ASU Electronic Theses and Dissertations
This collection includes most of the ASU Theses and Dissertations from 2011 to present. ASU Theses and Dissertations are available in downloadable PDF format; however, a small percentage of items are under embargo. Information about the dissertations/theses includes degree information, committee members, an abstract, supporting data or media.
In addition to the electronic theses found in the ASU Digital Repository, ASU Theses and Dissertations can be found in the ASU Library Catalog.
Dissertations and Theses granted by Arizona State University are archived and made available through a joint effort of the ASU Graduate College and the ASU Libraries. For more information or questions about this collection contact or visit the Digital Repository ETD Library Guide or contact the ASU Graduate College at gradformat@asu.edu.
Filtering by
- All Subjects: Statistics
- Creators: Rogers, Bradley
- Creators: Davulcu, Hasan
In this thesis, several data mining algorithms have been applied to analyze students’ code assignment submission data from a real classroom study. The goal of this work is to explore
and predict students’ performances. Multiple machine learning models and the model accuracy were evaluated based on the Shapley Additive Explanation.
The Cross-Validation shows the Gradient Boosting Decision Tree has the best precision 85.93% with average 82.90%. Features like Component grade, Due Date, Submission Times have higher impact than others. Baseline model received lower precision due to lack of non-linear fitting.
This research explores the problem of the why so few of the published algorithms enter production and furthermore, fewer end up generating sustained value. The dissertation proposes a ‘Design for Deployment’ (DFD) framework to successfully build machine learning analytics so they can be deployed to generate sustained value. The framework emphasizes and elaborates the often neglected but immensely important latter steps of an analytics process: ‘Evaluation’ and ‘Deployment’. A representative evaluation framework is proposed that incorporates the temporal-shifts and dynamism of real-world scenarios. Additionally, the recommended infrastructure allows analytics projects to pivot rapidly when a particular venture does not materialize. Deployment needs and apprehensions of the industry are identified and gaps addressed through a 4-step process for sustainable deployment. Lastly, the need for analytics as a functional area (like finance and IT) is identified to maximize the return on machine-learning deployment.
The framework and process is demonstrated in semiconductor manufacturing – it is highly complex process involving hundreds of optical, electrical, chemical, mechanical, thermal, electrochemical and software processes which makes it a highly dynamic non-stationary system. Due to the 24/7 uptime requirements in manufacturing, high-reliability and fail-safe are a must. Moreover, the ever growing volumes mean that the system must be highly scalable. Lastly, due to the high cost of change, sustained value proposition is a must for any proposed changes. Hence the context is ideal to explore the issues involved. The enterprise use-cases are used to demonstrate the robustness of the framework in addressing challenges encountered in the end-to-end process of productizing machine learning analytics in dynamic read-world scenarios.
Part 1 characterizes soiling losses using various techniques to understand the effect of soiling on photovoltaic modules. The higher the angle of incidence (AOI), the lower will be the photovoltaic (PV) module performance. Our research group has already reported the AOI investigation for cleaned modules of five different technologies with air/glass interface. However, the modules that are installed in the field would invariably develop a soil layer with varying thickness depending on the site condition, rainfall and tilt angle. The soiled module will have the air/soil/glass interface rather than air/glass interface. This study investigates the AOI variations on soiled modules of five different PV technologies. It is demonstrated that AOI effect is inversely proportional to the soil density. In other words, the power or current loss between clean and soiled modules would be much higher at a higher AOI than at a lower AOI leading to excessive energy production loss of soiled modules on cloudy days, early morning hours and late afternoon hours. Similarly, the spectral influence of soil on the performance of the module was investigated through reflectance and transmittance measurements. It was observed that the reflectance and transmittances losses vary linearly with soil density variation and the 600-700 nm band was identified as an ideal band for soil density measurements.
Part 2 of this thesis performs statistical risk analysis for a power plant through FMECA (Failure Mode, Effect, and Criticality Analysis) based on non-destructive field techniques and count data of the failure modes. Risk Priority Number is used for the grading guideline for criticality analysis. The analysis was done on a 19-year-old power plant in cold-dry climate to identify the most dominant failure and degradation modes. In addition, a comparison study was done on the current power plant (framed) along with another 18-year-old (frameless) from the same climate zone to understand the failure modes for cold-dry climatic condition.
Part 1 of this thesis determines the most dominant failure modes of field aged photovoltaic (PV) modules using experimental data and statistical analysis, FMECA (Failure Mode, Effect, and Criticality Analysis). The failure and degradation modes of about 5900 crystalline-Si glass/polymer modules fielded for 6 to 16 years in three different photovoltaic (PV) power plants with different mounting systems under the hot-dry desert climate of Arizona are evaluated. A statistical reliability tool, FMECA that uses Risk Priority Number (RPN) is performed for each PV power plant to determine the dominant failure modes in the modules by means of ranking and prioritizing the modes. This study on PV power plants considers all the failure and degradation modes from both safety and performance perspectives, and thus, comes to the conclusion that solder bond fatigue/failure with/without gridline/metallization contact fatigue/failure is the most dominant failure mode for these module types in the hot-dry desert climate of Arizona.
Part 2 of this thesis determines the best method to compute degradation rates of PV modules. Three different PV systems were evaluated to compute degradation rates using four methods and they are: I-V measurement, metered kWh, performance ratio (PR) and performance index (PI). I-V method, being an ideal method for degradation rate computation, were compared to the results from other three methods. The median degradation rates computed from kWh method were within ±0.15% from I-V measured degradation rates (0.9-1.37 %/year of three models). Degradation rates from the PI method were within ±0.05% from the I-V measured rates for two systems but the calculated degradation rate was remarkably different (±1%) from the I-V method for the third system. The degradation rate from the PR method was within ±0.16% from the I-V measured rate for only one system but were remarkably different (±1%) from the I-V measured rate for the other two systems. Thus, it was concluded that metered raw kWh method is the best practical method, after I-V method and PI method (if ground mounted POA insolation and other weather data are available) for degradation computation as this method was found to be fairly accurate, easy, inexpensive, fast and convenient.