Abstract: The insulation level of power system is based on the overvoltage protection level of arresters. ZnO varistor is widely used as the core element in power arrester and surge suppressor because of its excellent nonlinear voltage-current characteristics and high surge energy absorption capability. ZnO varistor's electrical characteristics are determined by its complex micro-structure and grain boundary characteristics. In this study, the microstructure model of ZnO varistor was established based on Voronoi model. And a large-scale nonlinear resistance network equation for solving the macroscopic electrical properties of ZnO varistors was established by using a grain boundary zoning model considering intergranular bypass effect. For it is difficult to solve the large-scale nonlinear resistance network equation by using the conventional Newton iterative method, a fast optimization algorithm based on piecewise linearization and differential reconstruction was proposed. The error index of the calculation result is less than 10^-3 orders of magnitude. Besides, the effects of microstructure parameters, such as spinel content, non-uniformity of grain size and porosity, on the electrical properties of ZnO varistors were obtained by calculation.