Feature Selection and Parameters Optimization of SVM Using Particle Swarm Optimization for Fault Classification in Power Distribution Systems

Table 3

Results of SVM classification without and with considering PSO optimization techniques.

SVM classifier	Number of features			Classification accuracy (%)	Training time (s)
Without PSO	12	181.0193	1.1212	93.00	134.8
With PSO	8	15.0381	0.0334	97.15	83.54