The present invention discloses an optimization scheduling method based on mixing power supply multi-time scale complementation. The method comprises: 1, bundling a wind turbine generator, a photovoltaic power station and a hydroelectric generating set into a VP for combined operation, defining a load tracking index, and allowing a load curve deducting the VP output to be an optimal load curve; 2, establishing a day-ahead 24-hour planning model comprising a VP optimal scheduling layer and a thermal power optimal scheduling layer, and taking the minimum of the load tracking index and the thermal power generation cost as target functions; 3, establishing an intra-day 4-hour rolling planning model, taking the invariance of the optimal load curve in this period and the minimum of the thermal power adjustment cost as targets, and setting wind-abandoning and photovoltaic-abandoning constraint of progressive correction; and 4, establishing a real-time 15-mit planning model, and taking the invariance of the optimal load curve in the last moment and the minimum of the thermal power adjustment cost, and setting the wind-abandoning and photovoltaic-abandoning constraint of the progressive correction. The optimization scheduling method based on the mixing power supply multi-time scale complementation realizes the maximization of regulation benefit, environmental protection benefit and economic benefit, and provides valuable solution schemes for a plurality of new energy grid-connected problems.