CONTENTS

 

Amelina N.O. (SPbSU), Fradkov A.L. (SPbSU, IPME)

The method of averaged models for consensus problem........................................................ 3

Vasilev V. I. (SPb) The new network-based approach for problems

without single solution ........................................................ 40

Kiyaev V.I. (SPbSU), Gerasimov R.V. Smart CRM based on multi-agent approach............................................................ 50

Krivokon D.S. (SPbSU) Simultaneous adjustment of camera parameters and positions of observed points ........................................................ 95

Shats V.N. (SPb) Learning model of the self-organizing system ............................................................ 101

Komarov S.N., Nesterov A.V., Serenryakova CM. (SPbSU)

The information system of the university for accounting purposes and

effective usage of real estate ........................................................ 115

Granichin O.N. (SPbSU) 48-50th International Conference "Conference on Decision and Control" (IEEE CDC/CCC 2009, CDC 2010, CDC-ECC 2011) ........................................................ 142

 

 

 

ABSTRACTS

Multi-Agent Systems

The Method of Averaged Models for Consensus Problem

N. O. Amelina, A. L. Fradkov

Saint Petersburg State University

Institute of Problems of Mechanical Engineering

leishe@mail.ru, fradkov@mail.ru

Key words: consensus problem, multi-agent systems, method of averaged models.

In this paper we study properties of multi-agent consensus control, which is formed by local voting protocol (algorithm) with step size which is not decreasing to zero, with the switching topology, random delay and noise in observations for agents with nonlinear state dynamics. To analyze closed loop system we propose to use the method of averaged models.

Bibliogr.: 38 refs.

 


The New Network-Based Approach for Problems Without Single Solution

V. I. Vasilev

Sankt-Petersburg State University

gnome@bk.ru

Key words: smart networks, multiagent, multiagent networks.

This paper describes a promising new approach in organization of the parallel computations. This approach inherits traditional merit neurocomputing effective parallelism and flexibility of multi-agent technology. The proposed concept allows to obtain a solution with accuracy that depends on the number of elementary processors and structure of the intellectual network. In this case, due to the fact that each processor divided into two separate processes: calculation (clean mathematical function), and control (any algorithm, whose influence on the computation and the network is limited on the given set of the elementary restrict actions), concept may give us a flexible way to describe the various non-deterministic system.

Bibliogr.: 12 refs.

 


Smart CRM Based on Multi-Agent Approach

V. I. Kiyaev, R.V. Gerasimov

Saint Petersburg State University

kiyaev@mail.ru, driven5@rambler.ru

Key words: multiagent approach, client network management, Customer Relationship Management.

In this paper the usage of multi-agent approach to customer service is described. It is proposed to supplement the widely used concept of client-oriented approach based on CRM (Customer Relationship Management) with multi-agent systems and mechanisms. This approach represents a new way of solving the management problems the client network.

Bibliogr.: 34 refs.

Pattern Recognition

Simultaneous Adjustment of Camera Parameters and Positions of Observed Points

D. S. Krivokon

Saint-Petersburg State University dmitry00@gmail.com

Key words: multidimensional optimization, camera's localization, point's triangulation.

The paper introduces a problem of adjustment of camera parameters and positions of points which they observe. Short introduction into subject area is given. Formalization of problem in question is presented. Major complications are concerned.

Bibliogr.: 16 refs.

 

Learning Model of the Self-Organizing System

V. N. Shats St. Petersburg vlnash@mail.ru

Key words: information environment, adaptation, branching chain, learning problem, stochastic model.

Theory of information environment allows to treat learning sample as external information of the self-organizing system. The system operates as a distributed parallel processor that provides its adaptation. Recognition of external influence runs stage of receiving the stimulus, transformations in sensation and then perception which is the classification of information based on accumulated experience in the system. Under this approach the matrix table is displayed in the ensemble matrices of perceptions. These matrices depend on system parameters and has stochastic character. Learning concludes with analysis frequencies of perceptions combinations of inherent in certain classes. The result of system is a probabilistic estimate of the index class.

Bibliogr.: 14 refs.

 

Information Systems

The Information System of the University for Accounting Purposes and Effective Usage of Real Estate

S. N. Komarov, A. V. Nesterov, G. M. Serebryakova Saint Petersburg State University stas@iti.spbu.ru

Key words: efficient usage of real estate, information systems.

The paper is considered the original approach to design the University information system for accounting and the efficient usage of real estate. It is developed as a part of complex general systems.

Bibliogr.: 9 refs.


Conferences

48-50th International Conference "Conference on Decision and Control" (IEEE CDC/CCC 2009, CDC 2010, CDC-ECC 2011

O. N. Granichin

Saint Petersburg State University

O1eg_granichin@mail.ru

Key words: conferences, automatic control, systems.

Control Systems Society (CSS) of the international Institute of Electrical and Electronics Engineers (IEEE) holds the annual conference "Conference on Decision and Control (CDC)" usually on December. It is the largest series of conferences on Automatic Control. The paper describes the experiences of the author's participation in the last three of them.

Bibliogr.: 10 refs.