Tuning parameter of kalman filter in a wheel inspection

• US 8,688,314 B2
• Filed: 01/23/2012
• Issued: 04/01/2014
• Est. Priority Date: 01/30/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Duan, Ning, Hu, Ke, Jiang, Zhong Bo, Mo, Wen Ting, Sun, Wei, Zhang, Xin
• Primary Examiner(s): Black, Thomas
• Assistant Examiner(s): CASTRO, PAUL A
• Application Number:
  US13/355,633
• Publication Number:
  US 20120197485A1
• Time in Patent Office: 799 Days
• Field of Search: None
• US Class Current: 701/31.4
• CPC Class Codes: G01M 17/013 Wheels

Abstract

A method for tuning a parameter of a Kalman filter in a wheel inspection for a vehicle. The method includes: associating wheel inspection data of the vehicle with locations of wheels in the vehicle; evaluating a wheel stable score of the vehicle based on the wheel inspection data and the association, wherein the wheel stable score indicates reliability of the wheel inspection; and tuning measurement error covariance of the Kalman filter according to the evaluated wheel stable score. By using the method, the parameter of the Kalman filter can be dynamically tuned to enable the Kalman filter to filter more accurately to ensure the correct result of the wheel inspection. In addition, an apparatus for tuning a parameter of a Kalman filter and a wheel inspection system for a vehicle are disclosed.

Claims

1. A method for tuning a parameter of a Kalman filter in a wheel inspection for a vehicle, comprising:

associating wheel inspection data of said vehicles with locations of corresponding wheels in said vehicle;

evaluating a wheel stable score of said vehicle based on said wheel inspection data and said association, wherein said wheel stable score indicates reliability of said wheel inspection, the wheel stable score calculated according to a wheel diameter differential of left and right wheels coupled to a common axis; and

dynamically tuning a measurement error covariance (R) of said Kalman filter according to said evaluated wheel stable score.

2. The method according to claim 1, wherein associating wheel inspection data of said vehicle with locations of said corresponding wheels in said vehicle comprises:

identifying said locations of said corresponding wheels in said vehicle;

determining said location in said vehicle for at least one of an axle, a bogie and a car where said wheel is located; and

establishing an association among said wheel inspection data, the locations of said wheels and the location of the at least one of said axle, said bogie and said car where said wheel is located.

3. The method according to claim 2, wherein evaluating a wheel stable score of said vehicle based on said wheel inspection data and said association comprises:

calculating at least one of an axle stable score, a bogie stable score and a car stable score of said wheels based on said association among said wheel inspection data, said locations of said wheels and said location of the at least one of said axle, said bogie and said car where said wheel is located; and

calculating said wheel stable score based on the at least one of said axle stable score, said bogie stable score and said car stable score.

4. The method according to claim 3, wherein calculating said axle stable score of said wheels comprises:

calculating an average value of coaxial wheel inspection data differences between the left and right wheels in the respective same axles;

calculating a current coaxial wheel inspection data difference deviation score based on said average value of the coaxial wheel inspection data differences and a mean value and a variance value of a predetermined normal distribution of coaxial wheel inspection data differences; and

determining the axle stable score of the wheels according to the current coaxial wheel inspectional data difference deviation score.

5. The method according to claim 3, wherein calculating said bogie stable score of said wheels comprises:

calculating, for each bogie, a wheel inspection data difference between the left wheels in said bogie and a wheel inspection data difference between the right wheels in said bogie as a co-bogie wheel inspection data difference for said bogie;

calculating an average value of said co-bogie wheel inspection data differences;

calculating a current co-bogie wheel inspection data difference deviation score based on the average value of the co-bogie wheel inspection data differences and a mean value and a variance value of a predetermined normal distribution of co-bogie wheel inspection data differences; and

determining said bogie stable score of said wheels according to said current co-bogie wheel inspection data difference deviation score.

6. The method according to claim 3, wherein calculating said car stable score of said wheels comprises:

calculating average values of said wheel inspection data of the left and right wheels in the respective bogies in the same car, and calculating wheel inspection data differences between the bogies as co-car wheel inspection data differences;

calculating an average value of said co-car wheel inspection data differences;

calculating a current co-car wheel inspection data difference deviation score based on the average value of the co-car wheel inspection data differences and a mean value and a variance value of a predetermined normal distribution of co-car wheel inspection data differences; and

determining said car stable score of said wheels according to said current co-car wheel inspection data difference deviation score.

7. The method according to claim 3, wherein said wheel stable score is calculated as a weight sum of said axle stable score, said bogie stable score and said car stable score, wherein each of said weights of said axle stable score, said bogie stable score and the car stable score is within a range of 0 to 1, and the sum of said weights is equal to 1.

8. The method according to claim 1, wherein tuning said measurement error covariance of said Kalman filter comprises:

comparing said wheel stable score with a predetermined stable threshold; and

provided that the wheel stable score is not less than the stable threshold, tuning said measurement error covariance.

9. The method according to claim 8, wherein tuning said measurement error covariance of said Kalman filter further comprises:

determining, provided that said wheel stable score is not less than said stable threshold, whether a number of times of consecutive occurrences when said wheel stable score is not less than the stable threshold exceeds a predetermined value; and

tuning said measurement error covariance provided that said number of times of consecutive occurrences exceeds said predetermined value.

10. The method according to 9, wherein said measurement error covariance is tuned according to the following formula:

R=A(ƒ

(StableIndex)−

ƒ

(StableIndex_threshold))+R₀  

 

(18)

where represents tuned measurement error covariance, A is a constant, represents a linear function, represents calculated wheel stable score, represents stable threshold, and represents initial measurement error covariance.

11. A non-transitory computer readable article of manufacture tangibly embodying computer readable instructions which, when executed, cause a computer to carry out the steps of a method for tuning a parameter of a Kalman filter in a wheel inspection for a vehicle, the method comprising:

associating wheel inspection data of said vehicle with locations of corresponding wheels in said vehicle;

evaluating a wheel stable score of said vehicle based on said wheel inspection data and said association, wherein said wheel stable score indicates reliability of said wheel inspection, the wheel stable score calculated according to a wheel diameter differential of left and right wheels coupled to a common axis; and

dynamically tuning a measurement error covariance (R) of said Kalman filter according to said evaluated wheel stable score.