SIMULASI PENINGKATAN KUALITAS TEGANGAN MENGGUNAKAN DYNAMIC VOLTAGE RESTORER (DVR) DENGAN KENDALI LEVENBERG MARQUARDT NEURAL NETWORK PADA TEGANGAN RENDAH

Abstract

In this thesis proposed a kontrol algorithm for dynamic voltage restorer (DVR). The proposed kontroller is using a neural network with Levenberg Marquardt method Neural Network (LMNN). The purpose of this kontrol is to obtain a DVR with a fast and accurate response while improving the quality of the voltage from the voltage sag. Levenberg Marquardt Neural Network (LMNN) used in making the detection of changes in conditions of stress, either in the form of fluctuations in amplitude and phase changes in the voltage at the load. After generating a signal which is detected Levenberg Marquardt Neural Network (LMNN) compared with the PWM carrier signal. To find out Levenberg Marquardt Neural Network (LMNN) kontroller performance, then the simulation is used as a comparison of conventional kontrollers. Based on simulation results is known that the kontroller Levenberg Marquardt Neural Network (LMNN), DVR is more stable with faster response