Application of Artificial Neural Networks - Digital Signal Processing in Power System Protection and Control - page 263

CT

Transmission line

Z SA

F

E SA

Load

System equivalent

Z O

Fig. 12.20 Schematic diagram of the system with CTs under study

with satlins activation function. This option was introduced to check how the
simplification of activation function influences the performance of ANN and
computational burden of the processor, provided the ANN is implemented in
hardware. The satlins function is defined as

8
<

1 ;
x ;
1 ;

x\ 1
j x j 1
x [ 1

satlins ð x Þ¼

ð 12 : 9 Þ

:

and both activation functions are compared in Fig. 12.19 .
The input vector of all ANNs consisted of a = 15 most recent samples of the
secondary CT current i s (3/4-cycle sliding window, sampling with 1 kHz). The
network ANN3 was trained without any scaling of the input signals, while for three
other ANNs their input vector was being scaled (all the samples of the input vector
were divided by the maximum over considered data window). Target values
adopted during training were primary current samples i p (n) related to the sec-
ondary side that were available from EMTP-ATP simulations. All the ANNs were
trained to be used during CT saturation periods only.
The testing of developed units for CT saturation, detection and correction was
performed with EMTP-ATP generated signals [ 8 ]. Simple 110 kV transmission
system fed from one side was modeled, as shown in Fig. 12.20 . It comprises the
equivalent of feeding system represented by an electromotive source behind an
impedance (positive and zero sequence values, X 0 /X 1 = 1.5) and transmission line
transferring power to the end user (load). The simulation model from Fig. 12.20
was equipped with suitable non-linear model of CT, [ 20 ]. The CT model devel-
oped was an equivalent of standard current transformer 5P20. The last point of the
flux-current curve was set at various levels (flux coordinate) to obtain CTs with a
number of different inductances and saturation time constants of the secondary
circuit. The transformation ratio of CT under study was # CT = 500:1 (A/A).
Thorough simulation study of the current transformer operation under various
conditions has been performed. Phase-to-ground or phase-to-phase faults in the
middle or at the end of transmission line were simulated. To obtain a wide variety
of cases the following parameters of the system and transient conditions were
being changed in each simulation run:

• Level of the primary current I p1 during fault (resulting from the strength of the
supplying system): [4.8, 5.2, 5.7, 6.2, 6.9, 7.7, 8.8, 10.2, 12.2, 15.0, 19.7,
28.5] 9 I n ,