Showing papers presented at "Computer Aided Systems Theory in 1994"

Improvements in the systems-based models generator sigem

Abstract: Program generators, for us, are computer programs that produce other computer programs. SIGEM is an expert system program generator that can help in the modeling process of real systems. It is associated with a methodology well adapted to modeling practice. In this paper, we present and compare this methodology with other similar ones. Static models (databases), dynamic models, rule-based expert systems, literal and/or numerical variables, probabilistic uncertainty in data and in functions, dimensioned variables, discrete event simulation, and other related problems can be treated with this methodology. We suggest a systems modeling methodology and a programming tool to increase generality of application and easiness of use.

Distributed inference in bayesian networks

Abstract: Bayesian networks originated as a framework for distributed reasoning. In singly connected networks, there exists an elegant inference algorithm that can be implemented in parallel having a processor for every node. It can be extended to take advantage of the OR-gate, a model of interaction among causes that simplifies knowledge acquisition and evidence propagation. We also discuss two exact and one approximate methods for dealing with general networks. It will be shown how all these algorithms admit distributed implementations.

Reduced-order kalman filter for alignment

Abstract: The use of a full-order Kalman filter for the alignment process of an inertial navigation system (INS) imposes an unacceptable burden on computers. In this paper we intend to design and analyze a low-order Kalman filter for the initial alignment of a ring laser gyro strapdown inertial system that results in a substantially lighter computer burden with a little significant reduction in system accuracy.

Multimedia environments: designs for simulation

Abstract: The importance of simulation in all fields is widely recognized due to the impact it has had on design, production, and engineering. Currently, computer technology offers rapid solutions and graphics facilities that contribute to the development and improvement of complex simulation programs and their user interfaces. This paper presents a methodology aimed at developing graphic interfaces for the simulation of industrial processes and with the possibility of extending the scope into the area of control. The first part summarizes the simulation process technique and the problem of the user interface design from an ergonomic perspective. The second part describes the design of the user interface. First, the kind of process to be simulated is identified. Focusing on the objects that are to be represented, these are named with their respective actions in the process and the interface. Finally, a prototype of process simulation of a nuclear plant is presented.

Image segmentation using a modified split and merge technique

Abstract: In this paper, a reformulation of the processes that composed the split and merge technique is proposed. A recursive split process is done first; then, after the merge process, a segmentation of the image is obtained. This is possible because the merging is implemented like a growth process, so the grouping has been eliminated. The use of a complete quadtree helps to achieve the reformulation. The determination of the region uniformity in each process is carried out with a different predicate, because of the different nature of the two processes. Experiments on a blocks world image and an industrial parts image are included to show the result of applying the algorithm.

Generation and codification of connected components in images of pixel diagnosis

Abstract: A methodology is introduced for the generation of a primary segmentation of a scene image and extraction of its connected components from several diagnostic maps. The segmentation is based on an adaptation of the Watershed transform. Also presented is an extension of the algorithm described in Mandler (1990) for the analysis of the connected components contained in the color map produced by the Watershed transform. The algorithm performs the connected components analysis, returning as results every component in the color map, its border code, and an ordered list of its neighbors including the coordinates of the first common border point. This is done efficiently, scanning every line in the color map only once. The original algorithm has been extended to deal with 8-connectivity.

Systems Science and Systems Technology: From Conceptual Framework to Applicable Solutions

Abstract: The paper points out the importance of systems technology for dealing with complex design tasks in real life problem solving especially in the field of engineering.

CASCADE: A Computer-Aided Tool for Low-Energy High-Performance Multi-DSP VLSI System Design

Abstract: The design of an energy efficient digital signal processing system benefits from trade-offs of system complexity and performance during the algorithmic design and architectural design. CAST methods are useful during these early phases of design exploration. The traditional CAD tools are used in the final implementation phases. CASCADE is a design environment consisting of method banks of CAST and CAD tools. The system modeling and simulation method called OPERAS is described in context with the CASCADE.

Impact of smooth surface perception capability on the structure of a vision system

Abstract: This article discusses some properties of an artificial visual system that follow from the fact that low-level image processing can produce (sparse) descriptors of a shape that is smooth in mathematical sense. A structural surface model suitable for model-based recognition is proposed and an efficient general structural matching algorithm is outlined. The problem of the relation between image properties and the differential surface properties is also discussed.

Evaluation of a shape identification system based on a signature function

Abstract: One of the commonest problems in artificial vision is how to identify an object by comparing its characteristics with those of the models included in the recognition system database. Relevant characteristics include properties both of the object's surface (size, shape, and reflectivity) and of its edges. In general, the solution to the problem depends (1) on the type of illumination used and whether its distribution across the scene is uniform or not and (2) on the surface characteristics (reflectivity, texture) of both the object and the background. In this first phase of research we have tried out a particular signature function: the distance of all points on the object's border from the object's center.

On the implementation of preference relational systems in a computer-aided decision support system

Abstract: Preference systems are models used in managing the preferences on a set of alternatives. Preference relational systems are preference systems based on a series of exhaustive and mutually exclusive binary relations. Preferences are best shown about paired comparisons, and binary relations are the most intuitive and suitable mathematical tool for modeling preferences based on paired comparisons. In addition to a preference relation P and an indifference relation I, a general preference relational system includes a semipreference relation S between the preference and an indifference and the incomparability relation X corresponding to the nonexistence of preference information to relate the alternatives. The system is simple when the semipreference is excluded and complete when the incomparability is excluded. The most important structures of these systems are introduced by mixed transitivity properties (conditions to infer a preference relation from a chain of related alternatives). The mixed transitivity is...

Computer-aided decision system for future strategies in old hydropower plants

Abstract: The Institute of Waterpower and Pumps at the Technical University Vienna has gained a lot of experience on uprating and refurbishing old hydroelectric power plants in the last decade. During this time a few experts have finished their work and have left the institute. One advantage of artificial expertise is its permanence. Once required, it is around forever, barring catastrophic accidents related to memory storage. This was the main point in the development of a knowledge-based system that stores human expertise. This paper describes a knowledge-based system to aid in the decision making for future strategies of old hydropower plants.

Towards a CAST Method for a Systematic Generation of Non-Orthogonal Complete Transformations

Abstract: 1) We have studied in depth the Theory of Receptive Field Functional Transformation, increasing their scope to situations of variable dimension partitions. 2) We have developed a theorem which allows the systematic generation of Complete Transform in the above case.