To improve the harmonic current detection performance of active power filter (APF), the author proposes an adaptive harmonic current detection method based on neural network. According to the basic principles of adaptive noise elimination technology, the main current is filtered from the load current to obtain a harmonic current. The principles of this method have been analyzed in detail, and the specific implementation of this method has been given, such as the selection of neural network reference, inputs, and weight updates. discuss the input of the measurement results reference. Assuming that the load changes abruptly after the third cycle, that is, the square wave amplitude changes from 1.0 A to 0.5 A, the circuit structure does not change, and the multilayer feedforward network training parameters remain the same. This harmonic real-time performance ensures the accuracy of current detection methods.
Cite this article as: P. Ji, X. Li, V. Jagota and G. Sekhar Sajja, "Study on an adaptive harmonic current detection method based on neural network," Electrica., 22(3), 358-364, 2022.