To achieve a fast approach to the dynamic reactive power optimization model, a mixed integer nonlinear programming model was set up, with consideration of the discrete reactive power equipment operation limit and power grid security constraints, to minimize the transmission losses. Furthermore, a method called heuristic search and variable correction was proposed that for the first stage, the discrete variables are relaxed into continuous ones, which leads to an optimization model with equilibrium constraints. According to the optimal value of relaxed model, a mixed integer quadratic programming model is formulated to minimize the objective of approximation variance, satisfied the discrete switching operation constraints in primary model. For the second stage, the continuous variables are redressed with respect to the determined discrete variables. The impact of adjustment step and operation limit on the optimal solution was fully studied in IEEE 14-bus test system. Besides, the result of IEEE 30, 57, 118 test systems show that an optimal solution can fast be obtained by proposed method, and the precision satisfies the actual engineering, which can also employ the parallel technology.