APPLICATION OF ARTIFICIAL NEURAL NETWORKS (ANN) AS MULTIPLE DEGRADATION CLASSIFIERS IN THERMAL AND FLOW DIAGNOSTICS

Abstract

Application of a neural network of the classifier type for diagnostic purposes is presented. The described ANN solves the task of recognizing causes of degradation of power units’ thermal cycle components. Verification of the applied ANN responses is based on the presented research in the numerical simulation of selected power installations. The obtained results could be used in diagnostics of power cycle being properly measured. Considerably good obtained results prove that the ANN technique can be applied as an automatic detector of operational faults. Thus an ANN can serve as a support tool for operational decisions. The present work also offers a way of reducing training time.