Examples of such advancements are application of wireless technology for communication, the utilisation of fully automated systems for long-term monitoring and the remote control of the sensing system over Internet. Although each of these technologies has been used in different structural health monitoring projects in the recent years, inclusion of an all-in-one sensing system represents the state-of-the-art in measurement techniques. This paper presents the integration of all of the above-mentioned advanced monitoring approaches in one sensing system for forensic quantification of an in-service tall building. The inclusive measurement and monitoring system along with advanced data analysis techniques enabled extraction of detailed information about dynamic characteristics of the building structure and development of reliable conclusions regarding its performance. It is shown that the performance of the investigated structural components is satisfactory in terms of strength demand. However, the level of vibration in some portions of the structure does not meet the limits of human comfort. In addition, wind-speed spectrum, acceleration response spectrum and the modes of lateral vibration are extracted to assist with evaluation of the structures performance.

Effects of measurement noise on modal parameter identification

In the past decade, much research has been conducted on data-driven structural health monitoring (SHM) algorithms with use of sensor measurements. A fundamental step in this SHM application is to identify the dynamic characteristics of structures. Despite the significant efforts devoted to development and enhancement of the modal parameter identification algorithms, there are still substantial uncertainties in the results obtained in real-life deployments. One of the sources of uncertainties in the results is the existence of noise in the measurement data. Depending on the subsequent application of the system identification, the level of uncertainty in the results and, consequently, the level of noise contamination can be very important. As an effort towards understanding the effect of measurement noise on the modal identification, this paper presents parameters that quantify the effects of measurement noise on the modal identification process and determine their influence on the accuracy of results. The performance of these parameters is validated by a numerically simulated example. They are then used to investigate the accuracy of identified modal properties of the Golden Gate Bridge using ambient data collected by wireless sensors. The vibration monitoring tests of the Golden Gate Bridge provided two synchronized data sets collected by two different sensor types. The influence of the sensor noise level on the accuracy of results is investigated throughout this work and it is shown that high quality sensors provide more accurate results as the physical contribution of response in their measured data is significantly higher. Additionally, higher purity and consistency of modal parameters, identified by higher quality sensors, is observed in the results.

An Iterative Modal Identification Algorithm for Structural Health Monitoring Using Wireless Sensor Networks

A novel modal identification approach for the use of a wireless sensor network (WSN) for structural health monitoring is presented, in which the computational task is distributed among remote nodes to reduce the communication burden of the network and, as a result, optimize the time and energy consumption of the monitoring system. Considering the need for having an agile system to capture the earthquake response and also the limited energy resource in WSN, such algorithms for speeding the analysis time and preserving energy are essential. The algorithm of this study, called iterative modal identification (IMID), relies on an iterative estimation method that solves for unknown parameters in the absence of complete information about the system. Applying IMID in WSN-based monitoring systems results in significant savings in time and energy. Validation through implementation of the algorithm on numerically simulated and experimental data illustrates the superior performance of this approach.

Localized Damage Detection Algorithm and Implementation on a Large-Scale Steel Beam-to-Column Moment Connection

Civil structures experience loading scenarios ranging from typical ambient excitations to extreme loads induced by natural events that, depending on their intensity, cause damage. It is important to detect damage before it propagates to become detrimental to integrity and functionality of the structure. Significant research efforts are focused on developing damage detection algorithms to diagnose damage from performance and response of the structure. A major challenge in many existing algorithms is in their validation and absence of real-scale implementation. This paper presents implementation of influence-based damage detection algorithm by implementation on a large-scale structural model (steel beam-to-column moment connection) which experiences progressive damage towards collapse of the system through increasing cyclic loading. IDDA utilizes statistical analysis of correlation functions between the structural responses at different locations. It is shown through this implementation that IDDA, accompanied by a statistical framework, can accurately identify structural changes and indicate the intensity of the damage.

Uncertainties in Identification of a Steel Bridge Dynamic Characteristic

Vibration of a steel truss bridge is monitored throughout a year in different environmental and operational conditions in order to identify the dynamic characteristics of the bridge through modal testing, and also to observe the variability of results due to the changes in environmental and operating conditions of the bridge. For this purpose, the vibration of the bridge is measured multiple times in different seasons while the temperature and the number of vehicles passing the bridge were recorded at the time of each test. Having the modal parameters identified from separate data sets, a statistical study could be performed to provide the mean and standard deviation of the identified parameters and also determine the possible dependency of the identified parameters on the mentioned variables. Furthermore, to observe the dependency of the estimated modal parameters on the selection of identification algorithm, one set of measured response is passed through different time- and frequency-domain algorithms and the obtained results are compared and their variation is extracted.

Validation of a wireless sensor network using local damage detection algorithm for beam-column connections

There has been a rapid advancement in wireless sensor network (WSN) technology in the past decade and its application in structural monitoring has been the focus of several research projects. The evaluation of the newly developed hardware platform and software system is an important aspect of such research efforts. Although much of this evaluation is done in the laboratories and using generic signal processing techniques, it is important to validate the system for its intended application as well. In this paper the performance of a newly developed accelerometer sensor board is evaluated by using the data from a beam-column connection specimen with a local damage detection algorithm. The sensor board is a part of a wireless node that consists of the Imote2 control/communication unit and an advanced antenna for improved connectivity. A scaled specimen of a steel beam-column connection is constructed in ATLSS center at Lehigh University and densely instrumented by synchronized networked systems of both traditional piezoelectric and wireless sensors. The column ends of the test specimen have fixed connections, and the beam cantilevers from the centerline of the column. The specimen is subjected to harmonic excitations in several test runs and its acceleration response is collected by both systems. The collected data is then used to estimate two sets of system influence coefficients with the wired one as the reference baseline. The performance of the WSN is evaluated by comparing the quality of the influence coefficients and the rate of convergence of the estimated parameters.

Pipelining in structural health monitoring wireless sensor network

Application of wireless sensor network (WSN) for structural health monitoring (SHM), is becoming widespread due to its implementation ease and economic advantage over traditional sensor networks. Beside advantages that have made wireless network preferable, there are some concerns regarding their performance in some applications. In long-span Bridge monitoring the need to transfer data over long distance causes some challenges in design of WSN platforms. Due to the geometry of bridge structures, using multi-hop data transfer between remote nodes and base station is essential. This paper focuses on the performances of pipelining algorithms. We summarize several prevent pipelining approaches, discuss their performances, and propose a new pipelining algorithm, which gives consideration to both boosting of channel usage and the simplicity in deployment.

Significance of sensor quality in modal identification of a bridge structure

Advancements in sensing technology have improved the practice of structural health monitoring in different aspects. One of the distinguished developments introduced to the monitoring systems is deployment of wireless technology for data communication in a sensing network. While researchers have shown the effective role of wireless sensor networks in improving the affordability of structural monitoring systems, their possible impact on the reliability and accuracy of the results is still a research question. Some challenges in the design of wireless sensor units, such as the trade-off between the functionality and the power consumption, and also attempts for minimizing the cost, have caused limitations in their architecture which do not necessarily exist in the design of wired systems. On the other hand, depending on the subsequent application of the results of sensing and monitoring, the accuracy of measurements and the level of uncertainty in results can be very important. Therefore, it is necessary to carefully investigate the impact of sensor quality on monitoring results. As an effort towards understanding the effects of sensor quality on the results of structural monitoring, this paper presents and validates a metric, called Physical Contribution Ratio (PCR), which can be used to investigate the influence of measurement noise on modal parameter identification. This parameter in applied for quantification of measurement noise effects on the quality of modal identification of a steel bridge structure. Bridge’s vibration is measured through use of wired and wireless sensors with different sensing qualities and the obtained results are compared through the use of the developed metric.

Network architecture design of an agile sensing system with sandwich wireless sensor nodes

Wireless sensor network (WSN) is recently emerged as a powerful tool in the structural health monitoring (SHM). Due to the limitations of wireless channel capacity and the heavy data traffic, the control on the network is usually not real time. On the other hand, many SHM applications require quick response when unexpected events, such as earthquake, happen. Realizing the need to have an agile monitoring system, an approach, called sandwich node, was proposed. Sandwich is a design of complex sensor node where two Imote2 nodes are connected with each other to enhance the capabilities of the sensing units. The extra channel and processing power, added into the nodes, enable agile responses of the sensing network, particularly in interrupting the network and altering the undergoing tasks for burst events. This paper presents the design of a testbed for examination of the performance of wireless sandwich nodes in a network. The designed elements of the network are the software architecture of remote and local nodes, and the triggering strategies for coordinating the sensing units. The performance of the designed network is evaluated through its implementation in a monitoring test in the laboratory. For both original Imote2 and the sandwich node, the response time is estimated. The results show that the sandwich node is an efficient solution to the collision issue in existing interrupt approaches and the latency in dense wireless sensor networks.