Generator sinkron memerlukan alat yang berfungsi untuk mengatur tegangan terminal tetap terjaga, alat tersebut dikenal sebagai Automatic Voltage Regulator (AVR) atau alat yang dapat mengontrol tegangan dan arus eksitasi yang disuplai ke generator sehingga tegangan terminal generator tetap stabil. Penelitian ini membandingkan dua kendali yang berbeda yaitu logika fuzzy dan jaringan saraf tiruan pada AVR yang diharapkan dapat mengatur tegangan terminal generator pada nominal 220V. Data pelatihan jaringan saraf tiruan diambil dari input dan output kendali PID sedangkan desain rule logika fuzzy menggunakan metode trial and error pada AVR yang sama. Hasil penelitian menunjukkan respon transient kinerja generator dengan hasil sebagai berikut. Logika fuzzy menghasilkan respon delay time (t_d) 1.058s, rise time (t_r) 2.576s, peak time (t_p) 20s, settling time (t_s) 5.354s, max overshoot (Mp) 0%, error steady state (e_ss) 0%. Jaringan saraf tiruan menghasilkan respon delay time (t_d) 1.031s, rise time (t_r) 1.518s, peak time (t_p) 3.128s, settling time (t_s) 2.3286s, max overshoot (Mp) 1.345%, error steady state (e_ss) 0.045%. Berdasarkan nilai respon transient logika fuzzy memiliki respon lebih lambat dibandingkan jaringan saraf tiruan, akan tetapi logika fuzzy dapat meminimalisir overshoot.

Kata kunci: AVR, Generator Sinkron, Jaringan Saraf Tiruan, Logika Fuzzy

Synchronous generators require a device that regulates the terminal voltage to remain stable; the device is called an Automatic Voltage Regulator (AVR) or a device that can control the voltage and excitation current supplied to the generator so that the generator terminal voltage remains stable. This research aimed to compare two different controls, namely fuzzy logic and artificial neural network on AVR, which is expected to regulate the generator terminal voltage at a nominal 220V. Artificial neural network training data was taken from the input and output of PID control, while fuzzy logic rule design used a trial and error method on the same AVR. The results showed that the transient response of the generator performance based on the following results, Fuzzy logic produces a response delay time (td) of 1.058s, rise time (tr) of 2.576s, peak time (tp) of 20s, settling time (ts) 5.354s, max overshoot (Mp) 0%, steady state error (ess) 0%. The artificial neural network produced response delay time (td) 1.031s, rise time (tr) 1.518s, peak time (tp) 3.128s, settling time (ts) 2.3286s, max overshoot (Mp) 1.345%, steady state error (ess) 0.045%. Based on the transient response value, fuzzy logic had a slower response than artificial neural networks, but fuzzy logic can minimize overshoot Keywords: AVR, Fuzzy Logic, Artificial Neural Network, Synchronous Generator