Using an artificial neural network to determine the wear level of the cylinder piston group of a marine engine

The article presents a detailed analysis of the peculiarities of using artificial neural network in the tasks of diagnosis and prediction of the technical condition of the elements of ship power plant. The prospects and peculiarities of using artificial neural network for determining the wear level of the cylinder-piston group of a ship engine are considered. The prognostic neural network model formalized in the article allows to determine the wear level of a ship engine cylinder-piston group on the basis of a neural predictor. Besides, it makes it possible to analyze functional relations between parameters and draw conclusions about serviceability of diagnosed elements. Separate attention is paid to the construction of the neural predictor circuit and the choice of diagnostic parameters. In order to put the proposed model into practice the list of ship engine cylinder-piston group operating parameters which can be fed into the model input is presented. In addition, the article specifies the learning algorithm of the neural network, the basis of which is the rule of the normalized least mean square.