The problem of detecting a useful signal in the presence of a strong background noise is considered. To solve it, a statistical approach is used, based on a change in the level of chaos in the system when an additional random or deterministic process occurs, which is probabilistically independent from a set of stochastic phenomena that form background noise. It is shown that the occurrence of this process changes the level of entropy of the measured signal; this fact is the basis of the applied mathematical algorithm. It is based on the elements of the Fourier transform apparatus for the probability density with an appropriate choice of a nonlinear function of the random process under study. The proposed approach, based on variations in the randomness in the system in the presence of a useful signal, makes it possible to record its presence against the background of noise components even at low signal-to-noise ratios. The effectiveness of the method is confirmed both by theoretical justification and by the calculations presented in this work. The condition for the implementation of the technique described in the article, which does not impose restrictions on the studied physical fields and frequency ranges, is the comparability of the width of the probabilistic distribution of the desired useful signal with several intervals of discreteness of the measuring equipment. One of the results of this work is a high sensitivity to the emergence of independent random components.