The key features of Active Distribution Network (ADN) which include active management and active control exacerbate the difficulty of its generation planning. Based on features of ADN, a multi-objective Bi-level programming model about distribution generation considering active management mode was proposed. The objective of upper level model was to minimize the total life cycle cost of power supply side in the system, thus the installation and configuration plan was obtained after optimization. Whereas the lower level model was aimed at minimizing active power curtailment value, and operation optimization of distributed generation (DG) was obtained through applying DG output control, on-load-tap-changing transformer adjustment, reactive power compensation device adjustment and other active management modes. The upper and lower model were solved by improved particle swarm optimization (PSO) algorithm with adaptive mutation, which could improve the solution speed and the global optimal effectiveness. Besides, the simulation of the IEEE 33-node system was carried out to verify the rationality of the model.