No description available.

No description available.

No description available.

Hydro-Quebec utilizes a one-dimensional hydrological model called H2RM to simulate the evolution of flow on harnessed rivers. The model produces reliable results but its application is restricted to so-called 'hydrological domains', that define the flow routing pattern in-between hydroelectric facilities. When many power stations are present on a river, each hydrological domain must be treated separately, then linked to the next one. No information is provided on storage in head reservoirs nor electric production from power stations. / A computer program (shell) has been developed to simulate a complete hydropower system comprising a number of head reservoirs, hydrological domains and power stations. In this program, head reservoir behavior is reproduced by computing a mass balance, flow routing within a hydrological domain is simulated using the H2RM model, and production from each power station is estimated through an optimization procedure. / The shell program can be used as a decision support tool by allowing the comparison between various water management schemes and by displaying stage and discharge at any point of the hydropower system, storage in head reservoirs, and optimal power output from the turbines for the head and flow conditions prevailing at each power station. Examples of application are provided.

<p> The need for performance measurement and evaluation in supply chain management is well recognized in the literature. The timeliness of delivery is a key concern to customers and numerous empirical studies have documented the importance that on time delivery plays in the operation of the supply chain. Supply chain delivery performance models are based on the concept of the delivery window, which is defined as the difference between the earliest acceptable delivery date and the latest acceptable delivery date. In the dissertation supply chain delivery performance is evaluated from a supplier's and a buyer's prospective. </p><p> The research introduces a concept of the optimal positioning of the delivery window in a serial supply chain. Optimally positioning the delivery window minimizes the expected penalty cost due to early and late delivery. The conditions for the optimal position of the delivery window are derived for the general form of a delivery time distribution. </p><p> The research herein addresses strategies of delivery performance improvement using a cost based delivery performance model and evaluates the effect of different parameters on the expected penalty cost. An understanding of these analytical properties provides a strong foundation for identifying and integrating strategies to improve delivery performance. </p><p> Furthermore, we investigate how the timeliness of the delivery will affect the inventory cost structure of a buyer in a two stage supply chain. From the perspective of the buyer, untimely delivery can impact inventory holding and stockout costs. We formulate the supply chain delivery window problem as a stochastic model with three possible delivery outcomes (early, on time, and late delivery) and integrate this feature with an inventory model with two levels of storage (owned warehouse and rented warehouse). </p><p> This comparison and supporting analysis bridges existing gaps found in the literature and contributes to linking and coordinating the delivery and inventory sub processes within supply chains. Theoretical and managerial implications of the findings are discussed.</p>

<p> Linear approaches for multivariate data analysis are popular due to their lower complexity, reduced computational time, and easier interpretation. In many cases, linear approaches produce adequate results; however, non-linear methods may generate more robust transformations, features, and decision boundaries. Of course, these non-linear methods present their own unique challenges that often inhibit their use. </p><p> In this research, improvements to existing non-linear techniques are investigated for the purposes of providing better, timely class separation and improved anomaly detection on various multivariate datasets, culminating in application to anomaly detection in hyperspectral imagery. Primarily, kernel-based methods are investigated, with some consideration towards other methods. Improvements to existing linear-based algorithms are also explored. Here, it is assumed that classes in the data have minimal overlap in the originating space or can be made to have minimal overlap in a transformed space, and that class information is unknown <i>a priori.</i> Further, improvements are demonstrated for global anomaly detection on a variety of hyperspectral imagery, utilizing fusion of spatial and spectral information, factor analysis, clustering, and screening. Additionally, new approaches for n-dimensional visualization of data and decision boundaries are developed.</p>

No description available.

<p> This dissertation presents research that focuses on efficient allocation of defense resources to minimize the damage inflicted on a spatially distributed physical network such as a pipeline, water system, or power distribution system from an attack by an active adversary. The allocation methodology recognizes the fundamental difference between preparing for natural disasters such as hurricanes, earthquakes, or even accidental systems failures and the problem of allocating resources to defend against an opponent who is aware of and anticipating the defender's efforts to mitigate the threat. </p><p> Conceptualizing the problem as a Stackelberg "leader-follower" game, the defender first places his assets to defend key areas of the network, and the attacker then seeks to inflict the maximum damage possible within the constraints of resources and network structure. The approach is to utilize a combination of integer programming and agent-based modeling to allocate the defensive resources. The criticality of arcs in the network is estimated by a deterministic network interdiction formulation, a maximum-flow linear program (LP), or a combination of both of these methods, which then inform an evolutionary agent-based simulation. The evolutionary agent-based simulation is used to determine the allocation of resources for attackers and defenders that results in evolutionarily stable strategies in which actions by either side alone cannot increase their share of victories. </p><p> These techniques are demonstrated on several example networks using several different methods of evaluating the value of the nodes and comparing the evolutionary agent-based results to a more traditional, Probabilistic Risk Analysis (PRA) approach. The results show that the agent-based allocation approach results in a greater percentage of defender victories than does the PRA-based allocation approach.</p>

The problem of vertex coloring holds an important place in engineering as it models situations in which a number of shared resources must be minimized and distributed across the sub-components of a system. The objective is to ensure a valid and cost effective implementation of the overall system. The problem surfaces in a great number of applications, many of which fall within the area of digital systems design. However, despite its wide range of applications, vertex coloring remains one of the most complex optimization problems known and to this day no efficient method has been shown to provide optimal answers in all instances of the general case. / This dissertation explores characteristics of optimality of vertex colorings, bounds on the chromatic number and coloring heuristics. The first few characteristics are based around the fundamental and residual nodes of an optimal coloring. An examination of the subset of fundamental nodes will reveal necessary subgraph properties for an optimal coloring and its graph; one of which is based on Kempe chains. In turn, this leads to a bound relating the chromatic number and the number of odd cycles in a graph. / Subsequently, a continuous variable formulation of the vertex coloring problem is presented along with an analysis of its solution space. The characterization of the space illustrates the problem's complexity and the nature of its local minima relates to the Gallai-Roy theorem. The results given will have algorithmic significance since their proofs are constructive. / The WWI pair of vertex coloring heuristics is then disclosed. The algorithms are based on successive compressions of pairs of non-adjacent nodes each reducing the problem instance by one node until a complete graph is obtained. The criteria for selecting pairs of nodes concentrate on the affinity and conflict values calculated from structural properties of the graphs. The heuristics are justified by upper bounds on the chromatic number and approximation arguments. It is demonstrated that some compressions preserve the chromatic number or the maximal clique size of a graph, thus resulting into some identifiably optimal selections by the algorithms. Ultimately, this leads to the characterization of a class of perfect graphs. Finally, a set of benchmarks for the WWI algorithms on random graphs and k-colorable random graphs is given and favorable comparisons are offered with existing algorithms.

<p> A local utility company processes a variety of jobs each day including meter reading, service shut-offs, emergency response, and customer service work. For the Company, a specific workflow begins with automated meter-reading (AMR) and ends with collections/service shut-offs (CSOs) for accounts with excessively late payments (AMR-CSO workflow). There are considerable and systemic sources of variability in both the workload and resource demands of the AMR-CSO workflow including order arrival, order release schedules, order batch-sizing and maintenance scheduling. This project draws on theory from the job-shop problem to explore possible means to mitigate this variability. We hypothesized that controlling various forms of input variability would lead to reduced downstream workload variability. Using discrete event simulation we tested a variety of measures to reduce input variability in the workflow. Consistent with other literature we find that various workload control tactics have limited impact on output measures and system performance. However, we found that system is much more sensitive to resource capacity variability. One input control tactic we call Targeted Release allowed us to reduce Company capacity variability which suggested significantly improved outcomes. These initial results are promising for both the Company and for future investigation of tactics to mitigate resource capacity variability.</p>