Logistic regression is a widely used statistical method in data analysis and machine learning. When the capacity of data is large, it is time-consuming and even infeasible to perform big data machine learning using the traditional approach. Therefore, it is crucial to come up with an efficient way to evaluate feature combinations and update learning models. With the approach proposed by Yang, Wang, Xu, and Zhang (2018) a system can be represented using small enough matrices, which can be hosted in memory. These working sufficient statistics matrices can be applied in updating models in logistic regression. This study applies the working sufficient statistics approach in logistic regression machine learning to examine how this new method improves the performance. This study investigated the difference between the performance of this new working sufficient statistics approach and performance of the traditional approach on Spark\rq s machine learning package. The experiments showed that the working sufficient statistics method could improve the performance of training the logistic regression models when the input size was large.

This work was on the effect of electric current on the processing of ceramics. The focus was on electromigration/electrochemistry and plasma effects. While there is no solid evidence that there is plasma in Spark Plasma Sintering, (SPS), newer techniques e.g. flash, use different conditions so there is an interest in understanding the conditions under which a plasma forms. The minimum arcing voltage was found from literature to be from 10-15V for materials of interest. This is above that found in SPS (10V). However, due to the many contact points in a powder compact much higher voltages (50V) were required in practical experiments. Optical spectroscopy was used to verify the formation of a plasma, and emission peaks from the powder compact material were visible implying they were vaporised and formed the plasma. Electromigration was exploited to alter the oxidation of zirconium diboride, by passing current through the oxide layer (120μm zirconia base grown at 1200°C) oxygen could be pumped either away or toward the diboride bulk. Small cubes (3mm) of diboride had platinum foil electrodes applied on both sides and oxidation was performed at 1400°C for 5hr. Without a field the oxide grew to 360μm, by applying 10V and 100mA the oxide grew to 150μm under the +ve electrode but 1400μm under the -ve electrode. Electrochemical reduction was believed to have occurred due to the electrical properties of the material changing during oxidation and visible blackening of the oxide. Combining the techniques from both earlier works, a contactless flash sintering setup was developed. This used two plasma arcs as electrodes to heat and pass current through the sample. Various materials, currents and times were used, but the best result was with SiC:B4C which was sintered in 3s with 6A, the microstructure showed sharp grains, no segregation and limited grain growth ( initially 0.7μm SiC and 0.5μm B4C, this grew to 1.1μm and 1.4μm). This was the first recorded case of contactless flash sintering and the technique has the potential to sinter ceramics in a continuous manner.

Monolithic composites are needed that combine low friction and wear, high mechanical hardness, and high fracture toughness. Thin films and coatings are often unable to meet this engineering challenge as they can delaminate and fracture during operation ceasing to provide beneficial properties during service life. Two material systems were synthesized by spark plasma sintering (SPS) and were studied for their ability to meet these criteria. A dual hybrid composite was fabricated and consisted of a nickel matrix for fracture toughness, TiC for hardness and graphite for solid/self‐lubrication. An in‐situ reaction during processing resulted in the formation of TiC from elemental Ti and C powders. The composition was varied to determine its effects on tribological behavior. Stellite 21, a cobalt‐chrome‐molybdenum alloy, was also produced by SPS. Stellite 21 has low stacking fault energy and a hexagonal phase which forms during sliding that both contribute to low interfacial shear and friction. Samples were investigated by x‐ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive x‐ray spectroscopy (EDS), and electron back‐scattered diffraction (EBSD). Tribological properties were characterized by pin on disc tribometry and wear rates were determined by profilometry and abrasion testing. Solid/self‐lubrication in the TiC/C/Ni system was investigated by Raman and Auger mapping. A tribofilm, which undergoes a stress‐induced phase transformation from polycrystalline graphite to amorphous carbon, was formed during sliding in the TiC/C/Ni system that is responsible for low friction and wear. TiC additions help to further decrease wear. Stellite 21 was also found to exhibit acceptably low friction and wear properties arising from the presence of Cr23C6 in the matrix and work hardening of the cobalt and chromium during sliding.

Typescript (photocopy). / Digitized by Kansas Correctional Industries

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 1982. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Includes bibliographical references. / by Stephen Gregory Poulos. / M.S.

Traditional lightweight armor ceramics such as boron carbide (B4C) and silicon carbide (SiC) are used alone or together in varying amounts to create monolithic protective plates. These materials exhibit relatively small differences in hardness, flexure strength, and fracture toughness. Many of the routes taken during the synthesis of the powder and sintering of the plates using traditional ceramic processing techniques have long processing times, tend to leave asperities within the microstructure, and have unwanted secondary phases that lower the performance of these materials. In lieu of the incremental changes in the above properties, it is thought that adding diamond particulates to the ceramic matrix will dramatically improve the mechanical properties and overall performance. With the reduced cost of synthetic diamond and the commercial development of more rapid spark plasma sintering (SPS), this work develops a novel reactive SPS process to fabricate near fully dense SiC-TiC-diamond composites at various processing temperatures with minimal graphitization and full adhesion to the ceramic matrix. It was found that samples with up to ~97% theoretical density can be fabricated with no quantifiable graphite content within the characterization ability using advanced X-ray diffraction and microscopy techniques.

In this thesis Direct Numerical Simulation (DNS) is used to investigate the development of turbulent premixed flame kernels during the early stages of growth typical of the period following spark ignition. Two distinct aspects of this phase are considered: the interaction of the expanding kernel with a field of decaying turbulence, and the chemical and thermo-diffusive response of the flame for different fresh-gas compositions. In the first part of the study, three-dimensional, repeated simulations with single-step chemistry are used to generate ensemble statistics of global flame growth. The surface-conditioned mean fluid-velocity magnitude is found to vary significantly across different isosurfaces of the reaction progress variable, and this is shown to lead to a bias in the distribution of the Surface Density Function (SDF) around the developing flame. Two-dimensional simulations in an extended domain indicate that this effect translates into a similar directional bias in the Flame Surface Density (FSD) at later stages in the kernel development. Properties of the fresh gas turbulence decay are assessed from an independent, non-reacting simulation database. In the second part of this study, two-dimensional simulations with a detailed 68-step reaction mechanism are used to investigate the thermo-diffusive response of pure methane-air, and hydrogen-enriched methane-air flames. The changes in local and global behaviour due to the different laminar flame characteristics, and the response of the flames to strain and curvature are examined at different equivalence ratios and turbulence intensities. Mechanisms leading to flame quenching are discussed and the effect of mean flame curvature is assessed through comparison with an equivalent planar flame. The effects of hydrogen addition are found to be particularly pronounced in flame kernels due to the higher positive stretch rates and reduced thermo-diffusive stability of hydrogen-enriched flames.

This thesis includes three chapters on electricity and natural gas prices. In the first chapter, we give a brief introduction to the characteristics of power prices and propose a mean reversion jump diffusion model, in which jump intensity depends on temperature data and overall system load, to model electricity prices. Compared to the models used in the literature, we find the model proposed in this chapter is better to capture the tail behavior in the electricity prices. In the second chapter, we use the model proposed in the first chapter to simulate the spark spread option and value the power generations. In order to simulate power generation, we first propose and estimate mean reversion jump diffusion model for natural gas prices, in which jump intensity is defined as a function of temperature and storage. Combing the model with the electricity models in chapter 1, we find that the value of power generation is closer to the real value of the power plants as reflected in the recent market transaction than one obtains from many other models used in literature. The third chapter investigates extremal dependence among the energy market. We find a tail dependence that exceeds the Pearson correlation ρ, which means the traditional Pearson correlation is not appropriate to model tail behavior of oil, natural gas and electricity prices. However, asymptotic dependence is rejected in all pairs except Henry Hub gas return and Houston Ship Channel gas return. We also find that extreme value dependence in energy market is stronger in bull market than that in bear market due to the special characteristics in energy market, which conflicts the accepted wisdom in equity market that tail correlation is much higher in periods of volatile markets from previous literature.

Biogas as a fuel is attractive from a greenhouse standpoint, since biogas is carbon neutral. To be used as such, increasingly stringent emission standards must be met. Current low-emission technologies meet said standards by precisely controlling the air-fuel ratio. Biogas composition can vary substantially, making air-fuel ratio control difficult. This research was conducted as part of a larger project to develop a sensor that accurately measures biogas composition. Biogas was simulated by fuel mixtures consisting of natural gas and CO2; the effects that fuel composition has on emissions and catalyst performance were investigated. Engine-out THC and NOx increased and decreased, respectively, with increasing CO2 in the fuel mixture. Doubling the catalyst residence time doubled the conversion of THC and CO emissions. The effectiveness of the catalyst at converting THC emissions was found to be dependent on the relative proportions of engine-out THC, NOx and CO emissions.

Biogas as a fuel is attractive from a greenhouse standpoint, since biogas is carbon neutral. To be used as such, increasingly stringent emission standards must be met. Current low-emission technologies meet said standards by precisely controlling the air-fuel ratio. Biogas composition can vary substantially, making air-fuel ratio control difficult. This research was conducted as part of a larger project to develop a sensor that accurately measures biogas composition. Biogas was simulated by fuel mixtures consisting of natural gas and CO2; the effects that fuel composition has on emissions and catalyst performance were investigated. Engine-out THC and NOx increased and decreased, respectively, with increasing CO2 in the fuel mixture. Doubling the catalyst residence time doubled the conversion of THC and CO emissions. The effectiveness of the catalyst at converting THC emissions was found to be dependent on the relative proportions of engine-out THC, NOx and CO emissions.