The Effect of Probability of Mutation on the Performance of Genetic Algorithms for Scheduling Optimization in Unimed Electrical Engineering Study Program Rudi Salman(1), Irfandi(2), Suprapto(3), Sayuti Rahman(4), Herdianto(5)
Universitas Negeri Medan(1)(2)(3)
Universitas Medan Area(4)
Universitas Panca Budi(5)
Abstract
Computing time determines the speed of the genetic algorithm (AG). Parameters such as population size (pop size), crossover probability (Pc), mutation probability (Pm), and selected selection methods significantly affect the computing time in AG to find the optimum value. Pm value is one of the crucial parameters in AG. The value of Pm dramatically affects the process of mutation of the parent chromosome, and the value of Pm shows how much the parent chromosome will experience mutations. Determination of the correct Pm value indicates how much the parent chromosome will experience mutations.
The method used to analyze the effect of Pm on AG performance is to change the Pm value between 0.01 and 0.1. The simulation uses Matlab R2012b for the best computing time for various Pm values. Other AG parameters were unchanged, namely Pc = 0.85 and population size = 100 for each change in Pm value.
Test results using Matlab R2012b show a value of Pm = 0.06 and an average computing time of 0.382 seconds. This shows that in the case of optimizing the Unimed electrical engineering study program schedule with a value of Pm = 0.06, it will provide the fastest computing time.
Keywords: Mutation probability, computing time, optimization, genetic algorithm, scheduling course
