Skip to Main content Skip to Navigation

Contribution à la surveillance d'un processus de forage pétrolier

Abstract : During the PhD thesis our works have been directed towards identification approaches serving as tools for on-line monitoring procedures dedicated particularly to oilfield drilling processes. Several approaches have been proposed. The role of the first proposed method denoted by GVFF-RLS (Gradient Variable Forgetting Factor Recursive Least Square) is to reduce the transient stage duration by providing a rapid convergence of the FF-RLS. Consequently, this approach allows fast fault detections. Then, an extension of the GVFF-RLS named SALR-GVFF-RLS has been developed. Note that, the SALR stands for (Stochastic Adaptive Learning Rate) and its specificity is to accelerate the GVFF-RLS convergence by rendering the learning rate adaptive. The SALR-GVFF-RLS provides better performances than GVFF-RLS do. In order to ensure the stability of SALR-GVFF-RLS, we have developed an approach giving the maximum value of the learning rate and leaded to a new algorithm called LST-GVFF-RLS where the LST accounts for (Lyapunov Stability Theory). The LST-GVFF-RLS, besides providing a fast convergence it guarantees the stability of the algorithm. In order to take into account the particularity of a drilling process we have explored approaches using Sequential Monte Carlo methods and we have used one of their variants (RBPF). Then, we have emphasized its possibility of exploitation for fault detection strategies. These approaches have been tested on data bases obtained from field tests and highlighted interesting performances in terms of fast and reliable fault detection of bit balling.
Document type :
Complete list of metadata

Cited literature [90 references]  Display  Hide  Download
Contributor : Ecole Arts et Métiers ParisTech Connect in order to contact the contributor
Submitted on : Thursday, April 15, 2010 - 8:00:00 AM
Last modification on : Saturday, August 6, 2022 - 3:07:09 AM
Long-term archiving on: : Thursday, March 30, 2017 - 5:49:56 AM


  • HAL Id : pastel-00005989, version 1


Amadou Ba. Contribution à la surveillance d'un processus de forage pétrolier. Physique [physics]. Arts et Métiers ParisTech, 2010. Français. ⟨NNT : 2010ENAM0007⟩. ⟨pastel-00005989⟩



Record views


Files downloads