OCR Text |
Show Thus, the genetic algorithm has been successful in demonstrating the fundamental requirement of an active control system as well: performance of the burner, under a static fuel load and burner geometry, has been steadily increased over time. Figure A 7 is a plot of both average performance of the direction-set and the genetic algorithm cases. Note that even after 83 iterations the average performance of the direction-set method has not surpassed the overall performance of the genetic algorithm. 0.75 ~ Q~) 0.65 ] ....... 0.55 Q) ~ e 0.45 ~ 0.35 Q) ~ 0.25 o 5 10 15 20 25 30 35 40 45 50 iteration, individual Figure A 7 . Average performance history of the co-swirl nozzle for both direction-set (black) and genetic algorithm (white) search techniques. The behavior of the two typical trials discussed so far deserve some "what-if' discussion. From other attempts, the results of which are not presented here, it is clear that the genetic algorithm will perfonn essentially the same as in this typical case: erratic start, converging after a few generations, but almost always with a few outliers. The direction-set technique, however, could perfonn quite differently. It should be obvious that this method depends heavily on two initial parameters: the starting location and the starting search direction. One can imagine the trivial search that starts in the peak region. In this case, the goal is achieved by default, attaining and maintaining peak operation without any search. Or, consider starting a search in a "poor" perfonning region, but initializing the search direction headed exactly for the region of peak performance. For this scenario, peak performance would be achieved much earlier than in the results presented. 22 |