At present, with the comprehensive promotion of distribution network construction, the demand of reliability for electricity is getting higher and higher. Therefore, the investment planning model for distribution network, which focuses on the overall load growth within the region, is not applicable. In this paper, a distribution network planning model that considering different customers’ reliability requirements is proposed. Firstly, the fuzzy binary contrast strategy is adopted to quantify the reliability requirements of different customers in the region. The neural network is used to mine the correlation model between reliability indices of the different power supply requirements and corresponding reconstruction measures. Then, the optimal model of distribution network planning with the target of reliability indices and the constraint of investment cost is established. As a result, the proposed model fully meets the reliability requirements of customers and forms the optimal investment planning scheme of the distribution network. Finally, the results of the IEEE 33 test system verify the rationality and feasibility of the proposed model.