Theses - Dept. of Electrical Engineering and Computer Sciences
http://hdl.handle.net/1721.1/7599
2015-06-15T13:37:03ZPower processing and active protection for photovoltaic energy extraction
http://hdl.handle.net/1721.1/97330
Power processing and active protection for photovoltaic energy extraction
Chan, Arthur Hsu Chen
Solar photovoltaic power generation is a promising clean and renewable energy technology that can draw upon the planet's most abundant power source - the sun. However, relatively high levelized cost of energy (LCOE), the ratio of the total cost of ownership to the total energy extracted over the lifetime of the generation system, has limited the grid penetration of solar power. Mismatch loss remains an important issue to address in PV systems, and a solar power system can lose as much as 30% of its energy generation capability over a year due to mismatch. Maximum power point tracking (MPPT) using power electronics converters can increase the overall solar energy extraction efficiency and thus reduce the LCOE. Many power electronics solutions have been proposed at the module and submodule levels, which only partially addresses the mismatch problem. However, scaling the existing solutions to finer optimization granularity has been cost-prohibitive. In the first part of this thesis, a new cell-level strategy, termed diffusion charge redistribution (DCR), is proposed to fully recover mismatch loss. The proposed technique processes power by leveraging the intrinsic solar cell capacitance rather than relying on externally added intermediate energy storage in order to drastically reduce to the cost of MPPT while enabling the finest optimization granularity. Moreover, strings balanced by this technique exhibit power versus current curves that are convex, which simplifies the required MPPT algorithm. Cell-level power balancing may also ease the testing and binning criteria during manufacturing, which leads to additional cost savings. Differential power processing (DPP) is a key concept to further improve energy efficiency by minimizing the amount of power conversion. In the second part of this thesis, the concept of differential power processing is introduced to the proposed cell-level power balancing technique by rethinking the string-level power electronics architecture. This enhancement can improve the overall efficiency of DCR by more than 3.5% while permitting the use of a slower DCR switching frequency. It can also be applied to many other cascaded converter architectures to reduce insertion loss. In particular, the proposed differential DCR (dDCR) architecture simultaneously achieves maximum power point tracking without any external passive components at the cell-level, and maintains differential power processing with zero insertion loss. This is accomplished by decoupling the MPPT functional block from the DPP functional block. The new power optimization aims to not only maximize energy extraction from each solar cell but also minimize the amount of processed power. The new multi-variable optimization space for the dDCR topology is evaluated and shown to be convex, which simplifies the required optimization algorithm. The inverter represents a large part of the overall cost and is often the most failure-prone component in a photovoltaic power system. In order to improve the cost and reliability of a grid-tie inverter, switched-capacitor techniques are adopted to reduce the required capacitance and rated voltage of the dc-link capacitor. The proposed switched-capacitor energy buffer can improve capacitor energy utilization by more than four times for a system with a 10% peak-to-peak ripple specification, and enable the use of film or ceramic capacitors to prolong the system lifetime to over a hundred years. The third part of this thesis explores the SC energy buffer design space and examines tradeoffs regarding circuit topology, switching configuration, and control complexity. Practical applications require control schemes capable of handling source and load transients. A two-step control methodology that mitigates undesirable transient responses is proposed and demonstrated in simulation. Finally, dc power system architectures have attracted interest as a means for achieving high overall efficiency and facilitating integration of renewable and distributed energy sources, such as a photovoltaic system. However, to enable widespread adoption of dc systems, the reliability of fault protection and interruption capability is essential. A new dc breaker topology, called the series-connected Z-source circuit breaker, is introduced to minimize the reflected fault current drawn from a source while retaining a common return ground path. Analogous in some respects to an ac thermal-magnetic breaker, the proposed Z-source breaker can be designed for considerations affecting both rate of fault current rise and absolute fault current level. The proposed manual tripping mechanism also enables protection against both instantaneous large surges in current and longer-term over-current conditions.
Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2015.; Cataloged from PDF version of thesis.; Includes bibliographical references (pages 203-206).
2015-01-01T00:00:00ZEnergy-efficient approximate computation in Topaz
http://hdl.handle.net/1721.1/97329
Energy-efficient approximate computation in Topaz
Achour, Sara
The increasing prominence of energy consumption as a first-order concern in contemporary computing systems has motivated the design of energy-efficient approximate computing platforms. These computing platforms feature energy-efficient computing mechanisms such as components that may occasionally produce incorrect results. We present Topaz, a new task-based language for computations that execute on approximate computing platforms that may occasionally produce arbitrarily inaccurate results. The Topaz implementation maps approximate tasks onto the approximate machine and integrates the approximate results into the main computation, deploying a novel outlier detection and reliable re-execution mechanism to prevent unacceptably inaccurate results from corrupting the overall computation. Because Topaz can work effectively with a very broad range of approximate hardware designs, it provides hardware developers with substantial freedom in the designs that they produce. In particular, Topaz does not impose the need for any specific restrictive reliability or accuracy guarantees. Experimental results from our set of benchmark applications demonstrate the effectiveness of Topaz in vastly improving the quality of the generated output while only incurring 0.2% to 3% energy overheard.
Thesis: S.M. in Computer Science and Engineering, Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2015.; Cataloged from PDF version of thesis.; Includes bibliographical references (pages 69-73).
2015-01-01T00:00:00ZBeatDB : an end-to-end approach to unveil saliencies from massive signal data sets
http://hdl.handle.net/1721.1/97328
BeatDB : an end-to-end approach to unveil saliencies from massive signal data sets
Dernoncourt, Franck
Prediction studies on physiological signals are time-consuming: a typical study, even with a modest number of patients, usually takes from 6 to 12 months. In response we design a large-scale machine learning and analytics framework, BeatDB, to scale and speed up mining knowledge from waveforms. BeatDB radically shrinks the time an investigation takes by: * supporting fast, flexible investigations by offering a multi-level parameterization, allowing the user to define the condition to predict, the features, and many other investigation parameters. * precomputing beat-level features that are likely to be frequently used while computing on-the-fly less used features and statistical aggregates. In this thesis, we present BeatDB and demonstrate how it supports flexible investigations on the entire set of arterial blood pressure data in the MIMIC II Waveform Database, which contains over 5000 patients and 1 billion of blood pressure beats. We focus on the usefulness of wavelets as features in the context of blood pressure prediction and use Gaussian process to accelerate the search of the feature yielding the highest AUROC.
Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, February 2015.; Cataloged from PDF version of thesis.; Includes bibliographical references (pages 109-114).
2015-01-01T00:00:00ZLP/SDP hierarchy lower bounds for decoding random LDPC codes
http://hdl.handle.net/1721.1/97327
LP/SDP hierarchy lower bounds for decoding random LDPC codes
Ghazi, Badih
Random (dv, dc)-regular LDPC codes (where each variable is involved in d, parity checks and each parity check involves d, variables) are well-known to achieve the Shannon capacity of the binary symmetric channel (for sufficiently large dv, and dc,) under exponential time decoding. However, polynomial time algorithms are only known to correct a much smaller fraction of errors. One of the most powerful polynomial-time algorithms with a formal analysis is the LP decoding algorithm of Feldman et al. which is known to correct an [omega](1/dc) fraction of errors. In this work, we show that fairly powerful extensions of LP decoding, based on the Sherali-Adams and Lasserre hierarchies, fail to correct much more errors than the basic LP-decoder. In particular, we show that: -- For any values of d, and de, a linear number of rounds of the Sherali-Adams LP hierarchy cannot correct more than an O(1/dc) fraction of errors on a random (dv, dc)-regular LDPC code. -- For any value of d, and infinitely many values of de, a linear number of rounds of the Lasserre SDP hierarchy cannot correct more than an O(1/dc) fraction of errors on a random (dv, dc)-regular LDPC code. Our proofs use a new streching and collapsing technique that allows us to leverage recent progress in the study of the limitations of LP/SDP hierarchies for Maximum Constraint Satisfaction Problems (Max-CSPs). The problem then reduces to the construction of special balanced pairwise independent distributions for Sherali-Adams and special cosets of balanced pairwise independent subgroups for Lasserre. Our (algebraic) construction for the Lasserre hierarchy is based on designing sets of points in Fq (for q any power of 2 and d = 2,3) with special hyperplane-incidence properties constructions that may be of independent interest. An intriguing consequence of our work is that expansion seems to be both the strength and the weakness of random regular LDPC codes. Our techniques are more generally applicable to a large class of Boolean CSPs called Min-Ones. In particular, for k-Hypergraph Vertex Cover, we obtain an improved integrality gap of k - 1 - e that holds after a linear number of rounds of the Lasserre hierarchy, for any k = q + 1 with q an arbitrary prime power. The best previous gap for a linear number of rounds was equal to 2-E and due to Schoenebeck.
Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2015.; Cataloged from PDF version of thesis.; Includes bibliographical references (pages 57-59).
2015-01-01T00:00:00Z