Production Systems and Information Engineering

Adaptive Approaches to Distributed Resource Allocation

Balázs Csanád Csáji — 2009-12-30

The problem of allocating scarce, reusable resources over time to interconnected tasks in uncertain and changing environments, in order to optimize a performance measure, arises in many real-world domains. The paper examines several recent distributed optimization approaches to this problem and compares their properties, such as the guarantees of finding (near-)optimal solutions, their robustness against disturbances or against imprecise, uncertain models, with a special emphasis on adaptive capabilities. The paper also presents a reinforcement learning based distributed resource control system and argues that this method represent one of the most promising approaches to handling resource allocation problems in the presence of uncertainties.

Semantic Representation of Natural Language with Extended Conceptual Graph

Erika Baksa-Varga — 2009-12-30

We intend to create an intelligent agent that is able to detect the objects and relations in its environment, and based on the received information is able to build up an internal knowledge base on the basis of which linguistic expressions can be formulated. The present investigations focus on modeling the agent’s ability to assign semantic representations to its observations. For this purpose we have developed a graphical conceptual modeling language. In this article we examine the model’s expressive power, that is to what extent it is able to model natural language semantics. The analysis is performed in comparison with predicate logic. In our view, first-order logic does not provide sufficient means for the translation of natural language expressions and we will argue that also higher-order logic needs some extensions.

Estimation of Misclassification Error Using Bayesian Classifiers

Péter Barabás — 2009-12-30

Bayesian classifiers provide relatively good performance compared with other more complex algorithms. Misclassification ratio is very low for trained samples, but in the case of outliers the misclassification error may increase significantly. The usage of ‘summation hack’ method in Bayesian classification algorithm can reduce the misclassifications rate for untrained samples. The goal of this paper is to analyze the applicability of summation hack in Bayesian classifiers in general.

Developing Models Based on Real Environments

Tamás Bákai — 2009-12-30

All of the applications of information engineering are based on a correct model of the environment the applications represent. Unlike artificial environments, real environments cannot be modelled correctly. Firstly because the behaviour of the real environments also depends on the behaviour of other environments also. These relations cannot be revealed at the time the model is constructed. Therefore the boundaries of a real environment cannot be defined correctly. The only thing the model-designer can do is to define a model describing the behaviour of the environment to be modelled with no inconsistencies at the time the model is constructed. Therefore each model based on real environments needs to be redesigned continuously as the time passes to provide their consistency. It is only feasible using adaptive knowledge-intensive modelling tools. This paper shows a new concept for modelling real environment.

Analysis of Dynamometer Signals of Interrupted Cutting by Time-Frequency Localization Methods

Tamás Kovács — 2009-12-30

In this work the force sensor signals of an interrupted metal cutting process were analyzed by continuous wavelets (CWT) and Hilbert-Huang Transformation (HHT). The main purpose was to characterize the signal through the typical behavior of its dominant frequencies and, at the same time, to compare the performances of the two frequency localizing methods applied. It was found that the low dominant frequencies are approximately constant while the values of the high dominant frequencies (above one kHz) show strong fluctuations. Here we show that the fluctuations revealed by the CWT method are caused partly by numeric effects of the method itself and partly by real variations in the frequency values. These variations are responsible for the non-stationary behavior of the signal. Regarding the methods applied, it was found that the wavelet analysis method is capable of tracing fast varying frequencies of the signal with values close to each other by the proper choice of the central frequency parameter of the mother wavelet. In comparison with the HHT, the wavelet method proved much more robust in this case.

Solving Multi-Objective Production Scheduling Problems Using a New Approach

Gyula Kulcsár — 2009-12-30

Most companies use a proactive approach to schedule production orders and jobs at shop floor level. To make a near-optimal schedule for a shop with different types of machines and many operations is a very important but complicated task because of the very large number of alternative solution in the searching space. Advanced scheduling models, good heuristics and fast improving algorithms are required to satisfy constraints and to optimize production performances. The aim of the paper is to outline a new modelling and solving approach connected to discrete production scheduling and rescheduling.

An Extended Newsvendor Model for Solving Capacity Constraint Problems in a Multi-Item, Multi-Period Enviroment

Péter Mileff — 2009-12-30

In the past few years, the effective management of inventory control problems has become an increasingly critical issue for supplier companies. In this paper, on the basis of the needs a major Hungarian mass production company, we present an extension of an analytical inventory control model considering the condition of global capacity constraint. In a previous paper [6] we elaborated a model regarding the one customer - one supplier relation. Our aim is to determine an optimal holding-production policy of the supplier, which makes a cost-optimum stockpiling policy possible for an arbitrary long production time. We intend to show that, on the basis of our former results, the global capacity constraint satisfying policy can be determined with a new heuristic method.

OpTol: Spatial Tolerance Analysis Application

Károly Nehéz — 2009-12-30

The analysis of manufacturing and assembly dimension chains is indispensable for performing up-to-date part manufacturing and assembly. It will both reduce the manufacturing and assembly costs and will result in a wellgrounded body of knowledge and improved level of design. The paper deals with mathematical models for calculating spatial dimension chains and introduces the OpTol Tolerance Calculator software. This application is capable of calculating planar (2D) and spatial (3D) dimensional chains by using the classical worst-case and statistical methods, as well as applying the modern six-sigma tolerancing method. OpTol system also contains a CAD module in order to support engineers in analysing their existing assemblies.

Quality Assessment of Motion Picture Transmission over Digital Channels

Attila K. Varga — 2009-12-30

Digital technology forms our environment day by day and we can enjoy its advantages in more and more fields everyday. As a matter of fact television and computer networking technology take over the leading role. The real error analysis of digital channels is the best way for modeling the behaviour of motion picture transmission over digital channels. In this paper digital transmission has been focused on considering such factors as noises, disturbances and movement as well as the mathematical modeling of digital transfer channels has been shown. The paper first presents the most important standardized subjective quality assessment methods described in the ITU-R BT.500 recommendation. We briefly summarise why these subjective tests are so important. Finally, we discuss the implementation of the new subjective video quality measurement related to impaired digital quality television programs. Our aim is to improve these subjective picture quality assessment methods to get sophisticated results, which correlate better with objective picture quality test results.

Behaviour Based Control with Fuzzy Automaton in Vehicle Navigation

Dávid Vincze — 2009-12-30

From the viewpoint of Behaviour based Control many control tasks can be divided into separate behaviour components. By defining the relevant behaviour components, the actual control action can be constructed based on the individual control actions of the component behaviours. In this case the control action is either related to an individual behaviour component or to a fusion of behaviour components based on their relevance to the actual situation. This paper adapts the concept of fuzzy automaton for achieving the decision related to the relevance of the behaviour components in the task of the navigation of an autonomous vehicle. In the structure applied, the relevance of the behaviour components is approximated by a fuzzy rule interpolation (FRI, namely the FIVE method) based fuzzy automaton. The main reason for the FRI application is the state-transition rule-base simplification of the fuzzy automaton. In case of FRI, sparse rule bases (incomplete rule bases) are acceptable, because derivable rules can be omitted intentionally, saving construction time and reducing the complexity of the state-transition rule-base. The paper also provides a brief overview of Behaviour based Control and fuzzy rule interpolation (FRI). For demonstration purposes the paper gives a simple example of state-transition rulebase construction in case of the vehicle navigation task mentioned.

Look Up Table Editor for Small Animal Pet Instrument

Ákos Szlávecz — 2009-12-30

Small animal PET is a widely used instrument for functional examinations in pharmaceutical industry research; its spatial resolution has reached the physically possible limit by now. One important issue of the imaging process is to determine the position of photons impacting the surface of the detectors, and to filter them based on their energy. In this paper we introduce the methods developed to complete these tasks for the small animal PET instrument of Mediso Ltd. All the introduced methods have been implemented in an application called LUT-QT. The efficiency of the methods has been evaluated in real experiments.

Gradient-Based Consequent Optimisation of a FRI Rule Base

Zoltán Krizsán — 2009-12-30

The main contribution of this paper is the extension of an existing Fuzzy Rule Interpolation (FRI) method by gradient-based consequent optimization. The targeted FRI method is an application oriented approach, called FIVE (Fuzzy Rule Interpolation based on the Vague Environment of the Fuzzy Rule Base [1]). The goal of the consequent optimization is the rule base parameter optimization based on input-output sample data of the modelled system.