Article
Engineering, Electrical & Electronic
Yuhao Liu, Marzieh Ajirak, Petar M. Djuric
Summary: This method tackles the problem of sequential estimation in state-space and deep state-space models. It utilizes Gaussian and deep Gaussian processes implemented via random feature-based Gaussian processes to estimate functions and latent processes. The method addresses both highly nonlinear and conditionally linear unknowns in the models.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Maoran Zhu, Yuanxin Wu
Summary: A new framework for continuous-time maximum a posteriori estimation based on Chebyshev polynomial optimization is proposed in this paper. It transforms the nonlinear continuous-time state estimation into a constant parameter optimization problem. The proposed method achieves significantly improved accuracy compared to existing filters and approaches the Cramer-Rao lower bound.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Robotics
Sven Lilge, Timothy D. Barfoot, Jessica Burgner-Kahrs
Summary: This paper addresses the problem of state estimation for continuum robots using a Gaussian process regression approach, resulting in accurate and continuous shape estimation. Through simulations and experiments, the effectiveness and feasibility of the method are demonstrated.
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH
(2022)
Article
Automation & Control Systems
Chiwoo Park
Summary: This paper presents a Gaussian process model for estimating piecewise continuous regression functions. Unlike conventional GP regression methods, this approach partitions the local data into pieces using a local data partitioning function to improve modeling flexibility. The advantages of using this approach over traditional methods are demonstrated through various experiments and data studies.
JOURNAL OF MACHINE LEARNING RESEARCH
(2022)
Article
Robotics
Jacob C. Johnson, Joshua G. Mangelson, Randal W. Beard
Summary: Continuous-time estimation using splines on Lie groups has gained attention in the literature, but their computational cost limits their use mainly to offline applications. Motivated by trajectory planning, we develop a new estimation technique that defines splines in the flat output space of dynamic systems, enabling simple and effective inclusion of system inputs.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Energy & Fuels
Xin Xiong, Yujie Wang, Kaiquan Li, Zonghai Chen
Summary: This paper proposes a method for state of health (SOH) estimation using random charging data. The missing charging data is accurately reconstructed by the Gaussian Process Regression (GPR) model. The Long-Short-Term Memory (LSTM) estimation model is used to select the charging time sequence as input, avoiding complex feature extraction. Experimental results show high accuracy in laboratory and real-world datasets.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Energy & Fuels
Jiwei Wang, Zhongwei Deng, Tao Yu, Akihiro Yoshida, Lijun Xu, Guoqing Guan, Abuliti Abudula
Summary: This study proposes a data-driven method for state of health estimation of lithium-ion batteries. The method utilizes Gaussian process regression models based on the analysis of aging characteristics and extraction of health indicators from battery charging and discharging curves. The experiments demonstrate that the proposed method achieves satisfactory estimation results in terms of accuracy, generalization ability, and robustness.
JOURNAL OF ENERGY STORAGE
(2022)
Article
Energy & Fuels
Xiaobing Chen, Xiaoping Chen, Xiwen Chen
Summary: The article proposes a novel non-parametric model for lithium battery based on Gaussian process regression, which is built by learning experimental battery datasets offline. The unscented Kalman filter algorithm is used to improve state of charge estimation accuracy by correcting voltage errors. The mean and covariance function of the prediction model are employed to indicate the uncertainty model.
INTERNATIONAL JOURNAL OF ENERGY RESEARCH
(2021)
Article
Green & Sustainable Science & Technology
Wai Hou Lio, Ang Li, Fanzhong Meng
Summary: This paper presents a framework of rotor effective wind speed estimator design that eliminates the need for pre-computed power coefficient mapping. By reconstructing the mapping using Gaussian process regression and implementing an extended Kalman filter, optimal estimation from standard turbine measurements is achieved, significantly reducing estimation errors.
Article
Materials Science, Multidisciplinary
Aihua Ran, Ming Cheng, Shuxiao Chen, Zheng Liang, Zihao Zhou, Guangmin Zhou, Feiyu Kang, Xuan Zhang, Baohua Li, Guodan Wei
Summary: The article introduces a method for effectively estimating the remaining capacity of secondary lithium-ion batteries using real-time short pulse tests combined with data-driven Gaussian process regression algorithm, with an average accuracy of up to 95%. Compared to traditional long charge/discharge tests, this method can greatly reduce testing time.
ENERGY & ENVIRONMENTAL MATERIALS
(2023)
Article
Robotics
Daniil Lisus, Charles Champagne Cossette, Mohammed Shalaby, James Richard Forbes
Summary: This letter demonstrates how to estimate robot heading using UWB range and RSS measurements, by learning a data-driven relationship and combining with a gyroscope and an invariant extended Kalman filter.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Thermodynamics
Haiyan Jin, Ningmin Cui, Lei Cai, Jinhao Meng, Junxin Li, Jichang Peng, Xinchao Zhao
Summary: This paper focuses on addressing the challenge of creating a high-performance model with a compact structure for battery State-of-Health (SOH) estimation. The proposed method utilizes an auto-configurable Gaussian Process Regression (GPR) with elastic feature construction and an evolutionary framework to eliminate the impact of kernels. Additionally, a hierarchical feature construction strategy reduces complexity.
Article
Thermodynamics
Yong Zhou, Guangzhong Dong, Qianqian Tan, Xueyuan Han, Chunlin Chen, Jingwen Wei
Summary: Due to the complex behaviors of lithium-ion batteries, accurately estimating their state-of-health remains a critical challenge. Most existing battery health prognosis methods focus on low-frequency sampled time-domain response, which may not fully reflect the battery health status in automotive applications. This paper proposes a data-driven method using high and medium frequency impedance spectroscopy data to estimate battery state-of-health. Experimental results demonstrate the high accuracy and robustness of the proposed method, with an estimation error of 1.12%.
Article
Optics
Richard J. E. Abrantes, Yun-Wen Mao, David D. W. Ren
Summary: This study introduces Gaussian process regression as an alternative method for calculating and storing transition rate coefficient values in collisional-radiative simulations. Compared to traditional methods, this approach has lower relative errors by observing and fitting the integrated rate coefficient profiles.
JOURNAL OF QUANTITATIVE SPECTROSCOPY & RADIATIVE TRANSFER
(2022)
Article
Chemistry, Multidisciplinary
Kanta Matsunaga, Takuto Harada, Shintaro Harada, Akinori Sato, Shota Terai, Mutsunori Uenuma, Tomoyuki Miyao, Yukiharu Uraoka
Summary: This study focuses on predicting the quality of the interface between an insulator and GaN semiconductors for potential application of GaN power semiconductor devices. The UV/O-3-treated interfaces showed improved performance when using data of untreated interfaces. The automatic relevance vector-based Gaussian process regression model exhibited high predictive performance and reasonable uncertainty.