Optimasi Suitability Fungsi Keanggotaan Fuzzy Menggunakan Algoritma Modified Particle Swarm Optimization

Abstract

The optimization of fuzzy membership function is a solution that is widely applied to determine the proper design of membership functions and optimal. In this research the optimization based Modified Particle Swarm Optimization (MPSO) algorithm is applied to optimize fuzzy membership functions. There are two methods MPSO which applied the method of Linear Decreasing Inertia Weight (LDIW) and constriction Factor Method (FCM). Each of these methods is tested with 10 trials on two types of particle number, namely 50 and 20 particles. From the test results showed that the amount of the same particle, CFM obtains the global best fitness value which is more optimal than LDIW method. The test is performed as many as 10 and apply 50 particles, the first experiment obtained the global best fitness value, namely 1.4;1.4; 2.36 and 3.28 for each productivity variable, isolation, social relations and accessibility. The test is performed as many as 10 attempts and applied 20 particles obtained global best fitness value namely 2.34; 2.40; 2.J7 and 3.36 for each variable. On the other hand CFM obtain faster results convergence than the method of LIDW. The experiment at 100 swarm, LDIW method is to find the global best fitness on swarm 91, 84, 54 and 38 for each variable, while the CFM method finds the global best fitness on the swarm 81. 23. 34 and 23.