An important problem in the Electrical Power System operation is the s
teady-state security prediction. In order to take into account the loa
d uncertainty, in this paper the authors apply a Monte-Carlo method to
gether with an opportune Security Index to evaluate in a preventive ma
nner the probability to fall in insecure operating state, by determini
ng the security index probability density function. For this aim, in a
previous paper proposed by the authors, it has been possible to take
advantage of an Artificial Neural Network, trained to evaluate the Sec
urity Index probability density function in presence of the optimal ec
onomical dispatching of the generation powers for the load forecast. I
n the present paper, a more complex scenario is considered where the s
ecurity analysis can suggest to the dispatcher to determine also non-o
ptimal economical operating conditions to improve security. So a new,
more complex, organization of the Artificial Neural Network training s
tage, necessary in order to obtain increased generalization capacity i
n the production stage, has been considered. In the first part of the
paper the used security index, the Monte-Carlo simulation and the neur
al network structure with its learning algorithm utilized by the autho
rs for the particular problem are briefly recalled. Finally, a numeric
al application on a simple electrical test system is shown pointing ou
t very encouraging results.