The deterioration of equipment is taken into consideration by applying in HMI/ SCADA application the RLS-based algorithm used to on-line learning the predictive models parameters at each time when the new failure has been acknowledged delivering the new data to modify the model. The user hase possibility to runtime change the maintenance strategy, modify the alarming conditions change and create fuzzy predictors for the same alarming event using different operating conditions and after validation removing those models which show the worst quality of prognoses