An Improved Power Quality Disturbance Detection Using Deep Learning Approach

Table 3

DNN training model performance.

Type of model	Time consumed during each epoch (s)	Total parameters	Best validation loss	Best validation accuracy	Training time (s)	Elapsed epoch
CNN + LSTM	77	95632	0.0032	0.9990	3619	47
CNN + GRU	78	100944	0.0037	0.9978	2652	34
DCNN	97	174820	0.0040	0.9970	4462	46