With the increasing growth of HVDC transmission projects, preliminary formation of AC-DC hybrid power grid brings new challenges to reactive power optimization of traditional power grid. Aiming at this problem, an improved bare-bones differential evolution algorithm (IBBDE) is proposed to solve the reactive power optimization of hybrid AC-DC system. Initialization of population using generalized opposition based learning and self-adaption of crossover factor are introduced to enhance the global convergence of algorithm. The IBBDE algorithm is tested on IEEE 30-bus system with HVDC, the simulation results show that, comparing with DE and BBDE algorithm, the proposed IBBDE algorithm can achieve better reactive power optimization effect and has superiority on optimization stability.