Abstract

The rising prevalence of dispersed energy-generating resources like wind and solar power systems degrades the quality of electricity. For the safety of maintenance staff and the equipment in such situations, detecting the power quality troubles in the power systems is highly significant. In order to cater for these issues, the stationary discrete wavelet transform (SDWT) and hybrid long short-term memory (LSTM) convolution neural network (CNN) based classifier is proposed in this work. Besides, the African vulture optimization algorithm (AVOA) is adopted for optimizing the weights of layers in hybrid LSTM-CNN (HLC). This work aims to detect the islanding operation under different faulty conditions in the network. The implementation results are analyzed for various conditions, showing the fidelity of a proposed method by detecting the islanding faults. Moreover, the proposed method is validated under different noise levels and shows better performance in noisy signal to noise ratio (SNR). The accuracy obtained for the proposed SDWT-HLC method without noise is 99.92%. The accuracy with 20, 30 and 40 dB noise for the proposed method is obtained as 99.84%, 99.87% and 99.89%, respectively. Furthermore, the accuracy, precision, recall and F1-score are 0.999, 0.995, 0.994 and 0.994, which shows the efficacy of the proposed method.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call