Showing papers by "Charles R. Farrar published in 2000"

Structural Health Monitoring Using Statistical Process Control

Abstract: This paper poses the process of structural health monitoring in the context of a statistical pattern recognition paradigm. This paper particularly focuses on applying a statistical process control ...

A statistical pattern recognition paradigm for vibration-based structural health monitoring

Abstract: The process of implementing a damage detection strategy for aerospace, civil and mechanical engineering systems is often referred to as structural health monitoring. Vibration-based damage detection is a tool that is receiving considerable attention from the research community for such monitoring. In this paper, the structural health monitoring problem is cast in the context of a statistical pattern recognition paradigm. This pattern recognition process is composed of four portions; (1) operational evaluation, (2) data acquisition & cleansing, (3) feature selection & data compression, and (4) statistical model development. A general discussion of each portion of the process is presented, and the application of this statistical paradigm to two different real world structures, such as a bridge column and a surface-effect fast boat, is studied focusing on the issues of data normalization and feature extraction. INTRODUCTION Many aerospace, civil, and mechanical engineering systems continue to be used despite aging and the associated potential for damage accumulation. Therefore, the ability to monitor the structural health of these systems is becoming increasingly important from both economic and life-safety viewpoints. Damage identification based upon changes in dynamic response is one of the few methods that monitor changes in the structure on a global basis. The basic premise of vibration-based damage detection is that damage will significantly alter the stiffness, mass or energy dissipation properties of a system, which, in turn, alter the measured dynamic response of that system. Although the basis for vibration-based damage detection appears intuitive, its actual application poses many significant technical challenges. Because all vibration-based damage detection processes rely on experimental data with inherent uncertainties, statistical analysis procedures are necessary if one is to state in a quantifiable manner that changes in the vibration response of a structure are indicative of damage as opposed to operational and/or environmental variability. Therefore, this paper poses the vibration-based damage detection process in the context of a problem in statistical pattern recognition. A STATISTICAL PATTERN RECOGNITION PARADIGM In the context of statistical pattern recognition the process of vibration-based damage detection can be broken down into four parts; (1) operational evaluation, (2) data acquisition & cleansing, (3) feature extraction & data compression, and (4) statistical model development. Operational evaluation answers four questions in the implementation of a structural health monitoring system; (1) What are the life safety and/or economic justifications for monitoring the structure?; (2) How is damage defined for the system being monitored?; (3) What are the operational and environmental conditions under which the system of Smart Engineering System Design, Neural Networks, Fuzzy Logic, Evolutionary Programming, Complex Systems, and Data Mining, St. Louis, MO, USA, November 5-8, 2000. interest functions?; and (4) What are the limitations on acquiring data in the operational environment? Operational evaluation begins to set the limitations on what will be monitored and how to perform the monitoring as well as tailoring the monitoring to unique aspects of the system and unique features of the damage that is to be detected. The data acquisition portion of the structural health monitoring process involves selecting the types of sensors to be used, the location where the sensors should be placed, the number of sensors to be used, and the data acquisition/storage/transmittal hardware. Other considerations that must be addressed include how often the data should be collected, how to normalize the data, and how to quantify the variability in the measurement process. Data cleansing is the process of selectively choosing data to accept for, or reject from, the feature selection process. Filtering is one of the most common methods for data cleansing. The area of the structural damage detection process that receives the most attention in the technical literature is feature extraction. Feature extraction is the process of the identifying damage-sensitive properties derived from the measured vibration response that allows one to distinguish between the undamaged and damaged structures. Doebling et al. (1998) review propose many different methods for extracting damage-sensitive features from vibration response measurements. However, few of the cited references take a statistical approach to quantifying the observed changes in these features. The diagnostic measurement needed to perform structural health monitoring typically produces a large amount of data. Data compression into small dimensional features is necessary if accurate estimates of the feature statistical distribution are to be obtained. The need for low dimensionality in the feature vectors is referred to as the "curse of dimensionality" and is discussed in detail in general texts on statistical pattern recognition (Bishop 1995). The portion of the structural health monitoring process that has received the least attention in the technical literature is the development of statistical models to enhance the damage detection process. Statistical model development is concerned with the implementation of the algorithms that analyze the distributions of the extracted features in an effort to determine the damage state of the structure. The algorithms used in statistical model development usually fall into the three general categories of; (1) group classification, (2) regression analysis, and (3) outlier detection. The appropriate algorithm to use will depend on the ability to perform supervised or unsupervised learning. Here, supervised learning refers to the case were examples of data from damaged and undamaged structures are available. Unsupervised learning refers to the case were data are only available from the undamaged structure. EXPERIMENTAL APPLICATIONS The statistical pattern recognition paradigm is applied to vibration test data obtained from two different structural systems; (1) acceleration time series obtained from a pier in its undamaged state and then after various levels of damaged had been introduced through cyclic loading, and (2) strain time measurements recorded under various operational and environmental conditions of a surface-effect fast patrol boat. The examples presented here emphasize on the issues of data normalization and feature extraction. A Bridge Column Test This test applies statistical process control methods referred to as "control charts" to vibration-based damage detection (Montgomery, 1996). In this study an X-bar control Smart Engineering System Design, Neural Networks, Fuzzy Logic, Evolutionary Programming, Complex Systems, and Data Mining, St. Louis, MO, USA, November 5-8, 2000. chart is employed to monitor the changes of the selected feature means and to identify samples that are inconsistent with the past data sets. Application of the S control chart, which measures the variability of the structure over time, to the current test structure is presented in Fugate et al (2000). First, an auto-regressive (AR) model is fit to the measured acceleration-time histories from an undamaged structure. Residual errors, which quantify the difference between the prediction from the AR model and the actual measured time history at each time interval, are used as the damage-sensitive features. (Note that correlated data lead to the underestimation of control limits. The use of residual errors as damage sensitive features removes correlation in the data being monitored by the control charts.) Next, the X-bar and S control charts are employed to monitor the mean and variance of the selected features. Control limits for the control charts are constructed based on the features obtained from the initial intact structure. The residual errors computed from the previous AR model and subsequent new data are then monitored relative to the control limits. A statistically significant number of error terms outside the control limits indicate a system transit from a healthy state to a damage state. The assumption here is that there will be a significant increase in the residual errors when an AR model developed from the undamaged linear system response is used to predict the response from a damaged system exhibiting nonlinear response. For demonstration, this statistical process control is applied to vibration test data acquired from a concrete bridge column (Figure 1) as the column is progressively damaged. The details of the test structure are given in Fugate et al (2000) Figure 2 shows the control charts for all damage levels. In these charts, CL, UCL, and LCL denote the centerline, upper and lower control limits, respectively. Outliers correspond to subgroup sample means outside the control limits and are marked by a “+”. Because the control limits represent a 99% confidence interval, approximately 20 charted values (= 1 % of total 2046 samples) are expected to be outside the control limits even when the system is in control. Therefore, the 13 outliers in Figure 2 (a) are not unusual and do not indicate any system anomaly for damage level 0. However, the control charts successfully indicated some system anomaly for damage levels 1 though 5 by showing a statistically significant number of outliers. In general, the observation of a large number of outliers does not necessarily indicate that a structure is damaged but only that the system has varied to cause statistically significant changes in its vibration signature. This variability could be caused by a variety of environmental and operational conditions that the system of interest is subject to. A data normalization procedure to account for these operational and environmental variations is presented in the next example. A Surface-Effect Fast Patrol Boat This example discusses on the data

Structural Health Monitoring Studies of the Alamosa Canyon and I-40 Bridges

Abstract: From 1994 to 1997 internal research grants from Los Alamos National Laboratory's Laboratory Direct Research and Development (LDRD) office were used to fund an effort aimed at studying global vibration-based damage detection methods. To support this work, several field tests of the Alamosa Canyon Bridge have been performed to study various aspects of applying vibration-based damage detection methods to a real world in situ structure. This report summarizes the data that has been collected from the various vibration tests performed on the Alamosa Canyon Bridge, analyses of these data, and the results that have been obtained. Initially, it was the investigators' intent to introduce various types of damage into this bridge and study several vibration-based damage detection methods. The feasibility of continuously monitoring such a structure for the onset of damage was also going to be studied. However, the restrictions that the damage must be relatively benign or repairable made it difficult to take the damage identification portion of the study to completion. Subsequently, this study focused on quantifying the variability in identified modal parameters caused by sources other than damage. These sources include variability in testing procedures, variability in test conditions, and environmental variability. These variabilities must be understood and their influence on identified modal properties quantified before vibration-based damage detection can be applied with unambiguous results. Quantifying the variability in the identified modal parameters led to the development of statistical analysis procedures that can be applied to the experimental modal analysis results. It is the authors' opinion that these statistical analysis procedures represent one of the major contributions of these studies to the vibration-based damage detection field. Another significant contribution that came from this portion of the study was the extension of a strain-energy-based damage detection method originally developed for structures that exhibit beam-bending response to structures that exhibit plate-like bending or bending in two directions. In addition, based on lessons learned from the Alamosa Canyon Bridge test, data from the I-40 Bridge tests have been re-analyzed using the statistical analysis procedures developed as part of this study. The application of these statistical procedures to the I-40 Bridge test results gives particular insight into how statistical analysis can be used to enhance the vibration-based damage detection process.

Pattern recognition for structural health monitoring

Abstract: The process of implementing a damage detection strategy for engineering systems is often referred to as structural health monitoring. Vibration-based damage detection is a tool that is receiving considerable attention from the research community for such monitoring. Recent research has recognized that the process of vibration-based structural health monitoring is fundamentally one of statistical pattern recognition and this paradigm is described in detail. This process is composed of four portions: (1) Operational evaluation; (2) Data acquisition and cleansing; (3) Feature selection and data compression, and (4) Statistical model development for feature discrimination. A general discussion of each portion of the process is presented.

Unsupervised learning methods for vibration-based damage detection

Abstract: The basic premise of vibration-based damage detection is that damage will significantly alter the stiffness, mass, or energy dissipation properties of a system, which, in turn, alter the measured dynamic response of the system. Although the basis for vibration-based damage detection appears intuitive, its actual application poses many significant technical challenges. A fundamental challenge is that in many situations vibration-based damage detection must be performed in an unsupervised learning mode. Here, the term unsupervised learning implies that data from damaged systems are not available. These challenges are supplemented by many practical issues associated with making accurate and repeatable vibration measurements at a limited number of locations on complex structures often operating in adverse environments. This paper will discuss two statistical methods for approaching the unsupervised learning damage detection problem. The first method is density estimation and significance testing. The second method is statistical process control. Examples of these methods are applied to data from an undamaged and subsequently damaged concrete column.

Modal parameter extraction of Z24 bridge data

Abstract: The vibration data obtained from ambient, drop-weight, and shaker excitation tests of the Z24 Bridge in Switzerland are analyzed to extract modal parameters such as natural frequencies, damping ratios, and mode shapes. Two system identification techniques including Frequency Domain Decomposition and Eigensystem Realization Algorithm are employed for the extraction of modal parameters and the stationarity of the bridge is also investigated using time-frequency analysis.

Statistical process control and projection techniques for structural health monitoring

Abstract: This paper focuses on applying statistical process control techniques to vibration-based damage diagnosis. First, an auto-regressive (AR) model is fit to the measured acceleration-time histories from an undamaged structure. Coefficients of the AR model are selected as the damage-sensitive features for the subsequent control chart analysis. Finally, the AR coefficients of the models fit to subsequent new data are monitored relative to the control limits. A unique aspect of this study is the coupling of various projection techniques such as principal component analysis, linear and quadratic discriminant operators with the statistical process control in an effort to enhance the discrimination between features from the undamaged and damaged structures. This combined statistical procedure is applied to vibration test data acquired from a concrete bridge column as the column is progressively damaged. The coupled approach captures a clearer distinction between undamaged and damaged vibration responses.

Damage diagnosis using statistical process control

Abstract: Structural health monitoring is described in the context of a statistical process control paradigm. This paper demonstrates the application of various statistical process control techniques such as the Shewhart, the exponentially weighted moving average, and the cumulative sum control charts to vibration-based damage diagnosis. The control limits are first constructed based on the measurements obtained from the initial intact structure. Then, new data are monitored against the control limits. A statistically significant number of outliers outside the control limits indicate a system transition from a healthy state to a damage state. Environmental and operation conditions, such as temperature change and the magnitude variation of the input forces, are also incorporated into the monitoring process. Blind tests of various damage cases are conducted without prior knowledge of the actual damage scenarios to evaluate the performance of the presented control chart techniques.

Continuous structural monitoring using statistical process control

Abstract: A damage detection problem is cast in the context of a statistical pattern recognition paradigm In particular, this paper focuses on applying statistical process control methods referred to as control charts to vibration-based damage detection First, an auto-regressive (AR) model is fitted to the measured time histories from an undamaged structure Residual errors, which quantify the difference between the prediction from the AR model and the actual measured time history at each time interval, are used as the damage-sensitive features Next, the average and variability of the selected features are monitored by the X-bar and S control charts A statistically significant number of error terms outside the control limits indicate a system transit from a healthy state to a damage state For demonstration, this statistical process control is applied to vibration test data acquired from a concrete bridge column as the column is progressively damaged

Time series analyses for locating damage sources in vibration systems

Abstract: A novel time series analysis procedure is presented to localize damage sources in a mechanical system An attempt is made to pinpoint the sources of nonlinear damage by solely analysing the vibration signatures recorded from a structure of interests First, a linear prediction model, combining Auto-Regressive (AR) and Auto-Regressive with eXogenous inputs (ARX) techniques, is estimated using a time series recorded under an undamaged stage of the structure Then, the residual error, which is the difference between the actual time measurement and the prediction from the previously estimated AR-ARX combined model, is defined as our damage-sensitive feature This study is based on the premise that if there were damage in the structure, the prediction model previously identified using the undamaged time history data would not be able to reproduce the newly obtained time series data measured under a damaged state of the structure Furthermore, the increase of the residual errors would be maximised at the sensors instrumented near the actual damage locations The applicability of this approach is demonstrated using the vibration test data obtained from an eight degrees-of-freedom (DOF) mass-spring system 1 505-667-6135 (Voice), 505-665-7836 (Fax), sohn@lanlgov (E-mail) 2 505-667-4551 (Voice), 505-665-2137 (Fax), farrar@lanlgov (E-mail)

Issues for the application of statistical models in damage detection

Abstract: Many aerospace, civil, and mechanical systems continue to be used despite aging and the associated potential for damage accumulation. Therefore, the ability to monitor the structural health of these systems is becoming increasingly important. A wide variety of highly effective local non-destructive evaluation tools are available. However, damage identification based upon changes in vibration characteristics is one of the few methods that monitor changes in the structure on a global basis. The process of vibration-based damage detection will be described as a problem in statistical pattern recognition. This process is composed of four portions: 1.) Operational Evaluation, 2.)Data acquisition and cleansing; 3.) Feature selection and data compression, and 4.) Statistical model development. Current studies regarding supervised learning methods for statistical model development are discussed and emphasized with the application of this technology to a laboratory test structure. Specifically, a comparison is made between a linear discriminant classifier and a general Bayesian classifier for the purpose of determining the existence of damage.

Data normalization issue for vibration-based structural health monitoring

Abstract: The process of implementing a damage detection strategy for aerospace, civil and mechanical engineering systems is often referred to as structural health monitoring. In this paper, the structural health monitoring problemis cast in the context of a statistical pattern recognition paradigm. This pattern recognition process is composed of four portions: 1.) Operational evaluation; 2.) Data acquisition & cleansing; 3.) Feature selection & data compression, and 4.) Statistical model development for feature classification. This paper mainly focuses on the discussion of feature extraction and classification issues using the fiber optic strain gauge data obtained from two different structural conditions of a surface-effect fast patrol boat. The main objective is to extract features and to construct a statistical model that distinguishes the signals recorded under the different structural conditions of the boat. The feature extraction process began by looking at relatively simple statistics of the signals and progressed to using the residual errors from auto-regressive (AR) models fit to the measured data as the damage-sensitive features. Data normalization proved to be the most challenging portion of this investigation. A novel approach to data normalization, where the residual errors in the AR model are considered to be an unmeasured input and an auto-regressive model with exogenous inputs (ARX) is then fit to portions of the data exhibiting similar waveforms, was successfully applied to this problem. With this proposed procedure, a clear distinction between the two different structural conditions was achieved.

Structural health monitoring activities at los alamos national laboratory

Abstract: The authors of this summary would like to thank the following people whose work is summarized in this paper: William Baker, Philip Cornwell, Tim Darling, Thomas Duffey, Norman Hunter, Albert Migliori. The managers, support personnel, university students, and collaborators who have been involved in these studies also deserve more credit than can be offered herein. Los Alamos National Laboratory is operated by the University of California for the United States Department of Energy.

The los alamos dynamics summer school

Abstract: A unique summer educational program focusing on engineering dynamics has been developed and implemented af Los Afamos National Laboratory. The purpose of this summer school is to expose a select group of students to the broad field of engineering dynamics with the hopes thaf they will be motivated to pursue this area of research in their graduate studies. The summer school activities included I) lectures on various engineering topics such as computational structural dynamics, experimental modal analysis, random vibrations, signal processing, etc., 2) a distinguished lecturer series in which prominent guest lecturers gave talks about “cutting edge research” in structural dynamics, 3) field trips and 4) an eight week project having both an analytical and an experimental component. This paper presents an overview of the program. 1. PROGRAM DESCRIPTION Over the last 20 years there has been a 20% decline in the number of engineering degrees granted while university degrees in general have increased approximately 20% [l]. Engineering dynamics, which encompasses areas such as flight dynamics, vibration isolation for precision manufacturing, earthquake engineering, blast loading, signal processing, experimental modal analysis, etc. is naturally affected by this decrease in numbers. The competition for talented individuals with the background necessary to replace those leaving the field of engineering dynamics necessitates a proactive approach of identifying, motivating, and educating students who are embarking on their graduate school career. The Los Alamos Dynamics Summer School was designed with this proactive approach in mind. The program is &signed not only to benefit the students through their educatibnal experience, but also to motivate them to attend graduate school and to make the students aware of career possibilities in the field of engineering dynamics after they have completed their graduate studies. The summer school had two focus areas. First, the multidisciplinary nature of research in engineering dynamics was emphasized throughout the summer school. To this end, the students were assigned to multi-disciplinary teams and assigned a project where a coupled analytical/experimental approach to the problem was required. Second, the program was designed to develop the students’ written and oral communications skills. To develop these skills, the students were required to give numerous informal oral presentations of their work as it progressed throughout the summer culminating in a formal oral presentation and a paper written for the International Modal Analysis Conference.