The use of powerline communications has attracted the attention of researchers increasingly, since this kind of data transmission allows to reduce the number of cables. This advantage is particularly appealing in railway, naval and aeronautical fields. In this paper a method to detect noisy events that occur in a powerline communication system in a naval environment is proposed. To this aim, a model of the powerline used to transmit wideband signals is employed to simulate different possible noisy conditions. The channel is analyzed in the presence of three types of noise which characterize the system. A supervised technique based on the use of principal component analysis and a k-mean classifier is developed to automatically detect the presence and classify the specific noise which occurs during the transmission. Numerical simulations show that the proposed approach enables to classify the occurring of noisy events reliably.