Neural network for contingency ranking dynamic security indices for use under fault conditions in a power distribution system

• US 5,625,751 A
• Filed: 08/30/1994
• Issued: 04/29/1997
• Est. Priority Date: 08/30/1994
• Status: Expired due to Fees

• Current Assignee: Electric Power Research Institute Incorporated
• Original Assignee: Electric Power Research Institute
• Inventors: Ipakchi, Ali, Cauley, Gerald W., Brandwajn, Vladimir, Kumar, A. B. Ranjit
• Primary Examiner(s): Hafiz, Tariq R.
• Application Number:
  US08/297,987
• Time in Patent Office: 973 Days
• Field of Search: 395/20-27, 395/907, 395/915, 382/155-159, 455/3.1, 455/3.3, 364/492-494
• US Class Current: 706/20
• CPC Class Codes: H02J 3/24 Arrangements for preventing or reducing oscillations of power in networks by control effected upon a single generator H02P9/00

Abstract

Analysis and evaluation of outage effects on the dynamic security of power systems is made with a neural network using composite contingency severity indices. A preferably small number of indices describes the power system characteristics immediately post-contingency. These indices are then used as classifiers of the safety of the power system. Using the values of the severity indices, an artificial neural network distinguishes between safe, stable contingencies and potentially unstable contingencies. The severity of the contingency is evaluated based upon a relatively small fixed set of severity indices that are calculated based on a partial time domain simulation. Because a fixed set of severity indices is used, the size and architecture of the neural network is problem independent, thus permitting its use with large scale power systems. Further, the amount of required time domain simulation for the selection of the potentially harmful unstable contingencies is reduced by screening out benign, stable appearing contingencies. The network is trained off-line using training cases that concentrate around the security boundary to reduce the number of cases required to train the neural network.

Claims

1. A method for real-time evaluation of fault contingencies upon the dynamic security of a power system that includes power generators and power-carrying lines, the method comprising the following steps:

(a) defining a set of indices describing a fault-induced deviation from a pre-fault steady state condition of said power system;

(b) forming from said set of indices a subset of composite indices for each fault contingency of interest, wherein a fault contingency of interest includes a fault experienced by said power system, for which fault contingency real-time evaluation is desired, wherein said subset of composite indices is formed as follows;

(b-1) calculating a change from pre-fault steady-state condition of said power system for each system variable defined by an index in a said set of indices;

(b-2) segregating changes calculated at step (b-1) into positive-signed values and negative-signed values;

(b-3) normalizing said positive-signed values and said negative-signed values so segregated at step (b-2);

(b-4) raising to a power n, n≧

4, values normalized at step b-3);

(b-5) combine values so power-raised in step (b-4) to yield at least one composite index selected from the group consisting of (a) terms corresponding to at least some of said positive-signed values, (b) terms corresponding to at least some of said negative-signed values, (c) terms corresponding to at least some of said positive-signed values and corresponding to at least some of said negative-signed values, and (d) terms corresponding to a difference between at least some of said positive-signed values and some of said negative-signed values;

(c) providing a computer system including a neural network to classify in terms of at least stability and instability each of said composite indices, said neural network receiving said composite indices as input; and

(d) providing from an output of said neural network at least one indication of relative stability of said power system in response to a said fault contingency of interest.

2. The method of claim 1, wherein at step (a), said set of indices includes power-carrying line indices and generator indices.

3. The method of claim 1, wherein step (a) includes providing at least one set of indices selected from the group consisting of (i) generator real power output normalized by inertia, (ii) generator reactive power output normalized by inertia, (iii) generator apparent power normalized by inertia, (iv) generator bus voltage, (v) generator rotor angle (DELTA) with respect to the center of inertia, and (vi) generator (Q-Qbias)/P where Q is reactive power, Qbias is a user-specified tuning parameter, and P is real power.

4. The method of claim 1, wherein step (a) includes providing at least one set of indices selected from the group consisting of (i) voltage line sending end real power (P) normalized by line reactance, (ii) voltage line sending end reactive power (Q) normalized by line reactance, (iii) line phase angle, and (iv) line sending end (Q-Qbias)/P where Q is reactive power, Qbias is a user-specified tuning parameter, and P is real power.

5. The method of claim 1, wherein step (a) includes providing at least one set of indices selected from the group consisting of (i) bus load normalized by admittance of voltage lines, (ii) bus voltage, (iii) real power flows in pre-specified interfaces normalized by admittance of available interface voltage lines), and (iv) reactive power flows in pre-specified interfaces normalized by admittance of available interface voltage lines.

6. The method of claim 1, wherein step (a) includes providing at least one set of indices selected from the group consisting of (i) power system reactive power generation (TOTQ) normalized by sum of P, and (ii) power system stress defined by sum of (P*DELTA)/(sum of P)-0.5*TOTQ.

7. The method of claim 1, wherein step (b) is carried out using a simulation program.

8. The method of claim 7, wherein said simulation program includes a program selected from the group consisting of (i) a transient stability simulation program, and (ii) an extended midterm stability program.

9. The method of claim 1, wherein step (b) is carried out using a simulation program and includes:

(i) initializing said simulation program with a description of said power system corresponding to a relevant state of interest of said power system;

(ii) providing said simulation with a description of contingency events said power system may experience;

(iii) initiating and operating said simulation until a last description of contingency events has been provided, whereupon said simulation is halted; and

(iv) calculating a set of said composite indices.

10. The method of claim 1, wherein at step (c), said neural network includes at least three layers and at least six nodes.

11. The method of claim 1, wherein at step (c), said neural network includes an off-line training model deviation mode, wherein a set of pairs of index values and corresponding power system stability classification data is input to said neural network, wherein training cases are concentrated at a boundary separating stable from unstable cases for said power system.

12. The method of claim 1, wherein step (c) includes screening-out from further processing contingencies associated with a safe and stable state of said power system.

13. The method of claim 1, wherein step (d) includes at least one step selected from the group consisting of (i) providing a visual indication of a post-contingency state of said power system for a human operator, (ii) initiating corrective action to alleviate effects of a present fault contingency on said power system, and (iii) initiating preventive action to alleviate effects of a present fault contingency on said power system.

14. The method of claim 1, including a further step of providing to said neural network a security knowledge base memorializing historical contingencies.

15. The method of claim 14, including a further step of updating said security knowledge base following analysis of at least selected contingencies presented to said power system.