Matching Items (103)
Filtering by
- Genre: Masters Thesis
- Creators: Yang, Yezhou
Description
In the recent past, due to regulatory hurdles and the inability to expand transmission systems, the bulk power system is increasingly being operated close to its limits. Among the various phenomenon encountered, static voltage stability has received increased attention among electric utilities. One approach to investigate static voltage stability is to run a set of power flow simulations and derive the voltage stability limit based on the analysis of power flow results. Power flow problems are formulated as a set of nonlinear algebraic equations usually solved by iterative methods. The most commonly used method is the Newton-Raphson method. However, at the static voltage stability limit, the Jacobian becomes singular. Hence, the power flow solution may fail to converge close to the true limit.
To carefully examine the limitations of conventional power flow software packages in determining voltage stability limits, two lines of research are pursued in this study. The first line of the research is to investigate the capability of different power flow solution techniques, such as conventional power flow and non-iterative power flow techniques to obtain the voltage collapse point. The software packages used in this study include Newton-based methods contained in PSSE, PSLF, PSAT, PowerWorld, VSAT and a non-iterative technique known as the holomorphic embedding method (HEM).
The second line is to investigate the impact of the available control options and solution parameter settings that can be utilized to obtain solutions closer to the voltage collapse point. Such as the starting point, generator reactive power limits, shunt device control modes, area interchange control, and other such parameters.
To carefully examine the limitations of conventional power flow software packages in determining voltage stability limits, two lines of research are pursued in this study. The first line of the research is to investigate the capability of different power flow solution techniques, such as conventional power flow and non-iterative power flow techniques to obtain the voltage collapse point. The software packages used in this study include Newton-based methods contained in PSSE, PSLF, PSAT, PowerWorld, VSAT and a non-iterative technique known as the holomorphic embedding method (HEM).
The second line is to investigate the impact of the available control options and solution parameter settings that can be utilized to obtain solutions closer to the voltage collapse point. Such as the starting point, generator reactive power limits, shunt device control modes, area interchange control, and other such parameters.
ContributorsYi, Weili (Author) / Vittal, Vijay (Thesis advisor) / Tylavsky, Daniel (Thesis advisor) / Qin, Jiangchao (Committee member) / Arizona State University (Publisher)
Created2017
Description
This work presents a communication paradigm, using a context-aware mixed reality approach, for instructing human workers when collaborating with robots. The main objective of this approach is to utilize the physical work environment as a canvas to communicate task-related instructions and robot intentions in the form of visual cues. A vision-based object tracking algorithm is used to precisely determine the pose and state of physical objects in and around the workspace. A projection mapping technique is used to overlay visual cues on tracked objects and the workspace. Simultaneous tracking and projection onto objects enables the system to provide just-in-time instructions for carrying out a procedural task. Additionally, the system can also inform and warn humans about the intentions of the robot and safety of the workspace. It was hypothesized that using this system for executing a human-robot collaborative task will improve the overall performance of the team and provide a positive experience to the human partner. To test this hypothesis, an experiment involving human subjects was conducted and the performance (both objective and subjective) of the presented system was compared with a conventional method based on printed instructions. It was found that projecting visual cues enabled human subjects to collaborate more effectively with the robot and resulted in higher efficiency in completing the task.
ContributorsKalpagam Ganesan, Ramsundar (Author) / Ben Amor, Hani (Thesis advisor) / Yang, Yezhou (Committee member) / Zhang, Yu (Committee member) / Arizona State University (Publisher)
Created2017
Description
In this thesis, I propose a new technique of Aligning English sentence words
with its Semantic Representation using Inductive Logic Programming(ILP). My
work focusses on Abstract Meaning Representation(AMR). AMR is a semantic
formalism to English natural language. It encodes meaning of a sentence in a rooted
graph. This representation has gained attention for its simplicity and expressive power.
An AMR Aligner aligns words in a sentence to nodes(concepts) in its AMR
graph. As AMR annotation has no explicit alignment with words in English sentence,
automatic alignment becomes a requirement for training AMR parsers. The aligner in
this work comprises of two components. First, rules are learnt using ILP that invoke
AMR concepts from sentence-AMR graph pairs in the training data. Second, the
learnt rules are then used to align English sentences with AMR graphs. The technique
is evaluated on publicly available test dataset and the results are comparable with
state-of-the-art aligner.
with its Semantic Representation using Inductive Logic Programming(ILP). My
work focusses on Abstract Meaning Representation(AMR). AMR is a semantic
formalism to English natural language. It encodes meaning of a sentence in a rooted
graph. This representation has gained attention for its simplicity and expressive power.
An AMR Aligner aligns words in a sentence to nodes(concepts) in its AMR
graph. As AMR annotation has no explicit alignment with words in English sentence,
automatic alignment becomes a requirement for training AMR parsers. The aligner in
this work comprises of two components. First, rules are learnt using ILP that invoke
AMR concepts from sentence-AMR graph pairs in the training data. Second, the
learnt rules are then used to align English sentences with AMR graphs. The technique
is evaluated on publicly available test dataset and the results are comparable with
state-of-the-art aligner.
ContributorsAgarwal, Shubham (Author) / Baral, Chitta (Thesis advisor) / Li, Baoxin (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2017
Description
LPMLN is a recent probabilistic logic programming language which combines both Answer Set Programming (ASP) and Markov Logic. It is a proper extension of Answer Set programs which allows for reasoning about uncertainty using weighted rules under the stable model semantics with a weight scheme that is adopted from Markov Logic. LPMLN has been shown to be related to several formalisms from the knowledge representation (KR) side such as ASP and P-Log, and the statistical relational learning (SRL) side such as Markov Logic Networks (MLN), Problog and Pearl’s causal models (PCM). Formalisms like ASP, P-Log, Problog, MLN, PCM have all been shown to embeddable in LPMLN which demonstrates the expressivity of the language. Interestingly, LPMLN has also been shown to reducible to ASP and MLN which is not only theoretically interesting, but also practically important from a computational point of view in that the reductions yield ways to compute LPMLN programs utilizing ASP and MLN solvers. Additionally, the reductions also allow the users to compute other formalisms which can be reduced to LPMLN.
This thesis realizes two implementations of LPMLN based on the reductions from LPMLN to ASP and LPMLN to MLN. This thesis first presents an implementation of LPMLN called LPMLN2ASP that uses standard ASP solvers for computing MAP inference using weak constraints, and marginal and conditional probabilities using stable models enumeration. Next, in this thesis, another implementation of LPMLN called LPMLN2MLN is presented that uses MLN solvers which apply completion to compute the tight fragment of LPMLN programs for MAP inference, marginal and conditional probabilities. The computation using ASP solvers yields exact inference as opposed to approximate inference using MLN solvers. Using these implementations, the usefulness of LPMLN for computing other formalisms is demonstrated by reducing them to LPMLN. The thesis also shows how the implementations are better than the native solvers of some of these formalisms on certain domains. The implementations make use of the current state of the art solving technologies in ASP and MLN, and therefore they benefit from any theoretical and practical advances in these technologies, thereby also benefiting the computation of other formalisms that can be reduced to LPMLN. Furthermore, the implementation also allows for certain SRL formalisms to be computed by ASP solvers, and certain KR formalisms to be computed by MLN solvers.
This thesis realizes two implementations of LPMLN based on the reductions from LPMLN to ASP and LPMLN to MLN. This thesis first presents an implementation of LPMLN called LPMLN2ASP that uses standard ASP solvers for computing MAP inference using weak constraints, and marginal and conditional probabilities using stable models enumeration. Next, in this thesis, another implementation of LPMLN called LPMLN2MLN is presented that uses MLN solvers which apply completion to compute the tight fragment of LPMLN programs for MAP inference, marginal and conditional probabilities. The computation using ASP solvers yields exact inference as opposed to approximate inference using MLN solvers. Using these implementations, the usefulness of LPMLN for computing other formalisms is demonstrated by reducing them to LPMLN. The thesis also shows how the implementations are better than the native solvers of some of these formalisms on certain domains. The implementations make use of the current state of the art solving technologies in ASP and MLN, and therefore they benefit from any theoretical and practical advances in these technologies, thereby also benefiting the computation of other formalisms that can be reduced to LPMLN. Furthermore, the implementation also allows for certain SRL formalisms to be computed by ASP solvers, and certain KR formalisms to be computed by MLN solvers.
ContributorsTalsania, Samidh (Author) / Lee, Joohyung (Thesis advisor, Committee member) / Baral, Chitta (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2017
Description
Compressive sensing theory allows to sense and reconstruct signals/images with lower sampling rate than Nyquist rate. Applications in resource constrained environment stand to benefit from this theory, opening up many possibilities for new applications at the same time. The traditional inference pipeline for computer vision sequence reconstructing the image from compressive measurements. However,the reconstruction process is a computationally expensive step that also provides poor results at high compression rate. There have been several successful attempts to perform inference tasks directly on compressive measurements such as activity recognition. In this thesis, I am interested to tackle a more challenging vision problem - Visual question answering (VQA) without reconstructing the compressive images. I investigate the feasibility of this problem with a series of experiments, and I evaluate proposed methods on a VQA dataset and discuss promising results and direction for future work.
ContributorsHuang, Li-Chin (Author) / Turaga, Pavan (Thesis advisor) / Yang, Yezhou (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2017
Description
Light field imaging is limited in its computational processing demands of high
sampling for both spatial and angular dimensions. Single-shot light field cameras
sacrifice spatial resolution to sample angular viewpoints, typically by multiplexing
incoming rays onto a 2D sensor array. While this resolution can be recovered using
compressive sensing, these iterative solutions are slow in processing a light field. We
present a deep learning approach using a new, two branch network architecture,
consisting jointly of an autoencoder and a 4D CNN, to recover a high resolution
4D light field from a single coded 2D image. This network decreases reconstruction
time significantly while achieving average PSNR values of 26-32 dB on a variety of
light fields. In particular, reconstruction time is decreased from 35 minutes to 6.7
minutes as compared to the dictionary method for equivalent visual quality. These
reconstructions are performed at small sampling/compression ratios as low as 8%,
allowing for cheaper coded light field cameras. We test our network reconstructions
on synthetic light fields, simulated coded measurements of real light fields captured
from a Lytro Illum camera, and real coded images from a custom CMOS diffractive
light field camera. The combination of compressive light field capture with deep
learning allows the potential for real-time light field video acquisition systems in the
future.
sampling for both spatial and angular dimensions. Single-shot light field cameras
sacrifice spatial resolution to sample angular viewpoints, typically by multiplexing
incoming rays onto a 2D sensor array. While this resolution can be recovered using
compressive sensing, these iterative solutions are slow in processing a light field. We
present a deep learning approach using a new, two branch network architecture,
consisting jointly of an autoencoder and a 4D CNN, to recover a high resolution
4D light field from a single coded 2D image. This network decreases reconstruction
time significantly while achieving average PSNR values of 26-32 dB on a variety of
light fields. In particular, reconstruction time is decreased from 35 minutes to 6.7
minutes as compared to the dictionary method for equivalent visual quality. These
reconstructions are performed at small sampling/compression ratios as low as 8%,
allowing for cheaper coded light field cameras. We test our network reconstructions
on synthetic light fields, simulated coded measurements of real light fields captured
from a Lytro Illum camera, and real coded images from a custom CMOS diffractive
light field camera. The combination of compressive light field capture with deep
learning allows the potential for real-time light field video acquisition systems in the
future.
ContributorsGupta, Mayank (Author) / Turaga, Pavan (Thesis advisor) / Yang, Yezhou (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2017
Description
In recent years, several methods have been proposed to encode sentences into fixed length continuous vectors called sentence representation or sentence embedding. With the recent advancements in various deep learning methods applied in Natural Language Processing (NLP), these representations play a crucial role in tasks such as named entity recognition, question answering and sentence classification.
Traditionally, sentence vector representations are learnt from its constituent word representations, also known as word embeddings. Various methods to learn the distributed representation (embedding) of words have been proposed using the notion of Distributional Semantics, i.e. “meaning of a word is characterized by the company it keeps”. However, principle of compositionality states that meaning of a sentence is a function of the meanings of words and also the way they are syntactically combined. In various recent methods for sentence representation, the syntactic information like dependency or relation between words have been largely ignored.
In this work, I have explored the effectiveness of sentence representations that are composed of the representation of both, its constituent words and the relations between the words in a sentence. The word and relation embeddings are learned based on their context. These general-purpose embeddings can also be used as off-the- shelf semantic and syntactic features for various NLP tasks. Similarity Evaluation tasks was performed on two datasets showing the usefulness of the learned word embeddings. Experiments were conducted on three different sentence classification tasks showing that our sentence representations outperform the original word-based sentence representations, when used with the state-of-the-art Neural Network architectures.
Traditionally, sentence vector representations are learnt from its constituent word representations, also known as word embeddings. Various methods to learn the distributed representation (embedding) of words have been proposed using the notion of Distributional Semantics, i.e. “meaning of a word is characterized by the company it keeps”. However, principle of compositionality states that meaning of a sentence is a function of the meanings of words and also the way they are syntactically combined. In various recent methods for sentence representation, the syntactic information like dependency or relation between words have been largely ignored.
In this work, I have explored the effectiveness of sentence representations that are composed of the representation of both, its constituent words and the relations between the words in a sentence. The word and relation embeddings are learned based on their context. These general-purpose embeddings can also be used as off-the- shelf semantic and syntactic features for various NLP tasks. Similarity Evaluation tasks was performed on two datasets showing the usefulness of the learned word embeddings. Experiments were conducted on three different sentence classification tasks showing that our sentence representations outperform the original word-based sentence representations, when used with the state-of-the-art Neural Network architectures.
ContributorsRath, Trideep (Author) / Baral, Chitta (Thesis advisor) / Li, Baoxin (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2017
Description
Composite insulators on overhead lines are frequently subjected to corona discharges due to increased electric field intensities under various conditions. These discharges can cause localized heating on the surface and affect the hydrophobicity of the insulator. A study has been undertaken to quantify and evaluate the thermal degradation that composite insulation is subjected to from corona discharges. This has been conducted primarily at the power frequency (60 Hz) and at the low frequency range (37 kHz). Point to plane corona discharge experiments have been performed in the laboratory at both the frequencies and varying levels of thermal degradation has been observed. The amplitude and the frequency of current spikes have been recorded at different voltage levels. A temperature model based on the amplitude and the frequency of current data has been formulated to calculate the maximum temperature attained due to these discharges. Visual thermal degradation has been found to set in at a low frequency range while there is no visual degradation observed at power frequency even when exposed to discharges for relatively much longer periods of time. However, microscopic experiments have been conducted which revealed degradation on the surface at 60 Hz. It has also been found that temperatures in excess of 300 Celsius have been obtained at 37 kHz. This corroborates the thermo gravimetric analysis data that proves thermal degradation in silicone rubber samples at temperatures greater than 300 Celsius. Using the above model, the maximum temperature rise can be evaluated due to discharges occurring on high voltage insulation. This model has also been used to calculate the temperature rise on medium voltage distribution equipment such as composite bushings and stand-off plugs. The samples were subjected to standard partial discharge tests and the corresponding discharge magnitudes have been recorded. The samples passed the tests and the corresponding temperatures plotted have been found to be within thermal limits of the respective insulation used on the samples. The experimental results concur with the theoretical model. A knowledge of the maximum temperatures attained due to these discharges can help in design of insulation with better thermal properties.
ContributorsSangaraju Venkateshwara, Pradeep Varma (Author) / Gorur, Ravi S (Thesis advisor) / Farmer, Richard (Committee member) / Vittal, Vijay (Committee member) / Arizona State University (Publisher)
Created2010
Description
This research work illustrates the use of software packages based on the concept of nu-merical analysis technique to evaluate the electric field and voltage distribution along composite insulators for system voltages ranging from 138 kV up to 1200 kV ac. A part of the calculations was made using the 3D software package, COULOMB 8.0, based on the concept of Boundary Element Method (BEM). The electric field was calculated under dry and wet conditions. Compo-site insulators experience more electrical stress when compared to porcelain and are also more prone to damage caused by corona activity. The work presented here investigates the effect of corona rings of specific dimensions and bundled conductors on the electric field along composite insulators. Inappropriate placement or dimensions of corona rings could enhance the electric field instead of mitigating it. Corona ring optimization for a 1000 kV composite insulator was per-formed by changing parameters of the ring, such as the diameter of the ring, thickness of the ring tube and the projection of the ring from the high voltage energized end fitting. Grading rings were designed for Ultra High Voltage (UHV) systems that use two units of composite insulators in pa-rallel. The insulation distance, which bears 50% of the total applied voltage, is raised by 61% with the grading ring installed, when compared to the distance without the grading ring. In other words, the electric field and voltage distribution was found to be more linear with the application of grad-ing rings. The second part of this project was carried out using the EPRI designed software EPIC. This is based on the concept of Charge Simulation method (CSM). Comparisons were made be-tween electric field magnitude along composite insulators used for suspension and dead end configuration for system voltages ranging from 138 kV to 500 kV. It was found that the dead end composite insulators experience significantly higher electrical stress when compared to their suspension counterpart. It was also concluded that this difference gets more prominent as the system voltage increases. A comparison made between electric field distribution along composite insulators used in single and double dead end structures suggested that the electric stress experienced by the single dead end composite insulators is relatively higher when compared to double dead end composite insulators.
ContributorsDoshi, Tanushri (Author) / Gorur, Ravi S (Thesis advisor) / Vittal, Vijay (Committee member) / Farmer, Richard (Committee member) / Arizona State University (Publisher)
Created2010
Description
Market acceptability of distributed energy resource (DER) technologies and the gradual and consistent increase in their depth of penetration have generated significant interest over the past few years. In particular, in Arizona and several other states there has been a substantial in-crease in distributed photovoltaic (PV) generation interfaced to the power distribution systems, and is expected to continue to grow at a significant rate. This has made integration, control and optimal operation of DER units a main area of focus in the design and operation of distribution systems. Grid-connected, distributed PV covers a wide range of power levels ranging from small, single phase residential roof-top systems to large three-phase, multi-megawatt systems. The focus of this work is on analyzing large, three-phase systems, with the power distribution system of the Arizona State University (ASU) Tempe campus used as the test bed for analysis and simulation. The Tempe campus of ASU has presently 4.5 MW of installed PV capacity, with another 4.5 MW expected to be added by 2011, which will represent about 22% of PV penetration. The PV systems are interfaced to the grid invariably by a power electronic inverter. Many of the important characteristics of the PV generation are influenced by the design and performance of the inverter, and hence suitable models of the inverter are needed to analyze PV systems. Several models of distributed generation (DG), including switching and average models, suitable for different study objectives, and different control modes of the inverter have been described in this thesis. A critical function of the inverters is to quickly detect and eliminate unintentional islands during grid failure. In this thesis, many active anti-islanding techniques with voltage and frequency positive feedback have been studied. Effectiveness of these techniques in terms of the tripping times specified in IEEE Std. 1547 for interconnecting distributed resources with electric power systems has been analyzed. The impact of distributed PV on the voltage profile of a distribution system has been ana-lyzed with ASU system as the test bed using power systems analysis tools namely PowerWorld and CYMDIST. The present inverters complying with IEEE 1547 do not regulate the system vol-tage. However, the future inverters especially at higher power levels are expected to perform sev-eral grid support functions including voltage regulation and reactive power support. Hence, the impact of inverters with the reactive power support capabilities is also analyzed. Various test sce-narios corresponding to different grid conditions are simulated and it is shown that distributed PV improves the voltage profile of the system. The improvements are more significant when the PV generators are capable of reactive power support. Detailed short circuit analyses are also per-formed on the system, and the impact of distributed PV on the fault current magnitude, with and without reactive power injection, have been studied.
ContributorsNarayanan, Anand (Author) / Ayyanar, Raja (Thesis advisor) / Vittal, Vijay (Committee member) / Heydt, Gerald T (Committee member) / Arizona State University (Publisher)
Created2010