METHOD FOR ESTIMATING STATES OF A POWER ELECTRONIC SYSTEM

• US 20100070247A1
• Filed: 09/11/2009
• Published: 03/18/2010
• Est. Priority Date: 09/11/2008
• Status: Active Grant

• Current Assignee: ABB Research Ltd. (ABB Limited)
• Original Assignee: ABB Research Ltd. (ABB Limited)
• Inventors: Gaulocher, Sebastian, Papafotiou, Georgios
• Granted Patent: US 8,364,459 B2
• Time in Patent Office: Days
• US Class Current: 703/2
• CPC Class Codes: G05B 13/048 using a predictor
• CPC Class Codes: H02P 23/12 Observer control, e.g. using Luenberger observers or Kalman filters

Abstract

A method is disclosed for estimating states of a power electronic system, the system having a converter circuit. An exemplary method includes varying a system state vector x(k) and a system state vector x(k+1) for each of sampling times k=−N+1 to k=0 in such a manner that a sum formed by an addition of a first vector norm obtained by subtracting a first system model function f(x(k), u(k)) from the system state vector x(k+1), and another vector norm obtained by subtracting a second system model function g(x(k), u(k)) from the output variable vector y(k), becomes minimal over the sampling times k=−N+1 to k=0. A desired system state vector x(k) at the sampling time k=0 can then be selected.

Claims

1. A method for estimating states of a power electronic system having a converter circuit, the method comprising:

(a) determining output variable vectors y(k) for sampling times k=−

N+1 to k=0, where N is a predefinable sampling horizon;

(b) determining manipulated variable vectors u(k) for the sampling times k=−

N+1 to k=0;

(c) determining a first system model function f(x(k), u(k)) at the sampling time k for describing the power electronic system, which function is dependent on a manipulated variable vector u(k) and a system state vector x(k) at the sampling time k;

(d) determining a second system model function g(x(k), u(k)) at the sampling time k for describing the power electronic system, which function is dependent on the manipulated variable vector u(k) and the system state vector x(k) at the sampling time k, the system state vector x(k) being estimated at the sampling time k=0 by;

(e) varying the system state vector x(k) and the system state vector x(k+1) for each of the sampling times k=−

N+1 to k=0, such that a sum formed by an addition of a first vector norm obtained by subtracting the first system model function f(x(k), u(k)) from the system state vector x(k+1), and another vector norm obtained by subtracting the second system model function from output variable vector y(k), becomes minimal over the sampling times k=−

N+1 to k=0; and

(f) selecting the system state vector x(k) at the sampling time k=0, wherein;

(g) the first system model function f(x(k), u(k)) and the second system model function g(x(k), u(k)) are each affine-linear.

3. The method as claimed in claim 1, wherein a first absolute value sum norm is selected using the first vector norm obtained by subtracting the first system model function f(x(k), u(k)) from the system state vector x(k+1), and another absolute value sum norm is obtained using the another vector norm obtained by subtracting the second system model function g(x(k),u(k)) from the output variable vector y(k).

4. The method as claimed in claim 1, wherein a first Euclidean norm is selected using the first vector norm, and another Euclidean norm is selected using the another vector norm.

5. The method as claimed in claim 1, wherein a first maximum norm is selected using the first vector norm, and another maximum is selected using the another vector norm obtained by substracting a right-hand side of the second system model function g(x(k),u(k)) from the output variable vector y(k).

6. The method as claimed in claim 1, comprising:

storing variations of the system state vector x(k) and the system state vector x(k+1) for each of the sampling times k=−

N+1 to k=0 in an allocation table, a system state vector x(k) and a system state vector x(k+1) being allocated to each output variable vector y(k) and manipulated variable vector u(k) for each of the sampling times k=−

N+1 to k=0.

7. The method as claimed in claim 1, comprising:

calculating variations in the system state vector x(k) and the system state vector x(k+1) for each of the sampling times k=−

N+1 to k=0 continuously.

2. A method for estimating states of a power electronic system having a converter circuit, the method comprising:

(a) determining output variable vectors y(k) for sampling times k=−

N+1 to k=0, where N is a predefinable sampling horizon;

(b) determining manipulated variable vectors u(k) for the sampling times k=−

N+1 to k=0;

(c) determining a first system model function f(x(k), u(k)) at the sampling time k for describing the power electronic system, which function is dependent on a manipulated variable vector u(k) and a system state vector x(k) at the sampling time k;

(d) determining a second system model function g(x(k), u(k)) at the sampling time k for describing the power electronic system, which function is dependent on the manipulated variable vector u(k) and the system state vector x(k) at the sampling time k, the system state vector x(k) being estimated at the sampling time k=0 by;

(e) varying the system state vector x(k) and the system state vector x(k+1) for each of the sampling times k=−

N+1 to k=0, such that a sum formed by an addition of a first vector norm obtained by subtracting the first system model function f(x(k), u(k)) from the system state vector x(k+1), and another vector norm obtained by subtracting the second system model function from the output variable vector y(k), becomes minimal over the sampling times k=−

N+1 to k=0; and

(f) selecting the system state vector x(k) at the sampling time k=0, wherein;

(g) the first system model function f(x(k), u(k)) and the second system model function g(x(k), u(k)) are each piecewise affine-linear.

8. The method as claimed in claim 2, wherein a first absolute value sum norm is selected using the first vector norm obtained by subtracting the first system model function f(x(k), u(k)) from the system state vector x(k+1), and another absolute value sum norm is obtained using the another vector norm obtained by subtracting the second system model function g(x(k),u(k)) from the output variable vector y(k).

9. The method as claimed in claim 2, wherein a first Euclidean norm is selected using the first vector norm, and another Euclidean norm is selected using the another vector norm.

10. The method as claimed in claim 2, wherein a first maximum norm is selected using the first vector norm, and another maximum is selected using the another vector norm obtained by substracting a right-hand side of the second system model function g(x(k),u(k)) from the output variable vector y(k).

11. The method as claimed in claim 2, comprising:

storing variations of the system state vector x(k) and the system state vector x(k+1) for each of the sampling times k=−

N+1 to k=0 in an allocation table, a system state vector x(k) and a system state vector x(k+1) being allocated to each output variable vector y(k) and manipulated variable vector u(k) for each of the sampling times k=−

N+1 to k=0.

12. The method as claimed in claim 2, comprising:

calculating variations in the system state vector x(k) and the system state vector x(k+1) for each of the sampling times k=−

N+1 to k=0 continuously.