Real time harmonic load identification based on fast fourier transform and artificial neural network

Dimas Okky Anggriawan, Aidin Amsyar, Aji Akbar Firdaus, Endro Wahjono, Indhana Sudiharto, Eka Prasetyono, Anang Tjahjono

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)


– Harmonics at nonlinear loads have different values for each load. These characteristics can be used as identification of the type of load from the harmonic value produced from nonlinear load. Therefore, this paper proposes harmonic load identification using Fast Fourier Transform (FFT) and Artificial Neural Network (ANN). In order to obtain harmonic values, a prototype of measuring instruments is used, while the method for obtaining harmonic values is FFT. The harmonic value obtained will be used as training data and testing data in ANN. Then, the type of training used as the classification of load type is the Levenberq Marquardt. The input of this method is a harmonic value of three types of nonlinear loads with seven combinations. The training process is carried out in MATLAB. Then, the weight and the bias values of each neuron are obtained and programmed in microcontroller. In order to validate the proposed algorithm, testing is conducted by three nonlinear load combinations. The results show that the proposed algorithm has good results in load identification with the highest accuracy of 99.94%.

Original languageEnglish
Pages (from-to)220-228
Number of pages9
JournalInternational Review of Electrical Engineering
Issue number3
Publication statusPublished - 2021


  • Artificial Neural Network
  • Fast Fourier Transform
  • Harmonic
  • Non-Linear Loads


Dive into the research topics of 'Real time harmonic load identification based on fast fourier transform and artificial neural network'. Together they form a unique fingerprint.

Cite this