一种临近空间艇载计算机故障检测与系统重构方法

摘要

Near space airship needs computer systems with fault tolerance for the requirment of high reliability. Based on the redundancy hardware of the computer used in near space airship, a kind of software management strategy is proposed. The data bus node fault detection method based on handshake mechanism, the CPU fault detection method based on watchdog and heartbeat information, the data bus reconstruction method based on node health matrix, and the adaptive reconstruction method of multi-CPU parallel processing system based on finite state machine are raised. The fault injection experiments show that, when a fault happens, the airship's redundant computer can detect and diagnose the fault real-timely, and it can realize the system reconstruction through data bus and CPU selection. The proposed fault detection and reconstruction methods can improve the reliability of airship's computer system and ensure the flight safety of airship.%平流层飞艇是可靠性要求很高的系统,需要由具有容错能力的艇载计算机来进行控制和管理.针对艇载计算机采用的余度结构,进行软件管理策略的研究和设计,提出基于异构总线的握手机制节点故障检测方法、基于"看门狗"与"心跳"相结合的CPU故障检测方法、基于节点健康矩阵的互援式总线重构方法及基于有限状态机的多CPU并行处理系统自适应重构方法.在故障注入试验中,艇载计算机在遇到故障时能够实时检测出故障,诊断故障类型,并对故障进行处理,实现系统重构,保证了平流层飞艇长期驻空时的安全飞行.