Study on parameter optimization and dynamic regulation method of polymer flooding reservoir based on artificial intelligence

This study focuses on the method of parameter optimization and dynamic regulation of polymer flooding reservoir based on artificial intelligence (AI). With the gradual depletion of conventional oil and gas resources, polymer flooding technology is very important to ensure energy security, but traditional optimization methods have many limitations. In this study, a closed-loop optimization system including data-driven modeling, multi-objective optimization and dynamic strengthening regulation is constructed. Firstly, the multi-source data are integrated, and after preprocessing, a proxy model is constructed by deep learning. With the improved ResNet-18 architecture, the prediction accuracy is improved by weighted mean square error (MSE), mean absolute error (MAE) and L2 regularization. Then NSGA-II algorithm is used for multi-objective optimization to maximize oil recovery and minimize cost, and constraints are set to ensure the feasibility of the results. Finally, PPO (Proximal Policy Optimization) algorithm is used to realize dynamic regulation, and the parameters are adjusted according to the real-time state of the reservoir. Taking a block in Shengli Oilfield as an example, the results show that after 12 months of AI optimization and regulation, the stage oil recovery increased by 34.1%, the comprehensive water cut decreased by 3.6%, the oil production per ton of polymer increased by 16.7%, the proportion of ineffective polymer injection decreased by 50%, the economic benefit was significantly improved, and the net income reached 26.3 million yuan.