Robust technology for calculation of the coefficients of the Fourier series of random signals

Estimators in spectral analysis are made robust by balancing the positive a nd negative microerrors caused by the influence of noise. However, the need to determine the sign of the noise for each reading makes it difficult to implement the algorithms, It is shown that there is a way of eliminating th is difficulty, Algorithms are proposed that take into account four possible variant relations between the mean values of the positive and negative pro ducts of the readings, on the one hand, and the sine and cosine curves, on the other hand, A methodology for robust spectral analysis of measurement i nformation is constructed on the basis of the set of the: curves.