Levenberg Marquardt Backpropagation Neural Network for Harmonic Detection

Nowadays, non-linier loads show declining effect for power quality in electric power system that be concerned by engineering and scientist community. In order to prohibit harmonic from deflate the power quality in power system caused by non linear loads, detecting harmonic for harmonic mitigations becomes critical issue. This paper presents an effective procedure based on levenberg marquardt backpropagation neural network is offered to find harmonic of the measured signal that is an input harmonics content sampling signal. The result from neural network harmonic detection method are compare to Fast Fourier Transform algorithm. To validate the proposed algorithm compared under varying neurons in hidden layer. The result show that the proposed algorithm has good accuracy for harmonic detection on the first harmonic signal with 5 neurons has average 0,00083 percent and 10 neurons has average 0,00033 percent and in the second harmonic signal with 5 neurons has average 0,00005 percent and 10 neurons has average 0 percent.