14. Start Preparation of
dataset
Training dataset
(80%)
Testing dataset
(20%)
Development of
accurate models
Selection of best
fitted model
Number of neurons
Number of hidden layer
Number of epoch
Activation function
Training Algorithm
Validation of the
selected model
End
Mean absolute error (MAE)
Root mean square error (RMSE)
Correlation coefficient (R)
Coefficient of determination (𝑅2
)
Implementation of
Artificial neural
network (ANN)
Splitting the dataset
No
Yes
15. Start
Define input & output pairs
Control the DC/DC inverter
using the trained ANN
Dataset extraction from
simulation
Define Number of hidden
layers
Adjust number of
hidden layers
Chose training method
(Levenberg Marquardt
technique)
Validation of the
selected model
End
Dataset distribution
Training= 60%
Testing=20%
Validation=20%
FALSE
TRUE
Check MSE
&
Confusion Matrix
FALSE
TRUE
17. Start
Define input & output pairs
Control the DC/DC inverter
using the trained ANN
Dataset extraction from
simulation
Define Number of hidden
layers
Adjust number of
hidden layers
Chose training method
(Levenberg Marquardt
technique)
Validation of the
selected model
End
Dataset distribution
Training= 60%
Testing=20%
Validation=20%
FALSE
TRUE
Check MSE
&
Confusion Matrix
FALSE
TRUE
22. Percentage
0.01 0.1 1 1000
Time period (seconds)
35.0%
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0.0%
10 100
Transmission and HV distribution systems LV distribution systems
Percentage Overvoltage
Must Stay Connected
Disconnection Permitted
33. Past
K K+1 K+2 K+p
Future
Prediction Horizon
Sample Time
Reference Trajectory
Predicted Output
Predicted Output
Predicted Control Input
Past control Input
40. Past
K K+1 K+2 K+p
Future
Prediction Horizon
Reference Trajectory
Predicted Output
Measured Output
Predicted Control Input
Past control Input
Sample Time
41. Yes
No
Measure
𝑉𝑑𝑐(𝑘 + 1)
x=0
x=x+1
g=𝑉
𝑠
∗
− 𝑉
𝑐
𝑝
(k+1)
𝑊𝑎𝑖𝑡 𝑓𝑜𝑟 𝑛𝑒𝑥𝑡
𝑠𝑎𝑚𝑝𝑙𝑖𝑛𝑔 𝑖𝑛𝑠𝑡𝑎𝑛𝑡
Start-up
x < 𝑁2
Store optimal
values
Apply optimal
Switch state s(k+1)
is
p
k + 1 = G1iL
p
k + 1
𝑉
𝑐
𝑝
𝑘 + 1 =
𝑇𝑠
2
𝐿1𝑐1
𝑠 𝑘 ∗ 𝑉
𝑠 − 𝑉
𝑐
𝑝
(𝐾) + 2𝑉
𝑐
𝑝
𝐾 − 𝑉
𝑐 𝑘 − 1