This article continues research on the pantograph-catenary system started in the previous papers [1], [2]. The main purpose of the study is to use a computer of high-speed train to optimize the pantograph-catenary system by reducing power consumption when basic parameters of pantograph and catenary (contact network) change over time randomly. A linear model of pantograph-catenary system is considered where the upper and lower blocks of pantograph and catenary are modelled using lumped masses, springs and shock absorbers. The input and output system signals are measured when the train moves. These signals are processed by parametric identification algorithms to determine current values of system matrices. State matrices are used in Riccati equation to calculate controller coefficients. Adaptive controller provides dynamic stability of the system when its parameters change over time, and the system is subject to random external perturbations.