Renewable energy technologies are central to emissions reduction and essential to achieve net-zero emission.Segmented thermoelectric generators(STEG)facilitate more efficient thermal energy recovery over a large tempe...Renewable energy technologies are central to emissions reduction and essential to achieve net-zero emission.Segmented thermoelectric generators(STEG)facilitate more efficient thermal energy recovery over a large temperature gradient.However,the additional design complexity has introduced challenges in the modelling and optimization of its performance.In this work,an artificial neural network(ANN)has been applied to build accurate and fast forward modelling of the STEG.More importantly,we adopt an iterative method in the ANN training process to improve accuracy without increasing the dataset size.This approach strengthens the proportion of the high-power performance in the STEG training dataset.Without increasing the size of the training dataset,the relative prediction error over high-power STEG designs decreases from 0.06 to 0.02,representing a threefold improvement.Coupling with a genetic algorithm,the trained artificial neural networks can perform design optimization within 10 s for each operating condition.It is over 5,000 times faster than the optimization performed by the conventional finite element method.Such an accurate and fast modeller also allows mapping of the STEG power against different parameters.The modelling approach demonstrated in this work indicates its future application in designing and optimizing complex energy harvesting technologies.展开更多
With lithium-ion energy density nearing its limits,next-generation storage requires an atomic-scale understanding of dendrites and the solid-electrolyte interphase evolution.Conventional simulations remain computation...With lithium-ion energy density nearing its limits,next-generation storage requires an atomic-scale understanding of dendrites and the solid-electrolyte interphase evolution.Conventional simulations remain computationally prohibitive,whereas machine learning typically predicts macroscopic metrics rather than ion dynamics.We present a deep learning framework that couples a one-dimensional convolutional network with iterative training and a physics-based voltage embedding to forecast ion positions,charge distributions,and dendriticmorphology over repeated charge and discharge cycles in lithium metal batteries.The model achieves amean error of 1.53%for atomic positions and reduces computation time from 18 h(molecular dynamics simulation)to 25 min(proposed framework).It preserves redox trends across cycles and reproduces electrolyte-dependent dendrite suppression(Dice similarity coefficient 0.90;mean absolute percentage error<2%).The approach offers a practical surrogate for time-series atomistic simulation and supports internal-state screening,failure diagnosis,and the design of next-generation systems.展开更多
基金supported by an EPSRC IAA funding.The authors acknowledge using the IRIDIS High-Performance Computing Facility and associated support services at the University of Southampton to complete this work.All data supporting this study are available from the University of Southampton repository at DOI:https://doi.org/10.5258/SOTON/D2454.
文摘Renewable energy technologies are central to emissions reduction and essential to achieve net-zero emission.Segmented thermoelectric generators(STEG)facilitate more efficient thermal energy recovery over a large temperature gradient.However,the additional design complexity has introduced challenges in the modelling and optimization of its performance.In this work,an artificial neural network(ANN)has been applied to build accurate and fast forward modelling of the STEG.More importantly,we adopt an iterative method in the ANN training process to improve accuracy without increasing the dataset size.This approach strengthens the proportion of the high-power performance in the STEG training dataset.Without increasing the size of the training dataset,the relative prediction error over high-power STEG designs decreases from 0.06 to 0.02,representing a threefold improvement.Coupling with a genetic algorithm,the trained artificial neural networks can perform design optimization within 10 s for each operating condition.It is over 5,000 times faster than the optimization performed by the conventional finite element method.Such an accurate and fast modeller also allows mapping of the STEG power against different parameters.The modelling approach demonstrated in this work indicates its future application in designing and optimizing complex energy harvesting technologies.
基金supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT)(RS-2022-NR070832).
文摘With lithium-ion energy density nearing its limits,next-generation storage requires an atomic-scale understanding of dendrites and the solid-electrolyte interphase evolution.Conventional simulations remain computationally prohibitive,whereas machine learning typically predicts macroscopic metrics rather than ion dynamics.We present a deep learning framework that couples a one-dimensional convolutional network with iterative training and a physics-based voltage embedding to forecast ion positions,charge distributions,and dendriticmorphology over repeated charge and discharge cycles in lithium metal batteries.The model achieves amean error of 1.53%for atomic positions and reduces computation time from 18 h(molecular dynamics simulation)to 25 min(proposed framework).It preserves redox trends across cycles and reproduces electrolyte-dependent dendrite suppression(Dice similarity coefficient 0.90;mean absolute percentage error<2%).The approach offers a practical surrogate for time-series atomistic simulation and supports internal-state screening,failure diagnosis,and the design of next-generation systems.
