Abstract: To reduce the instability of photovoltaic generation it is necessary to formulate a power dispatching scheme to conform to the multi-objective requirements. For this purpose, based on the principle of low carbon and economy and leading the carbon trading mechanism into power system dispatching, a probability distribution model-based dispatching scheme, in which the solar irradiance and photovoltaic generation output were taken into account, was established. The K-means clustering method was applied to photovoltaic generation output to curtail the scenario of photovoltaic generation and improve the predictive accuracy, and on this basis a multi-objective optimal dispatching model after the integration of large-scale photovoltaic generation was constructed. The objective function of the constructed model included the cost of thermal power generation, the cost of carbon trading and the amount of pollution discharge, meanwhile the economy of the dispatching and the low-carbon environmental protection performance were also considered. The entropy weight method was utilized to give weights to different objectives and turned them into single objective functions. In the analysis on calculating example, four dispatching schemes with different optimization purposes are established and the qualitative and quantitative contrastive analyses are performed to verify whether the constructed dispatching model possessing good economy, environmental protection property and reliability.