Mechanical Engineering Theses and Dissertations

Permanent URI for this collectionhttp://hdl.handle.net/1903/2795

Browse

Search Results

Now showing 1 - 10 of 12

Application of Diagnostics and Prognostics Techniques to Qualification Against Wear-Out Failure
(2022) Ram, Abhishek; Das, Diganta; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
QUALIFICATION IS A PROCESS THAT DEMONSTRATES WHETHER A PRODUCT MEETS OR EXCEEDS SPECIFIED REQUIREMENTS. TESTING AND DATA ANALYSIS PERFORMED WITHIN A QUALIFICATION PROCEDURE SHOULD VERIFY IF PRODUCTS SATISFY THOSE REQUIREMENTS, INCLUDING RELIABILITY REQUIREMENTS. MOST OF THE ELECTRONICS INDUSTRY QUALIFIES PRODUCTS USING PROCEDURES DICTATED WITHIN QUALIFICATION STANDARDS. A REVIEW OF COMMON QUALIFICATION STANDARDS REVEALS THAT THOSE STANDARDS DO NOT CONSIDER CUSTOMER REQUIREMENTS OR THE PRODUCT PHYSICS-OF-FAILURE IN THAT INTENDED APPLICATION. AS A RESULT, QUALIFICATION, AS REPRESENTED IN THE REVIEWED QUALIFICATION STANDARDS, WOULD NOT MEET OUR DEFINITION OF QUALIFICATION FOR RELIABILITY ASSESSMENT. THIS THESIS PROVIDES AN APPLICATION-SPECIFIC APPROACH FOR DEVELOPING A QUALIFICATION PROCEDURE THAT ACCOUNTS FOR CUSTOMER REQUIREMENTS, PRODUCT PHYSICS-OF-FAILURE, AND KNOWLEDGE OF PRODUCT BEHAVIOR UNDER LOADING. THIS THESIS PROVIDES A REVAMPED APPROACH FOR DEVELOPING A LIFE CYCLE PROFILE THAT ACCOUNTS FOR LOADING THROUGHOUT MANUFACTURING/ASSEMBLY, STORAGE AND TRANSPORTATION, AND OPERATION. THE THESIS ALSO DISCUSSES IDENTIFYING VARIATIONS IN THE LIFE CYCLE PROFILE THAT MAY ARISE THROUGHOUT THE PRODUCT LIFETIME AND METHODS FOR ESTIMATING LOADS. THIS UPDATED APPROACH FOR DEVELOPING A LIFE CYCLE PROFILE SUPPORTS BETTER FAILURE PRIORITIZATION, TEST SELECTION, AND TEST CONDITION AND DURATION REQUIREMENT ESTIMATION. ADDITIONALLY, THIS THESIS INTRODUCES THE APPLICATION OF DIAGNOSTICS AND PROGNOSTICS TECHNIQUES TO ANALYZE REAL-TIME DATA TRENDS WHILE CONDUCTING QUALIFICATION TESTS. DIAGNOSTICS TECHNIQUES IDENTIFY ANOMALOUS BEHAVIOR EXHIBITED BY THE PRODUCT, AND PROGNOSTICS TECHNIQUES FORECAST HOW THE PRODUCT WILL BEHAVE DURING THE REMAINDER OF THE QUALIFICATION TEST AND HOW THE PRODUCT WOULD HAVE BEHAVED IF THE TEST CONTINUED. AS A RESULT, COMBINING DIAGNOSTICS AND PROGNOSTICS TECHNIQUES CAN ENABLE THE PREDICTION OF THE REMAINING TIME-TO-FAILURE FOR THE PRODUCT UNDERGOING QUALIFICATION. SEVERAL ANCILLARY BENEFITS RELATED TO AN IMPROVED TESTING STRATEGY, PARTS SELECTION AND MANAGEMENT, AND SUPPORT OF A PROGNOSTICS AND HEALTH MANAGEMENT SYSTEM IN OPERATION ALSO ARISE FROM APPLYING PROGNOSTICS AND DIAGNOSTICS TECHNIQUES TO QUALIFICATION.
DEEP ADVERSARIAL APPROACHES IN RELIABILITY
(2020) Verstraete, David Benjamin; Modarres, Mohammad; Reliability Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Reliability engineering has long been proposed with the problem of predicting failures using all available data. As modeling techniques have become more sophisticated, so too have the data sources from which reliability engineers can draw conclusions. The Internet of Things (IoT) and cheap sensing technologies have ushered in a new expansive set of multi-dimensional big machinery data in which previous reliability engineering modeling techniques remain ill-equipped to handle. Therefore, the objective of this dissertation is to develop and advance reliability engineering research by proposing four comprehensive deep learning methodologies to handle these big machinery data sets. In this dissertation, a supervised fault diagnostic deep learning approach with applications to the rolling element bearings incorporating a deep convolutional neural network on time-frequency images was developed. A semi-supervised generative adversarial networks-based approach to fault diagnostics using the same time-frequency images was proposed. The time-frequency images were used again in the development of an unsupervised generative adversarial network-based methodology for fault diagnostics. Finally, to advance the studies of remaining useful life prediction, a mathematical formulation and subsequent methodology to combine variational autoencoders and generative adversarial networks within a state-space modeling framework to achieve both unsupervised and semi-supervised remaining useful life estimation was proposed. All four proposed contributions showed state of the art results for both fault diagnostics and remaining useful life estimation. While this research utilized publicly available rolling element bearings and turbofan engine data sets, this research is intended to be a comprehensive approach such that it can be applied to a data set of the engineer’s chosen field. This research highlights the potential for deep learning-based approaches within reliability engineering problems.
Damage Precursor Based Structural Health Monitoring and Prognostic Framework Using Dynamic Bayesian Network
(2016) Rabiei, Elaheh; Lopez Droguett, Enrique; Modarres, Mohammad; Reliability Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Structural health monitoring (SHM), as an essential tool to ensure the health integrity of aging structures, mostly focus on monitoring conventional observable damage markers such as fatigue crack size. However, degradation starts and progressively evolves at microstructural levels much earlier than detection of such indicators. This dissertation goes beyond classical approaches and presents a new SHM framework based on evolution of Damage Precursors, when conventional direct damage indicator, such as crack, is unobservable, inaccessible or difficult to measure. Damage precursor is defined in this research as “any detectable variation in material/ physical properties of the component that can be used to infer the evolution of the hidden/ inaccessible/ unmeasurable damage during the degradation”. Accordingly, the degradation process is to be expressed based on progression of damage precursor through time and the damage state assessment would be updated by incorporating multiple different evidences. Therefore, this research proposes a systematic integration approach through Dynamic Bayesian Network (DBN) to include all the evidences and their relationships. The implementation of augmented particle filtering as a stochastic inference method inside DBN enables estimating both model parameters and damage states simultaneously in light of various evidences. Incorporating different sources of information in DBN entails advance techniques to identify and formulate the possible interaction between potentially non-homogenous variables. This research uses the Support Vector Regression (SVR) in order to define generally unknown nonparametric and nonlinear correlation between some of the variables in the DBN structure. Additionally, the particle filtering algorithm is studied more fundamentally in this research and a modified approach called “fully adaptive particle filtering” is proposed with the idea of online updating not only the state process model but also the measurement model. This new approach improves the ability of SHM in real-time diagnostics and prognostics. The framework is successfully applied to damage estimation and prediction in two real-world case studies of 1) crack initiation in a metallic alloy under fatigue and, 2) damage estimation and prognostics in composite materials under fatigue. The proposed framework is intended to be general and comprehensive such that it can be implemented in different applications.
HEALTH ESTIMATION AND REMAINING USEFUL LIFE PREDICTION OF ELECTRONIC CIRCUIT WITH A PARAMETRIC FAULT
(2016) Sai Sarathi Vasan, Arvind; Pecht, Michael G; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Degradation of electronic components is typically accompanied by a deviation in their electrical parameters from their initial values. Such parametric drifts in turn will cause degradation in performance of the circuit they are part of, eventually leading to function failure due to parametric faults. The existing approaches for predicting failures resulting from electronic component parametric faults emphasize identifying monotonically deviating parameters and modeling their progression over time. However, in practical applications where the components are integrated into a complex electronic circuit assembly, product or system, it is generally not feasible to monitor component-level parameters. To address this problem, a prognostics method that exploits features extracted from responses of circuit-comprising components exhibiting parametric faults is developed in this dissertation. The developed prognostic method constitutes a circuit health estimation step followed by a degradation modeling and remaining useful life (RUL) prediction step. First, the circuit health estimation method was developed using a kernel-based machine learning technique that exploits features that are extracted from responses of circuit-comprising components exhibiting parametric faults, instead of the component-level parameters. The performance of kernel learning technique depends on the automatic adaptation of hyperparameters (i.e., regularization and kernel parameters) to the learning features. Thus, to achieve high accuracy in health estimation the developed method also includes an optimization method that employs a penalized likelihood function along with a stochastic filtering technique for automatic adaptation of hyperparameters. Second, the prediction of circuit’s RUL is realized by a model-based filtering method that relies on a first principles-based model and a stochastic filtering technique. The first principles-based model describes the degradation in circuit health with progression of parametric fault in a circuit component. The stochastic filtering technique on the other hand is used to first solve a joint ‘circuit health state—parametric fault’ estimation problem, followed by prediction problem in which the estimated ‘circuit health state—parametric fault’ is propagated forward in time to predict RUL. Evaluations of the data from simulation experiments on a benchmark Sallen–Key filter circuit and a DC–DC converter system demonstrate the ability of the developed prognostic method to estimate circuit health and predict RUL without having to monitor the individual component parameters.
An Analytical Model for Developing a Canary Device to Predict Solder Joint Fatigue Failure under Thermal Cycling Conditions
(2015) Mathew, Sony; Pecht, Michael G; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Solder joint fatigue failure is a prevalent failure mechanism for electronics subjected to thermal cycling loads. The failure is attributed to the thermo-mechanical stresses in the solder joints caused by differences in the coefficient of thermal expansion of the printed circuit board (PCB), electronic component, and solder. Physics of failure models incorporate the knowledge of a product's material properties, geometry, life-cycle loading and failure mechanisms to estimate the remaining useful life of the product. Engelmaier's model is widely used in the industry to estimate the fatigue life of electronics under thermal cycling conditions. However, for leadless electronic components, the Engelmaier strain metric does not consider the solder attachment area, the solder fillet thickness, and the thickness of the PCB. In this research a first principles model to estimate the strain in the solder interconnects has been developed. This new model considers the solder attachment area, and the geometry and material properties of the solder, component and PCB respectively. The developed model is further calibrated based on the results of finite element analysis. The calibrated model is validated by comparing its results with results of testing of test assemblies under different thermal cycling loading conditions. Further, the calibrated first principles model is used to design reduced solder attachment areas for electronic components so that under the same loading conditions they fail faster than components with regular solder attachment areas. Such structures are called expendable Canary devices and can be used to predict the solder joint fatigue failure of regular electronic components in the actual field conditions. The feasibility of using a leadless chip resistor with reduced solder attachment area as a canary device to predict the failure of ball grid array (BGA) component has been proven based on testing data. Further, a methodology for the developing and implementing canary device based prognostics has been developed in this research. Practical implementation issues, including estimating the number of canary devices required, determination of appropriate prognostic distance, and failure prediction schemes that may be used in the actual field conditions have also been addressed in this research.
PROGNOSTICS-BASED QUALIFICATION OF WHITE LIGHT-EMITTING DIODES (LEDS)
(2014) Chang, Moon-Hwan; Pecht, Michael G.; Ayyub, Bilal; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Light-emitting diode (LED) applications have expanded from display backlighting in computers and smart phones to more demanding applications including automotive headlights and street lightening. With these new applications, LED manufacturers must ensure that their products meet the performance requirements expected by end users, which in many cases require lifetimes of 10 years or more. The qualification tests traditionally conducted to assess such lifetimes are often as long as 6,000 hours, yet even this length of time does not guarantee that the lifetime requirements will be met. This research aims to reduce the qualification time by employing anomaly detection and prognostic methods utilizing optical, electrical, and thermal parameters of LEDs. The outcome of this research will be an in-situ monitoring approach that enables parameter sensing, data acquisition, and signal processing to identify the potential failure modes such as electrical, thermal, and optical degradation during the qualification test. To detect anomalies, a similarity-based-metric test has been developed to identify anomalies without utilizing historical libraries of healthy and unhealthy data. This similarity-based-metric test extracts features from the spectral power distributions using peak analysis, reduces the dimensionality of the features by using principal component analysis, and partitions the data set of principal components into groups using a KNN-kernel density-based clustering technique. A detection algorithm then evaluates the distances from the centroid of each cluster to each test point and detects anomalies when the distance is greater than the threshold. From this analysis, dominant degradation processes associated with the LED die and phosphors in the LED package can be identified. When implemented, the results of this research will enable a short qualification time. Prognostics of LEDs are developed with spectral power distribution (SPD) prediction for color failure. SPD is deconvoluted with die SPD and phosphor SPD with asymmetric double sigmoidal functions. Future SPD is predicted by using the particle filter algorithm to estimate the propagating parameters of the asymmetric double sigmoidal functions. Diagnostics is enabled by SPD prediction to indicate die degradation, phosphor degradation, or package degradation based on the nature of degradation shape of SPD. SPDs are converted to light output and 1976 CIE color coordinates using colorimetric conversion with color matching functions. Remaining useful life (RUL) is predicted using 7-step SDCM (standard deviation of color matching) threshold (i.e., 0.007 color distance in the CIE 1676 chromaticity coordinates). To conduct prognostics utilizing historical libraries of healthy and unhealthy data from other devices, this research employs similarity-based statistical measures for a prognostics-based qualification method using optical, electrical, and thermal covariates as health indices. Prognostics is conducted using the similarity-based statistical measure with relevance vector machine regression to capture degradation trends. Historical training data is used to extract features and define failure thresholds. Based on the relevance vector machine regression results, which construct the background health knowledge from historical training units, the similarity weight is used to measure the similarity between each training unit and test unit under the test. The weighted sum is then used to estimate the remaining useful life of the test unit.
FAULT DETECTION AND PROGNOSTICS OF INSULATED GATE BIPOLAR TRANSISTOR (IGBT) USING A K-NEAREST NEIGHBOR CLASSIFICATION ALGORITHM
(2013) Sutrisno, Edwin; Pecht, Michael; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Insulated Gate Bipolar Transistor (IGBT) is a power semiconductor device commonly used in medium to high power applications from household appliances, automotive, and renewable energy. Health assessment of IGBT under field use is of interest due to costly system downtime that may be associated with IGBT failures. Conventional reliability approaches were shown by experimental data to suffer from large uncertainties when predicting IGBT lifetimes, partly due to their inability to adapt to varying loading conditions and part-to-part differences. This study developed a data-driven prognostic method to individually assess IGBT health based on operating data obtained from run-to-failure experiments. IGBT health was classified into healthy and faulty using a K-Nearest Neighbor Centroid Distance classification algorithm. A feature weight optimization method was developed to determine the influence of each feature toward classifying IGBT's health states.
AN OPTIONS APPROACH TO QUANTIFY THE VALUE OF DECISIONS AFTER PROGNOSTIC INDICATION
(2011) Haddad, Gilbert; Sandborn, Peter A; Pecht, Michael G; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Safety, mission and infrastructure critical systems have started adopting prognostics and health management, a discipline consisting of technologies and methods to assess the reliability of a product in its actual life-cycle conditions to determine the advent of failure and mitigate system risks. The output from a prognostic system is the remaining useful life of the host system; it gives the decision-maker lead-time and flexibility in maintenance. Examples of flexibility include delaying maintenance actions to use up the remaining useful life and halting the operation of the system to avoid critical failure. Quantifying the value of flexibility enables decision support at the system level, and provides a solution to the fundamental tradeoff in maintenance of systems with prognostics: minimize the remaining useful life thrown while concurrently minimizing the risk of failure. While there are cost-benefit models to quantify the value of implementing prognostics, they are applicable to the fleet level, they do not incorporate the value of decisions after prognostic indication (value of flexibility or contingency actions), and do not use PHM information for dynamic maintenance scheduling. This dissertation develops a decision support model based on `options' theory- a financial derivative tool extended to real assets - to quantify maintenance decisions after a remaining useful life prediction. A hybrid methodology based on Monte Carlo simulations and decision trees is developed. The methodology incorporates the value of contingency actions when assessing the benefits of PHM. The model is extended and combined with least squares Monte Carlo methods to quantify the option to wait to perform maintenance; it represents the value obtained from PHM at the system level. The methodology also allows quantifying the benefits of PHM for individualized maintenance policies for systems in real-time, and to set a dynamic maintenance threshold based on PHM information. This work is the first known to quantify the flexibility enabled by PHM and to address the cost-benefit-risk ramifications after prognostic indication at the system level. The contributions of the dissertation are demonstrated on data for wind farms.
Prognostics of Insulated Gate Bipolar Transistors
(2011) Patil, Nishad Kashinath; Pecht, Michael; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Insulated gate bipolar transistors (IGBTs) are the devices of choice for medium and high power, low frequency applications. IGBTs have been reported to fail under excessive electrical and thermal stresses in variable speed drives and are considered as reliability problems in wind turbines, inverters in hybrid electric vehicles and railway traction motors. There is a need to develop methods to detect anomalous behavior and predict the remaining useful life (RUL) of IGBTs to prevent system downtime and costly failures. In this study, a framework for prognostics of IGBTs was developed to provide early warnings of failure and predict the remaining useful life. The prognostic framework was implemented on non punch through (NPT) IGBTs. Power cycling of IGBTs was performed and the gate-emitter voltage, collector-emitter voltage, collector-emitter current and case temperature was monitored in-situ during aging. The on-state collector-emitter current (ICE(ON)) and collector-emitter voltage (VCE(ON)) were identified as precursors to IGBT failure. Electrical characterization and X-ray analysis was performed before and after aging to map degradation in the devices to observed trends in the precursor parameters. A Mahalanobis distance based approach was used for anomaly detection. The initial ICE(ON) and VCE(ON) parameters were used to compute the healthy MD distance. This healthy MD distance was transformed and the mean and standard deviation of the transformed MD data was obtained. The μ+3σ upper bound obtained from the transformed healthy MD was then used as a threshold for anomaly detection. This approach was able to detect anomalous behavior in IGBTs before failure. Upon anomaly detection, a particle filter approach was used for predicting the remaining useful life of the IGBTs. A system model was developed using the degradation trend of the VCE(ON) parameter. This model was obtained by a least squares regression of the IGBT degradation curve. The tracking and prediction performance of the model with the particle filter was demonstrated.
Detection of Interconnect Failure Precursors using RF Impedance Analysis
(2010) Kwon, Daeil; Pecht, Michael G; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Many failures in electronics result from the loss of electrical continuity of common board-level interconnects such as solder joints. Measurement methods based on DC resistance such as event detectors and data-loggers have long been used by the electronics industry to monitor the reliability of interconnects during reliability testing. DC resistance is well-suited for characterizing electrical continuity, such as identifying an open circuit, but it is not useful for detecting a partially degraded interconnect. Degradation of interconnects, such as cracking of solder joints due to fatigue or shock loading, usually initiates at an exterior surface and propagates towards the interior. A partially degraded interconnect can cause the RF impedance to increase due to the skin effect, a phenomenon wherein signal propagation at frequencies above several hundred MHz is concentrated at the surface of a conductor. Therefore, RF impedance exhibits greater sensitivity compared to DC resistance in detecting early stages of interconnect degradation and provides a means to prevent and predict an important cause of electronics failures. This research identifies the applicability of RF impedance as a means of a failure precursor that allows for prognostics on interconnect degradation based on electrical measurement. It also compares the ability of RF impedance with that of DC resistance to detect early stages of interconnect degradation, and to predict the remaining life of an interconnect. To this end, RF impedance and DC resistance of a test circuit were simultaneously monitored during interconnect stress testing. The test vehicle included an impedance-controlled circuit board on which a surface mount component was soldered using two solder joints at the end terminations. During stress testing, the RF impedance exhibited a gradual non-linear increase in response to the early stages of solder joint cracking while the DC resistance remained constant. The gradual increase in RF impedance was trended using prognostic algorithms in order to predict the time to failure of solder joints. This prognostic approach successfully predicted solder joint remaining life with a prediction error of less than 3%. Furthermore, it was demonstrated both theoretically and experimentally that the RF impedance analysis was able to distinguish between two competing interconnect failure mechanisms: solder joint cracking and pad cratering. These results indicate that RF impedance provides reliable interconnect failure precursors that can be used to predict interconnect failures. Since the performance of high speed devices is adversely affected by early stages of interconnect degradation, RF impedance analysis has the potential to provide improved reliability assessment for these devices, as well as accurate failure prediction for current and future electronics.

Mechanical Engineering Theses and Dissertations

Browse

Filters

Settings

Sort By

Results per page

Search Results