抗辐射的三冗余高可靠星载计算机系统的研究

发布时间：2018-03-20 01:20

本文选题：星载计算机　切入点：三冗余　出处：《复旦大学》2014年硕士论文　论文类型：学位论文

【摘要】：星载嵌入式计算机长期工作在宇宙空间辐射环境中,各种辐射效应是导致星载计算机在运行期间发生故障的主要原因。各类容错技术,如冗余加固技术、机内故障自诊断技术、系统健康状态管理技术等纷纷被用于星载计算机中以提高系统的可靠性,对于保证整个卫星系统的长寿命准确稳定运行具有重要意义。本文针对星载嵌入式计算机遇到新的可靠性挑战,首先在传统的具有一活动节点一冷备份的两冗余卫星系统的基础上,改进并提出了一种具有三个硬件上完全相同的冗余运算硬件结构的星载计算机设计。这三个硬件上完全相同的运算节点的主次并不固定,而是通过故障检测模块选择出错误概率最低的节点作为主节点。并将相对错误概率高的节点进行冷藏,使其作为冷备份节点。这种三冗余结构与传统的两冗余冷备份系统相比具备更高的系统可靠性。同时,定期自主选举表决的策略也增强了系统稳定性。论文也开发了一套具有一定故障预测能力的故障检测系统,由故障标尺和基于故障树的故障检测系统组成。因为场效应管(metal oxide semiconductor, MOS)器件的阈值在接受了一定量的辐射照射后会发生漂移,且阈值漂移的幅度与接受的辐射总剂量正相关,所以论文利用这一特性,通过测量MOS器件的阈值电压漂移程度,来预测整系统接受到的辐射总剂量,并将辐射总剂量转化为对故障概率的预测。同时,论文使用基于故障树的故障检测方案替代了传统的基于专家系统的故障检测方案。相比传统方案而言,基于故障树的故障检测新方案可以更好地利用原有系统各模块之间的内在逻辑联系,利用故障树推理得到的系统故障概率较使用传统专家系统的方案具有更高的判决正确率。论文中也使用Matlab仿真软件比较了这两种故障检测方法的检测性能,仿真结果表明随着系统逻辑复杂度的增加,新方案具有更高的故障检测率。论文还针对目前星载计算机系统随机存取存储器(random access memory, RAM)模块中发生频率较高的单粒子多位翻转问题,提出采用具有更高纠检错能力的线性循环码来替换传统的汉明码,并在现场可编程门阵列(Field Programmable Gate Array, FPGA)器件中进行编解码算法实现,充分利用了FPGA器件丰富的存储资源,完成编解码算法的快速工程实现。
[Abstract]:The space-borne embedded computer works in the space radiation environment for a long time, and various radiation effects are the main causes leading to the failure of the spaceborne computer during operation. All kinds of fault-tolerant techniques, such as redundant reinforcement technology, in-machine fault self-diagnosis technology, System health condition management technology has been used in spaceborne computer to improve the reliability of the system. It is of great significance to ensure the accurate and stable operation of the whole satellite system. In this paper, a new reliability challenge is encountered for the spaceborne embedded computer. First of all, based on the traditional two redundant satellite systems with an active node and cold backup, In this paper, a design of spaceborne computer with three identical redundant computing hardware structures is proposed, in which the primary and secondary functions of the same operation nodes are not fixed. Instead, the node with the lowest error probability is selected as the master node by the fault detection module, and the node with relatively high error probability is refrigerated. This three-redundancy structure is more reliable than the traditional two-redundancy cold backup system. The strategy of periodic independent voting also enhances the stability of the system. A fault detection system with certain ability of fault prediction is also developed in this paper. It is composed of fault scale and fault detection system based on fault tree. Because the threshold of metal oxide semiconductors (moss) devices will drift after a certain amount of radiation exposure, and the amplitude of threshold drift is positively related to the total radiation dose. Therefore, by measuring the threshold voltage drift of MOS devices, the total radiation dose received by the whole system is predicted, and the total radiation dose is converted into the prediction of failure probability. In this paper, the fault detection scheme based on fault tree is used to replace the traditional fault detection scheme based on expert system. The new fault detection scheme based on fault tree can make better use of the inherent logical relationship between the modules of the original system. The fault probability obtained by fault tree reasoning is higher than that of traditional expert system. In this paper, Matlab simulation software is also used to compare the detection performance of the two fault detection methods. Simulation results show that with the increase of system logic complexity, The new scheme has higher fault detection rate. This paper also aims at the problem of high frequency single particle multibit flipping in random access memory (RAM) modules of spaceborne computer systems. A linear cyclic code with higher error-correcting ability is proposed to replace the traditional hamming code, and the coding and decoding algorithm is implemented in Field Programmable Gate array (FPGA) devices, which makes full use of the abundant storage resources of FPGA devices. Complete the fast engineering implementation of coding and decoding algorithm.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP302.7

【参考文献】