Demand responsive transit typically operates in chartered or ride-sharing modes. Due to the time-varying road information， using a single operation mode will reduce the operational efficiency. To maximize the use of vehicle resources， a flexible scheduling and fleet configuration method under limited bus resources and multiple operation modes is proposed. First， based on the characteristics of chartered and ride-sharing modes， combined with the new point-to-point ride-sharing mode and heterogeneous fleet characteristics， the above three modes are taken into account as a whole. By designing a double-layer time window and aiming to maximize the operational benefits， a multi-vehicle-type mixed operation scheduling model is constructed. Secondly， the heuristic algorithm of adaptive large neighborhood search algorithm is improved. By optimizing neighborhood search operations， the solution quality is enhanced. Finally， simulation experiments based on real road network data and travel data from Wuhan City are conducted. The results show that， compared with the chartered and ride-sharing models， when the fleet size reaches 100 vehicles， the multi-vehicle-type mixed operation model increases the operational revenue by 9% and 7%， improves the order response rates by 10% and 6%， and reduces the average passenger wait time by 30% and 25%， respectively. Overall， the model enhances both the operator’s and passengers’ benefits.