The accelerated movement towards smart and sustainable energy systems has established the need of sophisticated computational procedures to address issues at numerous levels of the energy system as more important. Here the paper gives a comprehensive analysis of AI-based computational methodologies (AI-CMs) to improve energy devices and systems, both at the materials level and grid-level optimization. On the materials and device scales, deep learning (DL) and machine learning (ML) models can be used to learn the functionality on complex structure-property-performance correlations to speed-up the search and optimization of functional materials in batteries, fuel cells, supercapacitors, and solar systems. Surrogate models created using data will save experimental and simulation costs by a significant margin, as well as enhance predictive precision and design effectiveness. AI-powered control and optimization systems are applied at the system level to enhance the efficiency and reliability and lifetime of power electronic converters and distributed energy sources. To handle the changing conditions related to operations, adaptive control, fault detection and predictive maintenance use improved approaches including reinforcement learning, evolutionary optimization, and hybrid physics-informed models. The optimization using AI contributes to smart energy management, load forecasting, demand response, and congestion-sensitive routing in smart grids and microgrids with high renewable energy. The predictive performance on materials of the presented AI-CMs framework was high R 2 of 0.91–0.94, and Mean Absolute Error (MAE) was lowered 0.028–0.032 with computation speed-ups of 35–48 × times compared to the performance of Density Functional Theory (DFT) and molecular dynamics simulations. The efficiency increased by 6.4–6.9% at different voltage, temperature and State of Charge (SOC) conditions with high reliability scores of 0.89–0.93 at long operational cycles. On the grid level, AI-CMs obtained load prediction precision of 92–95% in addition to facilitating 60–75% integration of renewable energy and excellent grid stability with the indexes in the range of 0.93–0.97 throughout network nodes. • AI-CMs improve energy devices at both the material and system levels. • DL and ML models make it easier to find materials in energy devices. • Surrogate models reduce operating costs and make predictions more accurate. • AI optimization makes energy systems work better and more reliably. • The AI-CMs framework is really good at making predictions and speeding things up.
Xu et al. (Thu,) studied this question.