ANALISIS KECEPATAN KERJA SACO SAAT MANUVER BEBAN DI BANDARA NGURAH RAI MENGGUNAKAN ARTIFICIAL NEURAL NETWORK

Abstrak

Ngurah Rai Airport has a power load of 10 380 kVA, supplied by two feeders namely
Gayatri feeder and Bandara feeder, Gayatri feeder is as the primary feeder and Bandara feeder
is as backup feeder. The use of these two feeders aims to overcome the interference on the
network, due to the fact that the sensitivity of electrical equipments at airports is high. However,
if there is an interruption in the supply load of Gayatri feeder then the load supply will
automatically be transferred to the Bandara feeder using automatic switch which is called
Switching Automatic Change Over (SACO). The high sensitivity level of the electrical equipment
of Ngurah Rai allows maximum voltage drop limit of 0.5 kV of feeder nominal voltage value.
Relating to the problems, an analysis is made on the voltage drop calculations and short circuit
on Gayatri feeder. This analysis uses the program Artificial Neural Network (ANN). Parameters
that are used included the number of iterations totaling 60000 epochs, the learning speed of 0.3
and hidden layer of 40, wherein the target value of the test is 0.0000l. The process of training
and testing that has been done to produce a voltage drop value at 100 (end feeder) is a 274.7
volt with the Squared Mean Error (MSE) value of 45.5. Value fastest time delay relay was 0.3
seconds at the point of interruption 5%, and the MSE with the best result is 0.00032 to 1 phase
to ground disturbance.