Effect of Genetic Algorithm in Optimizing Deep Structured Petroleum Reservoir Classifier

Abstract

Well-test analysis contributes to petroleum reservoir description for field development. The reservoir formation identification is the foremost step in characterizing petroleum reservoirs. This research aims to investigate the performance of evolutionary optimization assisted deep structured classifier to identify the homogeneous and fractured reservoirs. The classifier consists of OSTM-long short-term memory and dense neural networks. The hyper-parameters of the classifier have been fine tuned, using evolutionary optimization. The (GA)genetic algorithm conducts a rigorous problem space search to fine-tune the model. The proposed classifier has attained 95.53% accuracy in classifying the reservoirs and their external boundaries. An optimized classifier automatically detects the reservoir formations minimizing human efforts and costs.