The paper is devoted to the problems of the information task of the upper level control system of the NPP I&C system "Calculation and analysis of technical and economic indicators" and the construction of instantaneous predictive models for this task based on the critical parameters of the power unit. The article considers a modified algorithm for assessing the risk potential of a technological process based on data mining, analysis of approximating and detailing coefficients of the wavelet decomposition of the input and output parameters of the instantaneous predictive model, and stability conditions of the predictive model based on a multiple-scale transformation. Using the example of an instantaneous predictive model for the feed water temperature at the inlet of a high-pressure heater, Hurst indicators for the measured parameters of the model are given, and a conclusion is made about the persistence of the ongoing process with the effect of long-term memory, a sample of close states from the object database is shown on the example of one of the points. Also, for this model, a comparison of real data and the predictive model is given, which shows the adequacy of the developed instant predictive model. In this connection, it is concluded that this model can be used to monitor the state of the NPP power unit, which allows identifying changes of various nature in the processes occurring in the system to assess the risk potential of processes, despite the presence of unreliable or substituted signals.