基于不确定推理的军用软件故障诊断方法研究

Study on the Method of Military Software Fault Diagnosis Based on Uncertainty Reasoning

摘要:

针对基于构件的军用软件使用过程中故障频发,传统规则推理的方法进行故障诊断存在不确定性问题,提出了基于不确定规则知识库的故障诊断方法；首先根据不确定推理逻辑,分析传统规则推理的不确定因素,构建不确定规则知识库；然后基于该知识库和正向链推理策略,计算确定性因子；合理设置激活规则的阈值,并对中间结论的确定性因子进行排序以消解冲突；实验结果表明,基于不确定推理的诊断方法与传统精确推理方法相比,具有更高的诊断准确率。

Abstract:

This paper focuses on the frequent faults of Component-based military software during the process of the software using, the uncertainty problems of traditional methods of rule-based reasoning for software fault diagnosis, and proposes the fault diagnosis method based on uncertain rule knowledge base. First, analyze the uncertain factors of traditional rule-based reasoning according to the logic of uncertain reasoning, and build the knowledge base of uncertain rules. Then, based on this knowledge base and the forward chain of reasoning strategies, calculate the uncertainty factor, legitimately set the threshold to activate rules, and sort the certainty factors of intermediate conclusions to eliminate conflicts. Experiment results show that, the military software fault diagnosis method based on uncertain reasoning has a higher accuracy rate compared with traditional accurate reasoning method. 