Automatic Detection of Power Quality Disturbance Using Convolutional Neural Network Structure with Gated Recurrent Unit

Table 2

The performance comparison of deep CNN models in the detection of PQD.

Methods	Evaluation metrics (%)
	ACC	SEN	SPE	PRE	F1-score	MCC	Kappa
VGG-16	97.83	97.80	99.64	97.83	97.81	97.45	91.13
GoogleNet	96.73	96.68	99.45	96.76	96.71	96.17	86.63
ResNet-50	97.31	97.25	99.55	97.26	97.25	96.81	89.01
Proposed method	98.44	98.45	99.74	98.45	98.45	98.19	93.64

Bold values indicate the better performance over the other approaches.