Method for control and monitoring using a state estimator having variable forgetting factors

• US 20060284600A1
• Filed: 11/10/2005
• Published: 12/21/2006
• Est. Priority Date: 06/21/2005
• Status: Active Grant

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: GM Global Technology Operations Incorporated (General Motors Company)
• Inventors: Verbrugge, Mark W.
• Granted Patent: US 7,612,532 B2
• Time in Patent Office: Days
• US Class Current: 320/132
• CPC Class Codes: G01R 31/367 Software therefor, e.g. for battery testing using modelling or look-up tables
• CPC Class Codes: H01M 10/48 Accumulators combined with arrangements for measuring, testing or indicating the condition of cells, e.g. the level or density of the electrolyte constructional details of current conducting connections for detecting conditions inside cells or batteries, e.g. details of voltage sensing terminals, H01M50/569
• CPC Class Codes: Y02E 60/10 Energy storage using batteries

Abstract

A recursive algorithm is provided for adaptive multi-parameter regression enhanced with forgetting factors unique to each regressed parameter. Applications of this algorithm can include lead acid batteries, nickel-metal hydride batteries, and lithium-ion batteries, among others. A control algorithm is presented, having an arbitrary number of model parameters, each having its own time-weighting factor. A method to determine optimal values for the time-weighting factors is included, to give greater effect to recently obtained data for the determination of a system'"'"'s state. A methodology of weighted recursive least squares is employed, wherein the time weighting corresponds to the exponential-forgetting formalism. The derived mathematical result does not involve matrix inversion, and the method is iterative, i.e. each parameter is regressed individually at every time step.

Claims

1. Method to estimate a state of a device using an executable multivariate mathematical model comprising a summation of a plurality of inputs, each input factored by a corresponding weighting factor and by a corresponding parameter, each corresponding parameter determined by:

determining each input at a time-certain;

deriving a weighted recursive least squares equation from the executable multivariate mathematical model; and

,

executing the weighted recursive least squares equation based upon the inputs, the weighting factors, and, non-corresponding parameters.

2. The method of claim 1, wherein the method to estimate a state of the device is executed recursively.

3. The method of claim 1, wherein estimating a state of a device comprises estimating a state of charge of an electrical energy storage device.

4. The method of claim 1, wherein estimating a state of a device comprises estimating a state of health of an electrical energy storage device.

5. The method of claim 1, wherein estimating a state of a device comprises estimating a state of power of an electrical energy storage device.

6. The method of claim 1, wherein determining the input at a time-certain comprises sensing the input.

7. The method of claim 1, wherein executing the weighted recursive least squares equation further comprises:

executing the weighted recursive least squares equation using;

the inputs determined at the time-certain, the weighting factors, and, the non-corresponding parameters determined at a preceding time-certain.

8. The method of claim 7, further comprising decaying at least one of the weighting factors using exponential forgetting factors.

9. The method of claim 8, further comprising decaying at least one of the weighting factors using exponential forgetting factors based upon elapsed time.

10. The method of claim 7, further comprising decaying at least one of the weighting factors based upon a characteristic of the device.

11. Article of manufacture, comprising:

a storage medium having a computer program encoded therein for effecting a method to estimate a state of a device using an executable multivariate mathematical model comprising;

a summation of a plurality of inputs, each input factored by a corresponding weighting factor and factored by a corresponding parameter, said computer program comprising;

code for determining each input at a time-certain;

code for determining the corresponding parameter, comprising a weighted recursive least squares equation derived from the executable multivariate mathematical model;

the weighted recursive least squares equation operable to calculate the corresponding parameter based upon;

the inputs at the time-certain, the weighting factors, and, non-corresponding parameters.

12. The apparatus of claim 11, wherein the code for determining the corresponding parameter further comprises:

the weighted recursive least squares equation derived from the executable multivariate mathematical model operable to calculate the corresponding parameter based upon;

the inputs at the time-certain, the weighting factors, and, the non-corresponding parameters determined at a preceding time-certain.

13. The apparatus of claim 11, wherein the method to estimate a state of a device using an executable multivariate mathematical model comprises a method to estimate a state of charge for an electrical storage device.

14. The apparatus of claim 13, wherein the electrical storage device comprises a high voltage battery for a hybrid electric vehicle.

15. The apparatus of claim 11, wherein the computer program encoded for effecting the method to estimate state of a device is executed recursively.

16. The apparatus of claim 15, wherein the recursively executed code is cyclically executed on a predetermined time cycle.

17. The apparatus of claim 16, wherein the code for determining each corresponding parameter is executed within the predetermined time cycle.

18. The apparatus of claim 17, wherein at least one of the weighting factors is decayed each predetermined time cycle.

19. The apparatus of claim 17, wherein at least one of the weighting factors is decayed each predetermined time cycle based upon a characteristic of the device.

20. Method to recursively estimate a state of a device, comprising:

a) constructing an executable multivariate mathematical model of the device operable to estimate the state, comprising;

an equation summing a plurality of sensed data signals, each sensed data signal factored by;

one of a corresponding plurality of weighting factors; and

further factored by one of a corresponding plurality of parameters;

b) sensing each of the sensed data signals at a time-certain;

c) updating each of the plurality of parameters, comprising;

i) creating a weighted recursive least squares equation to calculate a parameter, based upon the executable multivariate mathematical model of the device; and

,

ii) executing the weighted recursive least squares equation to calculate each of the parameters, based upon;

the sensed data signals at the time-certain, the weighting factors, and, non-corresponding parameters determined at a preceding time-certain;

d) executing the multivariate mathematical model of the device operable to estimate the state, using the updated plurality of parameters, the sensed data signals at the time-certain, and the weighting factors.