Study on Detection Methods of Harmonic and Inter-Harmonic Based on Neural Network

When using artificial neural network(ANN) to detect harmonic, affected by the detection error of the fundamental frequency, the parameters of harmonic and inter-harmonic calculated by trained artificial neural network has greater error, and because the neural network generally cannot adjust the excitation function parameters in the training process, resulting in the frequency of harmonic and inter-harmonic also cannot be adjusted, considering the limitations of an excitation function, this paper presented a method that the parameters of excitation function can participate in the training process, realized the precise detection to various harmonic. Simulation examples show that the method was realized detection with high precision to frequency, amplitude and phase of harmonic and inter-harmonic, and has strong anti-noise ability.