Analysis of electrochemical noise by power spectral density applied to corrosion studies - Maximum entropy method or fast Fourier transform?

The fast Fourier transform and the maximum entropy method (MEM) provide algorithms for the estimation of the power spectral density (PSD) of fluctuations. Both have been employed for the analysis of electrochemical noise in corrosion studies, and claims have been made concerning the superiority of one method with respect to the other. in this paper, the two methods are compared to assess their relative advantages. A summary of the principles of the MEM is given and its main properties investigated. In particular, the effect on spectrum accuracy of varying the number of coefficients in computing the MEM and the validity of the low-frequency plateau in the PSD usually produced by this technique are examined. Also, the robustness of the two methods is compared when the random process is not completely stable, for instance, in the presence of signal drifts or slowly varying amplitude of the fluctuations. The results presented may be used as a guideline to choose the best computation method as a function of the measurement conditions.