| This dissertation introduces an enhanced Genetic Algorithm (GA) that is faster and more efficient. The proposed enhancements includes using the Daubechies D4 wavelet transform as a data pre-processing step, using multiple weighted roulettes in the selection process, using multiple normal distributions for creating the initial population, recombining using multiple sites and multiple mates, and the fitness measurement to be able to measure both objective and subjective results.;To test the new enhanced GA, it was used in three different applications. (1) To detect the angle of arrival of an approaching object by obtaining readings from an array of sixteen radars. (2) The enhanced GA was applied to a dynamic large optimization problem that uses symbolic data to differentiate between edible and poisonous mushrooms using twenty two different characteristics. (3) For optimal sizing of an off-grid hybrid microgrid (MG) system in order to achieve a certain load demand.;Each one of these applications was tested using both, the standard GA and the enhanced GA. Data was collected for comparing the two algorithms to know the accuracy of the result as well as the number of iterations used by each algorithm to achieve that result.;And since each of these applications had a different goal to achieve, a specific tailored fitness function was developed for each one of them to be able to measure the quality of the resulting answer.;A special graphical user interface (GUI) was developed to run the different applications and display the answers as well as any other needed details. Java was selected since it is an excellent language for developing cross-platform desktop applications, which enables the system to be migrated on any operating system that supports java.;The results obtained from the enhanced GA were compared to both, the standard GA and the very computationally expensive brute force. The brute force was used as a benchmark to know how much computing it will take without using any artificial intelligence. Of course when the GA was used it was much better than the brute force. But when the enhanced GA was used it almost cut the computing needed by the standard GA by half, which means it uses less resources and time to reach the result.;For the Radar system the enhanced GA was able to find the answer using 395 iterations, the Standard GA was able to reach the same answer using 771 iterations, while the Brute-Force used 3,067,916 iterations. For the Audubom system the enhanced GA was able to find the answer using 381 iterations, the Standard GA was able to reach the same answer using 767 iterations, while the Brute-Force used 186,852 iterations. And for the Micrigrid system the enhanced GA was able to find the answer using 76,406 iterations, the Standard GA was able to reach the same answer using 117,624 iterations, while the Brute-Force used 32,850,000 iterations. |
