Diagnosis of Cementing Oil Wells using Machine Learning
SVM are used to obtain relevant patterns which are validated by experts and become knowledge templates.
Category: Oil Industry Application
Technologies: Support Vector Machines (SVM), C++, C#, VTK.
Cementation is the process of filling with a special cement the separation between the casing and the rock formation. This process is evaluated with an ultrasonic tool called CBL-VDL, which generates the first two logs shown in the upper-left figure, the third log is produced by another tool that measures the acoustic impedance between the casing and bonding cement. A clustering algorithm (from the University of Massachusetts-Lowell) based on SVM machines is used to obtain the relevant patterns (classified patterns in the upper-right figure), these patterns are pre-classified by the SVM machine and them validated by experts and become knowledge templates. Knowledge templates are used as inputs to a SVM classifier. This classifier is used to diagnose the cementation quality in this borehole and other boreholes of the same oilfield. The last log of the upper-left figure is the result of the diagnosis process. This procedure is patent pending.
Related Projects
-
Linear, Nonlinear and Statistical Image Log Processing Image...
-
Cementation Evaluation using Cross-plot Graphs Using Cross-p...
-
Diagnosis of Cementing Oil Wells using Machine Learning SVM ...