Rate of Penetration Prediction using Batch Normalized Deep Elman Neural Network

Exploration, drilling, and extraction are basic components of the upstream process in the oil and gas industry. Rate of Penetration (ROP) is an essential component of the drilling process because it dictates the rate at which materials fracture to enlarge the aperture. For time and money savings, early ROP prediction is critical, but current methods are defective. To enhance ROP prediction, BG-DEL, an innovative method, combines Batch Normalization with a Deep Elman Neural Network based on the Gaussian Error Linear Unit. Following the proper preparation of the data, the subsequent operations are executed: clustering using WFFC, value generation utilizing LRF-GAN, and correlation calculation. To evaluate a classifier that has been trained to predict ROP using database values, this study employs L2CO method for parameter selection. The experimental evaluation yields substantial advantages for the upstream sector of the oil and gas industry. It demonstrates a 98.9% accuracy rate and a correlation coefficient of 0.95. Keywords-Rate of Penetration (ROP), Ward Farthest First Clustering (WFFC), Linear Regression Function based Generative Adversarial Network (LRF-GAN), Laplace Crossover Operator based Coot Optimization (L2CO), Batch Normalized.