Dynamic economic load dispatch by applying dynamic programming to a genetic algorithm

• US 7,752,150 B2
• Filed: 12/12/2006
• Issued: 07/06/2010
• Est. Priority Date: 09/29/2006
• Status: Active Grant

• Current Assignee: Honeywell Technology Solutions Incorporated (KBR Incorporated)
• Original Assignee: Honeywell International Inc.
• Inventors: Ye, Chen Zhou, Zhang, Dan Qing, He, JinWei, Xiang, Xunwei
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): Tran; Mai T
• Application Number:
  US11/637,406
• Publication Number:
  US 20080154810A1
• Time in Patent Office: 1,302 Days
• Field of Search: 706/13, 700/286, 700/295
• US Class Current: 706/13
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming
• CPC Class Codes: Y02E 40/70 Smart grids as climate change mitigation technology in the energy generation sector
• CPC Class Codes: Y04S 10/50 Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

In an embodiment, a system and method provide a constraint on a genetic algorithm, and further provide dynamic programming capability to the genetic algorithm. The system and method then allocate load demand over a finite number of time intervals among a number of power generating units as a function of the constraint on the genetic algorithm and the dynamic programming capability of the genetic algorithm.

Claims

1. A process comprising:

receiving at a computer processor a dynamic constraint on a genetic algorithm;

applying dynamic programming to the genetic algorithm in the computer processor such that the genetic algorithm accommodates the dynamic constraint; and

allocating load demand using the computer processor over a finite number of time intervals among a number of power generating units as a function of the constraint on the genetic algorithm and the dynamic programming of the genetic algorithm;

wherein the receiving the dynamic constraint and the applying the dynamic programming to the genetic algorithm comprises modifying one or more initialization, crossover, and mutation operators, thereby maintaining load dispatch solutions evolving freely near a feasible region; and

wherein the modifying of the mutation operator comprises heuristically maintaining locations of possible mutation points, and providing a positive mutation operator to reduce fuel cost and a negative mutation operator to achieve load equality.

2. The process of claim 1, wherein the constraint of the genetic algorithm comprises one or more of:

the load demand is equal to a sum of all outputs of the power generating units;

each power generating unit includes a lower output limit and an upper output limit;

a limit on ramp up and ramp down rates of the power generating units; and

one or more prohibited operating zones of the power generating units.

3. The process of claim 1, wherein a fuel cost function of one or more of the power generating units comprises one or more of a non-smooth characteristic and a non-convex characteristic.

4. The process of claim 1, wherein the modifying of the initialization operator comprises providing a biased and heuristic initialization to ensure a diversified initial population.

5. The process of claim 4, wherein the biased and heuristic initialization comprises:

assigning a highest priority to one or more power generating units operating in a prohibited operating zone;

assigning a normal priority to one or more power generating units selected randomly before the initialization;

assigning a lowest priority to any remaining power generating units not assigned the highest priority or the normal priority;

iteratively selecting a power generating unit according to the prioritized power generating units; and

allocating unfulfilled load demand to that power generating unit without violating the constraints on the genetic algorithm.

6. The process of claim 1, further comprising providing a discretized search strategy to one or more of the positive mutation operator and the negative mutation operator.

7. The process of claim 1, wherein the modifying of the crossover operator comprises selecting feasible positions for crossover to accommodate the constraints on the genetic algorithm.

8. A machine readable medium storing instructions, which when executed by a processor, cause the processor to perform a process comprising:

receiving a dynamic constraint on a genetic algorithm;

applying dynamic programming to the genetic algorithm such that the genetic algorithm accommodates the dynamic constraint; and

allocating load demand over a finite number of time intervals among a number of power generating units as a function of the constraint on the genetic algorithm and the dynamic programming of the genetic algorithm;

wherein the receiving a dynamic constraint and the applying dynamic programming to the genetic algorithm comprises modifying one or more initialization, crossover, and mutation operators, thereby maintaining load dispatch solutions evolving freely near a feasible region; and

wherein the modifying of the mutation operator comprises heuristically maintaining locations of possible mutation points, and providing a positive mutation operator to reduce fuel cost and a negative mutation operator to achieve load equality.

9. The machine readable medium of claim 8, wherein the constraint of the genetic algorithm comprises one or more of:

the load demand is equal to a sum of all outputs of the power generating units;

each power generating unit includes a lower output limit and an upper output limit;

a limit on ramp up and ramp down rates of the power generating units; and

one or more prohibited operating zones of the power generating units.

10. The machine readable medium of claim 8, wherein a fuel cost function of one or more of the power generating units comprises one or more of a non-smooth characteristic and a non-convex characteristic.

11. The machine readable medium of claim 8, wherein the modifying of the initialization operator comprises providing a biased and heuristic initialization to ensure a diversified initial population.

12. The machine readable medium of claim 11, wherein the biased and heuristic initialization comprises:

assigning a highest priority to one or more power generating units operating in a prohibited operating zone;

assigning a normal priority to one or more power generating units selected randomly before the initialization;

assigning a lowest priority to any remaining power generating units not assigned the highest priority or the normal priority;

iteratively selecting a power generating unit according to the prioritized power generating units; and

allocating unfulfilled load demand to that power generating unit without violating the constraints on the genetic algorithm.

13. The machine readable medium of claim 8, further comprising providing a discretized search strategy to one or more of the positive mutation operator and the negative mutation operator.

14. The machine readable medium of claim 8, wherein the modifying of the crossover operator comprises selecting feasible positions for crossover to accommodate the constraints on the genetic algorithm.

15. A system comprising:

a computer processor configured to receive a dynamic constraint on a genetic algorithm;

a computer processor configured to apply dynamic programming to the genetic algorithm, such that the genetic algorithm accommodates the dynamic constraint; and

a computer processor configured to allocate load demand over a finite number of time intervals among a number of power generating units as a function of the constraint on the genetic algorithm and the dynamic programming of the genetic algorithm;

wherein the reception of a dynamic constraint and the application of dynamic programming to the genetic algorithm comprises modifying one or more initialization, crossover, and mutation operators, thereby maintaining load dispatch solutions evolving freely near a feasible region; and

wherein the modifying of the mutation operator comprises heuristically maintaining locations of possible mutation points, and providing a positive mutation operator to reduce fuel cost and a negative mutation operator to achieve load equality.

16. The system of claim 15, wherein the constraint of the genetic algorithm comprises one or more of:

the load demand is equal to a sum of all outputs of the power generating units;

each power generating unit includes a lower output limit and an upper output limit;

a limit on ramp up and ramp down rates of the power generating units; and

one or more prohibited operating zones of the power generating units.