The vehicle routing problem of demand splitting under multiple time windows was studied, and the simultaneous delivery under multiple time windows was considered, and a multi-objective optimization model of related problems was constructed. It combined the two splitting methods in demand splitting according to the order and according to the type of goods, the unitized splitting method was used, and the calculation example was constructed based on the Solomon standard calculation example. The improved ant colony algorithm was used for testing, and compared the effects of the improved algorithm with other algorithms, and then compared the effects of the unitized splitting based on the order splitting and the splitting based on the product type on the basis of the classification cost. The results show that the improved ant colony algorithm is effective and unitized splitting is more cost-effective on some data sets.