CLASSIFICATION OF WEAR DEBRIS USING A NEURAL-NETWORK

In this work, an evaluation was made to prove the possibility of emplo ying a neural net method for the classification of debris and monitori ng of a lubricated contact pair. We trained a neural net to classify t he severity of wear into two types from the morphological features of the wear debris. The following procedures were used. First, the shape of wear particles was characterized by Fourier descriptors. The Fourie r descriptors were considered as coordinates of a point in multidimens ional feature space. A set of points form a cluster, and the location and structure of the cluster depend on the morphology of the wear part icles and the current conditions of the contact system. A distance dis tribution between the debris in the feature space was used to represen t the location of the cluster. Second, we trained a back-propagation n eural net. To train the neural net, we used the distance distribution corresponding to the different stages of the wear process as an input vector and encoded the wear rate as a desired response. The network wa s then further trained until the desired error goal was achieved. Fina lly, we tested the trained neural net. The ability of the neural net m ethod to monitor wear is shown. (C) 1997 Elsevier Science S.A.