The paper presents a modeling method using artificial neural networks and the possibility of implementing it in a software program. In this paper we present a supervised learning algorithm with back propagation. The algorithm consists of two phases: training phase and recall phase. In supervised learning, the purpose of a neural network is to change its weights according to the inputs/outputs data. After a network has established its input-output mapping with a defined minimum error value, the training task has been completed. The weights determined in the training phase are used in the recall phase to calculate the output for every input pattern.