Abstract: The uncertainty in photovoltaic (PV) power generation output makes forecasting challenging, resulting in significant errors that affect the dispatch decision-making process and lead to a relatively high curtailment rate in dispatch outcomes. To mitigate this issue, this study proposes an optimized operational dispatch method for PV power generation systems based on an improved NSGA-II algorithm. First,the probability density distribution function is employed to analyze irradiation conditions. Then, based on the PV operational mechanism, the corresponding generation current is calculated, and a multiple linear regression method is used to predict short-term PV power generation output. After eliminating error effects with the couple function, an improved NSGA-II algorithm with adjusted crossover and mutation rates is applied to achieve multi-objective optimization for economic and efficiency considerations, yielding the optimal PV power generation system operational dispatch scheme. Experimental results demonstrate that this method can reduce the curtailment rate to 0.62%, achieving superior dispatch performance and significantly improving system operational quality.