NobleBlocks

Research indexConference paper

An Adaptive Surrogate-Model-Based Reference Vector Guided Evolutionary Algorithm for CO2-EOR and Storage Optimization

Liang, Zhang, Yan and 1 other

1 Jan 2026

NobleIDark:/48914/w1/20261105/4E45BFD2DOI10.3997/2214-4609.202674024

Paywalled

The full text is at the publisher

You can read the abstract here, or open it at the publisher.

Read at the publisher

Abstract

CO2-enhanced oil recovery and storage (CO2-EOR and storage) requires efficient optimization under expensive simulations, high-dimensional well-control variables, and complex decision spaces. To address the limitations of fixed single-surrogate approaches, this study proposes an adaptive surrogate modeling framework that combines Gaussian process regression and radial basis function models and dynamically selects the best-performing surrogate during optimization. The framework is integrated with a reference-vector-guided evolutionary algorithm, forming the Adaptive Surrogate-model Based Reference Vector Guided Evolutionary Algorithm (ASRVEA). Initial samples are generated by Latin hypercube sampling, while simplex lattice design is used to construct reference vectors. During optimization, the database is periodically split into training and testing subsets for adaptive surrogate selection, and promising solutions are identified for expensive evaluation based on population distribution and angle-penalized distance. Benchmark results on the 100-dimensional ZDT1 and ZDT3 problems show that the proposed method achieves better convergence, diversity, and robustness than RVEA and KRVEA. These results demonstrate the potential of the proposed framework for efficient CO2-EOR and storage optimization with high-dimensional decision variables and expensive simulations.

Discussion

More

Suggest a correction or report