Within this work, a novel controller in terms of H infinity (H) and structured singular value decomposition has been presented to provide the robust performance of the Automatic Voltage Regulator (AVR) system Six real structured uncertainties in actuator, exciter and generator have been assumed for the linear transfer functions of the AVR system. Each uncertain parameter varies between a minimum and a maximum value due to the load variations in a period of time and aging effects over the life time. The efficiency of the presented design lies on two main reasons. The first is the simultaneous considering of the output disturbances, sensor noises and system uncertainties in the controller design approach. The second is the non-conservative modeling of all six structured parameters in the required μ-synthesis P-Δ-K configuration. By suboptimal H control design technique and μ-analysis theorem, a single input single output (SISO) controller comprising a closed loop system with μ<1 is obtained. The offered controller's supremacy is represented through comparison of its performance with some other optimized PID, PIDD fractional order PID (FOPID), fuzzy + PID and Interval Type-2 fuzzy logic controllers by heuristic optimization algorithms. The simulation outcomes indicate that the provided robust controller for the AVR system has the better performance than the other optimized and fuzzy controllers in a wide range of the uncertainties. Also, the better behavior of the intended robust controller was shown in two benchmarks: a single machine connected to a 230kV network, and a four-machine two-area test system.