Algorithms for improving functional robustness of diagnostic systems

Algorithms are developed for use in improving robust estimators of the vari ance and auto- and crosscorrelation functions and the coefficients of Fouri er series applied as descriptive features. By means of these: features, the degree of diagnostic validation may be increased beyond that which may be achieved with the well-known algorithms To predict malfunctions, algorithms for the analysis of noise that are carriers of information about hidden al terations that precede major faults are proposed. One possible alternative implementation of the proposed algorithms in diagnostic information systems is presented.