Novas metodologias para compressão de dados de processos e para o ajuste do Sistema PI

AUTOR(ES)
FONTE

IBICT - Instituto Brasileiro de Informação em Ciência e Tecnologia

DATA DE PUBLICAÇÃO

2012

RESUMO

The consolidation of the automation in the chemical industry has brought a challenge: translating the large volume of data in useful information. With the advent of digital systems, the expansion in the development of sensors and sampling methods allowed the acquisition of a large amount of data from the process field. Thus, nowadays, engineers and operators rarely suffer from lack of information arising from the several process variables. The algorithms for data compression appeared as an alternative to reducing the storage space demanded by these information. Data compressing means to record only a portion of the original information, preserving, however, the relevant features that they hold. Recently, is not only the disk space that must be prioritized when one talks about compression. It is necessary data to be reliable to the actual process information, and, moreover, must be retrived and transmitted at an aceptable speed. In this work, two methodologies for data compression of chemical processes are presented. Therefore, techniques to estimate derivatives of noisy signals and polynomial curves (cubic splines), which preserve the data features, are used. Also, systematic approaches to automate the tuning of the parameters in the OSIsoft® PI System® compression routines are proposed. In order to evaluate these proposals, some case studies are taken, composed by real signals, from a laboratorial plant, and artificial ones generated by computacional simulation, embracing diversified dynamics and peculiar features. The results show that the proposed algorithms are promising to store processes data without significant loss of information, which can be converted in knowledge without jeopardizing the quality in static and dynamic aspects.

ASSUNTO(S)

controle de processos químicos simulação computacional aquisicao de dados

Documentos Relacionados