The volume 1999

Table of Contents:

Nikolaos M. AVOURIS, Ciprian-Daniel NEAGU, Elias KALAPANIDAS: Distributed Artificial Intelligence Techniques in Environmental Problem Solving.

Abstract: The paper proposes use of artificial intelligence techniques through a distributed multi-agent architecture to environmental problems. In particular it is argued that machine learning techniques based on neuro-fuzzy knowledge representations, combined with heuristics are suitable for many environmental applications, while the distributed problem solving paradigm can handle effectively noisy environmental data collected through a distributed monitoring network. Performance robustness can be achieved through the proposed architecture. The developed techniques have been tested using air quality monitoring data from Athens, Greece.

Keywords: Distributed AI, neural networks, fuzzy logic, air quality monitoring.

Nicolae MARASESCU, Emil CEANGA: Intelligent Continuous Replacement Policy using Diagnosis and Prediction Techniques

Abstract: The paper presents the problem of time renewal determination using the real probabilities of the Markov model state. The diagnosis subsystem, based on neural networks, provides the probabilities for current states of Markov model of the equipment with positive wear. The Markov model’s parameters are adjusted using a neural network.

Keywords: reliability, positive wear, neural network, Markov model.

Viorel ARITON, Severin BUMBARU, Vasile PALADE: Bond Graphs and the Semi-Qualitative Encoding of the Faulty Behavior in Conductive Flow Systems

Abstract: The fault diagnosis in industry is difficult due to the imprecise and the incomplete information related to faults, as well as to the uncertain relations between the observed variables at fault. The faulty model of the target system is not available; only the human diagnostician knowledge on the faulty behavior of the system may be used, i.e. the shallow knowledge (as cases) and the deep knowledge (as qualitative relations between variables), all represented in a linguistic manner on the manifestations and the physical laws in the target system behavior. The paper presents an encoding approach of the power variables (pressure like and flow-rate like variables) in conductive flow systems, as fuzzy partition related to normal and abnormal situations. While in the knowledge acquisition phase some variables have known partition, but many other do not, the paper proposes a method to determine the unknown partitions from the known ones based on conductive flow laws in a qualitative manner, to preserve the semantic consistency of the derived partitions.

Keywords: fault diagnosis, fuzzy encoding, knowledge acquisition.

Viorel ARITON, Vasile MAZILESCU, Vasile PALADE: Elicitation Aspects on Fault Diagnosis of Multifunctional Coductive Flow Systems

Abstract: Fault diagnosis in industry often relies on human experts knowledge; however, the knowledge is mostly incomplete, due to the imprecise information available on target system's faulty behaviour. The paper shortly presents what kind of knowledge the human diagnostician deals with and the way he or she actually acts to manage the incomplete knowledge, aiming to stress how a computational model have to be built. An elicitation path is outlined for the normal and the faulty behaviour of multifunctional conductive flow systems, based on means-end models for the normal behaviour description, and based on generic flow anomalies for the faulty behaviour description for the three orthogonal aspects of flow conduction processes. The generic faulty model resulted, is then presented in an entity-relationship form, useful for the knowledge acquisition on the target system' faulty behaviour (e.g. through a FMEA technique). The paper argues the connectionst approach for the fault diagnosis, exploiting the excitatory/inhibitory relations between entities in a multifunctional conductive flow target system.

Keywords: connectionism, FMEA, knowledge elicitation, means-end models.

Ciprian-Daniel NEAGU, Severin BUMBARU: A Neural Approach of Fuzzy Operators

Abstract: Real world applications of fuzzy sets call for a variety of systems realizing fuzzy computation. A special focus is to develop some universal computing models, easy customizing to meet wide subjects of particular specifications. For this purpose, it is indispensable to identify a few generic-processing modules, which may be configured to perform general computations on fuzzy sets. A family of logic-based neurons emerges as a collection of processing operations whose role is to model logic oriented processing of fuzzy sets. With a generalized fuzzy neuron it is desirable to add yet another level of programmability, parametric learning. This fuzzy neuron utilizes in-situ learning, via fuzzy backpropagation, to adjust the interconnect strength between neurons. This combination of generalized fuzzy computation and adaptivity, creates a powerful processing element.

Keywords: Neuronal networks, fuzzy logic, fuzzy operators, matching, aggregation, projection, defuzzification.

Cornelia NOVAC: Hypertext Maintenance: an Actual Issue of Hypertext Quality

Abstract: This paper presents the actual problems of the maintenance of the hypertext documents and tries to give some solutions for these problems. Just as.software has grown extremely complex and difficult to maintain, hypertext is now approaching similar complexity, with many documents based on large and intricate component graphs. Since software maintenance is a major problem and since hypertext documents share many of the characteristics of software-structure, development process, and economic value-maintaining hypertext documents is also likely to become a major problem requiring immediate action.

Keywords: hypertext, maintenance, software quality.