Réconciliation de données en présence d'incertitudes de modèle. : application au convertisseur à oxygène

Abstract : In the steel industry, tuning the control system set-points in order to reach given product specification is a common problem. The control system objective is to obtain products satisfying the given specifications. There are however frequent changes in the behavior from one batch to another, thus requiring the modification of model parameters used for the following presets. This is particularly true for the Basic Oxygen Furnace (BOF) where the information collected during a specific batch serves to adjust the set-points of the next batch.In this thesis, we proposed a method allowing simultaneously robust data reconciliation and model parameter estimation. This approach is original because the current data validation methods make the assumption that the system model is known perfectly. Here, the uncertainties on the measurements of variables and on the knowledge of the model parameters are simultaneously taken into account to provide consistent estimates that can be utilized by the control system of the BOF, especially for the determination of the set-points. Parameter estimation being sensitive to the presence of outliers, which is common in this type of process operating in highly disturbed environment, the proposed method has been made robust to the presence of these errors by including in the estimation method, the notion of contaminated statistical distribution.The model of the converter was based on the principles of mass and energy conservation, but also on chemical equilibrium relationships for which some parameters are not well known that fully justifies the proposed approach. The variable and parameter estimates provide a more accurate picture of the actual state of the system which facilitates its control. Applied to the BOF, this method allows the updating of the model used for the preset. In addition, monitoring the evolution of some model parameters can be used to evaluate the degradation level of the system (eg, damage to the refractory of the converter)
Document type :
Theses
Complete list of metadatas

Cited literature [76 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/tel-01748864
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 11:51:34 AM
Last modification on : Wednesday, July 25, 2018 - 1:24:35 AM
Long-term archiving on : Friday, September 14, 2018 - 8:03:21 AM

File

2010_FRANCKEN_J.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-01748864, version 1

Collections

Citation

Julien Francken. Réconciliation de données en présence d'incertitudes de modèle. : application au convertisseur à oxygène. Autre. Institut National Polytechnique de Lorraine, 2010. Français. ⟨NNT : 2010INPL033N⟩. ⟨tel-01748864⟩

Share

Metrics

Record views

13

Files downloads

11