No description available.

In this theses techniques for high-speed digital delta-sigma modulator(DDSM) structures are considered. Four techniques are applied andevaluated: unfolding, increasing the number of delay elements in theinner loop, pipelining/retiming, and optimizations provided by thesynthesis tool. Of interest is to see the speed-area-power trade-offs.For implementation, three different modulators meeting the samerequirements are implemented. Each modulator has a 16-bit input andresults in a 3-bit output. The baseline case is a second-ordermodulator, which has one delay element in its inner loop. Throughoptimization, two new structures are found: to provide two delayelements in the inner loop, a fourth-order modulator is required,while to provide three delay elements, a thirteenth-order modulator isobtained.The results show that in general it is better to unfold the modulatorthan to obtain the speed-up through optimizing the arithmeticoperators with the synthesis tool. Using correct pipelining/retimingis also crucial. Finally, for very high-speed implementation, usingthe structures with more delay elements is required. Also, in manycases these are more area and power efficient compared to usingoptimized arithmetic operators, despite their higher computationalcomplexity.

Electrical power transmission companies have been inspecting underground power cables in a time consuming and destructive way. The current methodology used by Manitoba Hydro, is to remove the conductive material in the center of the cable, cutting the cable into wafers leaving behind the insulating polymer material known as XLPE, the area where many faults occur, and inspect the wafers manually with a microscope. The main goal of this work was to find a methodology to detect these cable faults in a non-destructive way so that the quality of the cable may be assessed, and its remaining lifetime be estimated and return it to use if possible. Two XLPE power cable samples were tested. Three small holes were drilled in one XLPE cable. A capacitive transducer with center frequency of 802.8 kHz was applied for transmitting receiving signal. For each sample, 48 scans were collected. Based on ultrasound images, we were able to detect these faults in this XLPE material from the peaks of the samples corresponding to the XLPE area by setting a threshold to 0.08 volts. Also, this detection technique was improved by using fourth-order cumulants. / May 2016

This thesis is concerned with the problem of learning comprehensible theories from structured data and covers primarily classification and regression learning. The basic knowledge representation language is set around a polymorphically-typed, higher-order logic. The general setup is closely related to the learning from propositionalized knowledge and learning from interpretations settings in Inductive Logic Programming. Individuals (also called instances) are represented as terms in the logic. A grammar-like construct called a predicate rewrite system is used to define features in the form of predicates that individuals may or may not satisfy. For learning, decision-tree algorithms of various kinds are adopted.¶ The scope of the thesis spans both theory and practice. On the theoretical side, I study in this thesis¶ 1. the representational power of different function classes and relationships between them;¶ 2. the sample complexity of some commonly-used predicate classes, particularly those involving sets and multisets;¶ 3. the computational complexity of various optimization problems associated with learning and algorithms for solving them; and¶ 4. the (efficient) learnability of different function classes in the PAC and agnostic PAC models.¶ On the practical side, the usefulness of the learning system developed is demontrated with applications in two important domains: bioinformatics and intelligent agents. Specifically, the following are covered in this thesis:¶ 1. a solution to a benchmark multiple-instance learning problem and some useful lessons that can be drawn from it;¶ 2. a successful attempt on a knowledge discovery problem in predictive toxicology, one that can serve as another proof-of-concept that real chemical knowledge can be obtained using symbolic learning;¶ 3. a reworking of an exercise in relational reinforcement learning and some new insights and techniques we learned for this interesting problem; and¶ 4. a general approach for personalizing user agents that takes full advantage of symbolic learning.

No description available.

Ice sheet modeling is an important tool both for reconstructing past ice sheets and predicting their future evolution, but is complex and computationally costly. It involves modeling a system including the ice sheet, ice shelves and ice streams, which all have different dynamical behavior. The governing equations are non-linear, and to capture a full glacial cycle more than 100,000 years need to be simulated. To reduce the problem size, approximations of the equations are introduced. The most common approximation, the Shallow Ice Approximation (SIA), works well in the ice bulk but fails in e.g. the modeling of ice streams and the ice sheet/ice shelf coupling. In recent years more accurate models, so-called higher order models, have been constructed to address these problems. However, these models are generally constructed in an ad hoc fashion, lacking rigor. In this thesis, so-called Second Order Shallow Ice Approximation (SOSIA) equations for pressure, vertical shear stress and velocity are implemented into the ice sheet model SICOPOLIS. The SOSIA is a rigorous model derived by Baral in 1999 [3]. The numerical solution for a simple model problem is compared to an analytical solution, and benchmark experiments, comparing the model to other higher order models, are carried out. The numerical and analytical solution agree well, but the results regarding vertical shear stress and velocity differ from other models. It is concluded that there are problems with the model implemented, most likely in the treatment of the relation between stress and strain rate.

Tremor is a neuro degenerative disease causing involuntary musclemovements in human limbs. There are many types of tremor that arecaused due to the damage of nerve cells that surrounds thalamus of thefront brain chamber. It is hard to distinguish or classify the tremors asthere are many reasons behind the formation of specific category, soevery tremor type is named behind its frequency type. Propermedication for the cure by physician is possible only when the disease isidentified.Because of the argument given in the above paragraph, there is a needof a device or a technique to analyze the tremor and for extracting theparameters associated with the signal. These extracted parameters canbe used to classify the tremor for onward identification of the disease.There are various diagnostic and treatment monitoring equipment areavailable for many neuromuscular diseases. This thesis is concernedwith the tremor analysis for the purpose of recognizing certain otherneurological disorders. A recording and analysis system for human’stremor is developed.The analysis was performed based on frequency and amplitudeparameters of the tremor. The Fast Fourier Transform (FFT) and higherorderspectra were used to extract frequency parameters (e.g., peakamplitude, fundamental frequency of tremor, etc). In order to diagnosesubjects’ condition, classification was implemented by statisticalsignificant tests (t‐test).

Vertex shader, one of the core parts in 3D graphics systems, is to speed up the operations of coordinate transformation and lighting in 3D graphics pipeline, and vector ALU is the key part of a vertex shader. This thesis proposes several unified architectures that integrate the floating-point vector arithmetic unit and special function unit in order to share some hardware resource. We propose three different architectures for the design of the unified vector ALU. The first architecture includes a single-instruction-multiple-data (SIMD) vector arithmetic unit, and uses table-based method with first-order approximation to calculate some special functions. The second architecture use higher-order approximation to reduce the table sizes and share the floating-point multipliers in the SIMD vector unit. The proposed third architecture has two copies of hardware that can compute two dot-product operations in parallel and thus increase the throughput of the matrix computation by a factor of two. Furthermore, the two dot-product units can be used to perform the interpolation for special function calculation.

Transverse-integrated nodal di usion methods currently represent the standard in full core neutronic simulation. The primary shortcoming of this approach is the utilization of the quadratic transverse leakage approximation. This approach, although proven to work well for typical LWR problems, is not consistent with the formulation of nodal methods and can cause accuracy and convergence problems. In this work, an improved, consistent quadratic leakage approximation is formulated, which derives from the class of higher-order nodal methods developed some years ago. In this thesis a number of iteration schemes are developed around this consistent quadratic leakage approximation which yields accurate node average results in much improved calculational times. The most promising of these iteration schemes results from utilizing the consistent leakage approximation as a correction method to the standard quadratic leakage approximation. Numerical results are demonstrated on a set of benchmark problems and further applied to realistic reactor problems for particularly the SAFARI-1 reactor operating at Necsa, South Africa. The nal optimal solution strategy is packaged into a standalone module which may be simply coupled to existing nodal di usion codes, illustrated via coupling of the module to the OSCAR-4 code system developed at Necsa and utilized for the calculational support of a number of operating research reactors around the world. / Thesis(PhD (Reactor Science))--North-West University, Potchefstroom Campus, 2013

Transverse-integrated nodal di usion methods currently represent the standard in full core neutronic simulation. The primary shortcoming of this approach is the utilization of the quadratic transverse leakage approximation. This approach, although proven to work well for typical LWR problems, is not consistent with the formulation of nodal methods and can cause accuracy and convergence problems. In this work, an improved, consistent quadratic leakage approximation is formulated, which derives from the class of higher-order nodal methods developed some years ago. In this thesis a number of iteration schemes are developed around this consistent quadratic leakage approximation which yields accurate node average results in much improved calculational times. The most promising of these iteration schemes results from utilizing the consistent leakage approximation as a correction method to the standard quadratic leakage approximation. Numerical results are demonstrated on a set of benchmark problems and further applied to realistic reactor problems for particularly the SAFARI-1 reactor operating at Necsa, South Africa. The nal optimal solution strategy is packaged into a standalone module which may be simply coupled to existing nodal di usion codes, illustrated via coupling of the module to the OSCAR-4 code system developed at Necsa and utilized for the calculational support of a number of operating research reactors around the world. / Thesis(PhD (Reactor Science))--North-West University, Potchefstroom Campus, 2013