The model of electric power consumption dynamic at the fuel and energy complex has been suggested which possesses properties of forecasting and generation of control actions. The model operation is based on the combined fuzzy neuron networks. The model allows to increase the accuracy of indexes in estimating of future periods for optimization problems solution in comparison to the existing realizations based on regression models. The paper has determined the objective function of optimization on minimum losses of active power together with limitations typical for fuel and energy complex facilities; suggested modified Newton-Raphson algorithm based on fuzzy neuron networks; obtained efficiency assessment of the developed algorithms applied to the simulation models of FECF functioning for short-, medium- and long-term forecasting.