Online Estimating State of Health of Lithium-Ion Batteries Using Hierarchical Extreme Learning Machine

Chen, Lin; Ding, Yunhui; Wang, Huimin; Wang, Yijue; Liu, Bohao; Wu, Shuxiao; Pan, Haihong

doi:10.1109/tte.2021.3107727

Cited by 39 publications

(8 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…34 Chen et al established a hierarchical extreme learning machine (HELM) framework to improve the robustness of the model, and firstly used the increase in average ohm resistance as health indicator to characterize battery aging. 35 The estimation framework based the gated recurrent unit neural network (RNN), the exogenous inputs neural network (EINN) and the deep neural network (DNN) also have been verified processing strong estimation reliability. 9,36,37 Furthermore, as more novel and powerful NNs appearing, estimation methods with better adaptability and less computational cost need to be explored.…”

Section: The Soh Estimation Methodsmentioning

confidence: 99%

A Polak-Ribière-Polyak Conjugate Gradient Algorithm Optimized Broad Learning System for Lithium-ion Battery State of Health Estimation

Gu¹,

Wang²,

Li³

2022

J. Electrochem. Soc.

View full text Add to dashboard Cite

Accurate state of health (SOH) estimation plays a significant role in battery management systems (BMS). This paper investigates a Polak-Ribière-Polyak conjugate gradient (PRPCG) algorithm optimized broad learning system (BLS) for lithium-ion battery (LIB) SOH estimation. First, effective health indicators (HIs) are extracted from the voltage curve in the constant current charge process. Second, a hybrid four layers BLS structure with mapped feature nodes and enhancement nodes connecting to the output is established to build both the linear and nonlinear relationships between the HIs and SOH, in which only the output weights require to be trained. Again, the PRPCG algorithm is adopted for searching optimal output weights without matrix inverse calculation during the training process. Furthermore, certain Gaussian noises are added to enhance the training data for solving the locally low accuracy problem. Finally, under the Oxford battery degradation data set, experiments validate the investigated algorithm has high accuracy in SOH estimation with the mean absolute error (MAE) below 1%. The enhanced data can efficiently improve the model generalization ability.

show abstract

Section: The Soh Estimation Methodsmentioning

confidence: 99%

A Polak-Ribière-Polyak Conjugate Gradient Algorithm Optimized Broad Learning System for Lithium-ion Battery State of Health Estimation

Gu¹,

Wang²,

Li³

2022

J. Electrochem. Soc.

View full text Add to dashboard Cite

show abstract

“…By constructing a recursive least-squares with a forgetting factor for the Thevenin model, rapid HI extraction can be realized and efficient real-time SOH prediction can be consequently achieved. [99] Though feature engineering is effective for data reduction, data quality and operation conditions need to be considered before feature extraction. Moreover, current feature engineering methods largely lean upon domain knowledge or experience, automated feature engineering is regarded as an important task in future research.…”

Section: Data Quality and Availabilitymentioning

confidence: 99%

Data‐Driven Online Prognosis of Rechargeable Batteries: Prospect and Perspective

Liu,

Wang,

Liu

et al. 2024

Batteries & Supercaps

View full text Add to dashboard Cite

Along with the growing popularity of electric vehicles (EVs) and smart grids, rechargeable batteries are playing an increasingly important role in the field of energy storage. To ensure a safe and stable operation, it remains essential to estimate the states of batteries accurately and efficiently in advance. Herein, we provide a perspective on data‐driven online prognosis of rechargeable batteries, where its scope and superiorities are first introduced. Four promising application scenarios in real world including battery manufacture, EVs, smart grid, and battery recycling are then discussed, followed by the challenges that require further investigation. We anticipate this perspective to attract more interest to this field, to illuminate potential directions for future researches, and to broaden the context of data‐driven physical and engineering sciences.

show abstract

“…36, the author proposed a new joint estimation algorithm based on the fractional order battery model through the electrochemical model, which was used to simultaneously estimate the internal resistance and capacity attenuation as indicators of the state of health (SOH). Data-driven SOH estimation algorithms have also been proposed in large numbers, such as extreme learning machine (ELM), 37,38 support vector machine (SVM), 39,40 Long short-term memory network (LSTM). [41][42][43] In particular, the generative adversarial network (GAN) used in image and speech recognition was also used in SOH estimation.…”

Section: List Of Symbolsmentioning

confidence: 99%

An Adaptive Double Extended Kalman Filter Algorithm Based on Incremental Change Rate for Co-estimation of Battery SOC and Capacity

Lei¹,

Wu²,

Li³

et al. 2023

J. Electrochem. Soc.

View full text Add to dashboard Cite

The battery state of charge (SOC) and capacity are important state management indicators of the battery management system, and their estimation accuracy directly affects the safety of power battery use and the driver's driving experience. Since the increment change rate of the estimated variable can reflect the changing trend of the estimated variable, an extended Kalman filter algorithm based on the increment change rate is proposed, on this basis, and an adaptive double-extended Kalman filter algorithm based on incremental change rate is constructed for the co-estimation of SOC and capacity of batteries. The tests under various operating conditions show that the target algorithm proposed in this paper has greater advantages over the traditional adaptive double-extended Kalman filter algorithm, and the maximum absolute error value (MAE) and root mean square error (RMSE) of the target algorithm can be reduced by 36.3% and 74.4% (SOC), 95.5% and 97.6% (capacity) compared with the traditional adaptive double-extended Kalman filter algorithm under DST operating conditions; The MAE and RMSE of the target algorithm can be reduced by 79.1% and 92.3% (SOC), 95.4% and 96.2% (capacity) under BBDST operating conditions.

show abstract

Online Estimating State of Health of Lithium-Ion Batteries Using Hierarchical Extreme Learning Machine

Cited by 39 publications

References 38 publications

A Polak-Ribière-Polyak Conjugate Gradient Algorithm Optimized Broad Learning System for Lithium-ion Battery State of Health Estimation

A Polak-Ribière-Polyak Conjugate Gradient Algorithm Optimized Broad Learning System for Lithium-ion Battery State of Health Estimation

Data‐Driven Online Prognosis of Rechargeable Batteries: Prospect and Perspective

An Adaptive Double Extended Kalman Filter Algorithm Based on Incremental Change Rate for Co-estimation of Battery SOC and Capacity

Contact Info

Product

Resources

About