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.
The first topic of this dissertation focuses on the data clustering. Data clustering is often the first step for analyzing a dataset without the label information. Clustering high-dimensional data with mixed categorical and numeric attributes remains a challenging, yet important task. A clustering algorithm based on tree ensembles, CRAFTER, is proposed to tackle this task in a scalable manner.
The second part of this dissertation aims to develop data representation methods for genome sequencing data, a special type of high-dimensional data in the biomedical domain. The proposed data representation method, Bag-of-Segments, can summarize the key characteristics of the genome sequence into a small number of features with good interpretability.
The third part of this dissertation introduces an end-to-end deep neural network model, GCRNN, for time series classification with emphasis on both the accuracy and the interpretation. GCRNN contains a convolutional network component to extract high-level features, and a recurrent network component to enhance the modeling of the temporal characteristics. A feed-forward fully connected network with the sparse group lasso regularization is used to generate the final classification and provide good interpretability.
The last topic centers around the dimensionality reduction methods for time series data. A good dimensionality reduction method is important for the storage, decision making and pattern visualization for time series data. The CRNN autoencoder is proposed to not only achieve low reconstruction error, but also generate discriminative features. A variational version of this autoencoder has great potential for applications such as anomaly detection and process control.
First, it is shown that exact, multi-factor D-optimal designs for the logistic regression model can be susceptible to separation. Several logistic regression models are specified, and exact D-optimal designs of fixed sizes are constructed for each model. Sets of simulated response data are generated to estimate the probability of separation in each design. This study proves through simulation that small-sample D-optimal designs are prone to separation and that separation risk is dependent on the specified model. Additionally, it is demonstrated that exact designs of equal size constructed for the same models may have significantly different chances of encountering separation.
The second portion of this research establishes an effective strategy for augmentation, where additional design runs are judiciously added to eliminate separation that has occurred in an initial design. A simulation study is used to demonstrate that augmenting runs in regions of maximum prediction variance (MPV), where the predicted probability of either response category is 50%, most reliably eliminates separation. However, it is also shown that MPV augmentation tends to yield augmented designs with lower D-efficiencies.
The final portion of this research proposes a novel compound optimality criterion, DMP, that is used to construct locally optimal and robust compromise designs. A two-phase coordinate exchange algorithm is implemented to construct exact locally DMP-optimal designs. To address design dependence issues, a maximin strategy is proposed for designating a robust DMP-optimal design. A case study demonstrates that the maximin DMP-optimal design maintains comparable D-efficiencies to a corresponding Bayesian D-optimal design while offering significantly improved separation performance.
important to increase the eciency and reliability of this emerging clean energy technologies.
This thesis focuses on modeling and reliability of solar micro inverters. In
order to make photovoltaics (PV) cost competitive with traditional energy sources,
the economies of scale have been guiding inverter design in two directions: large,
centralized, utility-scale (500 kW) inverters vs. small, modular, module level (300
W) power electronics (MLPE). MLPE, such as microinverters and DC power optimizers,
oer advantages in safety, system operations and maintenance, energy yield,
and component lifetime due to their smaller size, lower power handling requirements,
and module-level power point tracking and monitoring capability [1]. However, they
suer from two main disadvantages: rst, depending on array topology (especially
the proximity to the PV module), they can be subjected to more extreme environments
(i.e. temperature cycling) during the day, resulting in a negative impact to
reliability; second, since solar installations can have tens of thousands to millions of
modules (and as many MLPE units), it may be dicult or impossible to track and
repair units as they go out of service. Therefore identifying the weak links in this
system is of critical importance to develop more reliable micro inverters.
While an overwhelming majority of time and research has focused on PV module
eciency and reliability, these issues have been largely ignored for the balance
of system components. As a relatively nascent industry, the PV power electronics
industry does not have the extensive, standardized reliability design and testing procedures
that exist in the module industry or other more mature power electronics
industries (e.g. automotive). To do so, the critical components which are at risk and
their impact on the system performance has to be studied. This thesis identies and
addresses some of the issues related to reliability of solar micro inverters.
This thesis presents detailed discussions on various components of solar micro inverter
and their design. A micro inverter with very similar electrical specications in
comparison with commercial micro inverter is modeled in detail and veried. Components
in various stages of micro inverter are listed and their typical failure mechanisms
are reviewed. A detailed FMEA is conducted for a typical micro inverter to identify
the weak links of the system. Based on the S, O and D metrics, risk priority number
(RPN) is calculated to list the critical at-risk components. Degradation of DC bus
capacitor is identied as one the failure mechanism and the degradation model is built
to study its eect on the system performance. The system is tested for surge immunity
using standard ring and combinational surge waveforms as per IEEE 62.41 and
IEC 61000-4-5 standards. All the simulation presented in this thesis is performed
using PLECS simulation software.
For all three conductor arrangements, the shapes of the electric field distribution curves are different with the vertical arrangement best for minimizing right of way consideration, while the shapes of the magnetic field distributions curves are similar. Except for the horizontal arrangement, the maximum electric field magnitudes with shield conductors are larger than those without shield conductors. Among the three different arrangements, the maximum field value of the vertical arrangement is most vulnerable to the unbalanced conditions.
For both the electric and magnetic fields, increasing the heights of phase conductors gradually results in diminishing return in terms of the field reduction. In this work, both the maximum electric field magnitudes and the maximum magnetic field magnitudes produced by 500 kV power lines at 1 m height from the ground are all within the permissible exposure levels for the general public. At last, the dynamic trajectories of both fields with time are simulated and interpreted, with each field represented by a vector rotating in a plane describing an ellipse, where the vector values can be compared to high-speed vector measurements.
This dissertation introduces a real-time topology monitoring scheme for power systems intended to provide enhanced situational awareness during major system disturbances. The topology monitoring scheme requires accurate real-time topology information to be effective. This scheme is supported by advances in transmission line outage detection based on data-mining phasor measurement unit (PMU) measurements.
A network flow analysis scheme is proposed to track changes in user defined minimal cut sets within the system. This work introduces a new algorithm used to update a previous network flow solution after the loss of a single system branch. The proposed new algorithm provides a significantly decreased solution time that is desired in a real- time environment. This method of topology monitoring can provide system operators with visual indications of potential problems in the system caused by changes in topology.
This work also presents a method of determining all singleton cut sets within a given network topology called the one line remaining (OLR) algorithm. During operation, if a singleton cut set exists, then the system cannot withstand the loss of any one line and still remain connected. The OLR algorithm activates after the loss of a transmission line and determines if any singleton cut sets were created. These cut sets are found using properties of power transfer distribution factors and minimal cut sets.
The topology analysis algorithms proposed in this work are supported by line outage detection using PMU measurements aimed at providing accurate real-time topology information. This process uses a decision tree (DT) based data-mining approach to characterize a lost tie line in simulation. The trained DT is then used to analyze PMU measurements to detect line outages. The trained decision tree was applied to real PMU measurements to detect the loss of a 500 kV line and had no misclassifications.
The work presented has the objective of enhancing situational awareness during significant system disturbances in real time. This dissertation presents all parts of the proposed topology monitoring scheme and justifies and validates the methodology using a real system event.
The pseudo-Bayesian approach can be applied to the problem of optimal design construction under dependent observations. Often, correlation between observations exists due to restrictions on randomization. Several techniques for optimal design construction are proposed in the case of the conditional response distribution being a natural exponential family member but with a normally distributed block effect . The reviewed pseudo-Bayesian approach is compared to an approach based on substituting the marginal likelihood with the joint likelihood and an approach based on projections of the score function (often called quasi-likelihood). These approaches are compared for several models with normal, Poisson, and binomial conditional response distributions via the true determinant of the expected Fisher information matrix where the dispersion of the random blocks is considered a nuisance parameter. A case study using the developed methods is performed.
The joint and quasi-likelihood methods are then extended to address the case when the magnitude of random block dispersion is of concern. Again, a simulation study over several models is performed, followed by a case study when the conditional response distribution is a Poisson distribution.