The paper presents risk assessment processes, including data preparation, data mining, prediction instant model, risk potential assessment. Attention is paid to the issue of developing an instantaneous predictive model with stability conditions based on a multiple-scale wavelet transform as an integral part of the generalized risk assessment process. On the example of an instant predictive model of the feedwater temperature downstream of the high-pressure heater, which is the equipment of the 2nd circuit of the NPP power unit with a VVER-type reactor, an analysis of the detailing coefficients of the model parameters based on the object database was carried out and a conclusion was made about their normal distribution. Based on the analysis, the key process of assessing the risk potential “prediction instant model” is supplemented by the stage of the metric of compliance with the normal distribution of the distribution of the detailing coefficients of the model parameters.