摘要
综合模块化航空电子系统(integrated modular avionics, IMA)因其灵活性、易更改、高容错等特点广泛应用于航空领域。然而,IMA面临多变的环境和频繁变更的需求,现有的人工配置重构蓝图和传统算法生成重构蓝图方法存在自动化程度低、质量难以保证等问题,难以满足IMA任务切换时对资源调度的复杂度和难度不断提升的需求。针对IMA的多任务重构蓝图生成问题,提出了基于多步学习和网络噪声的DDQN-MS-NN重构算法,提高了资源调度的自动化水平和质量,从而提升IMA的稳定性和抗风险性。
The integrated modular avionics system is widely used in the aviation field due to its flexibility,ease of modification,and high fault tolerance.However,the system faces changing environments and evolving multi-tasking requirements.Existing manual configuration methods and traditional algorithms for generating reconstruction blueprints have limitations in terms of automation and quality assurance.They struggle to meet the increasing complexity and difficulty of resource scheduling during task switching in the comprehensive electronic system.In this paper,the DDQN-MS-NN reconstruction algorithm is proposed to address these challenges.The algorithm focuses on generating multi-tasking reconstruction blueprints for the comprehensive electronic system,improving the automation level and quality of resource scheduling,by introducing multi-step learning and noise network mechanism.Experimental results show that,the DDQN-MS-NN reconstruction algorithm can enhance system stability and resilience compared with traditional algorithms.
作者
李国栋
张翰谷
单鹏
王小辉
周炳林
赵岸荣
LI Guodong;ZHANG Hangu;SHAN Peng;WANG Xiaohui;ZHOU Binglin;ZHAO Anrong(Northwestern Polytechnical University,Xi'an 710072,China;The First Aircraft Institute,Xi'an 710089,China;Aeronautics Computing Technology Research Institute,Xi'an 710068,China;China Academy of Launch Vehicle Technology,Beijing 100076,China)
出处
《西北工业大学学报》
北大核心
2025年第4期821-830,共10页
Journal of Northwestern Polytechnical University
关键词
IMA
蓝图重构
强化学习
integrated modular avionics
task reconstruction
reinforcement learning
作者简介
李国栋(1982-),正高级工程师;通信作者:李国栋(1982-),e-mail:guodonglee@163.com。
