This article presents an up-to-date tutorial review of nonlinear Bayesian estimation. State estimation for nonlinear systems has been a challenge encountered in a wide range of engineering fields, attracting decades o...This article presents an up-to-date tutorial review of nonlinear Bayesian estimation. State estimation for nonlinear systems has been a challenge encountered in a wide range of engineering fields, attracting decades of research effort. To date,one of the most promising and popular approaches is to view and address the problem from a Bayesian probabilistic perspective,which enables estimation of the unknown state variables by tracking their probabilistic distribution or statistics(e.g., mean and covariance) conditioned on a system's measurement data.This article offers a systematic introduction to the Bayesian state estimation framework and reviews various Kalman filtering(KF)techniques, progressively from the standard KF for linear systems to extended KF, unscented KF and ensemble KF for nonlinear systems. It also overviews other prominent or emerging Bayesian estimation methods including Gaussian filtering, Gaussian-sum filtering, particle filtering and moving horizon estimation and extends the discussion of state estimation to more complicated problems such as simultaneous state and parameter/input estimation.展开更多
Dynamic time-varying operational conditions pose great challenge to the estimation of system remaining useful life （RUL） for the deteriorating systems. This paper presents a method based on probabilistic and stochas...Dynamic time-varying operational conditions pose great challenge to the estimation of system remaining useful life （RUL） for the deteriorating systems. This paper presents a method based on probabilistic and stochastic approaches to estimate system RUL for periodically moni- tored degradation processes with dynamic time-varying operational conditions and condition- specific failure zones. The method assumes that the degradation rate is influenced by specific oper- ational condition and moreover, the transition between different operational conditions plays the most important role in affecting the degradation process. These operational conditioqs are assumed to evolve as a discrete-time Markov chain （DTMC）. The failure thresholds are also determined by specific operational conditions and described as different failure zones. The 2008 PHM Conference Challenge Data is utilized to illustrate our method, which contains mass sensory signals related to the degradation process of a commercial turbofan engine. The RUE estimation method using the sensor measurements of a single sensor was first developed, and then multiple vital sensors were selected through a particular optimization procedure in order to increase the prediction accuracy. The effectiveness and advantages of the proposed method are presented in a comparison with exist- ing methods for the same dataset.展开更多
Efficient and accurate health state estimation is crucial for lithium-ion battery(LIB)performance monitoring and economic evaluation.Effectively estimating the health state of LIBs online is the key but is also the mo...Efficient and accurate health state estimation is crucial for lithium-ion battery(LIB)performance monitoring and economic evaluation.Effectively estimating the health state of LIBs online is the key but is also the most difficult task for energy storage systems.With high adaptability and applicability advantages,battery health state estimation based on data-driven techniques has attracted extensive attention from researchers around the world.Artificial neural network(ANN)-based methods are often used for state estimations of LIBs.As one of the ANN methods,the Elman neural network(ENN)model has been improved to estimate the battery state more efficiently and accurately.In this paper,an improved ENN estimation method based on electrochemical impedance spectroscopy(EIS)and cuckoo search(CS)is established as the EIS-CS-ENN model to estimate the health state of LIBs.Also,the paper conducts a critical review of various ANN models against the EIS-CS-ENN model.This demonstrates that the EIS-CS-ENN model outperforms other models.The review also proves that,under the same conditions,selecting appropriate health indicators(HIs)according to the mathematical modeling ability and state requirements are the keys in estimating the health state efficiently.In the calculation process,several evaluation indicators are adopted to analyze and compare the modeling accuracy with other existing methods.Through the analysis of the evaluation results and the selection of HIs,conclusions and suggestions are put forward.Also,the robustness of the EIS-CS-ENN model for the health state estimation of LIBs is verified.展开更多
Relative-risk models are often used to characterize the relationship between survival time and time-dependent covariates. When the covariates are observed, the estimation and asymptotic theory for parameters of intere...Relative-risk models are often used to characterize the relationship between survival time and time-dependent covariates. When the covariates are observed, the estimation and asymptotic theory for parameters of interest are available; challenges remain when missingness occurs. A popular approach at hand is to jointly model survival data and longitudinal data. This seems efficient, in making use of more information, but the rigorous theoretical studies have long been ignored. For both additive risk models and relative-risk models, we consider the missing data nonignorable. Under general regularity conditions, we prove asymptotic normality for the nonparametric maximum likelihood estimators.展开更多
文摘This article presents an up-to-date tutorial review of nonlinear Bayesian estimation. State estimation for nonlinear systems has been a challenge encountered in a wide range of engineering fields, attracting decades of research effort. To date,one of the most promising and popular approaches is to view and address the problem from a Bayesian probabilistic perspective,which enables estimation of the unknown state variables by tracking their probabilistic distribution or statistics(e.g., mean and covariance) conditioned on a system's measurement data.This article offers a systematic introduction to the Bayesian state estimation framework and reviews various Kalman filtering(KF)techniques, progressively from the standard KF for linear systems to extended KF, unscented KF and ensemble KF for nonlinear systems. It also overviews other prominent or emerging Bayesian estimation methods including Gaussian filtering, Gaussian-sum filtering, particle filtering and moving horizon estimation and extends the discussion of state estimation to more complicated problems such as simultaneous state and parameter/input estimation.
基金supported by the Fundamental Research Funds for the Central Universities(No.YWF-14-ZDHXY-16)
文摘Dynamic time-varying operational conditions pose great challenge to the estimation of system remaining useful life （RUL） for the deteriorating systems. This paper presents a method based on probabilistic and stochastic approaches to estimate system RUL for periodically moni- tored degradation processes with dynamic time-varying operational conditions and condition- specific failure zones. The method assumes that the degradation rate is influenced by specific oper- ational condition and moreover, the transition between different operational conditions plays the most important role in affecting the degradation process. These operational conditioqs are assumed to evolve as a discrete-time Markov chain （DTMC）. The failure thresholds are also determined by specific operational conditions and described as different failure zones. The 2008 PHM Conference Challenge Data is utilized to illustrate our method, which contains mass sensory signals related to the degradation process of a commercial turbofan engine. The RUE estimation method using the sensor measurements of a single sensor was first developed, and then multiple vital sensors were selected through a particular optimization procedure in order to increase the prediction accuracy. The effectiveness and advantages of the proposed method are presented in a comparison with exist- ing methods for the same dataset.
基金supported by the National Natural Science Foundation of China(No.62173281 and No.61801407)the Sichuan Science and Technology Pro-gram(No.2019YFG0427 and No.2023YFG0108)+1 种基金the China Scholarship Council(No.201908515099)the Fund of Robot Technology used for the Special Environment Key Laboratory of Sichuan Province(No.18kftk03).
文摘Efficient and accurate health state estimation is crucial for lithium-ion battery(LIB)performance monitoring and economic evaluation.Effectively estimating the health state of LIBs online is the key but is also the most difficult task for energy storage systems.With high adaptability and applicability advantages,battery health state estimation based on data-driven techniques has attracted extensive attention from researchers around the world.Artificial neural network(ANN)-based methods are often used for state estimations of LIBs.As one of the ANN methods,the Elman neural network(ENN)model has been improved to estimate the battery state more efficiently and accurately.In this paper,an improved ENN estimation method based on electrochemical impedance spectroscopy(EIS)and cuckoo search(CS)is established as the EIS-CS-ENN model to estimate the health state of LIBs.Also,the paper conducts a critical review of various ANN models against the EIS-CS-ENN model.This demonstrates that the EIS-CS-ENN model outperforms other models.The review also proves that,under the same conditions,selecting appropriate health indicators(HIs)according to the mathematical modeling ability and state requirements are the keys in estimating the health state efficiently.In the calculation process,several evaluation indicators are adopted to analyze and compare the modeling accuracy with other existing methods.Through the analysis of the evaluation results and the selection of HIs,conclusions and suggestions are put forward.Also,the robustness of the EIS-CS-ENN model for the health state estimation of LIBs is verified.
基金funded by National Natural Science Foundation of China(NSFC No.11771241)Natural Science Foundation of Anhui Province(No.1708085QA14)
文摘Relative-risk models are often used to characterize the relationship between survival time and time-dependent covariates. When the covariates are observed, the estimation and asymptotic theory for parameters of interest are available; challenges remain when missingness occurs. A popular approach at hand is to jointly model survival data and longitudinal data. This seems efficient, in making use of more information, but the rigorous theoretical studies have long been ignored. For both additive risk models and relative-risk models, we consider the missing data nonignorable. Under general regularity conditions, we prove asymptotic normality for the nonparametric maximum likelihood estimators.
