In this extended abstract, we briefly present the application of robust estimators of the autocorrelation function (ACF) in cyclostationarity detection methods which may be used for local damage detection. Such approach is devoted for signals with heavy-tailed, non-Gaussian behaviour, for which standard methods might fail. We consider the modified cyclostationarity detection algorithms, where the classical ACF estimator is replaced with a robust one. The results obtained for a simulated signal show the advantage of the proposed approach