To remedy the defect of traditional PI controller in static Var compensator nonlinear voltage control system, a method based on novel adaptive dynamic programming, goal representation heuristic dynamic programming (GrHDP), is proposed to optimize control of SVC voltage control system adaptively. According to the external enhancement signal and the internal strengthening signal, the GrHDP algorithm uses error backward propagation to adjust the weights of the three neural networks and obtain the optimal weights in order to realize the optimal control of the system by selecting the current and historical voltage errors as the system state feedback vector. The Action Dependent Heuristic Dynamic Programming (ADHDP) and the GrHDP algorithm are used to design the voltage regulator of the SVC voltage control system respectively, and simulation comparison is carried out in Matlab/Simulink platform. The results show that the SVC voltage control system based on GrHDP algorithm can compensate reactive power quickly, and has better voltage stability and control effect.The control system has faster response speed and stronger adaptive ability. This work is supported by National Natural Science Foundation of China (No. 61563006), Science and Technology Project of Guangxi (No. 1598008-2), and Natural Science Foundation of Guangxi (No. 2013GXNSFCA019020).