The low-carbon development goal brings huge challenges to the distribution network, and a bi-level planning model for low-carbon distribution network, considering on-load tap changer fuzzy control, is proposed to reduce the carbon emission while ensuring power supply. The location of micro gas turbines, new energy sources, energy storage and capacitors is determined by network loss sensitivity. The bi-level model is used to fulfill low carbon planning. The planning layer is aimed at minimizing the comprehensive cost, considering the investment and operation cost of micro gas turbines, new energy sources, energy storage and capacitors, and the model is analyzed by the improved whale algorithm. The operation layer is aimed at minimizing the operation cost and voltage offset, considering OLTC fuzzy control, capacitor dropout, new energy uncertainty, micro-gas turbine and energy storage scheduling. It adopts the adaptive ε-constraint method multi-objective particle swarm algorithm to obtain the uniform Pareto front and then uses TOPSIS decision method to select the optimal solution. The proposed method can achieve low carbon and economic operation, improve tide distribution, increase the voltage quality and reduce network loss.