Huang Jingde, Huang Zhangyu, Zhan Xin
Guangdong Intelligent Vision Precision Detection Engineering Technology Research Center, Zhuhai College of Science and Technology, Zhuhai, China.
Faculty of Innovation Engineering, Macau University of Science and Technology, Macau, China.
PeerJ Comput Sci. 2023 Jul 27;9:e1439. doi: 10.7717/peerj-cs.1439. eCollection 2023.
A high reliability system has the characteristics of complexity, modularization, high cost and small sample size. Throughout the entire lifecycle of system development, storage and use, the high reliability requirements and the risk analysis form a direct contradiction with the testing expenses. In order to ensure the system, module or component maintains good reliability status and effectively reduces the cost of sampling tests, it is necessary to make full use of multi-source prior information to evaluate its reliability. Therefore, in order to evaluate the reliability of highly reliable equipment under the condition of a small sample size correctly, the equipment reliability evaluation model should be built based on multi-source prior information and form scientific computing methods to meet the needs of condition evaluation and fund assurance of high reliability system. In engineering practice, high reliability system or module gradually develops from normal state to failure state, generally going through three working states of "safety-potential failure-functional failure". Firstly, the historical test data under the three states can be used for the data source for the reliability evaluation of the system at the current stage, which supplements the deficiency of the field data; secondly, due to the lack of accurate judgment on the working state of a high reliability system or modules and analysis of the health status, the unnecessary maintenance may aggravate the evolution speed from potential failure to functional failure; thirdly, when high reliability system or module operates under overload or harsh conditions, the potential failure will be worsened to a certain extent. Aiming at the difficulty of multi-state system reliability evaluation, a reliability evaluation method based on non-information prior distribution is proposed by fusing multi-source prior information, which provides ideas and methods for reliability evaluation and optimization analysis of high reliability system or module. The results show that the three-state reliability evaluation method proposed in this article is consistent with the actual engineering situation, providing a scientific theoretical basis for preventive maintenance of high reliability system. At the same time, the research method not only helps evaluate the reliability state of a high reliability system accurately, but also achieves the goal of effectively reducing test costs with good economic benefits and engineering application value.
高可靠性系统具有复杂性、模块化、高成本和小样本量的特点。在系统开发、存储和使用的整个生命周期中，高可靠性要求与风险分析同测试费用形成直接矛盾。为确保系统、模块或组件保持良好的可靠性状态并有效降低抽样测试成本，有必要充分利用多源先验信息来评估其可靠性。因此，为正确评估小样本量条件下高可靠性设备的可靠性，应基于多源先验信息构建设备可靠性评估模型，并形成科学的计算方法，以满足高可靠性系统状态评估和资金保障的需求。在工程实践中，高可靠性系统或模块逐渐从正常状态发展到故障状态，一般会经历“安全 - 潜在故障 - 功能故障”三个工作状态。首先，这三种状态下的历史测试数据可作为当前阶段系统可靠性评估的数据源，弥补现场数据的不足；其次，由于对高可靠性系统或模块的工作状态缺乏准确判断以及健康状态分析，不必要的维护可能会加剧从潜在故障到功能故障的演变速度；第三，当高可靠性系统或模块在过载或恶劣条件下运行时，潜在故障会在一定程度上恶化。针对多状态系统可靠性评估的难点，通过融合多源先验信息提出了一种基于非信息先验分布的可靠性评估方法，为高可靠性系统或模块的可靠性评估及优化分析提供了思路和方法。结果表明，本文提出的三状态可靠性评估方法与实际工程情况相符，为高可靠性系统的预防性维护提供了科学的理论依据。同时，该研究方法不仅有助于准确评估高可靠性系统的可靠性状态，还实现了有效降低测试成本的目标，具有良好的经济效益和工程应用价值。
