<p>The increasing penetration of distributed photovoltaic generation has intensified line losses and voltage limit violations in distribution networks. To address these operational security challenges caused by high photovoltaic penetration in distribution networks, this paper proposes a Proximal Policy Optimization - Energy Storage Dispatch method for line loss reduction and voltage regulation. First, the dispatching problem is reformulated as a Markov decision process with a 24-dimensional continuous action space. Second, an energy storage charging/discharging power update strategy is designed to enforce the feasibility of the state of charge at the action level, thereby restricting infeasible policy outputs that violate SOC constraints. Third, a multi-objective composite reward framework is designed, in which the negative line loss is adopted as the core reward to explicitly guide the agent toward line loss reduction, while voltage limit violation-related penalties are incorporated to achieve voltage regulation. The proposed method is validated on the standard IEEE 33-bus and IEEE 69-bus systems, and all experimental results are reported as the mean ± std over three random seeds. For the IEEE 33-bus system, the proposed method reduces line losses by 47.93 ± 0.10% (one energy storage) and 59.06 ± 0.14% (two energy storage, Multi-agent PPO), while maintaining all bus voltages within the range of [0.962, 1.033] p.u. For the IEEE 69-bus system, line losses are reduced by 47.81 ± 0.09% (one energy storage) and by 50.96 ± 0.20% (two energy storage, MAPPO), with all bus voltages maintained within the range of [0.961, 1.031] p.u.</p>