The article examines the process of building a mathematical model of intelligent decision support for determining the level of an enterprise's energy consumption while producing different quantities of products. To solve this problem, it is proposed to use a mathematical model based on the theory of neural networks, namely, a two-layer perceptron. The approximation of the nonlinear energy consumption function is carried out by the Scaled Conjugate Gradient method. As a result of training, a prognostic mathematical model was obtained that allows the determination of the level of the enterprise's energy consumption depending on the input factors. In particular, it permits forecasting the level of the enterprise's energy consumption due to a production plan, hours of planned power outages, and periods of using alternative sources.