Digital twins of lithium-ion batteries are increasingly used to enable predictive monitoring,control,and design at system scale.Increasing their capabilities involves improving their physical fidelity while maintainin...Digital twins of lithium-ion batteries are increasingly used to enable predictive monitoring,control,and design at system scale.Increasing their capabilities involves improving their physical fidelity while maintaining sub-millisecond computational speed.In this work,we introduce machine learning surrogates that learn physical dynamics.Specifically,we benchmark three operator-learning surrogates for the Single Particle Model(SPM):Deep Operator Networks(DeepONets),Fourier Neural Operators(FNOs)and a newly proposed parameter-embedded Fourier Neural Operator(PE-FNO),which conditions each spectral layer on particle radius and solid-phase diffusivity.We extend the comparison to classical machine-learning baselines by including U-Nets.Models are trained on simulated trajectories spanning four current families(constant,triangular,pulse-train,and Gaussian-random-field)and a full range of State-of-Charge(SOC)(0%to 100%).DeepONet accurately replicates constant-current behaviour but struggles with more dynamic loads.The basic FNO maintains mesh invariance and keeps concentration errors below 1%,with voltage mean-absolute errors under 1.7mV across all load types.Introducing parameter embedding marginally increases error but enables generalisation to varying radii and diffusivities.PE-FNO executes approximately 200 times faster than a 16-thread SPM solver.Consequently,PE-FNO’s capabilities in inverse tasks are explored in a parameter estimation task with Bayesian optimisation,recovering anode and cathode diffusivities with 1.14%and 8.4%mean absolute percentage error,respectively,and 0.5918 percentage points higher error in comparison with classical methods.These results pave the way for neural operators to meet the accuracy,speed and parametric flexibility demands of real-time battery management,design-of-experiments and large-scale inference.PE-FNO outperforms conventional neural surrogates,offering a practical path towards high-speed and high-fidelity electrochemical digital twins.展开更多
There is an increasing demand for real-time data-driven fault diagnosis of lithium-ion batteries that can predict battery faults at an early stage to avoid safety issues and improve battery reliability.However,such pr...There is an increasing demand for real-time data-driven fault diagnosis of lithium-ion batteries that can predict battery faults at an early stage to avoid safety issues and improve battery reliability.However,such prediction methods require large amounts of data,generally obtained through experiments or during the operation phase,resulting in substantial economic and time efforts.In this context,generating realistic battery pack data that covers all sensor values a battery management system receives,as well as including fault models,is of particular interest and can mitigate the need to perform extensive laboratory testing.This paper focuses on the systematic development of a data generation platform capable of simulating a large scale of battery packs with random battery faults and generating big data for the following battery fault diagnostics.Initially,the electrical,thermal,and aging modeling of a battery pack is performed.After this,four types of faults,namely hard short circuit,soft short circuit,abnormal internal resistance,and abnormal contact resistance,are modeled using equivalent circuit models.To generate realistic data,both cell-to-cell variations and pack-level variations are considered.Variations included are,for example,the manufacturing quality,temperatures,aging processes,road conditions,state of charge,and fault severity.By combining the battery pack models,fault models,and the different variations through Monte Carlo simulations,a large data set representing different packs with varying levels of inconsistencies is generated.展开更多
基金funding from the project“SPEED”(03XP0585)funded by the German Federal Ministry of ResearchTech-nology and Space(BMFTR)and the project“ADMirABLE”(03ETE053E)funded by the German Federal Ministry for Economic Affairs and Energy(BMWE)support of Shell Research UK Ltd.for the Ph.D.studentship of Amir Ali Panahi and the EPSRC Faraday Institution Multi-Scale Modelling Project(FIRG084).
文摘Digital twins of lithium-ion batteries are increasingly used to enable predictive monitoring,control,and design at system scale.Increasing their capabilities involves improving their physical fidelity while maintaining sub-millisecond computational speed.In this work,we introduce machine learning surrogates that learn physical dynamics.Specifically,we benchmark three operator-learning surrogates for the Single Particle Model(SPM):Deep Operator Networks(DeepONets),Fourier Neural Operators(FNOs)and a newly proposed parameter-embedded Fourier Neural Operator(PE-FNO),which conditions each spectral layer on particle radius and solid-phase diffusivity.We extend the comparison to classical machine-learning baselines by including U-Nets.Models are trained on simulated trajectories spanning four current families(constant,triangular,pulse-train,and Gaussian-random-field)and a full range of State-of-Charge(SOC)(0%to 100%).DeepONet accurately replicates constant-current behaviour but struggles with more dynamic loads.The basic FNO maintains mesh invariance and keeps concentration errors below 1%,with voltage mean-absolute errors under 1.7mV across all load types.Introducing parameter embedding marginally increases error but enables generalisation to varying radii and diffusivities.PE-FNO executes approximately 200 times faster than a 16-thread SPM solver.Consequently,PE-FNO’s capabilities in inverse tasks are explored in a parameter estimation task with Bayesian optimisation,recovering anode and cathode diffusivities with 1.14%and 8.4%mean absolute percentage error,respectively,and 0.5918 percentage points higher error in comparison with classical methods.These results pave the way for neural operators to meet the accuracy,speed and parametric flexibility demands of real-time battery management,design-of-experiments and large-scale inference.PE-FNO outperforms conventional neural surrogates,offering a practical path towards high-speed and high-fidelity electrochemical digital twins.
基金funding from the research project“Safe-DaBatt”(03EMF0409A)funded by the German Federal Ministry of Digital and Transport(BMDV).
文摘There is an increasing demand for real-time data-driven fault diagnosis of lithium-ion batteries that can predict battery faults at an early stage to avoid safety issues and improve battery reliability.However,such prediction methods require large amounts of data,generally obtained through experiments or during the operation phase,resulting in substantial economic and time efforts.In this context,generating realistic battery pack data that covers all sensor values a battery management system receives,as well as including fault models,is of particular interest and can mitigate the need to perform extensive laboratory testing.This paper focuses on the systematic development of a data generation platform capable of simulating a large scale of battery packs with random battery faults and generating big data for the following battery fault diagnostics.Initially,the electrical,thermal,and aging modeling of a battery pack is performed.After this,four types of faults,namely hard short circuit,soft short circuit,abnormal internal resistance,and abnormal contact resistance,are modeled using equivalent circuit models.To generate realistic data,both cell-to-cell variations and pack-level variations are considered.Variations included are,for example,the manufacturing quality,temperatures,aging processes,road conditions,state of charge,and fault severity.By combining the battery pack models,fault models,and the different variations through Monte Carlo simulations,a large data set representing different packs with varying levels of inconsistencies is generated.
