Thesis (Ph. D.)--University OF MICHIGAN.

Thesis--PlanB (M.S.)--University of Wisconsin--Stout, 1998. / Field project. Includes bibliographical references.

Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2008. / Source: Dissertation Abstracts International, Volume: 69-11, Section: B, page: 6920. Adviser: Sheldon Jacobson. Includes bibliographical references (leaves 120-128) Available on microfilm from Pro Quest Information and Learning.

DISSERTATION (PH.D.)--THE UNIVERSITY OF MICHIGAN

Thesis PlanB (M.S.)--University of Wisconsin--Stout, 2006. / Includes bibliographical references.

Master of Science / Department of Industrial & Manufacturing Systems Engineering / Todd W. Easton / Integer programming is a useful tool for modeling and optimizing real world problems. Unfortunately, the time required to solve integer programs is exponential, so real world problems often cannot be solved. The knapsack problem is a form of integer programming that has only one constraint and can be used to strengthen cutting planes for general integer programs. These facts make finding new classes of facet-defining inequalities for the knapsack problem an extremely important area of research. This thesis introduces three set inequalities (TSI) and an algorithm for finding them. Theoretical results show that these inequalities will be of dimension at least 2, and can be facet defining for the knapsack problem under certain conditions. Another interesting aspect of these inequalities is that TSIs are some of the first facet-defining inequalities for knapsack problems that are not based on covers. Furthermore, the algorithm can be extended to generate multiple inequalities by implementing an enumerative branching tree. A small computational study is provided to demonstrate the effectiveness of three set inequalities. The study compares running times of solving integer programs with and without three set inequalities, and is inconclusive.

In this thesis, we develop approaches for carrying out inference and model-based experimental design, under both internal and external sources of uncertainty. Specifically, in Chapter 1, we develop a stochastic growth model for the carbon-based super material, Graphene, and propose approaches for relating controllable experimental factors to the underlying growth mechanism. In Chapter 2 we develop a unified framework for carrying out response surface optimization when the input factors are noisy, and in Chapter 3, we explore the problem of designing optimal experiments, under the extra uncertainty generated by noisy inputs. Internal noise, a term used to describe the phenomenon of noisy inputs, is found to adversely affect optimization and model-based optimal designs. We show that accounting for this internal noise during the design and modeling stages significantly improve inference. In particular, we develop a modified optimality criterion for generating optimal experimental data, and show improvements in subsequent inference based on that data. In Chapter 4, a missing data perspective is used to improve inference on deformations along the profile of 3D printed products. We show that these deformations depend on missing angles, which can be used to infer global and local deformation patterns. We use the inferred deformation model to design compensation plans for minimizing deformations on future printed objects. / Statistics

The focus of this thesis is the development of methodologies for the analysis of two important issues in a manufacturing environment: human boredom and machine utilization. In particular, this study proposes a mathematical programming model for job rotation to cope with worker's boredom in a manufacturing cell. One of the important side effects of task rotation is the impact of rotation interval on workers' skill learning and forgetting which may cause productivity losses. For this reason, the proposed model simultaneously incorporates worker's skill and boredom variations. The developed formulation is a mixed integer nonlinear mathematical programming model. A linear version of the model is also presented by assuming that both worker's skill and boredom change linearly over time. Furthermore, given the complexity and uncertainty involving human emotions, a novel approach based on the state-of-the-art Bayesian Networks is also presented to measure and to predict human boredom at work. First, the static Bayesian Network is introduced to capture the static aspect of boredom modeling. The static boredom model is subsequently extended based on dynamic Bayesian Networks to account for temporal aspect of boredom modeling. The dynamic boredom model allows integrating boredom evidences spatially and temporally, thus providing a more robust and accurate boredom inference. The proposed boredom model is validated using case based scenarios. Moreover, a metaheuristic approach based on several well-known search algorithms is also presented to solve the nonlinear version of the job rotation model. The proposed algorithm integrates several ingredients including a simulated annealing module, three types of memory, an evolutionary operator, and a blockage removal feature in a generic framework. The main characteristic of the proposed model is the use of both positive and negative memory to intensify the search around good solutions as well as an evolution-based diversification approach. This generic metaheuristic can be tailored to solve a variety of hard optimization problems. Another important issue in a manufacturing environment is machine utilization which is largely affected by scheduling decisions. Though various scheduling problems have been investigated for several decades, the lack of efficient solution methods is still hindering the machine utilization and hence the manufacturing productivity. For this reason, the two most common scheduling problems, job shop scheduling and flow shop scheduling problems are investigated in this thesis. The proposed new metaheuristic is used to solve the scheduling problems. This study presents a methodology to explicitly deal with human's boredom and skill variations at work. It also introduces a probabilistic model to quantitatively measure and predict human boredom. Finally, it contributes to the development of computational intelligence by introducing a generic framework of a new metaheuristic to solve a class of combinatorial optimization problems.

Among various forms of disability, blindness has been regarded as one of the most devastating that can strike people of all ages and significantly affects the victim's life. In spite of the significant efforts made by the research community and industry over the past decades on developing electrical travel aids (ETAs) for the blind, their acceptance rate by the blind community has remained surprisingly low, due to a variety of reasons such as bulky size, heavy weight, high cost, lack of user-friendliness and incompatibility with daily walking situations. The majority of blind people nowadays are still using the conventional long cane as the primary travel aid, which is a simple walking stick that does not provide any protection against overhanging obstacles, such as tree branches, signposts, or improperly slung wires. As a result, head injuries of blind cane users often occur in daily walking situations. This research was motivated by the critical need for a new type of spatial obstacle localization technique that is particularly suited for integration into the long cane structure subject to both swing and tapping motion during its operation and that provides an orientation and travel aid for the blind. The research objective were three-fold: (1) to establish a theoretical framework for ultrasound-based ranging and spatial obstacle localization from a moving reference frame, represented by the long cane, using the time-of-flight (TOF) approach, (2) to design, analyze, implement, and prototype a miniaturized and self-contained ultrasonic transceiver module with wireless data transmission capability, integrate it into the shaft of a modified long cane, and develop a wireless data receiver with voice output display, and (3) to experimentally evaluate the sensor-integrated long cane with respect to sensor directionality, placement, measurement accuracy, repeatability, electromechanical coupling, and effect of mechanical vibration. A mechatronic approach was taken for the system-level design of the long cane based ultrasound measurement system. Sensor module coordination and ultrasound echo signal processing were controlled in real-time by a microcontroller, which was an integral part of the cane-embedded electronics. A dual-frequency technique has been proposed that introduced information redundancy to improve measurement quality. Quantitative analysis on the systematic and random errors of the embedded system was provided. In addition to providing information on the distance and height of the obstacles in front of the user, the system further incorporated self-test ability.

This research is designed to study the effects of highly developed information technologies and logistic strategies on freight transportation. A simulation model called TTMNet is formulated as a multi-level product supply chain system that integrates the financial, informational, logistic, and physical aspects of transportation networks, to address freight transportation problems within a much broader decisionmaking and policy sensitive environment. It simulates freight movement in a regional supply chain, given real-time information, to help understand the mechanisms or principles followed by system operation, freight flow patterns, and evolution and interaction of prices and costs across the networks over time. TTMNet is implemented using micro-simulation techniques and GIS tools, and several simulators are involved, including a dynamic freight traffic simulator, a supply chain decision making simulator, and a real-time information simulator. The construction of a prototype dynamic freight traffic simulation model called DyFTS is the focus of this research. DyFTS is designed as a discrete event simulation system, highly adaptable to more comprehensive transportation simulation models. Various decision-making processes are formulated within DyFTS, such as goods-to-vehicle assignment, departure time choice and pre-trip routing, and en-route vehicle redirection. Descriptive real-time traffic information is simulated to study its influence on freight traffic operations. A knowledge-based learning process is established to refine the perceptions of decision-makers to the transportation network based on past experience. The inclusion of the simulation of regional ITS system makes the DyFTS a powerful tool to evaluate the information. A preliminary study is conducted herein to construct an integrated logistic and transportation system, a simplified version of TTMNet. It simulates enterprise-wide freight movements in a comprehensive product supply chain system that integrates the logistic, informational, and physical aspects of transportation networks. The resulting simulation model developed in this dissertation can be applied widely in freight transportation industry, such as, the study of commercial vehicle operations with and without real-time information, freight transportation of different delivery time requirements and different fleet configurations, and freight traffic patterns by product demand. It may also help direct or justify the development of real-time surface transportation information systems.