Performance prediction of a supercritical CO2 Brayton cycle integrated with wind farm-based molten salt energy storage: Artificial intelligence (AI) approach

Abstract

Using energy storage technologies in combination with renewable energy sources can improve power generation efficiency and reliability. Here, we predict the performance of a wind farm-based molten salt energy storage system combined with a supercritical CO2 Brayton cycle (s-CO2BC) with the aid of artificial intelligence (AI). For this purpose, the Grasshopper Optimization Algorithm (GOA) and Particle Swarm Optimization (PSO) are applied to train a Multilayer Perceptron Artificial Neural Network (MLP-ANN). The proposed AI approaches model the complex interactions between the wind farm, the molten salt energy storage, and the s-CO2BC components. Integrated system simulations under various operating conditions are used for training the performance prediction model. Power generation, energy storage, and system efficiency can all be accurately predicted by a trained AI model. In addition to providing valuable insight into the optimal operation and control strategies of the integrated system, the proposed approach enables the efficient utilization of renewable energy resources and energy storage to generate sustainable power. This study contributes to the development of AI-based optimization approaches and predicts the performance of integrated renewable energy systems, resulting in improved grid integration and stability. According to the results, the PSO method yields average absolute percentage deviations (AAPDs) between 0.47% and 0.51% across different parameters, while GOA yields AAPDs between 0.05% and 0.27%.