Nuclear power units (NPP) assigns to significant objects of critical information infrastructure (SOCII). The SOCII state can define by the actual level of information security and the related risk potential. Within the intelligent operator support system of NPP is the problem of computing technical and economic indexes (TEI), the analysis of which will define the anomalies in the PU's operation. The computing TEI founded on multiscale wavelet transforms, data mining, and risk potential predicting models will enable the operating personnel to define the acceptable operating limits and anomalies' locations. To solve this problem, the paper proposes to use predictive instant models (for example, the feedwater temperature after the high-pressure heater 5 to assess the potential risk founded on the multiscale wavelet transform.