A neural network model for power system inter-harmonics estimation

A neural network model is proposed to estimate the parameters of inter-harmonics. As to this specific neural network, an adaptive learning algorithm based on improved Levenberg-Marquardt algorithm is derived. Because of the complete match between the neural network and the inter-harmonic model, the presented algorithm can effectively improve the precision in the process of inter-harmonics estimation and meanwhile, accelerate its convergence. Simulation results have testified its performance with a variety of generated inter-harmonics. If noise is not taken into consideration, the introduced algorithm can accurately measure the power system inter-harmonics. Furthermore, when the signal is polluted with Gaussian noise, our method can still maintain relatively high level of accuracy.