New starting points for the prediction of tool wear in hot forging

Formation of a sufficiently large database on tools for hot forging, which is necessary for successful prediction of wear at a given number of strokes, as well as for the prediction of the critical number of strokes when the acceptable tolerance of a forging is exceeded, is a relatively time-consuming process in the production practice. To overcome this problem, this article presents a starting point for quicker prediction of these quantities by means of conditional average estimator neural networks (CAE NN), namely by the so-called integral method and by the partial method. A comparison of the efficiency in prediction of these methods was carried out on the results of wear obtained in laboratory forging, which allowed a gradual and relatively quick tracing of wear contour progression on tools and thus the formation of a reliable database. The results presented show that in the case of a relatively small database, where, for instance, there are known data and wear parameters on at least three different tool steels, or, on differently heat treated steels, it is possible to effectively predict the wear of a fourth tool simply on the basis of the slightly perceivable wear profile of the tools. Here, the integral method gives better predictions. This conclusion is of great importance in practice: from intermediate control of gradual tool wear, we can predict its tool life.