Abstract
The paper explains a method for discerning the parts of a water supply system in need of renovation. The results are based on technical data collected over the last twenty one years, concerning more than two hundred sections of both renovated and nonrenovated pipelines. In the study, an appropriately prepared data set was used for training an artificial neural network (ANN) in the form of a multilayer perceptron (MLP). Further comparison between the responses of the trained MLP and the decisions made by human experts showed satisfactory consistency, although 15% of the database records produced certain discrepancies. The presented method can help create an expert system capable of supporting failure-free operation of a water distribution system.