In this paper we propose an algorithm of optimal control of melting process of copper nickel raw material in furnaces Vanyukov based on the method of Bellman's dynamic programming. To construct a multi-step management process developed by the authors used neural network model of the melting furnaces Vanyukov. The criterion selected for optimality for a minimum loss of non-ferrous metals smelting process. To reduce the amount of calculations the authors propose to enter the grouping of states of the process by loss of non-ferrous metals at a fixed time.