Catálogo de publicaciones - libros

Tail Assignment is the problem of assigning specific aircraft to flights, producing a fully operational, robust schedule which fulfills operational constraints, while minimizing a cost function. The costs are usually related to robustness and quality of the solution, e.g. to penalize short times between flights and reward “crew-friendly” connections. Carmen Systems’ Tail Assignment system utilizes a combination of optimization methods and constraint programming to solve this problem. An interesting aspect of the system is that it can solve several fleet types at once, moving towards an integration of the traditional two-step approach of first solving the Fleet Assignment problem and then Tail Assignment. The article intends to explain some of the methods used.

In der Regel wachsen die Transportkosten auf der letzten Meile mit der Enge des Lieferzeitfensters. Dies stellt insbesondere Anbieter vor ein Problem, die Aussagen über einen präzisen Liefertermin bei Bestellung abgeben müssen, obwohl noch nicht alle zu bedienenden Lieferungen bekannt sind.

Im folgenden Beitrag wird das Problem als D-VRSP-TW-UC-DO modelliert. Motiviert durch die raum-zeitliche Analyse der Solomoninstanzen wird der Begriff der Gleichzeitigkeit als wesentlicher Kostentreiber herausgearbeitet. Anschließend wird ein praxistaugliches Verfahren (ORA) vorgestellt, das es ermöglicht, eine hohe Avisgenauigkeit bei Bestellung zu erlangen. SchlieBlich werden typische Ergebnisse des Verfahrens dargestellt und diskutiert.

This contribution addresses the consideration of time windows in the optimization of multi-commodity network flows. For each node, one interval is specified in which the visitation is allowed. Applications in freight flow consolidation let this problem become interesting. An optimization model is proposed and a construction heuristic is presented. For improving the generated solutions, a genetic algorithm framework including several hill climbing procedures for local optimization, is configured.

We present two different approaches to minimize total delay in signalized fixed-time controlled inner city traffic networks. Firstly, we develop a time discrete model where all calculations are done pathwise and vehicles move on “time trajectories” on their routes. Secondly, an idea by , and (GLG) is extended to a continuous, linkwise operating model using “Link Performance Functions” to determine delays. Both models are formulated as mixed-integer linear programs and are compared and evaluated by PTV AG’s simulation tool VISSIM 3.70.

For the stochastic flow shop problem with machines, jobs and operation processing times being independent identically distributed random variables, we consider permutation schedules — those in which each machine processes the jobs in the same order. For fixed and increasing , the asymptotic behavior of an arbitrary permutation schedule length is researched. Let be the makespan of that schedule. Considering the maximum machine load as a lower bound of the makespan, we show that for a wide class of operation processing time distributions the average-case ratio converges to 1.

Thus, an arbitrary permutation flow shop schedule is asymptotically optimal with → ∞, and one can construct such schedule in a linear time. The same result is derived for the stochastic assembly line problem with machines and jobs.

We consider a generic workforce scheduling problem, where employees are characterized by qualifications. Given a set of shifts for each day, we have to determine for each employee his working days as well as a specific shift for each working day. The overall objective is to find a set of feasible schedules with respect to hard and soft restrictions. We propose an integer multi-commodity network flow formulation for the problem under consideration and show how branch-and-price can be used to solve this problem.

We consider a single processor scheduling problem with a common due window assignment. Jobs completed within the due window incur no penalty, while other jobs incur either earliness or tardiness penalties. Boundaries of the due window are decision variables. The objective is to minimize the sum of the total weighted earliness, the total weighted tardiness and due window width penalty. This problem is an extension of the classical Weighted Earliness and Tardiness problem (). We proved that our problem is NP-hard and presented some properties of an optimal solution. To solve the problem we constructed a dynamic programming algorithm and a fully polynomial time approximation scheme. We also presented a polynomial time algorithm for the case with unit job processing times.

Starting with a comprehensive analysis of empirical and theoretical papers on market entry mode choice, as well as by referring to the characteristics of small and medium sized enterprises (SMEs), we develop a new quantitative model that represents SMEs’ choice between joint venture (JV) and wholly owned foreign venture (WOFV). With the help of this model, we deduce some useful propositions for decision makers, both in companies concerned and in economic policy.

This paper introduces a new growing neural network for pattern detection which bears certain resemblances to the growing neural gas network suggested by Pritzke (1995) [2]. However, the algorithm at hand is more parsimonious with respect to the number of parameters to be specified a priori. Thus it is largely autonomous regarding the data-driven construction of the final network topology which unburdens the user significantly. To demonstrate its performance and adaptability the new algorithm is applied to real classification tasks in lifestyle analysis and media usage analysis.

Increasing attention has been devoted in recent years to the firm’s ability to adapt its marketing strategies to a rapidly changing environment. Given that the abundance of news, reports, and announcements found in new electronic environments such as the WWW hampers an extensive manual search, computer-based systems have become important supportive tools for business planning purposes. Several studies investigate the impact of managerial traits on this question, however the potential influence of an inadequate information structure in automatic information-seeking tools is rarely addressed.

In this paper, we examine the effect of the quality of the information structure in automated information-seeking tasks. We use a prototypic system that aims to detect and to evaluate relevant information about financial markets, and systematically contaminate the information structure by index terms referring to an adjacent but different task. Empirical evidence from an experimental evaluation of documents from the Reuters text collection substantiates the relevance of the prior information structure to the automated information search.