There are different weft insertion methods used in the weaving machines. Electromagnetic launcher can be used as accelerator to yarn which is attached to the ferromagnetic projectile. For optimization of the projectile velocity in this new weft insertion, the projectile velocity should be obtained by numerous states of inputs (different states of launcher parameters). Experimental tests are very time consuming and costly. Therefore, a validated model (Adaptive Neuro Fuzzy Interface System: ANFIS) which can predict the system behavior accurately and faster than experiments was implemented to predict the projectile velocity. The optimization result of the genetic algorithm shows that achieving the modern weaving machines speed for this weft insertion system is possible by the optimal values of launcher parameters. Also, using the ANFIS model for obtaining the system output is very useful approach. So, the system can be optimized for other desired targets by this procedure.