Abstract: The fluctuation of photovoltaic power generation and load randomness lead to the change of power quality evaluation indexes.The fixed weight comprehensive evaluation index and single objective optimization are usually used by the multifunctional gridconnected inverter,which may cause the deterioration of power quality control effect and the low utilization rate of residual capacity.Based on G1 method improved by mixed variable weight principle,a new comprehensive power quality index evaluation method is proposed. According to the dynamic changes of source and load,the weight of each power index is adjusted in real time to realize the objective evaluation of power quality. On this basis,the optimal comprehensive power quality index and the minimum compensation capacity of the inverter are taken as the objective functions,and the individual indexes are constrained. The nondominated sorting genetic algorithm Ⅱ（NSGA-Ⅱ）is used to realize the multi objective optimization. Finally,the feasibility and effectiveness of the proposed strategy are verified in Matlab.