Prediction of Petroleum Reservoir Properties using Different Versions of Adaptive Neuro-Fuzzy Inference System Hybrid Models

This paper presents a comparative study of the performance of three versions of Adaptive Neuro-Fuzzy Inference System (ANFIS) hybrid model and two innovative hybrid models in the prediction of oil and gas reservoir properties. ANFIS is a hybrid learning algorithm that combines the rule-based inferencing of fuzzy logic and the back-propagation learning procedure of Artificial Neural Networks. Functional Networks-Support Vector Machines (FN-SVM) and Functional Networks-Type-2 Fuzzy Logic (FN-T2FL) were proposed to improve the performance of the stand-alone SVM and T2FL models respectively. The FN component of the FN-T2FL hybrid model automatically extracts the most relevant attributes from the input data using the least square fitting algorithm as an improvement over the individual Functional Networks and Type-2 Fuzzy Logic models. The former is more promising as it combines two existing techniques that are very close in performance and well known for their computational stability and fast processing. The FN-SVM hybrid model also benefits from the excellent performance of the least-square-based model-selection algorithm of Functional Networks and the non-linear high-dimensional feature transformation capability that is based on structural risk minimization and Tikhonov regularization properties of SVM. Training and testing the SVM component of the hybrid model with the best and dimensionally-reduced variables from the input data resulted in better performance with higher correlation coefficients, lower root mean square errors and less execution time than the traditional SVM model. A comparison of FN-SVM and FN-T2FL with the three versions of ANFIS showed the superiority of the FN-SVM model over the others. The three ANFIS models still proved to be good in solving real industrial problems due to their speed of execution especially in dense data conditions.