Background:An increasing number of ecological processes have been incorporated into Earth system models.However,model evaluations usually lag behind the fast development of models,leading to a pervasive simulation unc...Background:An increasing number of ecological processes have been incorporated into Earth system models.However,model evaluations usually lag behind the fast development of models,leading to a pervasive simulation uncertainty in key ecological processes,especially the terrestrial carbon(C)cycle.Traceability analysis provides a theoretical basis for tracking and quantifying the structural uncertainty of simulated C storage in models.Thus,a new tool of model evaluation based on the traceability analysis is urgently needed to efficiently diagnose the sources of inter-model variations on the terrestrial C cycle in Earth system models.Methods:A new cloud-based model evaluation platform,i.e.,the online traceability analysis system for model evaluation(TraceME v1.0),was established.The TraceME was applied to analyze the uncertainties of seven models from the Coupled Model Intercomparison Project(CMIP6).Results:The TraceME can effectively diagnose the key sources of different land C dynamics among CMIIP6 models.For example,the analyses based on TraceME showed that the estimation of global land C storage varied about 2.4 folds across the seven CMIP6 models.Among all models,IPSL-CM6A-LR simulated the lowest land C storage,which mainly resulted from its shortest baseline C residence time.Over the historical period of 1850–2014,gross primary productivity and baseline C residence time were the major uncertainty contributors to the inter-model variation in ecosystem C storage in most land grid cells.Conclusion:TraceME can facilitate model evaluation by identifying sources of model uncertainty and provides a new tool for the next generation of model evaluation.展开更多
Recently,intelligent fault diagnosis based on deep learning has been extensively investigated,exhibiting state-of-the-art performance.However,the deep learning model is often not truly trusted by users due to the lack...Recently,intelligent fault diagnosis based on deep learning has been extensively investigated,exhibiting state-of-the-art performance.However,the deep learning model is often not truly trusted by users due to the lack of interpretability of“black box”,which limits its deployment in safety-critical applications.A trusted fault diagnosis system requires that the faults can be accurately diagnosed in most cases,and the human in the deci-sion-making loop can be found to deal with the abnormal situa-tion when the models fail.In this paper,we explore a simplified method for quantifying both aleatoric and epistemic uncertainty in deterministic networks,called SAEU.In SAEU,Multivariate Gaussian distribution is employed in the deep architecture to compensate for the shortcomings of complexity and applicability of Bayesian neural networks.Based on the SAEU,we propose a unified uncertainty-aware deep learning framework(UU-DLF)to realize the grand vision of trustworthy fault diagnosis.Moreover,our UU-DLF effectively embodies the idea of“humans in the loop”,which not only allows for manual intervention in abnor-mal situations of diagnostic models,but also makes correspond-ing improvements on existing models based on traceability analy-sis.Finally,two experiments conducted on the gearbox and aero-engine bevel gears are used to demonstrate the effectiveness of UU-DLF and explore the effective reasons behind.展开更多
A Leeway-Trace model was established for the traceability analysis of drifting objects at sea.The model was based on the Leeway model which is a Monte Carlo-based ensemble trajectory model,and a method of realistic tr...A Leeway-Trace model was established for the traceability analysis of drifting objects at sea.The model was based on the Leeway model which is a Monte Carlo-based ensemble trajectory model,and a method of realistic traceability analysis was proposed in this study by using virtual spatiotemporal drift trajectory prediction.Here,measured data from a drifting buoy observation experiment in the northern South China Sea in April 2019,combined with surface current data obtained from the finite volume community ocean model(FVCOM),were used for the traceability analysis of humanoid buoys.The results were basically consistent with the observations,and the assimilation of measured current data can significantly improve the accuracy of the traceability analysis.Several sensitive experiments were designed to discuss the effects of wind and tide on the traceability analysis,and their results showed that the wind-driven current and the wind-induced leeway drift are both important to the traceability analysis.The effect of tidal currents on traceability could not be ignored even though they were much weaker than the residual currents in the experimental area of the northern South China Sea.展开更多
基金supported by the National Key R&D Program of China(2017YFA0604600)National Natural Science Foundation of China(31722009).
文摘Background:An increasing number of ecological processes have been incorporated into Earth system models.However,model evaluations usually lag behind the fast development of models,leading to a pervasive simulation uncertainty in key ecological processes,especially the terrestrial carbon(C)cycle.Traceability analysis provides a theoretical basis for tracking and quantifying the structural uncertainty of simulated C storage in models.Thus,a new tool of model evaluation based on the traceability analysis is urgently needed to efficiently diagnose the sources of inter-model variations on the terrestrial C cycle in Earth system models.Methods:A new cloud-based model evaluation platform,i.e.,the online traceability analysis system for model evaluation(TraceME v1.0),was established.The TraceME was applied to analyze the uncertainties of seven models from the Coupled Model Intercomparison Project(CMIP6).Results:The TraceME can effectively diagnose the key sources of different land C dynamics among CMIIP6 models.For example,the analyses based on TraceME showed that the estimation of global land C storage varied about 2.4 folds across the seven CMIP6 models.Among all models,IPSL-CM6A-LR simulated the lowest land C storage,which mainly resulted from its shortest baseline C residence time.Over the historical period of 1850–2014,gross primary productivity and baseline C residence time were the major uncertainty contributors to the inter-model variation in ecosystem C storage in most land grid cells.Conclusion:TraceME can facilitate model evaluation by identifying sources of model uncertainty and provides a new tool for the next generation of model evaluation.
基金supported in part by the National Natural Science Foundation of China(52105116)Science Center for gas turbine project(P2022-DC-I-003-001)the Royal Society award(IEC\NSFC\223294)to Professor Asoke K.Nandi.
文摘Recently,intelligent fault diagnosis based on deep learning has been extensively investigated,exhibiting state-of-the-art performance.However,the deep learning model is often not truly trusted by users due to the lack of interpretability of“black box”,which limits its deployment in safety-critical applications.A trusted fault diagnosis system requires that the faults can be accurately diagnosed in most cases,and the human in the deci-sion-making loop can be found to deal with the abnormal situa-tion when the models fail.In this paper,we explore a simplified method for quantifying both aleatoric and epistemic uncertainty in deterministic networks,called SAEU.In SAEU,Multivariate Gaussian distribution is employed in the deep architecture to compensate for the shortcomings of complexity and applicability of Bayesian neural networks.Based on the SAEU,we propose a unified uncertainty-aware deep learning framework(UU-DLF)to realize the grand vision of trustworthy fault diagnosis.Moreover,our UU-DLF effectively embodies the idea of“humans in the loop”,which not only allows for manual intervention in abnor-mal situations of diagnostic models,but also makes correspond-ing improvements on existing models based on traceability analy-sis.Finally,two experiments conducted on the gearbox and aero-engine bevel gears are used to demonstrate the effectiveness of UU-DLF and explore the effective reasons behind.
基金The National Natural Science Foundation of China under contract Nos 41376012,41076048 and 41275029。
文摘A Leeway-Trace model was established for the traceability analysis of drifting objects at sea.The model was based on the Leeway model which is a Monte Carlo-based ensemble trajectory model,and a method of realistic traceability analysis was proposed in this study by using virtual spatiotemporal drift trajectory prediction.Here,measured data from a drifting buoy observation experiment in the northern South China Sea in April 2019,combined with surface current data obtained from the finite volume community ocean model(FVCOM),were used for the traceability analysis of humanoid buoys.The results were basically consistent with the observations,and the assimilation of measured current data can significantly improve the accuracy of the traceability analysis.Several sensitive experiments were designed to discuss the effects of wind and tide on the traceability analysis,and their results showed that the wind-driven current and the wind-induced leeway drift are both important to the traceability analysis.The effect of tidal currents on traceability could not be ignored even though they were much weaker than the residual currents in the experimental area of the northern South China Sea.
