With the continuous development of intelligent applications, large model technology has broken through with the increase in computing power, leading to an explosive growth in users of intelligent applications. The demand for resources such as computing power and storage is increasing. The huge demand for computing power and storage makes it impossible to deploy intelligent applications on a single computing node. Therefore, it is necessary to decompose the tasks of intelligent applications, divide the tasks of intelligent applications into multiple sub-tasks, and deploy them to multiple computing nodes for parallel execution. However, current research lacks studies on the deployment of components in multi-instance scenarios. Therefore, this paper proposes an intelligent application task decomposition algorithm based on DDQN for component multi-instance deployment. It adjusts the allocation of computing pow-er between instances after task decomposition to shorten the total task execution time. Simulation results show that the task decomposition algorithm proposed in this paper has good performance in terms of total task execution time, algorithm execution time, and computing power utilization.