Fault detection with Stacked Autoencoders and pattern recognition techniques in gas lift operated oil wells
Palavras-chave:
Fault detection, Oil well monitoring, Multivariate time series classification, Cross validation, Pattern recognitionResumo
The offshore industry is responsible for most of the oil and gas production in Brazil. When the level of
complexity in this industry is high, it has been a precursor to new technologies in recent years. The main objective
of the present work is the development of a system for the detection and classification of failures in oil produc-
tion wells operated with elevation by gas lift. Stacked autoencoders are used and pattern recognition techniques
for fault classification, verifying performance metrics and applying cross-validation to check the generalization of
the models for the available observations. After the development of the classifiers, high recall values were ob-
tained (much higher than 0.88), which shows the great applicability of the proposed system in detecting failures in
offshore production wells.
