Along with the increase of a massive number of power electronic equipment, the problem of harmonics in the distribution network is becoming increasingly serious. The accuracy of harmonic state estimation directly affects the subsequent harmonic control effect. The phasor measurement unit (PMU) can measure the node voltage and branch current in real time, and can be used to estimate the harmonic state. However, at present, the price of PMU is truly high. How to carry out reasonable optimization placement to ensure that the harmonic state of the whole distribution network is considerable, while reduce errors of harmonic state estimation, is an urgent problem to be solved. Firstly, a PMU optimal configuration model with the highest accuracy of PMU economic configuration and harmonic state estimation accuracy was constructed, and an improved binary particle swarm-genetic mixing algorithm was proposed for solving. Then, the IEEE14 node model was built in the real-time simulator. The mean interpolation method and Vondrak filter method were used for data processing and the influence of optimized PMU configuration scenarios on harmonic state estimation was analyzed. The results show that the proposed algorithm can give a reasonable PMU configuration scheme from the perspective of reducing investment cost and reducing harmonic state estimation error, which is helpful to support engineering decision-making.