System and method for upgrading metamodel for friction coefficient prediction of brake pad, and brake control system using the metamodel

Patent metadata

Field	Value
Publication number	US-11951965-B2
Application number	US-202117491948-A
Country	US
Kind code	B2
Filing date	Oct 1, 2021
Priority date	May 12, 2021
Publication date	Apr 9, 2024
Grant date	Apr 9, 2024

Abstract

A system and method includes upgrading a metamodel for friction coefficient prediction of a brake, in which the metamodel for friction coefficient prediction may be constructed using various derivative parameters relating to the speed, temperature and pressure of a brake disc in addition to basic parameters, such as the speed, temperature and pressure of the brake disc, to greatly improve performance and accuracy in friction coefficient prediction using the metamodel for friction coefficient prediction and to improve accuracy in evaluation of the driving performance of a vehicle through an increase in accuracy of determination of brake torque.

First claim

What is claimed is:

1. A system of upgrading a metamodel for friction coefficient prediction of a brake pad, the system comprising:
a data preprocessor configured to preprocess raw data including a rotation speed, a temperature and a pressure of a brake disc to validate the raw data;
a derivative parameter preprocessor configured to generate derivative parameters from basic parameters, as output by the data preprocessor, including the rotation speed, the temperature and the pressure of the brake disc; and
a metamodel processor configured to generate the metamodel for friction coefficient prediction by performing machine learning according to the derivative parameters output by the derivative parameter preprocessor,
wherein the derivative parameter preprocessor is configured to define the derivative parameters as principal factors influencing a change in a friction coefficient by analyzing relations between the basic parameters and the friction coefficient at a point in time when a friction coefficient difference occurs under an identical condition of each of the basic parameters preprocessed by the data preprocessor.

2. The system claim 1 , wherein the data preprocessor is configured to acquire valid data from the raw data by performing preprocessing of the raw data including classification of data stabilization sections for reliability of the raw data, removal of negative values and abnormal values deviating from a reference range, and removal of redundant data.

3. The system of claim 1 , wherein the metamodel processor includes:
a machine learning unit configured to perform the machine learning according to the derivative parameters generated by the derivative parameter preprocessor in addition to the basic parameters; and
a friction coefficient training model constructed through the machine learning performed by the machine learning unit,
wherein, when the machine learning according to the basic parameters and the derivative parameters is completed, the friction coefficient training model is generated as the metamodel for friction coefficient prediction.

4. The system of claim 1 , wherein the metamodel for friction coefficient prediction includes the derivative parameter preprocessor configured to generate the derivative parameters from the basic parameters, and a machine learning unit configured to output a friction coefficient predicted by performing the machine learning according to the derivative parameters in addition to the basic parameters.

5. A brake control system using the metamodel for friction coefficient prediction of claim 1 , the brake control system including:
the metamodel for friction coefficient prediction; and
a brake controller configured to determine a target brake torque according to a friction coefficient output by the metamodel for friction coefficient prediction and to apply a hydraulic brake pressure control signal corresponding to the determined target brake torque to a brake system.

6. The brake control system of claim 5 , wherein the metamodel for friction coefficient prediction includes the derivative parameter preprocessor configured to generate the derivative parameters from the basic parameters, and a machine learning unit configured to output the friction coefficient predicted by performing the machine learning according to the derivative parameters in addition to the basic parameters.

7. The system of claim 1 , wherein the derivative parameter preprocessor is configured to generate the derivative parameters defined through smoothing and lag processing in time series analysis of each of the parameters, among data analysis methods.

8. The system of claim 1 , wherein the derivative parameters defined by the derivative parameter preprocessor include a moving average of the pressure of the brake disc, a moving average of the rotation speed of the brake disc, a moving average of the temperature of the brake disc, a square of the moving average of the temperature of the brake disc, a deceleration of the brake disc, a change in the temperature of the brake disc, a change in the temperature change of the brake disc, a change in the pressure of the brake disc, an estimated torque value using the deceleration, a correlation value between the estimated torque value and the temperature, kinetic energy of the brake disc, and cumulative kinetic energy of the brake disc.

9. A method of upgrading a metamodel for friction coefficient prediction of a brake pad, the method comprising:
preprocessing, by a data preprocessor, raw data including a rotation speed, a temperature and a pressure of a brake disc to validate the raw data;
generating, by a derivative parameter preprocessor, derivative parameters from basic parameters, as output by the data preprocessor, including the rotation speed, the temperature and the pressure of the brake disc; and
generating, by a machine learning processor, the metamodel for friction coefficient prediction by performing machine learning according to the derivative parameters output by the derivative parameter preprocessor,
wherein, in the generating of the derivative parameters, the derivative parameters are defined as principal factors influencing a change in a friction coefficient by analyzing relations between the basic parameters and the friction coefficient at a point in time when a friction coefficient difference occurs under an identical condition of each of the basic parameters preprocessed by the data preprocessor.

10. The method claim 9 , wherein, in the preprocessing of the raw data, classification of data stabilization sections for reliability of the raw data, removal of negative values and abnormal values deviating from a reference range, and removal of redundant data are performed to acquire valid data from the raw data.

11. The method of claim 9 , wherein the generating of the metamodel for friction coefficient prediction by performing the machine learning includes:
performing the machine learning according to the derivative parameters generated by the derivative parameter preprocessor in addition to the basic parameters; and
constructing a friction coefficient training model through the machine learning,
wherein, when the machine learning according to the basic parameters and the derivative parameters is completed, the friction coefficient training model is generated as the metamodel for friction coefficient prediction.

12. The method of claim 9 , wherein the metamodel for friction coefficient prediction includes the derivative parameter preprocessor configured to generate the derivative parameters from the basic parameters, and a machine learning unit configured to output a friction coefficient predicted by performing the machine learning according to the derivative parameters in addition to the basic parameters.

13. The method of claim 9 , wherein, in the generating of the derivative parameters, the derivative parameters defined through smoothing and lag processing in time series analysis of each of the parameters, among data analysis methods are generated.

14. The method of claim 9 , wherein, in the generating of the derivative parameters, the generated derivative parameters include a moving average of the pressure of the brake disc, a moving average of the rotation speed of the brake disc, a moving average of the temperature of the brake disc, a square of the moving average of the temperature of the brake disc, a deceleration of the brake disc, a change in the temperature of the brake disc, a change in the temperature change of the brake disc, a change in the pressure of the brake disc, an estimated torque value using the deceleration, a correlation value between

Assignees

• Hyundai Motor Co Ltd
• Kia Corp

Inventors

• Cho Sung Hyun
• Bang Sung Hoon

Classifications

• B60T8/172 Primary

  Determining control parameters used in the regulation, e.g. by calculations involving measured or detected parameters {(B60T8/17551 takes precedence)} · CPC title

• B60T8/171

  Detecting parameters used in the regulation; Measuring values used in the regulation · CPC title

• B60T8/58

  responsive to speed and another condition or to plural speed conditions · CPC title

• B60T17/221

  Procedure or apparatus for checking or keeping in a correct functioning condition of brake systems (hydraulic pressure systems in general F15B19/00, F15B21/04; testing structures or apparatus G01M) · CPC title

• G06N20/00

  Machine learning · CPC title