In order to solve the problem of instability and high cost of photovoltaic power generation, the optimal scheduling model of a photovoltaic power system was established by minimizing power generation scheduling cost. The model increases the ability of load controlling by virtue of the demand response of price to reduce the influence of the instability of photovoltaic generation on optimal scheduling results. For solving the model, the classical genetic algorithm was improved. The simulation results show that it is beneficial to consider demand response, which can cut peaks and valleys of the load and improve the stability of photovoltaic power generating units. Furthermore, it will help to reduce the generation scheduling cost of the photovoltaic power system. The results also show the rightness and effectiveness of the model.