A novel extended Kalman filter-based battery internal and surface temperature estimation based on an improved electro-thermal model

2021 ◽  
Vol 41 ◽  
pp. 102854
Author(s):  
Hui Pang ◽  
Long Guo ◽  
Longxing Wu ◽  
Jiamin Jin ◽  
Fengqi Zhang ◽  
...  
Author(s):  
Song Chen ◽  
Fengjun Yan

The in-cylinder temperature information is critical for auto-ignition combustion control in diesel engines, but difficult to be directly accessed at low cost in production engines. Through investigating the thermodynamics of Tivc, cycle-by-cycle models are proposed in this paper for the estimation of in-cylinder temperature at the crank angle of intake valve closing (IVC), referred to as Tivc. An extended Kalman filter (EKF) based method was devised by utilizing the measurable temperature information from the intake and exhaust manifolds. Due to the fact that measured temperature signals by typical thermocouples have slow responses which can be modeled as first-order lags with varying time-constants, temperature signals need to be reconstructed in transient conditions. In the proposed EKF estimation method, this issue can be effectively addressed by analyzing the measurement errors and properly selecting the noises covariance matrices. The proposed estimation method was validated through a high-fidelity GT-power engine model.


Author(s):  
Sumukh Surya ◽  
Amit Bhesaniya ◽  
Aditya Gogate ◽  
Raghav Ankur ◽  
Vineeth Patil

AbstractCore temperature (Tc) estimation plays an important in role in establishing an effective thermal management system of a battery. In the present work, Tc of a lead acid (Pb) battery was estimated using a Kalman filter, based on a thermal model of the battery using convection resistances and capacitances. The governing equations based on measured surface temperature (Ts) and ambient temperature (Tamb) were derived. Since Tc cannot be measured directly, estimation technique was used to predict the same using measured Ts and Tamb. Five test cases for which the profiles of Tc versus time were available were analyzed. It was found that the errors in the predictions varied from 0.25 °C to 3.5 °C., depending on the nature of Tc profiles, with minimum errors when Tc has slow variations with time.


Sign in / Sign up

Export Citation Format

Share Document