The access of large-scale distributed photovoltaic energy seriously increases the harmonic pollution of distribution networks. The effect of harmonic control is related to the network topology and other nearby nodes "hitchhike" phenomena exist. To improve the consumption efficiency of users and enhance the economic benefits of the distribution network, a two-layer optimization model of value-added service considering power quality is proposed. First, the support vector machine algorithm is used to predict the harmonic voltage distortion rate of load nodes in different output conditions of harmonic sources. Then, a two-layer optimization model of purchasing and selling electricity between the distribution network and demand users is established. The upper layer, which considers the influence of the network topology structure on the power quality governance effect and cost, takes the maximum of distribution network operator profit as the target. It takes the distribution network operator as the main body, builds the harmonic voltage control model and establishes the electricity price for different power quality levels. The lower layer, which is based on the price of a different power quality, takes the users aim to minimize the operational cost as the target. It takes the users with differentiated power quality requirements as the main body and establishes the power quality purchasing strategy model. Finally, this paper uses the IEEE14-node example analysis for verification, and the results show that the proposed model can improve the power quality and the economic benefits of the distribution network, as well as providing a solution for the differentiated consumption of users.