Servicio de Información Comunitario sobre Investigación y Desarrollo - CORDIS


CHEM Informe resumido

Project ID: G1RD-CT-2001-00466
Financiado con arreglo a: FP5-GROWTH
País: Spain

Temporal bounds generation

The aim of the CHEM project was to develop and implement advanced Decision Support Systems (DSS) for process monitoring, data and event analysis, and operation support in industrial processes. The systems are synergistic integration of innovative software tools, which improve the safety, product quality and operation reliability as well as reduce the economic losses due to faulty states, mainly in refining, chemical and petrochemical processes.

The CHEM applications consist of integrated sets of software toolboxes that provide robust detection and diagnosis of process problems in real-time. The systems assist operators in assessing process status and responding to abnormal events. The project provides a flexible architecture and a methodology in order to facilitate the development of such applications on many processes.

The “Temporal bounds generation” is a fault detection system the objective of the toolbox is to detect internal faults in dynamic processes. The methodology is based on analytical redundancy: the behaviour of the real process and the behaviour of a model of the process are compared. When the behaviour of a variable is abnormal, an internal fault is detected and an alarm is launched. The uncertainty of the process, either on parameters’ values or on measurements, is expressed by means of interval values.

This toolbox requires a discrete-time interval model of the process and interval measurements of the process variables. The behaviour of the model of the process is obtained by simulation, reformulating the simulation problem as a problem of computation of the range of a function in a parameter space. This problem is solved by means of global optimisation methods based on interval computations, so it is a semi-qualitative approach. The simulation propagates the intervals and obtains envelopes. The method is also based on multiple sliding time windows, i.e. it uses several time horizons simultaneously. If the process is complex, the model is divided in sub models and each one is dealt with independently.

Reported by

Campus Montilivi, Edifici P4
17071 Girona
Síganos en: RSS Facebook Twitter YouTube Gestionado por la Oficina de Publicaciones de la UE Arriba