At present

capacitive reactive power reversal is becoming increasingly prominent in the period of light load in large urban network

which endangers the safe and stable operation of power grid. Based on this background

the reactive power optimization method of distribution network with access of large-scale cables is studied. The influence of reactive power reversal on nodal power factor is analyzed and an integrated reactive power optimization model with consideration of power factor constraints（containing reactor and capacitor）is established. The minimization of the real power loss is the objective function of the model. The constrains of node power factor and switching capacity of reactor are added based on traditional reactive power constraints. Then

an improved genetic algorithm with loss sensitivity is used to solve the model. By coordinating the switching of reactors and capacitors in distribution network

the optimized scheme can avoid reactive power reversal and power factor exceeding the limit with large-scale cables access. Finally

a typical high-voltage distribution network is used as an example to verify the effectiveness of the model and algorithm.