Volume 2, No. 10 (2004)
Articles
FUZZY CONTROL FOR LAUNDRY PROCESS
ABABEI ŞTEFAN, CULEA GEORGE
This paper present a fuzzy logic control model of the laundry process at a washing machine is presented, having as input variables the total weight of the clothes and their degree of dirtiness. This way less electrical energy is consumed and the appliances are less worn-out.
UNDERGROUND ACTIVITY DETECTION IN COMMUNICATIONS NETWORKS USING NEURAL NETWORKS
BĂLĂŞOIU LEONARD
Underground activity detection refers to the attempt to detect illegitimate usage of a communications network. Three methods to detect underground activity are presented. Firstly, a feed-forward neural network based on supervised learning is used to learn a discriminative function to classify subscribers using summary statistics. Secondly, Gaussian mixture model is used to model the probability density of subscribers` past behavior so that the probability of current behavior can be calculated to detect any abnormalities from the past behavior. Lastly, Bayesian networks are used to describe the statistics of a particular user and the statistics of different underground activity scenarios. The Bayesian networks can be used to infer the probability of fraud given the subscribers` behavior. The data features are derived from toll tickets. The experiments show that the methods detect over 82% of the fraudsters in our testing set without causing false alarms.
ARTIFICIAL NEURAL NETWORK SYSTEM FOR CARDIAC RHYTHMS CLASSIFICATION
CULEA CATALINA
A system of two simulated Kohonen networks was used in patients undergoing electrophysiology studies for classification of the cardiac rhythm. The system was successful in detecting wide complex tachycardia and ventricular fibrillation after only a short period of training. Narrow complex tachycardia couldn't be easily distinguished from sinus tachycardia or atrial fibrillation with a rapid ventricular response. The system appears promissing for distinguishing the generally benign narrow complex tachycardias from the wide complex tachycardias that need in control of electric therapy by the device or physical therapy.
THE NPN INTELLIGENT MODEL FOR DIAGNOSIS OF MECHANIC WEAR
CULEA GEORGE, ABABEI ŞTEFAN, GRIGORE ROXANA
The paper presents a new modeling method called Neural Petri Network; it also presents an application regarding the intelligent diagnosis of mechanic wear of a processing centre. The diagnosis is realized viewing the modeling of the wear's evolution in time, using the NPN method. This evolution of the wear is controlled by measuring each axis' time of functioning. By analyzing these time measures, it was modeled the decision of stopping the processing centre for technical revising. The modeling of taking the decision of stopping the functioning will be realized with the help of a neural network which will be inserted in the Petri network.
EXPERT SYSTEM BASED ON GRID INFRASTRUCTURES
CULEA GEORGE, CULEA CATALINA
The realisation of performant expert system their and accessing by various users at distance represents a goal which is to be achieved. The solutions found until the present moment involve that, generally, the utilisation of supercomputers, which are very expensive, also implies a hard accessibility. An alternative solution would be the using of some computable systems connected to the network instead of this expensive resource. The developing of digital communication allowed the realisation of the GRID technology, one that is capable to make up integrated / unitary systems mode of networks of computers by means of which distributed applications can be performed.
FUZZY MODELS FOR AN ELECTRICAL ENERGY TARIFFING
DUGAN VIOREL, SOLEA RAZVAN, GHINIŢĂ DANIELA
The paper is a decision - support application which design and use two fuzzy models to estimation an electrical energy tariff, as it to be sell at consumers. The fuzzy tariff estimation model integrate not only the S.C Electrica S.A. rate position, but and some constraints/compulsions of National Authority of Settlements from Energy (NASE), beginning with 1999, in this transition period from Romania. The paper not refer to a price concrete case (internal tariff used in certain year, production price, transport price, distribution price, spot price, or an external price to be sold electrical energy - EE, etc). The paper shows how, by changing the parameters of S.C Electrica S.A and NASE, it is possible to can perform sensitivity tests on the tariff function model until we obtain an acceptable price. Much more: the two fuzzy models use different rules (conservative and aggressive, respectively) for pricing. Finally, the paper not finished all fuzzy possibilities (rules) which can influences the expected value of a some EE tariff but, with certitude, can create a discussion base, about the way of approximate/ fuzzy reasoning, as a modality to find and to refine an EE price.
SOME THEORETIC CONSIDERATIONS ABOUT USING NEURAL NETWORKS FOR INCREASE THE SAFETY LEVEL OF A POWER PLANT
GRIGORE ROXANA
The implementation of any maintenance policy involve a good performance of measurement apparatus, monitorial techniques of elements states. It can used neural networks for reduce maintenance costs, for minimize chances of catastrophic failures through early detection and trend analysis.
APLICATION FOR THE CALCULUS REGIMES OF THE ELECTRICAL POWER SYSTEMS - RP V3.0
HAZI GHEORGHE
The paper presents the principles of realization, facilities and performances of the application RP V3.0. The author presents the system of administration of data, possibilities of migration from the version 2.2 to version 2.3, procedures of validation of information. The general reports of application and examples for using are presented, too.
MINING ASSOCIATION RULES IN GEOGRAPHIC INFORMATION SYSTEMS USING INTELLIGENT AGENTS
LEON FLORIN, SOVA IOAN, GALEA DAN
A geographical information system (GIS) is a framework capable to assemble, keep, process and display specific information, identified by geographical location, which can combines layers of information to give the user a better understanding about that location. Intelligent agents prove to be a useful tool for data mining in geographical distributed knowledge bases. In this paper we present a method of using intelligent agents for discovering association rules in geographical datasets.
ON THE CONCEPT OF EMERGENCE AND ITS APPLICATIONS
NECHITA ELENA, CRISAN GLORIA-CERASELA
Emergence is a concept widely used in science, arts and engineering. Complex systems (societies, economies, bio-systems) exhibit a behavior structured at two levels: the individual one and the colony one. While the behavior of individual members is unpredictable, the colony as a whole is self-organized and displays order. System grammars are a method of formalization such behavior. The aim of our paper is to outline recently discovered aspects concerning emergence and to survey some possible applications.
INTELIGENT TECHNICS-BASED GIS APPLICATIONS FOR CONTROL AND MONITORING SYSTEMS
NIŢU COSTICĂ, COSTACHE GABRIEL
Geographic Information Science or what is so called GIScience is an applied science, which involves spatial or geo data collection and analysis. It consists of three major fields: remote sensing, GIS and GPS. GIS is one of the most important techniques used to analyze spatial data. Internet-based GIS in some ways resembles the classic "mainframe and terminal" computer hardware model. That is, one computer does the actual GIS processing and the other is connected to it through a network—in this case, the Internet—and acts as the display terminal for the maps, images, and data tables which are produced by the GIS. The above-mentioned approach is known as "server-side" processing since a server will produce the results and send them back to the user's Web browser, also known as the "client".
MULTIPLE NEURAL NETWORK ARCHITECTURES FOR POWER SYSTEMS
NIŢU COSTICĂ, ENESCU ALEXANDRU
Artificial Neural Networks proved to be capable of finding internal representations of interdependencies within row data not explicitly given or even known by human experts. This typical characteristic together with the simplicity of building and training Neural Networks and their very short prediction delay time encourages researchers to apply neural networks to different tasks for power systems, such as load forecasting, fault diagnosis, alarm handling and state estimation. By combining more than one of these neural networks, better performance can be achieved with these new architectures.
NEW APPROACH FOR THE CONTACT RESISTANCE EVALUATION OF THE SWITCHGEARS
PANCU C. *, BARTIC P. **, BARABOI A. *, ADAM M. *
In this paper is proposed a system based on the artificial intelligence which will evaluates the contact resistance of the switchgears. This method is much more efficient than classical methods in the measure in which we have a behavior history of the analyzed equipment.
ASPECTS REGARDING APPLICATIVE AND THEORETICAL STUDY OF ELECTROMAGNETIC COMPATIBILITY IN ELECTRIC DEVICES OPERATION. ELECTROMAGNETIC IMPRINT OF AN ARC BLOW OUT CHAMBER
POPA SORIN
In this paper it is presented the activity of the digital simulation of the electromagnetic behavior of an arc blow out chamber of a circuit breaker, rather the attempt to determination an electromagnetic fingerprints of an arc chute, which can be use, in future, to diagnostic the state of electrical circuit breaker. The latest new digital techniques for modeling and simulation of the electromagnetic behavior of electrical systems are the Finite Element Method (FEM) and the Boundary Element Method (BEM). In this paper I use the FEM to modeling and simulates an arc chute of a circuit breaker.
NEURO-FUZZY METHOD FOR ECOLOGICAL MODEL DESIGN
PRUTEANU EUSEBIU
The rapid growth in the number and variety of applications of the fuzzy logic controller and neural networks reflects the advantages of the intelligent techniques applications. Neuro-fuzzy modeling is aimed at solving real world decision making, modeling and control system problems, witch are usually imprecisely designed and require human operator intervention. In this paper neuro-fuzzy method with their ability to incorporate human knowledge are used in the conception and design of hybrid intelligent systems and models. Implementation of the neuro-fuzzy models and algorithms by using the weighting sums method is proposed.
FUZZY INDEX FOR STEADY STATE POWER SYSTEMS OPERATION
PRUTEANU EUSEBIU, BOTEZATU VASILE LIVIU
This work is about the practical aspect of applying the model based diagnosis (MBD) to steady state power system operation. The results provide design criteria for expert system methodology implementation, which can be used by power system operators for taking optimal on-line decisions.
NEURO-FUZZY METHODS FOR MENTAL MODELS IMPROVING
PRUTEANU EUSEBIU, BOTEZATU VASILE LIVIU
The mathematical model of the human operator behavior in control systems is nonlinear because the parameters of transfer function are modifying in time. It has a complex serial-parallel structure with elements of commutation in function of the mental calculation results and of the process conditions and because in many applications the thinking process can be modified in time. In the paper the use of the neuro-fuzzy methods for mental model obtaining is described.
IMPROVEMENT OF THE NEURAL NETWORK TRAINING
ROTAR DAN
The paper shows a new method for improvement the neural network training. Based on the drawings programs this method allows in advance determination of weight and bias of the neural network. This is very useful for the neural network design, selection and optimization and the time reduce of the neural network training. The method described is based on graphical representation of the input domain of data and the vector representation of the decision surfaces.
SOME CONSIDERATIONS ABOUT SPICE-ASSISTED SIMULATION OF THE NEURAL NETWORKS
ROTAR DAN
The paper reports the possibilities of the neural network simulation using SPICE models. In the design and testing of the experimental neural networks it is very useful to tray the behavior of neural network in many reel conditions. SPICE can provide the initial training sets and also assist in network development, both at the circuit level and the theoretical level.
CONTROL AND COMMAND SYSTEM FOR AN ELEVATOR
TIMIS MIHAI, COSMICI ADRIAN, GALU CĂTĂLIN
The implementation of a such system implies a PLC (Programmable Logic Controller) specific circuit. It has many new add on facilities, like: 10 BaseT networking port, RS232 port, RS485 port, high processing speed, many I/O digital ports, many I/O analog ports. By using 10 BaseT networking port, through the internet we can permanently watch how the application works. Also we can detect and fix the errors that can appear. The RS232 port is used for connecting a TFT Touch Screen who will allow the user to input or modify the application settings. Also, by using the PLC system, we can control the application security functions. The access to application settings is done through specific passwords. The RS485 port is used for interconnection of the command bloc with the elevator up/down motors.