Abstract: In order to solve the problem of weight calculation of multi-dimensional indicators in power quality evaluation of distributed photovoltaic system, power quality indicators are classified based on national standards, and improved genetic projection tracing method and improved analytic hierarchy process are proposed to determine the objective weight and subjective weight of power quality evaluation model, respectively. By combining subjective and objective weights, the final comprehensive weight is obtained. Then, the boundary value calculation is used to divide the grade interval, and the power quality assessment is carried out on the five groups of power data collected from the distributed photovoltaic system. The results show that the power quality management level of distributed photovoltaic system in Hebei Key Laboratory of Power Machinery Equipment Health Maintenance and Failure Prevention is good; High system stability; Compared with the power quality comprehensive evaluation model based on entropy method and improved analytic hierarchy process (AHP), the power quality comprehensive evaluation model based on improved genetic projection tracing method and improved analytic hierarchy process (AHP) is more scientific and practical. The research results are of great significance for maintaining the stability of power grid, protecting power equipment and improving the overall safety of power grid.