Overview
Français
ABSTRACT
In this present paper, a bond graph tool as a multidisciplinary tool is used for integrated design of robust surveillance systems from modeling step to supervision algorithms generation.
Structural and causal properties of the bond graph are exploited for monitoring ability conditions analysis (ability to detect and isolate faults) before industrial design, and further on line implementation of generated robust algorithms.
The methodology is illustrated using an academic example represented by a DC motor.
Read this article from a comprehensive knowledge base, updated and supplemented with articles reviewed by scientific committees.
Read the articleAUTHOR
-
Belkacem OULD BOUAMAMA: Professor-Director of Research, École polytechnique universitaire de Lille (Polytech) - Head of the MOCIS team at the Laboratoire d'automatique génie informatique et signal (LAGIS) UMR CNRS8219
INTRODUCTION
While automatic control and regulation have been widely mastered in the industrial world, on-line supervision has been little developed. An ambiguity in its definition often reduces it to the task of monitoring parameters or managing alarms by thresholding variables. Whereas the protection of personnel and the environment, and the improvement of industrial systems' operating safety, rely essentially on on-line fault detection and isolation algorithms.
The bond graph tool, which has proven its effectiveness in building knowledge models of multiphysics systems, is used here for the integrated design of model-based monitoring systems ranging from :
modeling ;
-
off-line determination of monitoring conditions and means before industrial production;
generation of robust online monitoring algorithms.
In order to improve decision performance in the detection stage, adaptive thresholds that take into account the parametric uncertainties of the physical system are generated directly from the bond graph model. This systematic approach to monitoring system design is based on the causal, structural and behavioral properties of the bond graph. Its graphical aspect enables the automation of monitoring algorithm generation procedures, using a dedicated software tool. This methodological approach to modeling, analysis and generation of deterministic, then robust, fault indicators is illustrated step by step by a pedagogical application to a DC motor.
The first part presents the principles, definition and role of supervision in the industry.
The second part presents the state of the art in online monitoring methods for continuous systems, and the value of bond graphs for this task.
After a brief presentation of bond graph modeling (as already detailed in other articles in Techniques de l'ingénieur) and the tool's properties, the third chapter describes the bond graph methodology for designing on-line monitoring systems.
Finally, the fourth part is devoted to improving the detection robustness of monitoring systems, by introducing uncertain bond graph models in the form of Linear Fractional Transformation (LFT).
The fifth section concludes the article.
The Ultimate Scientific and Technical Reference
KEYWORDS
inspection | monitoring service | control theory | bond graph
This article is included in
Design and production
This offer includes:
Knowledge Base
Updated and enriched with articles validated by our scientific committees
Services
A set of exclusive tools to complement the resources
Practical Path
Operational and didactic, to guarantee the acquisition of transversal skills
Doc & Quiz
Interactive articles with quizzes, for constructive reading
Integrated design for robust system monitoring
On-line supervision of continuous systems
Bibliography
The Ultimate Scientific and Technical Reference
