Databases reconstruction from operating modes recognition in dynamic processes

This work presents a methodology for data reconstruction based in operational modes recognition in dynamic processes, maintaining dynamic properties of registered variables in such database. To do this, an introduction of process and system is made, characterizing the source of databases. Also, a re...

Full description

Autores:: Obando Montoya, Andrés Felipe

Tipo de recurso:

Fecha de publicación:: 2015

Institución:: Universidad Nacional de Colombia

Repositorio:: Universidad Nacional de Colombia

Idioma:: eng

Description
Summary:	This work presents a methodology for data reconstruction based in operational modes recognition in dynamic processes, maintaining dynamic properties of registered variables in such database. To do this, an introduction of process and system is made, characterizing the source of databases. Also, a review of data imputation methodology is presented, highlighting the main features of their procedures. Despite of count with several imputation methodologies, any of them are focused into conserving dynamic properties of variables contained in databases, only proposing different identification models sketchers without considering of a previous data selection step to assure the accuracy of predictive models. Taking into account this fact, the proposed data imputation methodology is based into Dynamical Operational Mode (DOM) recognition of processes, grouping data in clusters with similar dynamic properties, allowing the usage of correct information for auxiliary identification models. Under this considerations, Artificial Resonance Theory (ART2) is introduced as the algorithm for DOM recognition. Additionally, the proposed methodology verifies that imputations do not add uncertainty to original data, conserving initial dynamic information. (Tomado de la fuente)

Databases reconstruction from operating modes recognition in dynamic processes

Publicaciones similares