<p>Mobile edge computing (MEC) decentralizes some of the task processing and data management to edge nodes nearby user devices, and thus can provide low latency services. However, the high mobility of user devices may disrupt communications between users and edge nodes, which can lead to a failed task offloading. Therefore, we study the joint optimization of task scheduling and edge caching to improve the service quality for mobile users. First, we propose a selection method to decide which edge node to be assigned to each task for its offloading and result return. Then, considering the cooperation among edge nodes, we formulate the task scheduling problem with task migration as an integer nonlinear programming problem, taking into account the device mobility and the edge caching mechanism. And to solve the problem within a polynomial time complexity, we propose a hybrid algorithm combining the advantages of particle swarm optimization and genetic algorithm. In the proposed hybrid algorithm, particles have a better chance of moving to better or even best positions by exploiting the crossover operator and an improved mutation operator that retains good genes with a high probability. The extensive experiment results show that our method improves the task completion rate by 16%-34% and achieves a higher cache hit rate and a better load balance, compared to seven classical and state-of-the-art algorithms including two up-to-date hybrid intelligent algorithms, Immune Particle Swarm Optimization Algorithm and the hybrid of Manta Ray Foraging Optimization and Salp Swarm Algorithm.</p>