PhD, University of Cincinnati, 2013, Engineering and Applied Science: Mechanical Engineering

Monitoring the health condition of machinery has been an area of research for quite some time. Despites several advancements, the application of conventional signal analysis and pattern recognition methods face several challenges when the operating variables such as load, speed, and temperature vary considerably for the monitored asset. The residual clustering approach addresses the multi-regime monitoring challenge by first modeling the baseline non-linear correlation relationship in the measured signal features and by providing predicted signal features. Calculating the residual signal features allows one to normalize the effect of the operating variables, since one is considering how the response of the system compares with the predicted response based on the baseline behavior. In many instances the degradation signature of a component or system is more pronounced under certain operating conditions. The clustering portion of the residual clustering method specifically addresses the regime dependent signature aspect and bases the health value on the monitoring regime in which the degradation signature is more prevalent. This dissertation work highlights the mathematical framework and provides guidance on the appropriate processing methods for each portion of the approach. From simulation studies and wind speed data, the results highlight that the auto-associative neural network method provides the lowest prediction error when compared with regression, neural network, and principal component analysis methods. The results from this dissertation work also imply that the selection of the clustering algorithm does not significantly affect the calculated health value, and in general, most clustering algorithms appear suitable for detecting the problem using the residual clustering approach. The feasibility of the residual clustering approach is demonstrated in three case studies. For the wind speed sensor health monitoring case study, the residual clusterin (open full item for complete abstract)

Committee: Jay Lee Ph.D. (Committee Chair); Canh Ly Ph.D. (Committee Member); Teik Lim Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Mechanics

MS, University of Cincinnati, 2012, Engineering and Applied Science: Mechanical Engineering

In-situ monitoring of materials is a great problem in the field of structural health monitoring. The ability to receive real time data relaying the condition of a body is an elusive but invaluable goal. Even more difficult is monitoring the continuous body rather than a small subset of discrete points which may or may not represent the health of the whole body. The final challenge, specific to elastomeric materials, is to develop a sensor capable of surviving a great deal of strain as the body bends, flexes, and stretches during typical operation. This thesis provides a solution to these problems by exploring the development and performance of a continuous sensor skin. This skin has been carefully developed to survive the operational metrics of steel reinforced hydraulic hoses. This thesis explores several avenues for development with a focus on thosewhichshowpromiseinhydraulichoseapplications. Severaldifferenttheoriesofhowthesensor may operate are discussed in detail while the three most common failure modes are tested: puncture, tear, and foreign object damage.

Committee: Mark Schulz PhD (Committee Chair); Jay Lee PhD (Committee Member); Vesselin Shanov PhD (Committee Member) Subjects: Mechanics

Master of Science, Miami University, 2020, Computational Science and Engineering

Power electronics are critical components in society's modern infrastructure. In electrified vehicles and aircraft, losing power jeopardizes personal safety and incur financial penalties. Because of these concerns, many researchers explore condition monitoring (CM) methods that provide real-time information about a system';s health. This thesis develops a CM method that determines the health of a DC-link capacitor in a three-phase inverter. The approach uses measurements from a current transducer in two Machine Learning (ML) algorithms, a Support Vector Machine (SVM), and an Artificial Neural Network (ANN), that classify the data into groups corresponding to the capacitor's health. This research evaluates six sets of data: time-domain, frequency-domain, and frequency-domain data subjected to four smoothing filters: the moving average with a rectangular window (MARF) and a Hanning window, the locally weighted linear regression, and the Savitzky-Golay filter. The results show that both ML algorithms estimate the DC-link capacitor health with the highest accuracy being 91.8% for the SVM and 90.7% for the ANN. The MARF-smoothed data is an optimal input data type for the ML classifiers due to its low computational cost and high accuracy. Additionally, a Hidden Markov Model increases the classification accuracy up to 98% when utilized with the ANN.

Committee: Mark Scott Dr. (Advisor); Chi-Hao Cheng Dr. (Committee Member); Peter Jamieson Dr. (Committee Member) Subjects: Computer Engineering; Electrical Engineering

Master of Sciences (Engineering), Case Western Reserve University, 2019, EMC - Mechanical Engineering

The objective of this research is to further document and bring feasibility to milling tool condition monitoring using acoustic signals. In order to accomplish this objective, a sound signal model is developed which characterizes the acoustic signals of the milling process. Using this model, two machine learning methods are developed to detect tool wear. One method utilizes data from all tool wear classes available for learner training and the other utilizes only a single class for training. The latter technique solves a data availability issue regarding running milling machines under suboptimal conditions, which is discussed herein. Each machine learning model is shown to be effective at tool wear detection tasks. This research demonstrates the power of machine learning in acoustic tool condition monitoring and makes significant novel contributions to the field. This research demonstrates the feasibility of the monitoring technique and lays a groundwork for future work in the field.

Committee: Robert Gao (Committee Chair); Michael Lewicki (Committee Member); Chris Yuan (Committee Member) Subjects: Acoustics; Engineering; Mechanical Engineering

Master of Science in Engineering, University of Akron, 2019, Electrical Engineering

This thesis proposes a superior magnetic field based energy harvesting system, with a wide voltage range operation and low start-up voltage. Design parameters have been analyzed to determine their impact on the amount of power that can be harvested. An efficient and novel power processing unit which produces a regulated DC voltage and distributes the power between the load and the backup energy storage element is proposed. Experimental results have demonstrated that the proposed nanocrystalline harvester core achieves very high power density and can produce as much as 55W of power from a power line carrying 615A of 60Hz current, which is much higher than that reported in many research papers, and when implemented with the associated power processing unit results in a complete energy harvesting system that can start to operate at a low harvester coil AC voltage of 2.7V and processes the power to obtain a regulated DC voltage of 12.5V. The proposed energy harvester system can be used as a power source for high power condition monitoring systems, weather stations, and systems where the use of conventional mains supply, batteries, or renewable energy sources such as solar or wind may not be convenient, practical, or economical.

Committee: J. Alexis De Abreu García Dr. (Advisor); Yilmaz Sozer Dr. (Advisor); Robert J. Veillette Dr. (Committee Member) Subjects: Electrical Engineering; Electromagnetics; Energy

PhD, University of Cincinnati, 2016, Engineering and Applied Science: Mechanical Engineering

Machine health monitoring has advanced significantly for improving machine uptime and efficiency by providing proper fault detection and remaining useful life (RUL) prediction information to machine users. Despite these advancements, conventional condition monitoring (CM) techniques face several challenges in machine prognostics, including the ineffective RUL prediction modeling for machine under dynamic working regimes, and the lack of complete lifecycle data for modeling and validation, among others. To address these issues, this research introduces Accelerated Degradation Tests (ADT) with a deep learning technique, which is a novel method to improve machine life prediction accuracy under different working regimes for Prognostics and Health Management applications. This dissertation work highlights the mathematical framework of deep learning based machine life modeling under an ADT environment, including Constant Stress Accelerated Degradation Testing (CSADT) and Step-Stress ADT (SSADT) conditions. Since most CM features show no trend or indication of failure until a machine is approaching the end of its life, current RUL prediction techniques are not applicable in that they are only effective when incipient degradation is detected. This dissertation work applies feature enhancement to condition-based features using the enhanced Restricted Boltzmann Machine (RBM) method with a prognosability regularization term; afterwards, a similarity-based method is applied to predict machine life with the enhanced RBM features. In addition, this research has added varying stress conditions during experiments to replicate dynamic operation regimes. The stress variable, a type of regime variables, is input into Mixed-Variate RBM (MV-RBM) model. Therefore, a Regime Matrix based RBM (RM-RBM) is proposed to improve the feature prognosability and reduce the impact that the working stresses have on the features. Then the RBM features can be fused into a single health value which ref (open full item for complete abstract)

Committee: Jay Lee Ph.D. (Committee Chair); Linxia Liao Ph.D. (Committee Member); Teik Lim Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Mechanical Engineering; Mechanics

MS, University of Cincinnati, 2016, Engineering and Applied Science: Mechanical Engineering

The global installed capacity of wind turbines has been growing rapidly during the past decade. Along with the fast-growing number of wind turbines, the concerns for their maintenance and health management are also accumulating. The repair and maintenance for wind turbines are very expensive and time-consuming due to various reasons including logistics difficulties, distant locations, costly spare parts, and expensive labor force, etc. Prognostics and health management (PHM) technologies are of vital importance to wind turbines operation and maintenance since it can detect incipient faults in early time and predict the trend of their propagation so that the maintenance activities can be planned ahead of time to reduce the downtime and maintenance cost. Drivetrain systems are of the most concern in maintenance as they contribute the most downtime and repair costs. While there are various condition monitoring techniques available for drivetrain systems, vibration-based techniques have been most widely adopted due to its direct access to structure response and capability for early detection of incipient faults. However, there are impeding challenges of its application in wind turbine PHM: how to extract meaningful features from vibration signal when the rotating speed is unknown; how to detect and enhance incipient fault features under dynamic operation regimes and harsh environment; how to convert the multidimensional feature vectors into actionable health indicator to plan maintenance; and how to align these analytical techniques to enable smart, self-contained and unsupervised condition monitoring systems in big data environment. This thesis presents a systematic framework for unsupervised feature mining and fault detection for drivetrain systems. It consists several novel techniques that address the critical issues for vibration-based condition monitoring: A novel method for instantaneous angular speed estimation under non-stationary operation conditions based on (open full item for complete abstract)

Committee: Jay Lee Ph.D. (Committee Chair); Jay Kim Ph.D. (Committee Member); Allyn Phillips Ph.D. (Committee Member) Subjects: Mechanical Engineering; Mechanics

Doctor of Philosophy, Case Western Reserve University, 2016, EECS - System and Control Engineering

In this dissertation, we propose a novel Swarm Intelligence based approach for condition monitoring of dynamic systems. Systems can be viewed as collections of interconnected elements that communicate with one another through physical phenomena. Therefore, the behaviors of individual elements can be observed in the behaviors of other communicating system elements. We develop distributed algorithms for extracting the intrinsic communication topologies within a dynamic system and employ them for condition monitoring of the system. The structure of this communication topology is itself directly applicable to monitoring as it necessarily reflects the behavior of the system's elements as well as environmental effects and thus captures aspects of the system's operational state. The proposed approach in this dissertation provides the operators with graphs instead of tables or correlation matrices to efficiently identify the presence of a fault and request appropriate actions to mitigate it. Specifically, this dissertation considers dynamic systems and the available observations from a foraging perspective and discusses the application of ant foraging behavior-based search techniques to the discovery of the intrinsic communication topology of systems using information-theoretic measures of information flow within systems. Graph similarity measure algorithms are also used to quantify changes in communication topologies and, thus, for detecting the change points in the state of the system reflected by the topology changes. In this work, we present the mathematical framework for our proposed approach. Then, we consider a system consisting of multiple networks each with different topologies and connection strengths, where the individual networks composing the system can be switched at different times. The proposed algorithm is applied to this system to detect the switches between the networks and discover the communication topologies associated with network configura (open full item for complete abstract)

Committee: Kenneth A. Lopao (Advisor) Subjects: Electrical Engineering

Master of Science in Engineering, Youngstown State University, 2013, Department of Civil/Environmental and Chemical Engineering

With a large inventory of deficient and aging bridges in the United States, this research focused on developing dynamic response based health monitoring system of prestressed box beam (PSBB) bridges that will provide more realistic and cost-efficient results. The hypothesis is based on the assumption that the dynamic response is a sensitive and important indicator of the physical integrity and condition of a structure. Two wireless sensor networks (WSNs) were deployed for the collection of real-time acceleration response of a 25-year old PSBB bridge under trucks with variable loads and speeds. The acceleration response of the bridge at its newest condition was collected from the dynamic simulations of its full-scale finite element (FE) models mimicking field conditions. The FE model was validated using experimental and theoretical methods. The acceleration data in time domain were transformed into frequency domain using Fast Fourier Transform to determine peak amplitudes and their corresponding fundamental frequencies for the newest and the current condition of the bridge. The analyses and comparisons of the bridge dynamic response between the newest and the current bridge interestingly indicate a 37% reduction in its fundamental frequency over its 25 years of service life. This reduction has been correlated to the current condition rating of the bridge to develop application software for quick and efficient condition assessment of a PSBB bridge. The application software can instantly estimate overall bridge condition rating when used with the WSN deployed on a PSBB bridge under vehicular loads. The research outcome and the software is expected to provide a cost-effective solution for assessing the overall condition of a PSBB bridge, which helps to reduce maintenance costs and provide technologically improved bridge maintenance service.

Committee: AKM Anwarul Isam Ph.D. (Advisor); Javed Alam Ph.D. (Committee Member); Frank Li Ph.D. (Committee Member) Subjects: Civil Engineering; Engineering

Master of Science in Engineering, University of Akron, 2010, Mechanical Engineering

Condition monitoring of lubricating oil is a preventive tool that can be used to schedule machine maintenance downtime and predict impending machine failure. Techniques and apparatus for continuous on-line debris detection and oil analysis are become more and more sought after in these modern times. During normal machine operation small wear debris particles of on the order of 1 to 10 microns are generated. When abnormal wear begins, large debris particles in the range of 10 to 150 microns are produced. The particle size and concentration will increase gradually until machine failure. Also, during prolonged or extreme usage conditions various forms of contamination and additive depletion can begin to degrade the physical properties of the lubricant to an unsatisfactory level. Therefore, continuous monitoring of wear debris and critical oil properties is essential to prevent catastrophic system failure of machines. This thesis demonstrates the development of a comprehensive procedure for detecting debris and analyzing physical parameters associated with lubricating oil degradation using ultrasonic and capacitance based measurements. It was found that both ultrasonic and capacitance measurements can detect particles as small as 1.75 mils (44.5 μm) in diameter. The ultrasonic system detects debris by measuring the decrease in ultrasonic intensity caused by scattering of the wave by the presence of debris in the oil. As lubricating oil is non-conductive, the capacitance based system monitors increases in the effective capacitance of the system brought on by the presence of conductive wear debris. In an effort to expand this technique to include analysis of critical oil properties, ultrasonic and capacitance measurements are done to examine changes in the viscosity and pH of the oil. Comparative viscosity measurements were taken using the ultrasound equipment. A relationship between differences in viscosity and the amplitude and transit time of an ultrasonic wave wer (open full item for complete abstract)

Committee: Fred Choy Dr. (Advisor) Subjects: Mechanical Engineering