An active–adaptive control system for power grids with distributed architecture of data acquisition and processing is considered. The advantages of the proposed control principle are compared with commonly used methods. A domestic apparatus—a programmable recording bay controller (PRBC)—is described. This hardware was designed to measure the basic electrical parameters and create distributed systems for data acquisition and processing. The procedure is considered of supervisory control and data acquisition (SCADA) by the SONATA system based on a multicore distributed architecture having high reliability and supporting a rigid real-time mode. A full-function full-scale model of active–adaptive voltage control system (AAVC) was set up using proposed software and hardware. The AAVC makes it possible to use the results of calculating–measuring procedures to select a corresponding on-load tap-changer (OLTC) on the actual voltage levels in the nodes of distribution grid. This approach to centralized voltage control in distribution grids makes it possible to ensure the required level of voltage in the greatest possible number of power consumers owing to prediction of voltage change. The effect of emergencies (random failure of control equipment) during the operation of an active–adaptive voltage control system is analyzed. A qualitative evaluation of the effect of faults on regulation quality and choice of OLTC connection is carried out. It is shown that, for undisturbed operation of an active–adaptive voltage control system, additional diagnostic tools, backup, and data loss compensation are needed.