In response to the cost and stability issues in distribution networks caused by uncertainty of distributed PV output, this paper proposes a multi-objective probabilistic planning method for distribution network with high proportion of distributed PV. PV output data is reduced through K-means clustering to obtain typical scenarios and their probability models. Uncertainty scenarios are generated to simulate the actual operation of distributed PV through using Monte Carlo probabilistic power flow. A bi-level probabilistic planning model is established. The upper level minimizes costs and maximizes PV penetration, determining the location and capacity of distributed PV and energy storage. Considering the uncertainty of distributed photovoltaic output, the lower level minimizes operational and maintenance costs, network loss costs, power purchase costs, and voltage deviation index under probabilistic power flow, optimizing distributed PV and energy storage operation. An improved particle swarm optimization (PSO) algorithm is used to solve the model. Case study is conducted using IEEE 33-node system and actual PV output data from a certain country in Anhui. Results show that the proposed method improves PV penetration and operational stability while reducing costs compared to traditional planning methods.