IEEE Trans Cybern. 2022 May;52(5):4027-4038. doi: 10.1109/TCYB.2020.3015664. Epub 2022 May 19.
In current studies of safety assessment for complex systems with the evidential reasoning (ER) rule, the evidence reliability is generally given by experts, which makes the observation data by sensors ignored. However, sensors are inevitably affected by such various uncertainties as perturbations in engineering practice, which can reduce their quality and tracking ability. As such, the observation data may become unreliable, and the modeling accuracy of the ER rule is decreased. In this article, a new ER rule-based safety assessment method with sensor reliability for complex systems is proposed, where sensor reliability and perturbation are considered. The coefficient of the variation-based weighting (CVBW) method is employed to obtain sensor weight. The sensor reliability is calculated by static reliability and dynamic reliability, which are determined by experts and the distance-based method, respectively. The perturbation is quantified as a bounded parameter defined as the perturbation factor, which is used to describe uncertainties and aggregate static reliability and dynamic reliability. The performance analysis of safety assessment is conducted to demonstrate the rationality of perturbation and position poor sensors, followed by a safety assessment algorithm. A case study is carried out to validate the effectiveness of the proposed method.
在使用证据推理（ER）规则对复杂系统进行安全评估的当前研究中，证据可靠性通常由专家给出，这忽略了传感器的观测数据。然而，传感器在工程实践中不可避免地会受到各种不确定性的影响，如干扰，这会降低它们的质量和跟踪能力。因此，观测数据可能变得不可靠，并且 ER 规则的建模准确性会降低。本文提出了一种新的基于 ER 规则的复杂系统传感器可靠性安全评估方法，其中考虑了传感器可靠性和干扰。基于变化系数的加权（CVBW）方法用于获得传感器权重。传感器可靠性由专家和基于距离的方法分别确定的静态可靠性和动态可靠性来计算。干扰被量化为一个有界参数，定义为干扰因子，用于描述不确定性并聚合静态可靠性和动态可靠性。进行安全评估性能分析以验证干扰和位置不良传感器的合理性，然后提出安全评估算法。进行了案例研究以验证所提出方法的有效性。
