Akademisk avhandling--Uppsala. / Extra t.p., with thesis statement, inserted. Bibliography: p. [224]-238.

Thesis (Ph. D.)--University of Wisconsin, 1967. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliography.

The master thesis is dealing with Internet host localization methods, more exactly with determining geographical position of the unknown Internet host connected to the network using RTT delay measuring. The first part is dealing with description of RTT delays that may occur in the network and tools for their measurement. The next is part of thesis is devoted to description of two kinds of localization methods. Ones that are using existing data to determine the position of Internet host also called passive methods, and others that are using RTT delay measurement, also called active methods. The main part is focused on GeoWeight method which is based on geographical localization estimation of Internet host. It is based on RTT delay measurement using the principles of CBG method, enhanced by introduction of the theory of weights according to the probability of the target Internet host. The last part is describing the application that was made to determine the geographic localization of the target Internet host using GeoWeight method. The application was afterwards tested by measuring RTT delay in PlanetLab experimental network. At the end the final measured results were compared with other localization methods (CBG, Octant, SOI, GeoIP).

No description available.

No description available.

Artificial Neural networks have found many applications in various fields such as function approximation, time-series prediction, and adaptive control. The performance of a neural network depends on many factors, including the network structure, the selection of activation functions, the learning rate of the training algorithm, and initial synaptic weight values, etc. Genetic algorithms are inspired by Charles Darwin’s theory of natural selection (“survival of the fittest”). They are heuristic search techniques that are based on aspects of natural evolution, such as inheritance, mutation, selection, and crossover. This research utilizes a genetic algorithm to optimize multi-layer feedforward neural network performance and structure. The goal is to minimize both the function of output errors and the number of connections of network. The algorithm is modeled in C++ and tested on several different data sets. Computer simulation results show that the proposed algorithm can successfully determine the appropriate network size for optimal performance. This research also includes studies of the effects of population size, crossover type, probability of bit mutation, and the error scaling factor.

No description available.

The purpose of this paper was to design a simplified wind tunnel force balance for use in elementary aerodynamics/wind tunnel laboratory courses. The applied loads and moments were determined to be as follows: The design lift force is ± 20 lbs, the design drag force is ± 10 pounds, the design pitching moment is ± 20 inch-pounds. The force balance output accuracy was arbitrarily set at ± 5% since this would be sufficient for preliminary student work. The results of this work are as follows: The force balance is fabricated from a single bar of 2024 aluminum, machines and bent into an "L" shape, the applied forces and moments are sensed by strain gages bonded to machined surfaces on the bar, the output of the strain indicator equipment is transformed into uncorrected forces and moments by a system of three equations, the uncorrected forces and moments are transformed into true forces and moments by a system of force balance interaction equations, the design values for lift, drag, and pitching moment remain the same as originally proposed, the output error in lift is determined to be ± 3.5%, the output error in drag and pitching moment are determined to be ± 10%. The prototype has been in use for several months and its operation has been completely satisfactory.

A study of the relation between mechanical properties and molecular weight distribution of polyvinyl alcohol was carried out. The polyvinyl alcohol was fractionated with respect to degree of polymerization (D.P.) by precipitation of the fractions from 2% solution in water with a n-propyl alcohol-water nonsolvent. The initial small scale procedure was modified so as to accumulate fractions in sufficient quantity for the testing of mechanical properties and for the preparation of blends. The fractions were refractionated until further refractionation gave no further change in D.P. The homogeneous fractions were cast into films using water as the solvent. These films were then used in determining mechanical properties. A series of three blends, all with normal distributions and the same D.P. at the maximum of the differential distribution curve but with variable heights at the maximum D.P., were prepared and the mechanical properties were determined. A second series of six blends, all with constant height and constant D.P. at the maximum, but all being skewed from the normal distribution, were prepared and their mechanical properties were determined. A correlation of tensile strength with the shape of the distribution curve was found to have the form: T = (8200-l0^{6 .87-.00476P_m}) + (2.786 x 10⁶H^4.767-1000) + (1450-10.4B), Where T is tensile strength; P_m is D.P. at maximum of the differential distribution curve; H is height at the maximum; B is area skew. The second term in brackets is dropped when H is greater than 0.190 and the entire last term is dropped when B = 0. A graphical correlation of mechanical properties with total percentage of low D.P. material gives fairly good results for polyvinyl alcohol and, using the data of other investigators, for cellulose nitrate, cellulose acetate, and polyvinyl acetate. / Ph. D.

M.S.