**5. Results**

The results from the applied faulty classification algorithm on obtaining a signal from the experimental setup in the case of a fault on the transmission line and transformer is shown in Tables 2 and 3, respectively. From the table, it can be seen that the coefficient values were different when different mother wavelets were used. If the coefficients of the wavelet transform were in accordance

with the conditions in Figure 9, the faults inside the transformer would be detected under condition 1, while the faults occurring in the external transformer would be detected under condition 2. For this reason, the first case was an internal fault. By considering the coefficients of relay 1 (Daubechies (db2)), the faults that occurred in phase attained maximum values, and the unfaulted phases also attained maximum values. Both values had coefficients that were larger than 5 × <sup>10</sup>−3, while the coefficients under normal conditions were smaller than 5 × <sup>10</sup>−3, as shown in Figure 7a. Thus, the coefficients changed more than five times, and hence, an internal fault had occurred. The four mother wavelets exhibited the same behavioral characteristics of the coefficients. For the case of external faults, the faulted phases attained maximum values, and the unfaulted phases also attained maximum values. If these coefficients were smaller than 5 × <sup>10</sup>−3, external faults would occur. For fault classification with relay 2, positive sequences (Daubechies (db2)) were detected, which attained maximum values. Its coefficients were larger than 1 × <sup>10</sup>−2, and the pre-fault coefficients were smaller than 1 × <sup>10</sup>−2, as shown in Figure 7a. By considering the coefficients, they changed more than two times, and therefore, there were faults in the electrical systems. Additionally, all four mother wavelets provided similar results. For all data, these algorithms could discriminate the fault types with an average accuracy of 97.75%. The mother wavelet types of db2, sym2, and coif1 provided the highest accuracy at 98.65%, and the mother wavelet type of bior31 had the lowest accuracy at 95.05%, as summarized in Table 3.


**Table 2.** Coefficients from the DWT of the current signals for fault classification (relay 2).


