At present, along with the wide deployment of UHVDC in the power system and the connections of the high penetration of renewable energies for example large-scale wind farm, great challenges on reactive power optimization of UHVDC neighboring grid have appeared. In order to solve the reactive power optimization control issue for AC/DC hybrid power grid with flexible DC, this paper presents a dynamic reactive power optimization model with voltage source converter and considering reactive power support of the Doubly-Fed Induction Generator (DFIG). The objective function in the presented model is the minimum all-day network loss of the AC-DC grid. And also the model mainly considers four types of the constraints including the AC-DC system power flow constraints, the DC variables controlling constraints, the dynamic adjustment constraints of the discrete control variables, and the safety constraints of the node voltage. However, the model is a multi-period nonlinear mixed integer programming problem, which is tremendously difficult to be solved. As a result, by employing the linearization technology, the original model can be converted in to a second-order cone programming problem which can be efficiently solved. The evaluations have been carried out using the IEEE-30 bus system. The experiment results show the effectiveness of the presented model for dealing with the reactive power optimization task. This work is supported by National Key Research and Development Program of China (No. 2017YFB0903300) and Science and Technology Project of State Grid Corporation of China (No. DZ71-17-023).