[Objective] Solve the problems of poor control accuracy and adaptability in the Delta robot flexible gripper grasping method. [Methods] This article took the Delta robot flexible gripper as the research object and proposed a Delta robot flexible gripper grasping method that combined multiple models and improved whale algorithm. Established a stable grasping optimization model to seek the optimal contact position on the surface of the grasping object. Established a non-destructive grasping optimization model to minimize contact force while ensuring stable grasping of objects. Combine particle swarm optimization algorithm and whale algorithm to solve the model. The superiority of the proposed grasping method had been verified through experiments. [Results] The proposed method not only had good control accuracy, but also could adapt to objects of different shapes and sizes, with high flexibility and adaptability, success rate of grasping was greater than 96%, grasping damage rate was 0. [Conclusion] The proposed method effectively improves the performance of the Delta robot flexible gripper gripping method and is suitable for sorting fruits, vegetables, and fragile items.