loading page

Impact of Sampling Interval Variation using Interpolation Techniques on Lithium-Ion Battery Parameter Estimation and Modelling for Real Driving Profile to Facilitate Remote Data Processing
  • Pratik Pradhan,
  • Dr. Aurobinda Panda,
  • Dipanjan Pradhan
Pratik Pradhan
National Institute of Technology Sikkim

Corresponding Author:[email protected]

Author Profile
Dr. Aurobinda Panda
National Institute of Technology Sikkim
Author Profile
Dipanjan Pradhan
Sikkim Manipal Institute of Technology - Majitar Technical Campus
Author Profile


The Battery Management System’s (BMS) precision in estimating battery state depends on the accuracy of the Equivalent Circuit Model (ECM) parameters, which is affected by the sampling interval. This study investigates the impact of sampling interval on Lithium-ion parameter estimation and modelling using interpolation techniques. Five simple interpolation methods (Linear, S-Linear, Quadratic, Cubic, and Akima) are applied to interpolate Hybrid Pulse Power Characterization (HPPC) data from 1s interval to 0.5s and 0.25s intervals and down sampled to 2s. The interpolated and down sampled data is used to estimate ECM parameters, which are then used to fit battery aging data from the Electric Vehicle Urban Dynamometer Driving Schedule (UDDS) for 1RC and 2RC models. Statistical analysis is employed to compare the results. The findings demonstrate that not all interpolation techniques perform equally. Quadratic interpolation exhibits poor performance, particularly in the 1RC model with a 0.5s interval, resulting in significant errors. Linear and S-Linear interpolation show comparable performance to non-interpolated data with R2 score difference in the range of 0.000361 for 1RC model and 0.000061 for 2RC model closely followed by Akima. The error distribution of Quadratic and Cubic interpolation is right skewed, indicating a higher frequency of negative errors. Overall, Linear and S-Linear interpolation perform well especially for smaller interpolated intervals, providing accurate results with minimal processing time effectively capturing slow-changing charge cycle and fast-changing variations in the UDDS cycle for both models. Hence, low size acquired data can be transferred to cloud efficiently to be interpolated with proper interpolation technique at a suitable sampling period, complex and novel parameter estimation algorithms to give accurate internal parameter estimation reducing unit hardware cost in vehicles.