为了提高仿真实验系统的主控计算机分系统的软件可靠性, 建立了系统级软件可靠性屋(HOSR)。以主控计算机分系统下的运行时间框架管理模块为例, 分析了故障模式和故障原因的相关性、故障模式间的相关性、和故障原因间的相关性。介绍了故障模式及影响分析法(FMEA)、质量功能展开法(QFD)、及将QFD中的系统屋概念融入FMEA而形成的HOSR。根据主控计算机分系统软件层次图, 利用已建立的故障模式和故障原因库对HOSR进行填充; 最后, 根据HOSR的分析矩阵对RTI管理模块进行故障分析。分析结果表明: RTI管理模块中存在潜在故障模式23个, 潜在故障原因35个, 故障模式和故障原因的相关性关联116个, 故障模式间的相关性关联9个, 故障原因的相关性关联14个。该分析法能更全面地识别故障模式和故障原因, 尤其是共模故障和共因故障, 同时减少头脑风暴法带来的缺陷遗漏, 减少工作人员的工作量。

To improve the software reliability of the host computer subsystem in a simulation experimental system, a system-level House of Software Reliability (HOSR) was established. The Run-time Infrastructure(RTI) management module was taken as an example, the dependence of failure modes on the failure causes was analyzed and the relationships between the failure modes and the failure causes were disscussed. The Failure Mode and Effects Analysis (FMEA) and Quality Function Deployment (QFD) were introduced. And then, the HOSR formed by FMEA with the system concept house from the QFD was described. According to the software level diagram of the host computer subsystem, the established library of failure modes and failure causes was used to fill HOSR. Finally, the failure of RTI management module was analyzed on the HOSR analysis matrix. The analytical results indicate that, in RTI management module, there are 23 potential failure modes, 35 potential causes of failure, 116 relationships between failure modes and failure causes, 9 relationships between the failure modes, 14 relationships between the failure causes. analysis method can distinguish failure modes and failure causes much more comprehensive, especially for common-mode failures and common-cause failures. Meanwhile, it can significantly reduce the defect omission of brainstorming and staff workloads.