A preliminary system design and human factors analysis of a home computer workstation has been presented. Systems are proposed for families modifying existing furniture, as well as those purchasing new workstations. The systems are designed to meet standard human factors criteria to minimize potential injuries and discomfort. In both cases, the design assures enough flexibility to suit all members of the family. Environmental factors such as noise and light are discussed, as well as personal preference, safety, and budget. A description of the workspace requirements, the human factors criteria, and the cost analysis included. / Master of Science

This research studies the process of acquiring knowledge from experts; that is, studies knowledge-acquisition methods to acquire expert knowledge. Forty subjects used a machine-aided knowledge-acquisition tool to model a word processing task. By using the tool, the subjects developed models that were on average 72.8% accurate with a baseline model of the task and 88.5 % consistent among themselves. This research makes four contributions: 1) a complete review of thirty-one knowledge-acquisition methods from manual to machine learning, 2) an evaluation methodology and metrics to evaluate knowledge-acquisition methods, 3) an evaluation of an automated knowledge-acquisition tool called Cognitive Analysis Tool (CAT) developed for this research, and 4) suggested improvements to the current version of the tool. This research describes, develops a taxonomy of, and evaluates thirty-one knowledge-acquisition methods to determine which method matches a defined set of criteria A method is chosen, extended, and automated in the form of a machine-aided knowledge-acquisition tool. The method is chosen based on five criteria including a connection between the chosen, method and the information processing model of problem solving as defined by Newell and Simon (1972). This research evaluates the performance of the tool in terms of the accuracy and consistency of the knowledge bases generated by using the tool. A baseline is derived from this study to which other knowledge-acquisition tools' performance can be compared. The evaluation methodology and metrics developed in this research can be used to evaluate other knowledge-acquisition tools. From this research, four groups of changes to the automated knowledge-acquisition tool are suggested to improve the usability and performance of the tool. The changes are suggested for the user interface and the modes of operation of the tool. / Master of Science

A computationally efficient analysis approach is developed to predict buckling load of geodesically stiffened composite panels under in-plane loads. The analysis procedure accounts for the contribution of the in-plane extensional and out-of-plane bending stiffnesses of the stiffeners through the use of Lagrange multipliers in an energy method solution. The analysis is used to isolate the effect of various stiffener deformation modes on the buckling load and skin deformation patterns of geodesically stiffened panels under various load combinations. The analysis routines are then coupled with the numerical optimizer ADS to create a package for the design of minimum-mass stiffened panels, subject to constraints on buckling of the panel assembly and material strength failure. Material failure in the skin and stiffeners are estimated using a maximum strain criterion. The design variables that can be used for optimization include thickness of the skin laminate, stiffener thickness and height, and positions of straight stiffeners. Applied loads are uniaxial compression, pure shear, and combined compression-shear. / Master of Science

All aircraft must meet controllability requirements to be certified for commercial use or adopted by the mi1itary. Aircraft maneuverability is often 1imited by control authority. Thus, it is essential for designers to evaluate a candidate concept's control authority early in the conceptual design phase. In this thesis, a methodology for rapid control power evaluation of preliminary design configurations against requirements at the key flight conditions is established. First, a collection of critical flight conditions to be considered using this methodology is identified. To examine a variety of aircraft configurations and accelerate the process of estimating stability and control derivatives, a FORTRAN program using the Vortex-Lattice Method was written to estimate subsonic, low angle-of-attack aerodynamics. Then, a spreadsheet processes the aerodynamic data to check whether the design configuration possesses adequate control power to satisfy the requirements of the critical flight conditions. / Master of Science

Gray leaf spot (GLS) of corn (Zea mays L.), caused by the fungus Cercospora zeaemaydis (CZM) (Tehon and Daniels) has increased in incidence and severity with increasing use of no-tillage and continuous corn practices. This disease can be yield limiting. Corn hybrids were evaluated under natural disease pressure for three years (1989, 90, and 91) at two locations (Montgomery and Wythe Co., VA). Yield losses ranged from 2127.4 kg/ha (Wythe Co., 1991) to 4242.2 kg/ha (Wythe Co., 1990). It was estimated that 77% of the variability in yield was due to GLS. Fungicides were evaluated for the control of GLS over three years on a susceptible hybrid, Pioneer Brand 3320. All fungicides, with the exception of mancozeb, provided significant control over nontreated check in all years. Benomyl, propiconazole and terbutrazole were the most effective fungicides. As much as 93% of the variablilty in yield was attributed to blighting. Reduction in blighting also increased the kernel weight. The toxin, cercosporin, produced by CZM was evaluated for its ability to elicit differential responses in corn germplasm by three methods, ie., vein inoculation, root, and shoot uptake. No consistant differential reponses were found with vein inoculation, but 31-day old plants were significantly more sensitive to the toxin than 21-day old plants, as measured by lesion width. Root and shoot uptake of the toxin by inbred germplasm produced lesions that resembled those produced by CZM in the field. Microscopic, yellow fluorescing crystals were found associated with necrotic tissue from toxin-treated inbreds. Significantly more injury occurred to toxin-treated inbreds exposed to light than to darkness. By chromatographic analysis, 407.1-1076.7 ng of toxin/g of tissue was recovered from leaf lesion extracts of plants exposed to light. Five inbreds (B73, H99, Va59, NC250a, and NC264) showed consistent and differential responses to the toxin. H99 and NC250a showed differential responses to the same concentration of toxin, thus suggesting that some germplasm are more sensitive to the toxin than others. Tests using the toxin as a means to identify resistant germplasm did not provide reliable predictions of germplasm response to CZM in the field. / Master of Science

Brand loyalty measurement is important because of the extensive use of branding in the lodging industry during recent years. As the industry has matured, segmentation strategies have been adopted by lodging companies. Brands were supposed to protect market share,build brand loyalty, and differentiate the product (Withiam 1985). / Master of Science

Two new power conversion techniques, the constant-frequency zero-voltage-switching multiresonant-converter (CF ZVS-MRC) technique and the zero-voltage-switching technique that uses the magnetizing inductance of the power transformer as a resonant element {ZVS {L_M)) are proposed. analyzed, and evaluated for high-frequency applications. In addition, a novel design optimization approach for resonant type converters is introduced. Complete dc analysis of CF forward and half-bridge {HB) ZVS-MRCs are given, and the dc voltage-conversion-ratio characteristics for each of these two converters are derived. Graphic design procedures that maximize the efficiency and minimize current and voltage stresses are established. The design guidelines are verified on a 50 W CF forward ZVS-MRC operating with a switching frequency above 2 MHz, and on a 100 W HB ZVS-MRc operating with a switching frequency of 750 kHz. The ZVS (LM) technique is developed to eliminate the need for a large, inefficient external resonant inductor in ZVS resonant converters. This new family of isolated converters can operate with zero-voltage-switching of the primary active switches only (quasi-resonant (QR) operation) or with soft-switching of all semiconductor devices (multi-resonant (MR) operation). Furthermore, variable and constant frequency operation of all topologies in this new family of dc/dc converters are possible. A complete dc analysis of the HB ZVS-MRC (L_M) is given, and the dc voltage-conversion-ratio characteristics are derived. Design guidelines are defined using the same graphic method employed in the design of CF ZVS-MRCs. Constant frequency implementation of the HB ZVS-MRC (L_M) using controllable saturable inductors is also proposed. Finally, a novel approach to evaluate and design resonant converters based on the minimization of reactive power is developed. / Ph. D.

A method for the incorporation of continuum modeling in the optimization of large discrete structures is presented. The use of a continuum model facilitates decomposition of optimization problems and augments the scope and applicability of the multilevel decomposition method. This new concept is demonstrated by the optimization of slender, multi-bay, beam-like trusses with large numbers of members. An algorithm for the continuum model optimization of the truss is developed and tested against a traditional algorithm that might be used to solve the problem. Data are presented that reflect the advantages of the continuum model method over the traditional in the areas of computational efficiency and robustness. Additionally, design results for the beam-like truss are presented. / Master of Science

This dissertation concentrates on the three principal problems facing experimentalists during their attempts to identify and control lightly damped, large space structures (LSS). The problems are low damping, high modal density, and low natural frequencies of oscillation. They present a blend of difficulties which lead the experimentalist to turn to multiple-input multiple-output (MIMO) identification techniques and high performance compensators. Presented here are two MIMO modal identification techniques: Polyreference, and the Eigensystem Realization Algorithm, as well as two types of compensator-based controllers: Linear Quadratic Gaussian, and Independent Modal Space Control. The various techniques are described in the context of controlling lightly damped LSS. Because the research in this dissertation is primarily applications oriented, problems which experimentalists encounter in the laboratory are addressed as well as the performance of the different identification and control techniques on the test articles. Polyreference and ERA are both shown to perform very well in identifying modal frequencies while overestimating model damping ratios. Simulations show that high modal density combined with noisy data results in standard deviations that increase linearly with respect to mode separation. Pseudo- Inverse IMSC is shown to be robust with respect to system uncertainties. Block Independent Control is shown to possess minor coupling between blocks and provides a powerful control approach to overcome actuator bandwidth limitations. / Ph. D.

This dissertation deals with the application of object-oriented database techniques to the problem of storage and access of information retrieval (IR) data, especially data that can be organized as a graph, such as a thesaurus encoded in semantic networks, or hypertext collections. Even traditional IR models can use graph representations of documents and concepts. This dissertation reports the development of an object-oriented model called the LEND (Large External object-oriented Network Database) model. This model contains not only features found in a typical object-oriented model but also those that specifically are designed for graph-structured data. A query language is provided facilitating the specification of graph-oriented queries. A prototype LEND system has been implemented to test the model on realistic graph-structured data. It adopts an open system architecture and design, and is easily extensible, like the LEND model itself. The research result of suitable data structures and algorithms (a class of minimal perfect hashing functions) for the efficient implementation of the LEND model is also reported. These data structures and algorithms enable retrieval of a node or a set of nodes in an optimal fashion. Placement of a large graph on a disk is studied as well. The method developed permits efficient traversal of graphs. / Ph. D.