It is defined in the ARINC653-2 specification that the Integrated Modular Avionics （IMA） real-time system adopts the two-level scheduling scheme of partitions and tasks. The scheduling analysis of tasks based on this scheduling algorithm is a key issue to ensure the real time and reliability of the avionics system. Current schedulability analysis methods cannot be adapted to the multi-window partition situation. Therefore， based on the encapsulation of service capabilities of the network calculus model， this paper defines the processing platform service curve as the service function of the computing resources that the platform can provide， and defines the partition task arrival curve as the demand function of partition tasks for platform computing resources. By analyzing the service curve of the processing platform and the arrival curve of partition tasks， the Worst-Case Respone Time （WCRT） of the task is calculated， and then the schedulability is judged. Thus， the physical significance of interpreting the maximum response time of partitioned system from the perspectives of service capability and service demand is developed. A validation case of partition containing multiple activation windows is designed under the main time frame. The result shows that the proposed method can obtain the same accuracy as the traditional WCRT analysis method. It can also accurately calculate the maximum response time of the task when partition contains multiple activation windows. The schedulability of the partition system is explained from the perspective of network calculation.