Vedran Mornar

Solving timetable scheduling problem using genetic algorithms

A genetic algorithm for solving a timetable scheduling problem is described. The algorithm was tested on small and large instances of the problem. Algorithm performance was significantly enhanced with modification of basic genetic operators. Intelligent operators restrain the creation of new conflicts in the individual and improve the overall algorithm s behavior.

CASE STUDY BASED ON A MULTI-PERIOD MULTI-CRITERIA PRODUCTION PLANNING MODEL

The aim of the described model and procedures is to enable production planning by using multi-criteria linear programming in a manufacturing environment. It is assumed that linearity holds within acceptable tolerances and the problem of choice is present, i.e. the majority of products can be manufactured in a number of technically and financially different processes. The same resources, raw materials, machines, tools, and manpower can be assigned to different processes. The planning objectives include calculating the optimum production mix and achieving the capacity and material balance, while maximizing the contribution and minimizing the duration of the longest resource engagement. Multiple periods are applied to achieve approximated time scheduling. The described software also calculates the direct costs, enables easy locating of infeasibility cause and offers sensitivity analysis. A goal programming variation is introduced for maximum satisfaction of the market requirements. The described procedures are in practical use in a thermoplastics factory.

Application of smart cards in distributed information systems

Since their appearance in 1990, utilization of smart cards has been in constant growth. It is quite evident that any magnetic card can be replaced by a smart card bringing in all the benefits of this technology. The power of contemporary smart cards, which contain an embedded powerful computer, makes their area of application very extensive, especially in the field of information systems

An electronic journal management system

A Web-based information system for management of scientific electronic journals is presented. Journal management requires exchange of information between many participants in the publishing process, including authors, reviewers, editors, and collectors. Each participant in the process plays a different role, and cooperates with other participants. The Web system is implemented using open source solutions on the Linux platform. A three-tiered information system architecture has been used. An object oriented PHP (PHP hypertext preprocessor) is used as a server side scripting language to implement the business logic. A database management system is provided by PostgreSQL, although database connection abstraction is utilized. Apache Web server is used together with SSL encryption mechanism to provide HTML content to end-user. The system is currently in use for editorial support of an international scientific journal and the initial results have been encouraging.

A cutting stock procedure for printed circuit board production

Optimal cutting of raw materials to produce printed circuit boards is a special cutting stock problem with its own particular constraints. This paper describes an integrated system that includes a data base for parts and initial materials, a cutting stock procedure based on a heuristic algorithm, a production planning module which uses linear programming, and a graphical user interface for possible editing of generated plans and manual layout design.

INTRODUCTION OF SAP ERP SYSTEM INTO A HETEROGENEOUS ACADEMIC COMMUNITY

Introduction of a complex ERP system like SAP into a heterogeneous academic environment like the University of Zagreb is far from being a trivial task. The University comprises more than 30 constituents, called faculties or academies, geographically dispersed, with long and specific traditions. Financing according to the lump sum principle, enforced in Croatia as a side effect of the in Europe obligatory and omnipresent Bologna process, requires a unified view on the educational institutions in order to provide a more just and appropriate financing scheme than the current one. After the experience with own development to support educational tasks and student administration, for standard financial and administration tasks SAP has been chosen as the most appropriate platform. The developer was selected after public bidding and the authors’ institution was chosen for the pilot project. The authors were playing principal roles in the process of successful deployment and still expect to offer their expertise for implementation in the rest of the University. However, serious risks stemming from lack of motivation by some constituents are present.

A software infrastructure for distributed computing based on DCOM

To simplify the usage of a computer network as a virtual parallel machine, as well as to introduce the opportunity of distributed parallel computing to Visual Basic users, a DCOM-based software infrastructure is proposed in this article. This infrastructure is based on a master-slave model. It sends a broadcast call over a subnet as an invitation for cooperation, automatically distributes new program versions, and manages the unfinished jobs on the slave machines. As a test application, a simple solution of the traveling salesman problem is implemented.

General object-oriented framework for iterative optimization algorithms

It is usually impossible to exactly solve hard optimization problems. One is thus directed to iterative algorithms. In the implementation of these iterative algorithms, some common characteristics can be observed, which can be generalized in an object-oriented framework. This can significantly reduce the time needed for implementation of an iterative algorithm. This paper presents a class hierarchy, which can be used as an object-oriented framework for the solutions of iterative problems. Utilizing the options available in object-oriented languages, it separates the class interface from the implementation.

General Object-oriented Framework for Iterative Optimization Algorithms

It is usually impossible to exactly solve the hard optimization problems. One is thus directed to iterative algorithms. In implementation of these iterative algorithms, some common characteristics can be observed, which can be generalized in an object-oriented framework. This can significantly reduce the time needed for implementation of an iterative algorithm.