No description available.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Mechanical Engineering, 1980 / Vita. / Includes bibliographical references. / by Michael Benjamin Packer. / Ph. D. / Ph. D. Massachusetts Institute of Technology, Department of Mechanical Engineering

Catalogs of seismic source parameters (hypocenter locations, origin times, and magnitudes) are vital for studying various Earth processes, greatly enhancing our understanding of the nature of seismic events, the structure of the Earth, and the dynamics of fault systems. Modern seismic analyses utilize supervised machine learning (ML) to build enhanced catalogs based on millions of examples of analyst-picked phase-arrivals in waveforms, yet the ability to characterize the time-varying spectral content of the waveforms underlying those catalogs remains lacking. Unsupervised machine learning (UML) methods provide powerful tools for inferring patterns from musical spectrograms with little a priori information, yet has been relatively underutilized in the field of seismology. In this thesis, I leverage advanced tools from UML to analyze the temporal spectral content of large sets of spectrograms generated by different mechanisms in two distinct geologic settings: icequakes and tremors at Gornergletscher (a Swiss temperate glacier) and repeating earthquakes from a 10-km-long creeping segment of the San Andreas Fault. The core algorithm in this work, now known as Spectral Unsupervised Feature Extraction, or SpecUFEx, extracts time-varying frequency patterns from spectrograms and reduces them into low-dimensionality fingerprints via a combination of non-negative matrix factorization and hidden Markov Modeling (Holtzman et al. 2018), optimized for large data sets via stochastic variational inference. This work describes the SpecUFEx algorithm and the suite of preprocessing, clustering, and visualization tools developed to create an UML workflow, SpecUFEx+, that is widely-accessible and applicable for many seismic settings. I apply theSpecUFEx+ workflow to single- and multi-station seismic data from Gornergletscher, and demonstrate how some fingerprint-clusters track diurnal tremor related to subglacial water flow, while others correspond to the onset of the subglacial and englacial components of a glacial lake outburst flood. I also discover periods of harmonic tremor localized near the ice-bed interface that may be related to glacial stick-slip sliding. I additionally apply the SpecUFEx+ workflow to earthquakes on the San Andreas Fault to unveil far more repeating earthquake sequences than previously inferred, leading to enhanced slip-rate estimates at seismogenic depths and providing a more detailed image of seismic gaps along the fault interface. Unsupervised feature extraction is a novel tool to the field of seismology. This work demonstrates how scientific insight can be gained through the characterization of the spectral-temporal patterns of large seismic datasets within an UML-framework.

With the continuing spread of misinformation and disinformation online, it is of increasing importance to develop combating mechanisms at scale in the form of automated systems that can find checkworthy information, detect fallacious argumentation of online content, retrieve relevant evidence from authoritative sources and analyze the veracity of claims given the retrieved evidence. The robustness and applicability of these systems depend on the availability of annotated resources to train machine learning models in a supervised fashion, as well as machine learning models that capture patterns beyond domain-specific lexical clues or genre-specific stylistic insights. In this thesis, we investigate the role of models for argument structure and argument quality in improving tasks relevant to fact-checking and furthering our understanding of misinformation and disinformation. We contribute to argumentation mining, misinformation detection, and fact-checking by releasing multiple annotated datasets, developing unified models across datasets and task formulations, and analyzing the vulnerabilities of such models in adversarial settings. We start by studying the argument structure's role in two downstream tasks related to fact-checking. As it is essential to differentiate factual knowledge from opinionated text, we develop a model for detecting the type of news articles (factual or opinionated) using highly transferable argumentation-based features. We also show the potential of argumentation features to predict the checkworthiness of information in news articles and provide the first multi-layer annotated corpus for argumentation and fact-checking. We then study qualitative aspects of arguments through models for fallacy recognition. To understand the reasoning behind checkworthiness and the relation of argumentative fallacies to fake content, we develop an annotation scheme of fallacies in fact-checked content and investigate avenues for automating the detection of such fallacies considering single- and multi-dataset training. Using instruction-based prompting, we introduce a unified model for recognizing twenty-eight fallacies across five fallacy datasets. We also use this model to explain the checkworthiness of statements in two domains. Next, we show our models for end-to-end fact-checking of statements that include finding the relevant evidence document and sentence from a collection of documents and then predicting the veracity of the given statements using the retrieved evidence. We also analyze the robustness of end-to-end fact extraction and verification by generating adversarial statements and addressing areas for improvements for models under adversarial attacks. Finally, we show that evidence-based verification is essential for fine-grained claim verification by modeling the human-provided justifications with the gold veracity labels.

Heavy vehicle weights must be closely monitored to prevent fatigue-induced deterioration and critical fracture to civil infrastructure, among many other purposes. Developing a cost-effective weigh-in-motion (WIM) system remains challenging. This doctoral research describes the creation and experimental validations of a computer vision-based non-contact vehicle WIM system. The underlining physics is that the force exerted by each tire onto the roadway is the product of the two key vehicle parameters: tire-roadway contact pressure and contact area. Computer vision is applied (1) to measure the several tire parameters so that the tire-roadway contact area can be accurately estimated; and (2) to recognize the marking texts on the tire sidewall so that the manufacturer-recommended tire pneumatic pressure can be found. Consequently, a computer vision system is developed in this research. The computer vision system comprises a camera and computer vision software/techniques for measuring the tire parameters and recognizing the tire sidewall markings from individual tire images of a moving vehicle. Computer vision techniques, such as edge detection and optical character recognition (OCR), are applied to enhance the measurements and recognition accuracy. Numerous laboratory and field experiments were conducted on a sport utility vehicle and fully loaded or empty concrete trucks to demonstrate the feasibility of this novel method. The vehicle weights estimated by this novel computer vision-based non-contact vehicle WIM system agreed well with the curb weights verified by static weighing, demonstrating the potential of this computer vision-based method as a non-contact means for weighing vehicles in motion. To further illustrate and exemplify the versatility of this novel computer vision-based WIM system, this research explores the potential application capability of the system for structural health monitoring (SHM) in civil engineering. This work aims to investigate the potential of this proposed and prototyped computer vision-based vehicle WIM system to acquire vehicle weight and location information as well as to obtain corresponding bridge responses simultaneously for later structural model updating analysis and damage detection and identification framework. In order to validate the concept, a laboratory vehicle-bridge model was constructed. Subsequently, numerous experiments were carried out to demonstrate how the computer vision-based WIM system can be utilized as a resourceful application to (1) extract bridge responses, (2) estimate vehicle weight, and (3) localize the input force simultaneously. This doctoral research delivers a unique, pioneering, and innovative design and development of a computer vision-based non-contact vehicle WIM method and system that can remotely perform vehicle weight estimation. It also demonstrates a novel application of computer vision technology to solve challenging weigh-in-motion (WIM) and civil engineering problems.

Organic waste-fueled carbonaceous resource recovery using approaches such as arrested anaerobic digestion generates economically attractive products such as volatile fatty acid (VFA). The production of VFA expands the applications of anaerobic biotechnologies beyond the traditionally produced biogas. Compared to biogas, VFA is produced and recovered in a concentrated form in the aqueous phase, which is more conducive to direct utilization in downstream bioplastic, biodiesel production, and nitrogen/phosphorus removal in water resource-recovery facilities. However, this application is limited by the variability in VFA yield and composition as obtained from different complex solids streams. Additionally, the lack of understanding of the nexus between the performance-structure-function of the microbial community within the arrested anaerobic digestion process leads to the massive gap between the optimized engineering regulations and the high-throughput VFA production. Consequently, this dissertation aimed to unlock the potential of VFA production with maximized yield and regulated composition through the manipulation of the operational parameter (hydraulic retention time (HRT)) and the feedstock condition (thermal hydrolysis pretreatment (THP)). In response, meta-omics-derived approaches were applied to elucidate the dynamic changes of microbial structure, potential, and extant functionality in terms of the two processes (hydrolysis and acidification) within arrested anaerobic digestion of food waste. Specifically, the objectives were (1) Performance: Evaluate the hydrolysis and acidification performance changes including hydrolysis yield, VFA yield, VFA composition, methane yield, etc. under different HRTs and feeding THP or non-THP food waste. (2) Microbial structure: characterize and compare the significance of HRT and feedstock condition in shaping microbial structures. (3) Functional analysis: Interpret the community-level dynamic changes of potential and extant functions within the (3.1) customized acidification metabolic networks and the (3.2) carbohydrate hydrolysis niches. The highlighted findings are as follows: (1) Performance of the arrested anaerobic digestion (including hydrolysis and acidification processes): Neither the hydrolysis yield nor the VFA yield was improved by the extended HRT from 4 to 8 days (P > .05). The inclusion of THP on feedstock didn’t improve the hydrolysis yield (P > .05) while the VFA yield was significantly decreased (P = .003). Among all conditions, the methane production was less than 5% of the chemical oxygen demand (COD) and a propionic acid-dominant type product was robustly formed. (2) Microbial structures in the arrested anaerobic digestors (including core hydrolyzers and acidification microbial communities): Both HRT and the inclusion of THP on feedstock shaped distinct microbial structures in the arrested anaerobic digestors (P = .02 and .01). Although the extension of HRT didn’t change the Shannon diversity Index (P > .05), it was significantly decreased after feeding with THP food waste (P = .03), which might stem from the reduced indigenous microbes in the initial food waste feedstock. Prevotella was always the most abundant genus under all conditions, which might contribute to the dominantly produced propionic acid among all conditions. The successfully suppressed growth of methanogenic archaea was reflected in terms of the low relative abundance (<1.5%) among all conditions. (3.1) Functional analysis of the customized acidification metabolic networks: Under the two selected HRTs, the potential and extant functions of acidification were unchanged between the two reactors (P > .05), which indicated a community-level redundancy in convergent potential and extant acidification functions even under a completely shifted microbial structure. However, the inclusion of THP diminished the potential and extant functions of acidification, in the meantime, shifting the main producer of butyric acid from Bacteroides to Prevotella through the expression of gene buk2. Among all conditions, the highest potential and extant functions in propionic acid production corresponded to the propionic acid-dominant acid profile in all reactors. The prevalently enriched Prevotella contributed to the stable propionic acid-dominant production via the acryloyl-CoA to propionyl phosphate to the propionic acid pathway. (3.2) Functional analysis of the carbohydrate hydrolysis niches: The extension of HRT from 4 days to 8 days didn’t impact the potential and extant functions of carbohydrate-activated enzymes (CAZys) and the hydrolysis of polysaccharides. Only two intermediate steps (gene malQ and lplD) during the hydrolysis of starch and pectin were enhanced with higher absolute transcriptional activities (mRNA/DNA RPKM) under HRT 8 days. The abundance ratio of the two main hydrolysis phyla Firmicutes: Bacteroidetes was unchanged between the two HRTs. When feeding with THP feedstock, the potential and extant functions of CAZys were both enhanced. All steps within the hydrolysis of cellulose (polysaccharides) exhibited increased absolute transcriptional activities (mRNA/DNA RPKM). The abundance ratio of Firmicutes: Bacteroidetes was decreased after the inclusion of THP on feedstock, which corresponded to the increased hydrolysis of polysaccharides- cellulose. Although the carbohydrate hydrolysis functions were improved after feeding with THP food waste, the total hydrolysis yield was not enhanced. The hydrolysis of other compounds such as proteins and lipids could also contribute to the total hydrolysis yield. The taxonomic analysis revealed that in all four conditions, the genus Prevotella presented with the highest potential functions in CAZys, while the genus Pararhodospirillum exhibited the highest extant functions in CAZys. This indicated that distinct bacteria were endowed with different functional potentials of CAZys and mobilized these functions differently. Overall, this research provides practical suggestions for engineering designs to maximize the VFA production profits from arrested anaerobic digestion of food waste: (1) A properly controlled HRT enables a long-term high-throughput production of VFA with stable yield and the unchanged dominant acid type (2) The inclusion of THP to the feedstock was not suggested to be applied to maximize the VFA yield even the dominant acid type may not change. (3) The dominantly produced propionic acid could be targeted by enriching the Prevotella genus to produce the propionic acid through the acryloyl-CoA to propionyl phosphate to the propionic acid pathway. Besides the engineering aspect, this research also specifically elucidates the long-time lumped and simplified acidification and carbohydrate hydrolysis processes with the extended metabolic databases including each reaction, key intermediates, enzymes, and corresponding genes. This expanded database served as an essential upstream process, which could be integrated into the current anaerobic digestion model. Additional applications could be extended to the human digestion systems' microbiome and be exploited commercially for other mixed-culture biosynthesis processes such as bioplastic and biodiesel production. Finally, the application of meta-omics-derived methodology revealed the functional redundancy and the potential discrepancy between the most abundant group and the most actively functional group underlying the formed black box of VFA production performance. This discussion of the nexus of performance-structure-function suggested the importance of applying meta-omics approaches in engineering practice, especially when feeding the mixed-culture community with real complex solid streams. The targeted VFA profiles cannot be reached without identifying the actual functional bacteria under selected engineering conditions.

To fully optimize the value of material produced from a log requires information about the log's internal defects prior to log breakdown. Studies have shown that a 7 to 21 percent improvement in log value recovery can be achieved if the location and identity of internal defects are known. Recent developments in advanced nondestructive testing methods such as CT and MRI offer, for the first time, the possibility of finding internal defects in logs prior to breakdown. Our ability to detect and recognize defects using this data depends Critically on our understanding of wood structure and our ability to devise reliable method for automated image interpretation. While a lot of work has gone into demonstrating that certain types of defects manifest themselves in such sensor imagery, there has not been a systematic approach toward making the automatic inspection of logs a practical reality. This dissertation describes work aimed at creating a viable automated technology for locating and identifying log defects. The imaging modality used in this dissertation is CT. An important first step is to establish a data base of imagery and the ground truth information to determine how the various defects manifest themselves in this imagery. The second step is to study defect characterization and determine exactly which defects are detectable. The final step is to develop a basic method of approach to automated image analysis. A data base has been created from two hardwood species. It is representative of hardwood logs in the sense that it contains almost all the major defects. Visual inspection and analysis of these CT images have shown that most defects manifest themselves in CT imagery. These defects can be detected by features such as intensity, 3-d shape, and texture. As a means of automated image analysis, a knowledge-based vision system has been developed. It consists of three components: a data acquisition unit, an image segmentation module, and scene analysis module. A 3-d adaptive LS filter has been developed in the segmentation module that is efficient in removing annual rings while preserving other needed high frequency detail. Images are segmented using a multiple threshold scheme and regions are grouped using a 3-d connected volume growing algorithm. To represent the 3-d nature of wood defects, a set of basic features have been defined and used to design a set of hypothesis tests. These features seem to be adequate for defect recognition. To cope with imprecision and ambiguity the Dempster-Shaffer model for knowledge representation is used in the vision system. As a viable alternative to Bayesian-based theory, the Dempster's method of evidential reasoning is employed that uses previously unavailable information such as the amount of ignorance and ambiguity a hypothesis exhibits. As such, the proposed vision system seems to be able to recognize a number of hardwood defects. This dissertation also explores wood texture as an additional feature in defect recognition, and contributes the first application of robust Spatial AutoRegressive modeling to wood texture analysis. Based on a correlation measure, two simple but efficient texture discrimination schemes are proposed. Incorporating a texture test in the scene analysis should improve the vision system's recognition power. As a pilot research, this dissertation has explored a number of important issues in creating a vision system for automated log inspection. Clearly, more work is needed to make the system more robust with additional species. Nevertheless, preliminary results seem to indicate that a machine vision system for automated hardwood log inspection can be developed. / Ph. D.

Wood pulp mill effluents continue to attract much attention due to environmental consequences. However, in comparison, very little work has been published on wood pulp extracts themselves. In this investigation, chemithermomechanical (CTMP) pulps as well as Kraft (BKP) pulps were Soxhlet extracted with solvents of different polarity. These two types of pulp extracts were then compared qualitatively using GC-FID and GC-MSD as well as quantitatively based on the percent of extractives obtained. For all the pulps studied, the percent extractives of water > ethyl acetate > cyclohexane. The CTMP extracts exhibited many more components as compared to BKP extracts for all the extractions solvents. The presence of trace chlorinated phenolics in the above wood pulp extracts was also addressed utilizing GC-ECD, GC-EIMS and GC-NCIMS. 4-MCG, 4,5-DCG, 4,5,6-TCG, 3,4,5-TCG, 2,4,6- TCP, 2,3,4,6-TeCP, PCP and 6-MCVN were discovered. Due to a lack of knowledge of the complete history of the wood pulps studied, the exact causes for their discoveries are unknown. Attempts were also made to study the feasibility of Supercritical Fluid Extraction of the above mentioned wood pulps due to the difficulties faced with Soxhlet extractions. The percent extractives obtained using SF-CO₂ and cyclohexane were found to be comparable. / Ph. D.

A presentation is given of an acousto-ultrasonic system for the evaluation of composite materials. First, a brief statement will be made on the acousto-ultrasonic technique and its relative worth compared to other nondestructive testing techniques as applied to composite materials. The following two chapters describe the system instrumentation and system software, respectively. Next, comments are given regarding the implementation of the system for research on graphite/epoxy laminates, with additional remarks concerning efforts to evaluate aluminum/graphite tubes with the system. This includes physical descriptions of the composite systems. Subsequently, results are presented comparing parameters and forms of presentation which can be employed to relate results. Finally, conclusions are made on the application of the acousto-ultrasonic system to nondestructive testing of composite materials, with specific results on its application to graphite/epoxy plates. More specifically, comments are made on the variation of SWF factors with azimuthal angle on the graphite/epoxy plates, the identification of specific frequency peaks, and the relationships these may have to certain modes of vibration and material properties. For example, a low frequency mode was found to vary in a manner reminiscent of E_x and to show characteristics of an extensional Lamb wave. In general, results are presented and discussed in order to show how the system can be implemented to gain physical information on composite materials, such as the property of anisotropy. / Master of Science

This investigation was conducted to provide laboratory scale equipment that will facilitate an accurate study of the affect of ultrasonic energy on mass transfer in two-phase multiple-component liquid systems. A liquid-liquid extractor incorporating an ultrasonic generator and transducer was designed and constructed to the following specifications: (1) insonation frequency of 400 kilocycles per second, (2) insonation intensities equivalent to plate currents of 0 to 200 milliamperes, and (3) flow rate of solvent and feed through the reactor ranging approximately from one-half to 24 pounds per minute, in varying solvent-to-feed ratios. A special glass reactor, or contactor, was constructed from a standard, 60° pyrexx glass funnel and fitted with an acoustical window of 0.001-inch sheet nickel. Photographic studies were made of the two-phase mixing taking place inside the reactor in both the presence and absence of ultrasonic insonation. An all metal reactor, of the same general design as the glass reactor, was constructed for use with the extractor when investigations were to be made that would involve high pressures or sudden liquid surges through the reactor. An evaluation of the extractor was conducted employing the system moetons-water-1,1,2-trichloroethane. Stage efficiencies calculated for the individual tests, eleven in all, varied from 94.3 to 110.0 percent. An observed yellow color in the extract samples, probably due to dissolved impurities in the 1,1,2-trichloroethane, could have been responsible for the observation of refractive index readings that did not give true representation of the acetone concentration of the sample. Cavitation was observed in the reactor which the test system was undergoing ultrasonic insonation. The gross stirring effects resulting from cavitation in the liquids caused a mixing of the two phases that was more intense than that taking place in the reactor without insonation. / Master of Science