Search

US-12623570-B2 - Method and system for predicting battery capacity degradation for electric vehicle

US12623570B2US 12623570 B2US12623570 B2US 12623570B2US-12623570-B2

Abstract

A method and a system for predicting a battery capacity degradation for an electric vehicle having a battery are provided. The method comprises extracting and pre-processing a raw dataset comprising a plurality of battery loss indicators and a plurality of battery loss values each corresponding to a plurality of time steps to obtain a pre-processed dataset. The method further comprises selecting a loss indicator subset from the pre-processed dataset at a first time step and a second time step based on a smart feature selection (SFS) algorithm. The method further comprises training a machine learning model with each battery loss indicator at the first and second time steps and the battery loss value at the first time step. The method further comprises determining the battery loss value at the second time step with the machine learning model to predict the battery capacity degradation.

Inventors

  • Muhammad Khalid
  • Huzaifa RAUF
  • Naveed ARSHAD

Assignees

  • KING FAHD UNIVERSITY OF PETROLEUM AND MINERALS

Dates

Publication Date
20260512
Application Date
20240507

Claims (14)

  1. 1 . A computer-implemented method of predicting a battery capacity degradation for an electric vehicle having a battery, comprising: extracting and pre-processing a raw dataset comprising a plurality of battery loss indicators and a plurality of battery loss values each corresponding to a plurality of time steps to obtain a pre-processed dataset; selecting a loss indicator subset from the pre-processed dataset at a first time step and a second time step of the plurality of time steps based on a smart feature selection (SFS) algorithm, wherein the first time step is immediately prior to the second time step; training a machine learning model with each battery loss indicator of the plurality of battery loss indicators in the loss indicator subset at the first and second time steps of the plurality of time steps and the battery loss value at the first time step; and determining the battery loss value at the second time step with the machine learning model to predict the battery capacity degradation; wherein the plurality of battery loss values includes a battery cyclic loss value and a battery calendar loss value.
  2. 2 . The method of claim 1 , wherein the SFS algorithm comprises: extrapolating the loss indicator subset to fill missing values in the pre-processed dataset; extracting a mapping relationship between the plurality of battery loss indicators and the plurality of battery loss values from the pre-processed dataset at each time steps of the plurality of time steps based on a quantitative correlation analysis; and selecting one or more battery loss indicators from the plurality of battery loss indicators in the pre-processed dataset based on the mapping relationship to obtain the loss indicator subset.
  3. 3 . The method of claim 1 , wherein the plurality of battery loss indicators includes a distance travelled by an electronic vehicle having the lithium-ion battery, a charging efficiency of the lithium-ion battery, a discharging efficiency of the lithium-ion battery, an energy consumption at the first and second time steps, an internal resistance of the lithium-ion battery, and a temperature.
  4. 4 . The method of claim 1 , wherein the training further comprises: splitting the loss indicator subset into a training data and a testing data; training the machine learning model with the training data; and validating the machine learning model with the testing data.
  5. 5 . The method of claim 4 , wherein the machine learning model is selected from Linear Regression, Ridge Regression, Lasso Regression, Support Vector Regression, Gaussian Process Regression, Random Forest, ElasticNet, and XGBoost.
  6. 6 . The method of claim 1 , wherein the battery is a lithium-ion battery.
  7. 7 . The method of claim 1 , wherein the raw dataset includes a real-time data including a plurality of operating conditions obtained from the electric vehicle while operating.
  8. 8 . A battery health management system to predict a battery capacity degradation for an electric vehicle having a battery, comprising: a system processor communicatively connected to a vehicle control unit of the electric vehicle and configured to execute a program instruction; and a memory connected to the system processor and configured to store a raw data; wherein the program instruction comprises: extracting and pre-processing the raw dataset comprising a plurality of battery loss indicators and a plurality of battery loss values each corresponding to a plurality of time steps to obtain a pre-processed dataset; selecting a loss indicator subset from the pre-processed dataset at a first time step and a second time step of the plurality of time steps based on a smart feature selection (SFS) algorithm, wherein the first time step is immediately prior to the second time step; training a machine learning model with each battery loss indicator of the plurality of battery loss indicators in the loss indicator subset at the first and second time steps of the plurality of time steps and the battery loss value at the first time step; and determining the battery loss value at the second time step with the machine learning model to predict the battery capacity degradation; wherein the plurality of battery loss values includes a battery cyclic loss value and a battery calendar loss value.
  9. 9 . The system of claim 8 , wherein the SFS algorithm comprises: extrapolating the loss indicator subset to fill missing values in the pre-processed dataset; extracting a mapping relationship between the plurality of battery loss indicators and the plurality of battery loss values from the pre-processed dataset at each time steps of the plurality of time steps based on a quantitative correlation analysis; and selecting one or more battery loss indicators from the plurality of battery loss indicators in the pre-processed dataset based on the mapping relationship to obtain the loss indicator subset.
  10. 10 . The system of claim 8 , wherein the plurality of battery loss indicators includes a distance travelled by an electronic vehicle having the lithium-ion battery, a charging efficiency of the lithium-ion battery, a discharging efficiency of the lithium-ion battery, an energy consumption at the first and second time steps, an internal resistance of the lithium-ion battery, and a temperature.
  11. 11 . The system of claim 8 , wherein the training further comprises: splitting the loss indicator subset into a training data and a testing data; training the machine learning model with the training data; and validating the machine learning model with the testing data.
  12. 12 . The system of claim 11 , wherein the machine learning model is selected from Linear Regression, Ridge Regression, Lasso Regression, Support Vector Regression, Gaussian Process Regression, Random Forest, ElasticNet, and XGBoost.
  13. 13 . The system of claim 8 , wherein the battery is a lithium-ion battery.
  14. 14 . The system of claim 8 , wherein the raw dataset includes a real-time data including a plurality of operating conditions obtained from the electric vehicle while operating.

Description

STATEMENT REGARDING PRIOR DISCLOSURE BY THE INVENTORS Aspects of the present disclosure are described in “A novel smart feature selection strategy of lithium-ion battery degradation modelling for electric vehicles based on modern machine learning algorithms”, published in Journal of Energy Storage, Volume 68, 107577 which is incorporated herein by reference in its entirety. STATEMENT OF ACKNOWLEDGEMENT Support provided by the Saudi Data and AI Authority (SDAIA), Saudi Arabia and King Fahd University of Petroleum and Minerals (KFUPM) under SDAIA-KFUPM Joint Research Center for Artificial Intelligence, Saudi Arabia, grant No. JRC-AI-RFP-08, and the LUMS Energy Institute (LEI) at Lahore University of Management Sciences (LUMS), and the National Center of Big Data and Cloud Computing (NCBC) of the Higher Education Commission (HEC), Pakistan, is gratefully acknowledged. BACKGROUND Technical Field The present disclosure relates generally to the field of battery health management for electric vehicles (EVs), and more specifically to a method and a system for predicting battery capacity degradation, particularly for lithium-ion batteries, utilizing machine learning techniques. Description of Related Art The “background” description provided herein is for the purpose of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description which may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present invention. The transition of personal transportation from internal combustion engine (ICE) vehicles to electric vehicles (EV) is a vital step in achieving lower carbon emissions from the transportation sector. EVs and renewable energy systems are widely promoted as clean alternatives to conventional vehicles and power generation, and as promising solutions to effectively reduce GHG emissions and other environmental problems. The rapid development of the EV and renewable energy industry as a clean alternative to fossil-fuel-based vehicles and power generation sources has increased the demand for energy storage technologies. Among the available energy storage technologies, lithium-ion (Li-ion/LIB) batteries have detached as one of the solutions, which can meet the requirements imposed by both power grids and transportation sectors. In recent years, a significant interest in battery-related applications has arisen globally due to reducing fuel consumption, mitigating dependence on imported oil, and decreasing greenhouse gas emissions. Over the last few decades, battery technology has made significant progress in the area of energy storage and plays a key role in EVs and renewable energy systems. The advancements in LIBs have attracted considerable attention due to their high energy density, low maintenance, and the best performance. Meanwhile, the reliability and safety assessment of LIBs has become an important issue, in particular for future EV performance. The energy provision and consumption in LIB-related applications are highly dependent on the health condition of batteries and one main limitation of LIBs resides in battery ageing. LIBs are increasing in popularity, and there is an increased need to study and model their capacity degradation. The classical problem associated with the EV battery is that it undergoes a sophisticated degradation process during EV operations. Battery degradation gradually happens over time under specific driving conditions and affects EV power consumption due to battery ageing. LIBs undergo operation periods that are substantially shorter than the idle intervals and have different stresses of C-rate, depth-of-discharge (DOD), temperature, and state-of-charge (SOC). LIBs undergo a process to store and provide electrical energy which can last over different time scales. This stationary and transient operation of the Li-ion battery causes calendar and cyclic loss, respectively. Battery degradation takes place in every condition, but in different proportions as usage and external conditions interact to provoke degradation. When a defined degradation level is reached, the battery reaches its end-of-life (EOL) and has to be replaced. To address these difficulties, precise battery degradation models capable of accurately predicting the performance and lifetime of LIBs need to develop. Battery lifetime models are used to predict the long-term degradation behavior of LIB performance metrics such as capacity and internal resistance. Generally, the phenomena of battery degradation can be classified into two categories: the calendar loss, which refers to the irreversible loss of battery capacity during storage, and the cyclic loss, which occurs due to battery charge and discharge cycles. Cyclic ageing is one of the two main aspects used to model the battery degradation of a LIB. Battery cyclic loss is m