Table 20 The performance of the proposed scheme in fault and non-fault detection under specific conditions.

From: A fault classification scheme based on protective agents for microgrid with parameters impact analysis

Specific conditions

The performance of the proposed scheme for test events under specific conditions

CDR% (The DNNs have been trained for similar conditions)

CDR% (The DNNs have NOT been trained for similar conditions)

Fault cases

External faults at various distances outside the microgrid (MG) boundary

99.6

99.6% of the external faults tested were accurately classified.

92.5

92.5% of the external faults tested were accurately classified.

Fault detection in the microgrid, operating in islanded mode (the MG with IBDGs)

MG with PV (PV:1, SG:0, BESS:0)

MG with BESS (PV:0, SG:0, BESS:1)

MG with PV and BESS (PV:1, SG:0, BESS:1)

MG with PV (PV:1, SG:0, BESS:0)

MG with BESS (PV:0, SG:0, BESS:1)

MG with PV and BESS (PV:1, SG:0, BESS:1)

99.21

99.02

99.41

98.63

98.43

98.82

Severe non-fault cases

Capacitor switching

99.72

98.18

Measurement noise (30 Db)

99.65

99.15

DGs connection

PV

SG

BESS

PV and BESS

PV

SG

BESS

PV and BESS

100

100

100

100

100

100

100

100

Changes in sampling frequency

Accuracy of validation data in the training process of DNNs during data collection with different sampling frequencies

1.2 kHZ

3.2 kHZ

5 kHZ

95.5

99.895

99.95