TY - JOUR
T1 - Comparison of Kalman Filters for State Estimation Based on Computational Complexity of Li-Ion Cells
AU - Khalid, Areeb
AU - Kashif, Syed Abdul Rahman
AU - Ain, Noor Ul
AU - Awais, Muhammad
AU - Ali, Majid
AU - Carreño, Jorge El Mariachet
AU - Vasquez, Juan C.
AU - Guerrero, Josep M.
AU - Khan, Baseem
N1 - Publisher Copyright:
© 2023 by the authors.
PY - 2023/3/14
Y1 - 2023/3/14
N2 - Over the last few decades, lithium-ion batteries have grown in importance for the use of many portable devices and vehicular applications. It has been seen that their life expectancy is much more effective if the required conditions are met. In one of the required conditions, accurately estimating the battery’s state of charge (SOC) is one of the important factors. The purpose of this research paper is to implement the probabilistic filter algorithms for SOC estimation; however, there are challenges associated with that. Generally, for the battery to be effective the Bayesian estimation algorithms are required, which are recursively updating the probability density function of the system states. To address the challenges associated with SOC estimation, the research paper goes further into the functions of the extended Kalman filter (EKF) and sigma point Kalman filter (SPKF). The function of both of these filters will be able to provide an accurate estimation. Further studies are required for these filters’ performance, robustness, and computational complexity. For example, some filters might be accurate, might not be robust, and/or not implementable on a simple microcontroller in a vehicle’s battery management system (BMS). A comparison is made between the EKF and SPKF by running simulations in MATLAB. It is found that the SPKF has an obvious advantage over the EKF in state estimation. Within the SPKF, the sub-filter, the central difference Kalman filter (CDKF), can be considered as an alternative to the EKF for state estimation in battery management systems for electric vehicles. However, there are implications to this which include the compromise of computational complexity in which a more sophisticated micro-controller is required.
AB - Over the last few decades, lithium-ion batteries have grown in importance for the use of many portable devices and vehicular applications. It has been seen that their life expectancy is much more effective if the required conditions are met. In one of the required conditions, accurately estimating the battery’s state of charge (SOC) is one of the important factors. The purpose of this research paper is to implement the probabilistic filter algorithms for SOC estimation; however, there are challenges associated with that. Generally, for the battery to be effective the Bayesian estimation algorithms are required, which are recursively updating the probability density function of the system states. To address the challenges associated with SOC estimation, the research paper goes further into the functions of the extended Kalman filter (EKF) and sigma point Kalman filter (SPKF). The function of both of these filters will be able to provide an accurate estimation. Further studies are required for these filters’ performance, robustness, and computational complexity. For example, some filters might be accurate, might not be robust, and/or not implementable on a simple microcontroller in a vehicle’s battery management system (BMS). A comparison is made between the EKF and SPKF by running simulations in MATLAB. It is found that the SPKF has an obvious advantage over the EKF in state estimation. Within the SPKF, the sub-filter, the central difference Kalman filter (CDKF), can be considered as an alternative to the EKF for state estimation in battery management systems for electric vehicles. However, there are implications to this which include the compromise of computational complexity in which a more sophisticated micro-controller is required.
KW - Computational complexity
KW - Electric vehicle (EV's)
KW - Hybrid Electric Vehicle
KW - Kalman Filter
KW - State Estimation
KW - State of charge (SOC)
KW - Kalman filter
KW - unscented Kalman filter
KW - state of charge
KW - electric vehicle
KW - extended Kalman filter
KW - hybrid electric vehicle
KW - state estimation
KW - central difference Kalman filter
KW - computational complexity
UR - http://www.scopus.com/inward/record.url?scp=85151469546&partnerID=8YFLogxK
U2 - 10.3390/en16062710
DO - 10.3390/en16062710
M3 - Journal article
AN - SCOPUS:85151469546
SN - 1996-1073
VL - 16
JO - Energies
JF - Energies
IS - 6
M1 - 2710
ER -