Development of an intelligent framework for photocatalytic CO2 reduction to CH4 with enhanced selectivity using 3D/2D perovskite bilayer composites

Perovskite bilayer composites offer an effective approach for enhancing photocatalytic CO 2 reduction performance and operational stability under visible light irradiation. However, their practical implementation is limited by complex material design processes and difficulty in optimizing catalytic parameters for high selectivity. To address these issues, this paper gives an intelligent technique for photocatalytic CO 2 reduction to CH 4 with enhanced selectivity using 3D/2D perovskite bilayer composites. The proposed method integrates the Wasserstein Deep Convolutional Generative Adversarial Network (WDCGAN) with a Starfish Optimization Algorithm (SFOA). The main objective is to maximize CH 4 yield and selectivity while minimizing charge recombination losses in the photocatalytic process. The WDCGAN model is employed to predict photocatalytic activity based on material and structural features, while SFOA optimizes key catalyst parameters to achieve improved reaction performance. The proposed method demonstrates better performance than existing techniques, including the Genetic Algorithm (GA), Deep Neural Network (DNN) and Artificial Neural Network (ANN), as evaluated and compared using the MATLAB platform. The proposed SFOA-WDCGAN method demonstrates the highest R 2 of 0.99, accuracy of 99.5%, and efficiency of 98.5%, indicating improved performance compared to existing methods. The proposed strategy enables the design of stable and high-performance 3D/2D perovskite bilayer catalysts, significantly enhancing photocatalytic CO 2 reduction efficiency, CH 4 yield, and selectivity, making it promising for sustainable fuel generation applications.