Structural health monitoring (SHM) is an anomaly detection process. Data-driven SHM has gained much attention compared to the model-based strategy, specifically with the current state-of-the-art machine learning routines. Model-based methods require structural information, time-consuming model updating, and may fail with noisy data, a persistent condition in real-time SHM problems. However, there are several hindrances in supervised and unsupervised settings in machine learning-based SHM. This study identifies and addresses such hindrances with the versatility of state-of-the-art deep learning strategies. While managing those complications, we aim at proposing a general, structure-independent (ie requires no prior information) SHM framework. Developing such techniques plays a crucial role in the SHM of smart cities. In the supervised SHM and sensor output validation (SOV) category, data class imbalance results from the lack of data from nuanced structural states. Employing Long Short-Term Memory (LSTM) units, we developed a general technique that manages both SHM and SOV. The developed architecture accepts high-dimensional features, enabling the train of Generative Adversarial Networks for data generation, addressing the complications of data imbalance. GAN-generated SHM data improved accuracy for low-sampled classes from 44.77% to 64.58% and from 73.39% to 90.84% in two SOV and SHM case studies, respectively. Arguing the unsupervised SHM as a practical category since it identifies novelties (ie unseen states), the current application of dimensionality reduction (DR) in unsupervised SHM is investigated. Due to the curse of dimensionality, classical unsupervised techniques cannot function with high-dimensional features, driving the use of DR techniques. Investigations highlighted the importance of avoiding DR in unsupervised SHM, as data dimensions that DR suppresses may contain damage-sensitive features for novelties. With DR, novelty detection accuracy declined up to 60% in two benchmark SHM datasets. Other obstacles in the unsupervised SHM area are case-dependent features, lack of dynamic-class novelty detection, and the impact of user-defined detection parameters on novelty detection accuracy. We chose the fast Fourier transform-based (FFT) of raw signals with no dimensionality reduction to develop the SHM framework. A deep neural network scheme is developed to perform the pattern recognition of that high-dimensional data. The framework does not require prior information, with GAN models implemented, offering robustness to sensor placement in structures. These characteristics make the framework suitable for developing general unsupervised SHM techniques. / Doctor of Philosophy / Detecting abnormal behaviors in structures from the input signals of sensors is called Structural health monitoring (SHM). The vibrational characteristics of signals or direct pattern recognition techniques can be applied to detect anomalies in a data-driven scheme. Machine learning (ML) tools are suitable for data-driven methods; However, challenges exist on both supervised and unsupervised ML-based SHM. Recent improvements in deep learning are employed in this study to address such obstacles after their identification. In supervised learning, the data points for the normal state of structures are abundant, and datasets are usually imbalanced, which is the same issue for the sensor output validation (SOV). SOV must be present before SHM takes place to remove anomalous sensor outputs. First, a unified decision-making system for SHM and SOV problems is proposed, and then data imbalance is alleviated by generating new data objects from low-sampled classes. The proposed unified method is based on the recurrent neural networks, and the generation mechanism is Generative Adversarial Network (GAN). Results indicate improvements in accuracy metrics for data classes in the minority. For the unsupervised SHM, four major issues are identified, including data loss during feature extraction, case-dependency of such extraction schemes. These two issues are solved with the fast Fourier transform (FFT) of signals to be the features with no reduction in their dimensionality. The other obstacles are the lack of discrimination between different novel classes (ie only two classes of damage and undamaged) and the effect of the detection parameters, defined by users, on the SHM analysis. The latter two predicaments are also addressed by online generating new data objects from the incoming signal stream with GAN and tuning the detection system to have the same performance regarding user-defined parameters regarding GAN-generated data. The proposed unsupervised technique is further improved to be insensitive to the sensor placement on structures by employing recurrent neural network layers in the GAN architecture, with the GAN that has overfitted discriminator.

Maintenance of the structural integrity and damage detection are critical for all massive and complicated new and aging structures. A structural health monitoring (SHM) system intends to identify damage on the structure under monitoring, so that necessary action can be taken in advance to avoid catastrophic results. Impedance-based SHM utilizes a piezoelectric ceramic as a collocated actuator and sensor, which measures the electrical impedance of the piezoelectric ceramic over a certain frequency range. The impedance profile of a structure under monitoring is compared against a reference profile obtained from the healthy structure. An existing approach called the sinc method adopts a sinc wave excitation and performs traditional discrete Fourier transform (DFT) based structural condition assessment. The sinc method requires rather intensive computing and a digital-to-analog converter (DAC) to generate a sinc excitation signal. It also needs an analog-to-digital converter (ADC) to measure the response voltage, from which impedance profile is obtained through a DFT. This dissertation investigates system design approaches for impedance-based structural health monitoring (SHM), in which a primary goal is low power dissipation. First, we investigated behaviors of piezoelectric ceramics and proposed an electrical model in order to enable us to conduct system level analysis and evaluation of an SHM system. Unloaded and loaded piezoelectric ceramics were electrically modeled with lumped linear circuit components, which allowed us to perform system level simulations for various environmental conditions. Next, we explored a signaling method called the wideband method, which uses a pseudorandom noise (PN) sequence for excitation of the structure rather than a signal with a particular waveform. The wideband method simplifies generation of the excitation signal and eliminates a digital-to-analog converter (DAC). The system form factor and power dissipation is decreased compared to the previously existing system based on a sinc signal. A prototype system was implemented on a digital signal processor (DSP) board to validate its approach. Third, we studied another low-power design approach which employs binary signals for structural excitation and structural response measurement was proposed. The binary method measures only the polarity of a response signal to acquire the admittance phase, and compares the measured phase against that of a healthy structure. The binary method eliminates the need for a DAC and an ADC. Two prototypes were developed: one with a DSP board and the other with a microcontroller board. Both prototypes demonstrated reduction of power dissipation compared with those for the sinc method and for the wideband method. The microcontroller based prototype achieved an on-board SHM system. Finally, we proposed an analytical method to assess the quality of the damage detection for the binary method. Using our method, one can obtain the confidence level of a damage detection for a given damage distance. / Ph. D.

Structural health monitoring (SHM) has become a strong national interest because of the need of reliable and accurate damage detection methods for aerospace, civil and mechanical engineering infrastructure. Health monitoring of these structures usually requires the sensors to have such features as large area coverage, maintenance free or minimum maintenance, ultra-low cost per measurement point, and capability of operation in harsh environments. Fiber Bragg grating (FBG) has attracted considerable interest for this application because of its compactness, electromagnetic immunity, and excellent multiplexing capability. Several FBG multiplexing techniques have been developed to increase the multiplexing number and further reduce the unit cost. To the author's best knowledge, the current demonstrated maximum multiplexing number are 800 FBG sensors in a single array using optical frequency domain reflectometry (OFDR), whose maximum fiber span is limited by the coherence length of light source. In this work, we proposed and demonstrated a wavelength-scanning time-division multiplexing (WSTDM) of 1000 ultra-weak FBGs for distributed temperature sensing. In comparison with the OFDR method, the WSTDM method distinguishes the sensors by different time delays, and its maximum operation distance, which is limited by the transmission loss of the fiber, can be as high as tens of kilometers. The strong multiplexing capability and low crosstalk of the ultra-weak FBG sensors was investigated through both theoretical analysis and experiment. An automated FBG fabrication system was developed for fast FBG fabrication. With this WSTDM method, we multiplexed 1000 ultra-weak FBGs for distributed temperature sensing. Besides the demonstrated temperature measurement, the reported method can also be applied to measure other parameters, such as strain, pressure. / Ph. D.

The research presented in this thesis is all based on the impedance method for structural health monitoring. The impedance method is an electro-mechanical technique which utilizes a single piezoelectric transducer as both a sensor and actuator. Due to the high frequencies of excitation used for the method, the sensing area for damage detection can be very localized. Previous work has shown that wave propagation can be added to systems already equipped with hardware for impedance-based structural health monitoring. The work in this thesis shows what happens under varying temperature conditions for a structure being monitored with wave propagation. A technique to compensate for temperature fluctuations is also presented. The work presented here is an initial study to directly correlate the actual amount of damage in a composite specimen with a damage metric indicated by impedance-based structural health monitoring. Two different damage mechanisms are examined: transverse matrix cracking and edge delamination. With both composite defects, a sample is interrogated with the impedance method before and after damage is introduced. The exact amount of damage in each specimen is found using radiography and compared with the health monitoring results. Traditional impedance techniques require the use of a bulky and expensive impedance analyzer. With the trend of structural health monitoring moving towards unobtrusive sensors which can be permanently placed on a structure, an impedance analyzer does not lend itself to these small, low power consuming requirements. In this thesis, an initial attempt to miniaturize the hardware is described. A prototype impedance-based structural health monitoring system, incorporating wireless based communications, is fabricated and validated with experimental testing on a number of different structures. The first steps towards a complete self-contained, robust structural health monitoring sensor are presented. / Master of Science

Structural Health Monitoring (SHM) is the science and technology of monitoring and can assess the condition of aerospace, civil, and mechanical infrastructures using a sensing system integrated into the structure. SHM is capable of detecting, locating, and quantifying various types of damage such as cracks, holes, corrosion, delamination, and loose joints, and can be applied to various kinds of infrastructures such as buildings, railroads, windmills, bridges, and aircraft. A major technical challenge for existing SHM systems is high power consumption, which severely limits the range of its applications. In this thesis, we investigated adoption of acoustic emission detection to reduce power dissipation of SHM systems employing the impedance and the Lamb wave methods. An acoustic emission sensor of the proposed system continuously monitors acoustic events, while the SHM system is in sleep mode. The SHM system is evoked to perform the SHM operation only when there is an acoustic event detected by the acoustic emission sensor. The proposed system avoids unnecessary operation of SHM operations, which saves power, and the system is effective for certain applications such as spacecraft and wind turbine blades. We developed prototype systems using a Texas Instruments TMS320F2812 DSP evaluation board for the Lamb wave method and an MSP430 evaluation board for the impedance method. / Master of Science

Structural Health Monitoring (SHM) is the technology of monitoring and assessing the condition of aerospace, civil, and mechanical infrastructures using a sensing system integrated into the structure. Among variety of SHM approaches, impedance-based method is efficient for local damage detection. This thesis focuses on system level concerns for impedance-based SHM. Two essential requirements are reached in the thesis: reduction of power consumption of wireless SHM sensor, and compensation of temperature dependency on impedance. The proposed design minimizes power by employing on-board signal processing, and by eliminating power hungry components such as ADC and DAC. The prototype implemented with MSP430 micro controller is verified to be able to handle SHM operation and wireless communication with extremely low-power: 0.15 mW during the inactive mode and 18 mW during the active mode. Each SHM operation takes about 13 seconds to consume 236 mJ. When our ASN-2 operates once in every four hours, it can run for about 2.5 years with two AAA-size batteries. To compensate for temperature change, we proposed an algorithm to select a small subset of baseline profiles for some critical temperatures and to estimate the baseline profile for a given ambient temperature through interpolation. Experimental results show that our method reduces the number of baseline profiles to be stored by 45%, and estimates the baseline profile of a given temperature accurately. / Master of Science

Smart infrastructure has become increasingly prevalent in recent decades due to the emergence of sophisticated and affordable sensing technologies. As sensors are deployed more widely and higher sampling rates are feasible, managing the massive scale of real-time data collected by these systems has become fundamental to providing relevant and timely information to decision-makers. To address this task, a novel open-source framework has been developed to manage and intuitively present high-volume data in near real-time. This design is centered around the goals of making data accessible, supporting decision-making, and providing flexibility to modify and reuse this framework in the future. In this work, the framework is tailored to vibration-based structural health monitoring, which can be used in near real-time to screen building condition. To promote timely intervention, distributed computing technologies are employed to accelerate the processing, storage, and visualization of data. Vibration data is processed in parallel using a publish-subscribe messaging queue and then inserted into a NoSQL database that stores heterogeneous data across several nodes. A REST-based web application allows interaction with this stored data via customizable visualization interfaces. To illustrate the utility of this framework design, it has been implemented to support a frequency domain monitoring dashboard for a 5-story classroom building instrumented with 224 accelerometers. A simulated scenario is presented to capture how the dashboard can aid decisions about occupant safety and structural maintenance. / Master of Science / Advances in technology have made it affordable and accessible to collect information about the world around us using sensors. When sensors are used to aid decision-making about structures, it is frequently referred to as Structural Health Monitoring (SHM). SHM can be used to monitor long-term structural health, inform maintenance decisions, and rapidly screen structural conditions following extreme events. Accelerometers can be used in SHM to capture vibration data that give insight into deflection patterns and natural frequencies in a structure. The challenge with vibration-based SHM and many other applications that leverage sensors is that the amount of data collected has the potential to grow to massive scales. To communicate relevant information to decision-makers, data must be processed quickly and presented intuitively. To facilitate this process, a novel open-source framework was developed for processing, storing, and visualizing high-volume data in near real-time. This framework combines multiple computers to extend the processing and storage capacity of our system. Data is processed in parallel and stored in a database that supports efficient data retrieval. A web application enables interaction with stored data via customizable visualization interfaces. To demonstrate the framework functionality, it was implemented in a 5-story classroom building instrumented with 224 accelerometers. A frequency-domain dashboard was developed for the building, and a simulated scenario was conducted to capture how the dashboard can aid decisions about occupant safety and structural maintenance.

In recent years, composite materials have been widely used in several applications due to their superior mechanical properties including high strength, high stiffness, and low density. Despite the remarkable advancements in theoretical and computational methods for analyzing composites, investigating the effect of lamina properties and lay-up configurations on the strength of composites still remains an active field of research. Finite Element Method (FEM) and Extended Finite Element Method (XFEM) are powerful tools for solving the boundary value problems. One of the objectives of this work is to employ XFEM as a defect identification tool for predicting the crack initiation and propagation in composites. Another major objective of this study is to investigate the damage development in hyperelastic materials. Two Finite Element models are adopted to study this phenomenon: multiscale modeling of the cord-rubber composites in tires and modeling of intelligent tires for evaluating the feasibility of the proposed defect detection technique. A new three-dimensional finite element approach based on the multiscale progressive failure analysis is employed to provide the theoretical predictions for damage development in the cord-rubber composites in tires. This new three-dimensional model of the cord-rubber composite is proposed to predict the different types of damage including matrix cracking, delamination, and fiber failure based on the micro-scale analysis. This process is iterative and data is shared between the finite element and multiscale progressive failure analysis. It is shown that the proposed cord-rubber composite model solves the problems corresponding to embedding the rebar elements to the solid elements and also increases the fidelity of numerical analysis of composite parts since the laminate characteristic variables are determined from the microscopic parameters. A tire rolling analysis is then conducted to evaluate the effects of different variables corresponding to the cord-rubber composite on the performance of tires. Tires operate on the principle of safe life and are the only parts of the vehicle which are in contact with the road surface. Establishing a computational method for defect detection in tire structures will help manufacturers to fix and develop more reliable tire designs. A Finite Element model of a tire with a tri-axial accelerometer attached to its inner-liner was developed and the effects of changing the normal load, longitudinal velocity and tire-road contact friction on the acceleration signal were investigated. Additionally, using the model, the acceleration signals obtained from several accelerometers placed in different locations around the inner-liner of the intelligent tire were analyzed and the defected areas were successfully identified. Using the new intelligent tire model, the lengths, locations, and the minimum number of accelerometers in damage detection in tires are determined. Comparing the acceleration signals obtained from the damaged and original tire models results in detecting defects in tire structures. / PHD / In recent years, composite materials have been widely used in several applications due to their superior mechanical properties. Studying the effect of different configurations and thicknesses on the strength of composites still remains an active field of research. Finite Element Method (FEM) is a powerful tool for simulating real problems. One of the objectives of this work is to employ FEM to show the damage development in the composite and rubber-based materials. Two Finite Element models are adopted to study this phenomenon: multiscale modeling of the cord-rubber composites in tires and modeling of intelligent tires, which are tires with sensors attached to the inner-liner, for evaluating the feasibility of the proposed defect detection technique. A new three-dimensional finite element approach based on the multiscale progressive failure analysis is employed to provide the theoretical predictions for damage development in the cord-rubber composites in tires. This new three-dimensional model of the cord-rubber composite is proposed to predict the different types of damage based on the micro-scale analysis. This process goes through the damage prediction formulations in each step to check whether damage happened or not. If damage happened, the stiffness of materials will be decreased. The fidelity of analysis is increased since the macro-scale mechanical properties are calculated based on the micro-scale properties. A tire rolling analysis is then conducted to evaluate the effects of different variables corresponding to the cord-rubber composite on the performance of tires. Tires operate on the principle of safe life and are the only parts of the vehicle which are in contact with the road surface. Establishing a computational method for defect detection in tire structures will help manufacturers to fix and develop more reliable tire designs. A tire with a sensor attached to its inner-liner was developed and the effects of changing the normal load, velocity and tire-road contact friction on the acceleration signal were investigated. Additionally, using the model, the acceleration signals obtained from several sensors placed in different locations around the inner-liner of the tire were analyzed. The defected areas were successfully identified by comparing the acceleration signals obtained from the damaged and original tire models.

Yes / Ultra-high performance concrete (UHPC) is an innovative cementitious composite, that has been widely applied in numerous structural projects because of its superior mechanical properties and durability. However, ensuring the safety of UHPC structures necessitates an urgent need for technology to continuously monitor and evaluate their condition during their extended periods of service. Self-sensing ultra-high performance concrete (SSUHPC) extends the functionality of UHPC system by integrating conductive fillers into the UHPC matrix, allowing it to address above demands with great potential and superiority. By measuring and analyzing the relationship between fraction change in resistivity (FCR) and external stimulates (force, stress, strain), SSUHPC can effectively monitor the crack initiation and propagation as well as damage events in UHPC structures, thus offering a promising pathway for structural health monitoring (SHM). Research on SSUHPC has attracted substantial interests from both academic and engineering practitioners in recent years, this paper aims to provide a comprehensive review on the state of the art of SSUHPC. It offers a detailed overview of material composition, mechanical properties and self-sensing capabilities, and the underlying mechanisms involved of SSUHPC with various functional fillers. Furthermore, based on the recent advancements in SSUHPC technology, the paper concludes that SSUHPC has superior self-sensing performance under tensile load but poor self-sensing performance under compressive load. The mechanical and self-sensing properties of UHPC are substantially dependent on the type and dosage of functional fillers. In addition, the practical engineering SHM application of SSUHPC, particularly in the context of large-scale structure, is met with certain challenges, such as environment effects on the response of SSUHPC. Therefore, it still requires further extensive investigation and empirical validation to bridge the gap between laboratory research and real engineering application of SSUHPC.

The primary focus of this research is evaluation of feasibility, applicability, and accuracy of Doppler Light Detection And Ranging (LIDAR) sensors as non-contact means for measuring track speed, distance traveled, and curvature. Speed histories, currently measured with a rotary, wheel-mounted encoder, serve a number of useful purposes, one significant use involving derailment investigations. Distance calculation provides a spatial reference system for operators to locate track sections of interest. Railroad curves, using an IMU to measure curvature, are monitored to maintain track infrastructure within regulations. Speed measured with high accuracy leads to high-fidelity distance and curvature data through utilization of processor clock rate and left-and right-rail speed differentials during curve navigation, respectively. Wheel-mounted encoders, or tachometers, provide a relatively low-resolution speed profile, exhibit increased noise with increasing speed, and are subject to the inertial behavior of the rail car which affects output data. The IMU used to measure curvature is dependent on acceleration and yaw rate sensitivity and experiences difficulty in low-speed conditions. Preliminary system tests onboard a 'Hy-Rail' utility vehicle capable of traveling on rail show speed capture is possible using the rails as the reference moving target and furthermore, obtaining speed profiles from both rails allows for the calculation of speed differentials in curves to estimate degrees curvature. Ground truth distance calibration and curve measurement were also carried out. Distance calibration involved placement of spatial landmarks detected by a sensor to synchronize distance measurements as a pre-processing procedure. Curvature ground truth measurements provided a reference system to confirm measurement results and observe alignment variation throughout a curve. Primary testing occurred onboard a track geometry rail car, measuring rail speed over substantial mileage in various weather conditions, providing high-accuracy data to further calculate distance and curvature along the test routes. Tests results indicate the LIDAR system measures speed at higher accuracy than the encoder, absent of noise influenced by increasing speed. Distance calculation is also high in accuracy, results showing high correlation with encoder and ground truth data. Finally, curvature calculation using speed data is shown to have good correlation with IMU measurements and a resolution capable of revealing localized track alignments. Further investigations involve a curve measurement algorithm and speed calibration method independent from external reference systems, namely encoder and ground truth data. The speed calibration results show a high correlation with speed data from the track geometry vehicle. It is recommended that the study be extended to provide assessment of the LIDAR's sensitivity to car body motion in order to better isolate the embedded behavior in the speed and curvature profiles. Furthermore, in the interest of progressing the system toward a commercially viable unit, methods for self-calibration and pre-processing to allow for fully independent operation is highly encouraged. / Ph. D.