An Intelligent Fault Diagnosis Method for Transformer Based on IPSO-gcForest

Table 3

Comparison of diagnostic results of different models.

Evaluation index	Fault type	Diagnostic results of different models /%
		RF	DBN	gcForest	PSO-gcForest	IPSO-gcForest
Precision	N	80.00	80.56	85.29	93.75	96.88
	D1	80.85	82.11	90.80	98.77	98.80
	D2	71.43	80.00	82.14	90.00	96.55
	D3	88.24	87.50	73.68	71.43	84.21
	T1	88.68	90.57	91.38	89.66	91.38
	T2	87.65	88.89	92.41	94.87	94.94
	T3	85.00	85.00	87.50	95.24	95.24
Recall rate	N	84.85	87.88	87.88	90.91	93.94
	D1	90.48	92.86	94.05	95.24	97.62
	D2	50.00	53.33	76.67	90.00	93.33
	D3	78.95	73.68	73.68	78.95	84.21
	T1	83.93	85.71	94.64	92.86	94.64
	T2	93.42	94.74	96.05	97.37	98.68
	T3	73.91	73.91	60.87	86.96	86.96