Prediction of power battery health state based on data-driven Gaussian process regression algorithm

As one of the key functions of the battery management system, the prediction of the state of health (SOH) of the power battery of new energy vehicles is crucial for the management and maintenance of the power battery system and its safe use. In order to improve the prediction accuracy and generaliza...

Full description

Saved in:
Bibliographic Details
Published in:International journal of computers, communications & control Vol. 20; no. 3
Main Authors: Li, Wei, Cheng, Chen, Wang, Qian, Ma, Chang-song
Format: Journal Article
Language:English
Published: Oradea Agora University of Oradea 01.06.2025
Subjects:
Algorithms
Data correlation
Discharge
Electric potential
Gaussian process
Lithium batteries
Power management
Root-mean-square errors
Voltage
ISSN:1841-9836, 1841-9844
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:As one of the key functions of the battery management system, the prediction of the state of health (SOH) of the power battery of new energy vehicles is crucial for the management and maintenance of the power battery system and its safe use. In order to improve the prediction accuracy and generalization ability of power battery health state, a SOH prediction method combining data-driven and gaussian process regression (GPR) based on data correlation was proposed. The normalized values of discharge capacity, voltage range, voltage variance, internal resistance range, internal resistance variance and final discharge temperature of lithium battery were selected to analyze the relevant characteristics of SOH. The aging model of power battery was constructed by GPR algorithm, and 6 health indexes were input into the model to predict the test data set of power battery. mean absolute error (MAE), root mean square error (RSME) and fitting coefficient R2 were used as model evaluation indexes. Input 6 normalized values of the test set under different working conditions to verify the generalization ability of the model. The experimental results show that the error of the model is less than 1%. Both MAE and RSME values were within 0.04, and R2 values were greater than 0.95. In generalization verification, the average error is less than 1%. MAE was 0.0238; The RMSE value was 0.0239. The R2 value is 0.9241. The model has good generalization and application ability.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1841-9836
1841-9844
DOI:10.15837/ijccc.2025.3.7063