Optimization of neural-network model using a meta-heuristic algorithm for the estimation of dynamic Poisson’s ratio of selected rock types

Abstract

This research focuses on the predictive modeling between rocks' dynamic properties and the optimization of neural network models. For this purpose, the rocks' dynamic properties were measured in terms of quality factor (Q), resonance frequency (FR), acoustic impedance (Z), oscillation decay factor (α), and dynamic Poisson’s ratio (v). Rock samples were tested in both longitudinal and torsion modes. Their ratios were taken to reduce data variability and make them dimensionless for analysis. Results showed that with the increase in excitation frequencies, the stiffness of the rocks got increased because of the plastic deformation of pre-existing cracks and then started to decrease due to the development of new microcracks. After the evaluation of the rocks’ dynamic behavior, the v was estimated by the prediction modeling. Overall, 15 models were developed by using the backpropagation neural network algorithms including feed-forward, cascade-forward, and Elman. Among all models, the feed-forward model with 40 neurons was considered as best one due to its comparatively good performance in the learning and validation phases. The value of the coefficient of determination (R2 = 0.797) for the feed-forward model was found higher than the rest of the models. To further improve its quality, the model was optimized using the meta-heuristic algorithm (i.e. particle swarm optimizer). The optimizer ameliorated its R2 values from 0.797 to 0.954. The outcomes of this study exhibit the effective utilization of a meta-heuristic algorithm to improve model quality that can be used as a reference to solve several problems regarding data modeling, pattern recognition, data classification, etc.

Estimating elastic properties of sediments by pseudo-wells generation utilizing simulated annealing optimization method

“…Metaheuristic algorithms are optimization algorithms designed to find good enough solutions in in limited computational time by evaluating potential solutions and performing strategies and operations to refine them (Niri et al, 2018;Waqas et al, 2023). These methods, excel in addressing highly non-linear optimization problems and thus supply a promising approach for applications in oil and gas exploration (Abdullahi Mu'azu, 2023;Yan et al, 2020).…”

“…The regularization approach, offered initially by Bell (1978) and Doicu et al (2010), is typically employed to enhance the inversion stability. Recently, Wang et al (2021) used the dictionary learning algorithm to learn the formation characteristics of elastic parameters from logging data and then took this information as a prior constraint in AVO inversion (Guo et al, 2020;Waqas et al, 2023). Afterward, the sparse representation of the dictionary is employed as preliminary information to constrain AVO inversion (Hosseini Shoar et al, 2014;Kianoush et al, 2023d;Kianoush et al, 2023e;Kianoush et al, 2022a;Kianoush et al, 2023a;Pirhadi et al, 2023;Rashidi et al, 2020;Shakiba et al, 2018;Stork et al, 2011).…”

Estimation of elastic properties of sediments utilizing simulated annealing algorithm – a pseudo-wells generation approach

“…Metaheuristic algorithms are optimization algorithms designed to find good enough solutions in a small amount of computing time by evaluating potential solutions and performing strategies and operations on them (Niri et al, 2018;Waqas et al, 2023). Metaheuristic search techniques, e.g., bioinspired optimization algorithms such as genetic algorithms, can handle highly non-linear optimization concerns and thus supply a promising approach for oil and gas exploration (Abdullahi Mu'azu, 2023;Yan et al, 2020).…”

“…The training and learning of neural networks with a powerful optimizer such as particle swarm optimization (PSO) overwhelms slow learning rate problems and ameliorates the performance of neural network models. Recently, a hybrid orthogonal learning particle swarm optimization (HOLPSO) was introduced to solve the prestack seismic inversion issue with the special purpose of mitigating the instability of the outcomes and the premature convergence of the algorithm (Guo et al, 2020;Waqas et al, 2023). Metaheuristic algorithms are global methods that intend to solve the problem by heuristic search.…”

Probabilistic sunspot predictions with a gated recurrent units-based combined model guided by pinball loss

“…Notably, the employment of the GRU model for sunspot number prediction has yielded valuable insights 34 . The employment of meta-heuristic optimization algorithms has demonstrated promise in efficiently optimizing neural network parameters 35 , 36 . These algorithms provide an effective and flexible means of exploring high-dimensional search spaces while avoiding local optima, thereby preventing suboptimal solutions 37 .…”