Considering the active management, a multi-objective distributed generation planning model was proposed, the upper level planning objective of which was to minimize both the investment and operating cost of distributed generation and the power loss, while the lower level planning objective of which was to minimize the curtailment value of distributed generation. It took into account the randomness of intermittent generation, such as wind turbine generation and photovoltaic power generation, as well as the uncertainty of loads. The Latin hypercube sampling-based Monte Carlo simulation was used to sample the wind speed, the illumination intensity and the load. The upper planning model was solved by the non-dominated sorting genetic algorithm while the lower planning model was solved by the prime-dual interior point method. Finally, the Pareto-optimal solutions were obtained, which could avoid the subjective impact of the traditional weighted methods on determining the weights. The feasibility of the model and the effectiveness of the algorithm were proved by the simulation and analysis of a 33-bus distribution system.