Harmonic Detection Using Neural Networks with Conjugate Gradient Algorithm

Abstract

In this study, the Elman’s recurrent neural networks using conjugate gradient algorithm is used for harmonic detection. The feed forward neural networks are also used for comparison. The conjugate gradient algorithm is compared with back propagation (BP) and resilient BP (RP) for training of the neural networks. The distorted wave including 5th, 7th, 11th, 13th harmonics were simulated and used for training of the neural networks. The distorted wave including up to 25th harmonics were prepared for testing of the neural networks. The Elman’s recurrent and feed forward neural networks were used to recognize each harmonic. The results of the Elman’s recurrent neural networks are better than those of the feed forward neural networks. The conjugate gradient algorithm provides faster convergence than BP and RP algorithms in the harmonics detection