GA-vs-NeuralNetwork

Comparative research well log prediction: Genetic algorithm vs Neural Network

Data

Ikpikpuk 1 Log Data (LAS)

Ikpikpuk 1 Drilling and Geologic Reports

Result

Abstract

P-wave velocity is a very important parameter in exploration activities. P-wave velocity (Vp) can be determined from wireline logging data. Generally, the industry only does logging at certain depths which are considered to have prospects in order to save exploration costs. Missing wireline logging data will certainly be a serious problem because it requires complete and accurate data so that the chances of exploration success are high. A method is needed to estimate Vp using data other than sonic log. This research aims to estimate Vp based on available log data using Genetic Algorithm (GA) method and Neural Network (NN). The inversion process is carried out using the method in the Ikpikpuk1 well until the relationship of Vp is obtained with the gamma ray log, resistivity log and density log. The next process estimating Vp by blind test on the same well but the depth is different from inversion. The results showed that the Neural Network method was superior to the Genetic Algorithm method. In the three formations that are the object of research the Neural Network method is consistent because the estimation error is smaller than the Genetic Algorithm method.

requirement for this code

• MATLAB 2017 or later
• 4GB RAM or more
• processor 2 cores or more

References

DOI: 10.13140/RG.2.2.32222.59209