Reliability prediction of oil production wells by particle swarm optimized support vector machines - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2010

Reliability prediction of oil production wells by particle swarm optimized support vector machines

Résumé

In the oil industry, well failures lead to interruption of oil production, with significant economical losses. In order to avoid such undesirable situations, reliability prediction may be a valuable tool. Generally, system reliability depends on a number of factors, e.g. aging, operational and maintenance conditions. The consideration of all the involved aspects and their interdependencies renders difficult an analytical treatment. In these cases, empirical modeling may be an effective way for reliability prediction. In this paper, the powerful empirical modeling paradigm of Support Vector Machines (SVMs) is investigated, with respect to the dependence of their performance depends on a set of real-valued parameters that govern the learning phase. This leads to the so called SVM model selection problem, which is here tackled via Particle Swarm Optimization (PSO). The developed PSO?SVM approach is applied to the reliability prediction of Brazilian oil production wells.
Fichier non déposé

Dates et versions

hal-04366341 , version 1 (28-12-2023)

Identifiants

  • HAL Id : hal-04366341 , version 1

Citer

I.D. Lins, M.C. Moura, E.L. Droguett, E. Zio, C.M. Jacinto. Reliability prediction of oil production wells by particle swarm optimized support vector machines. European Safety and Reliability Annual Conference: Reliability, Risk and Safety: Back to the Future, ESREL 2010, Sep 2010, Rhodes, France. pp.914-922. ⟨hal-04366341⟩
7 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More