The new power system and dual carbon goals have driven the development of intelligent scheduling methods for active distribution networks. A new intelligent scheduling method for active distribution networks based on day before day to day scheduling is proposed to address the issues of poor economy and flexibility in existing intelligent scheduling methods. The current distribution network scheduling model adopts a double-layer structure, with the upper layer being constructed based on the comprehensive optimization of costs such as main network electricity purchase, network loss, energy storage, operation and maintenance, carbon emissions, and peak valley penalties. The lower level adopts a comprehensive optimal construction of peak valley fines, carbon emissions, and electric vehicle charging costs. The intraday distribution network optimization scheduling model is constructed by comprehensively optimizing the upper level model objectives and interruptible load compensation costs. Solve the model through an improved NSGA-III algorithm. Verify the superiority of the proposed method through numerical examples. The results indicate that, proposed method takes into account economy, low-carbon, and flexibility. Through joint scheduling in the past and within the day, the problem of photovoltaic output fluctuations has been effectively solved, and the operational performance of the distribution network has been optimized. The total solution time is 412.11 seconds, the total cost is 50665.60 yuan, and the carbon emission cost is 1583.50 yuan. It can provide certain assistance for achieving the dual carbon goal.