Article
Environmental Sciences
Marco Dal Molin, Dmitri Kavetski, Carlo Albert, Fabrizio Fenicia
Summary: Calibration of precipitation-streamflow models to streamflow signatures is an effective way to predict streamflow in ungauged basins. This study proposes a signature-based calibration method using an Approximate Bayesian Computation framework, which reliably estimates the uncertainty in ungauged catchments.
WATER RESOURCES RESEARCH
(2023)
Article
Physics, Multidisciplinary
Xuyou Li, Yanda Guo, Qingwen Meng
Summary: In this work, an improved maximum mixture correntropy Kalman filter (IMMCKF) is proposed to solve the non-Gaussian filtering problem for linear systems. By introducing the mixture correntropy method and a variational Bayesian approach, the proposed filter shows better performance in non-stationary noises.
Article
Engineering, Civil
Irina Georgievova, Martin Hanel, Petr Pavlik, Adam Vizina
Summary: In this study, the Kalman filter, a sequential data assimilation method, was used to estimate streamflow in an ungauged basin by combining observations from neighboring basins and model simulations. The methodology utilized the concept of concatenated upstream catchments to estimate unobserved states. Linear regression was applied to allocate the streamflow estimate to unknown sub-catchments, and the resulting error statistics were used in the Kalman filter application. Evaluation on 165 catchments in the Czech Republic showed that the proposed methodology improved streamflow estimations by an average of 40% compared to original simulations.
JOURNAL OF HYDROLOGY
(2023)
Article
Automation & Control Systems
Hongpo Fu, Zhenwei Li, Wei Huang, Yongmei Cheng, Tianyi Zhang
Summary: This work investigates the problem of state estimation for a class of nonlinear systems subject to randomly occurring measurement anomalies without prior statistical information. A novel measurement model is proposed, where the anomalous measurements and anomaly probability are modeled as Gaussian mixture distribution and Beta distribution, respectively. The model does not require a priori statistical knowledge of anomalous measurements and achieves the same performance as the classical cubature Kalman filter in the absence of measurement anomalies through adaptive learning of the anomaly probability. Variational Bayesian inference is employed to approximate the joint posterior distribution of the system state and unknown parameters, resulting in a robust filter. Numerical simulations demonstrate the effectiveness of the proposed filter.
Article
Environmental Sciences
Shailesh Kumar Singh, George A. Griffiths
Summary: The study addresses the challenging problem of predicting the time for a natural basin's outflow to decline from the average to a low flow value, offering new methods for accurate predictions. The developed models show high accuracy in predicting recession time and can be confidently applied elsewhere, with further testing recommended.
WATER RESOURCES RESEARCH
(2021)
Article
Automation & Control Systems
Shuanghu Qiao, Yunsheng Fan, Guofeng Wang, Dongdong Mu, Zhiping He
Summary: An improved strong tracking maximum correntropy criterion variational-Bayesian adaptive Kalman filter is proposed in this paper to address the performance degradation of current state estimation in the presence of slow-varying noise statistics. The inverse-Wishart distribution is adopted to model the unknown and time-varying measurement and process noise covariances. The varia-tional Bayesian method is used to estimate the noise covariances and system state, and the maximum correntropy criterion is employed to correct the filtering gain, improving the filtering performance of the proposed algorithm. Simulation results demonstrate that the proposed filter outperforms existing algorithms in terms of accuracy and convergence performance.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2023)
Article
Engineering, Electrical & Electronic
Haoshen Lin, Chen Hu
Summary: This paper investigates a noise covariance adaptive distributed Bayesian filter based on variational Bayesian inference method, approximating the joint posterior distribution of state and noises by recursively performing variational Bayesian expectation and maximization steps. It proposes a variational Bayesian based distributed adaptive cubature information filter to approximate Gaussian interval for effective estimation in cooperative object tracking problem.
Article
Engineering, Civil
Javier Senent-Aparicio, Patricia Jimeno-Saez, Raquel Martinez-Espana, Julio Perez-Sanchez
Summary: This study focuses on streamflow prediction in the Mino River basin in northwest Spain. A novel regionalisation approach is developed, which utilizes hydrological similarities between gauged and ungauged subbasins, as well as physiographic and climatic attributes, to predict streamflow. The results demonstrate satisfactory performance in the streamflow prediction, indicating the effectiveness of the regionalisation approach.
WATER RESOURCES MANAGEMENT
(2023)
Article
Multidisciplinary Sciences
Kevin Course, Prasanth B. Nair
Summary: This study presents a state estimation technique based on approximate Bayesian approach, which learns the missing terms and state estimation in the mathematical model. It enables state estimation for physical systems with partially or completely unknown dynamical equations.
Review
Engineering, Mechanical
Xiaoxu Wang, Haoran Cui, Tiancheng Li, Yan Liang, Zhengtao Ding
Summary: A new variational Gaussian regression filter is proposed in this paper by incorporating variational parameters into a linear parametric Gaussian regression process. The filtering evidence lower bound serves as a quantitative evaluation rule for different filters, and a relationship between F-ELBO and M-ELBO is identified. The accuracy performance improvement of VGRF can be theoretically explained based on these findings.
NONLINEAR DYNAMICS
(2021)
Article
Engineering, Electrical & Electronic
Baojian Yang, Binhan Du, Ning Li, Siyu Li, Zhiyong Shi
Summary: This paper proposes a centered error entropy based variational Bayesian adaptive and robust Kalman filter (CEEVBKF) to suppress outlier noise and estimate the unknown noise covariance adaptively. It improves the iterative efficiency and reduces the parameter sensitivity by jointly estimating the centered error entropy and variational Bayesian.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS
(2022)
Article
Computer Science, Interdisciplinary Applications
Alban de Lavenne, Tom Loree, Herve Squividant, Christophe Cudennec
Summary: This R package gathers methods developed and lessons learnt for estimating discharge of ungauged outlets using a runoff-runoff approach. It utilizes observed discharge from nearby gauged basins and a geomorphology-based deconvolution-convolution modeling approach. The package allows for the estimation and simulation of discharge series in targeted ungauged basins. The methodology has been tested and further evaluation, improvement, and operational applications are encouraged.
ENVIRONMENTAL MODELLING & SOFTWARE
(2023)
Article
Geosciences, Multidisciplinary
Qiong Wu, Yunfeng Ma, Dongsheng Li, Yuan Wang, Yanju Ji
Summary: The GATEM method is an efficient electromagnetic detection technique for geological exploration and groundwater exploration. However, the field data are often disturbed by various noises, which degrade the data quality. This study proposes a denoising algorithm based on the variational Bayesian-based adaptive Kalman filter (VBAKF) to effectively suppress noise in GATEM data.
JOURNAL OF APPLIED GEOPHYSICS
(2022)
Article
Automation & Control Systems
Zhengya Ma, Xiaoxu Wang, Mingyong Liu, Lixin Wang, Pu Gao, Gongmin Yan
Summary: This article introduces a new enabling variational inference filter (NEVIF) for the filtering problem with nonlinear measurements, which obtains the variational posterior by minimizing the Kullback-Leibler divergence. It analyzes the accuracy improvement and robustness of NEVIF compared to traditional methods, and develops a filtering evidence lower bound to evaluate the estimation accuracy performance. Numerical simulations demonstrate the efficiency and superiority of the proposed filters over existing filters.
INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL
(2022)
Article
Engineering, Electrical & Electronic
Wonkeun Youn, Nak Yong Ko, Stephen Gadsden, Hyun Myung
Summary: This article proposes a novel adaptive Kalman filter algorithm, which improves filtering performance by estimating the unknown measurement loss probability.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)