The work aims to propose a dynamic order insertion strategy and a time window assignment strategy to study the optimization problem of express logistics on-demand delivery considering dynamic new order demands, so as to solve problems of difficulty in providing timely services to dynamic customers and poor delivery timeliness in traditional express logistics on-demand delivery. Firstly, a bi-objective vehicle routing optimization model was constructed to minimize the logistics operation cost and the number of vehicles used, considering the periodic demand of the express logistics on-demand delivery network and the new order demand. Secondly, an improved multi-objective ant colony algorithm was designed to solve the optimization model. This algorithm enhanced the quality of Pareto optimal solutions through local optimization strategies and external archive update mechanisms. And a dynamic order insertion strategy and a time window assignment strategy was further proposed to improve the overall search performance of the algorithm. Thirdly, the improved multi-objective ant colony algorithm was compared and analyzed with the multi-objective particle swarm optimization algorithm, the multi-objective grey wolf algorithm, and the multi-objective multi-verse algorithm, verifying the effectiveness of the proposed algorithm. Finally, an instance optimization study was conducted based on a certain express logistics on-demand delivery network in Chongqing, and the impact of different service time period divisions on indicators such as logistics operation cost, and the number of vehicles used, and penalty cost were analyzed and discussed. The results showed that the logistics operation cost was reduced by 48% after optimization, and the number of vehicles used was reduced by 12. The optimization scheme that divided the service time of the distribution center into three time periods had the best effect. In conclusion, the model and algorithm proposed in this paper are helpful to reduce the logistics operation cost and the number of vehicles used, providing method support and decision-making reference for the optimization of express logistics on-demand delivery considering dynamic new order demands.