基于Z语言和贝叶斯网络的存储系统可靠性评估方法

现有的存储系统可靠性评估方法难以动态地描述其可靠性特征，也难以反映其与时间的关系；针对此问题，提出了一种结合Z语言、连续时间马尔科夫链以及贝叶斯网络的可靠性评估方法；该方法将存储系统抽象为两级，分为功能单元和存储系统整体两部分；在针对功能单元进行可靠性评估时，建立了一种基于Z语言的可靠性模型，由于其关于时间的状态转移符合Markov链的性质，因此引入连续时间马尔科夫链进行评估；为了更清晰地表达存储系统与各个功能单元之间的多状态关系，构建了功能单元失效率与存储系统整体失效率之间的贝叶斯网络；基于贝叶斯网络计算出存储系统的整体可靠性，并得到其可靠性关键模块；最后，通过实例分析与计算表明，基于提出的可靠性评估方法能够准确描述存储系统的状态转移，可以量化计算存储系统的可靠性，得到其可靠度，具有一定的实用价值。

Memory System Reliability Evaluation Model Based on Z and Bayesian Network

The existing reliability evaluation methods of storage systems are difficult to describe its reliability characteristics dynamically, and it is difficult to reflect its relationship with time. To solve this problem, a method combining Z language, continuous-time Markov chain and Bayesian network is proposed. This method abstracts the storage system into two levels, which are divided into two parts, functional unit and storage system as a whole. When evaluating the reliability of functional units, a reliability model based on Z language is established. Since its state transition with respect to time conforms to the properties of the Markov chain, a continuous-time Markov chain is introduced for evaluation. In order to more clearly express the multi-state relationship between the storage system and each functional unit, the functional unit failure rate and The Bayesian network between the overall failure rates of the storage system. The overall reliability of the storage system is calculated based on the Bayesian network, and the key reliability modules are obtained. Finally, through the example analysis and calculation, it is shown that the proposed reliability evaluation method can accurately describe the state transition of the storage system, quantify the reliability of the storage system, and obtain its reliability, which has certain practical value.