Abstract: To address issues such as intensified voltage fluctuations and increased network losses in distribution networks with high-penetration photovoltaic (PV) systems, a two-stage voltage and reactive power optimization strategy considering voltage resilience is proposed. First, a voltage resilience index based on an equilibrium coefficient is established to reflect the balance of voltage resilience across all nodes in the network. A day-ahead optimization model is subsequently constructed by integrating traditional indicators such as voltage deviation, voltage over-limit risk, and network losses. Secondly, considering the strongly coupling and mixed-integer nonlinear characteristics between discrete and continuous devices, an exponential increasing particle swarm optimization (EIPSO) algorithm is proposed to solve the day-ahead scheduling problem, thereby improving convergence accuracy and speed. Furthermore, an intra-day real-time autonomous optimization strategy is proposed to address strong uncertainties such as PV output prediction errors and real-time load fluctuations. This strategy dynamically adjusts the PV inverter reactive power compensation based on prediction errors and incorporates voltage sensitivity for real-time autonomous correction. Finally, the effectiveness and superiority of the proposed method are verified through the simulations on a modified IEEE 33-bus system. The results indicate that the proposed method reduces day-ahead network losses by 17.6% in the day-ahead stage with fewer discrete device operations. In the real-time stage, voltage deviation and over-limit risk are reduced by 18.4% and 23.2%, respectively. The proposed strategy effectively suppresses voltage fluctuations and enhances the operational stability and economic efficiency of renewable energy distribution networks.