Development of the Product Lifecycle Management Support System on the Basis of Intelligent Agent Technology and Data Mining Methods
2011
Sergejs Paršutins

Defending
05.09.2011. 14:30, DITF, Meža 1/3, 202

Supervisor
Arkādijs Borisovs

Reviewers
Jānis Grundspeņķis, Oļegs Užga-Rebrovs, Mihails Kovaļovs

This PhD thesis presents research into the application of the intelligent agent technology and data mining methods for supporting the process of product lifecycle management. The purposes of this study were to investigate the short time series clustering methods and to develop a model of the multiagent system and methods for product life cycle phase transition point forecasting on the basis of intelligent agent technology and data mining methods. Product lifecycle management is a complex process that includes different tasks like product design, forecasting and planning, manufacturing, storage and transport network organization, product and client support, utilisation etc. The developed multiagent system is aimed at supporting the forecasting and planning processes. The product life cycle consists of several related phases representing the product growth on the market. Each phase differs in level of demand, its growth or decline trend and expenses related to product evolution, advertisement and other related fields, which makes the task of product life cycle phase transition point forecasting important and topical. A model of the product life cycle management support system is proposed in this PhD thesis. The system is capable of forecasting a product life cycle phase transition point using only first several demand curve points. The suitability of different clustering methods for clustering demand data was analysed and the suitable ones were chosen – self-organizing maps and hierarchical gravitational clustering algorithm (G-Algorithm). The classical variants of the algorithms were studied and modified versions were developed and described. A distance measure MEuclidean is proposed, which enables calculation of the distance between time series with different number of periods. The proposed system was tested using the demand data received within the scope of the international project ECLIPS, and also using the synthetic dataset. The results obtained have shown the high efficiency of the proposed product lifecycle phase transition point forecasting system. To evaluate the efficiency of the proposed system, a software tool was developed. A comparative analysis of the obtained results was performed and conclusions are stated regarding the efficiency of the proposed multiagent system in a task of forecasting a PLC phase transition point. The thesis is written in Latvian. It contains an introduction, 5 chapters, conclusions, the list of references, 5 appendixes, 47 pictures, 131 pages. The list of references contains 61 records.


Keywords
Intelligent agent technology, product lifecycle management, clustering short time series, forecasting PLC phase transition points

Paršutins, Sergejs. Development of the Product Lifecycle Management Support System on the Basis of Intelligent Agent Technology and Data Mining Methods. PhD Thesis. Rīga: [RTU], 2011. 131 p.

Publication language
Latvian (lv)
The Scientific Library of the Riga Technical University.
E-mail: uzzinas@rtu.lv; Phone: +371 28399196