Yin Xiuxian, Xu Bing, Hu Laihong, Li Hongyu, He Wei
Harbin Normal University, Harbin, 150025, China.
Rocket Force University of Engineering, Xi'an, 710025, China.
Sci Rep. 2023 Aug 28;13(1):14066. doi: 10.1038/s41598-023-41305-z.
Health state assessment is an important measure to maintain the safety of aerospace relays. Due to the uncertainty within the relay system, the accuracy of the model assessment is challenged. In addition, the opaqueness of the process and incomprehensibility of the results tend to lose trust in the model, especially in high security fields, so it is crucial to maintain the interpretability of the model. Thus, this paper proposes a new interpretable belief rule base model with step-length convergence strategy (IBRB-Sc) for aerospace relay health state assessment. First, general interpretability criteria for BRB are considered, and strategies for maintaining model interpretability are designed. Second, the evidential reasoning (ER) method is used as the inference machine. Then, optimization is performed based on the Interpretable Projection Covariance Matrix Adaptive Evolution Strategy (IP-CMA-ES). Finally, the validity of the model is verified using the JRC-7M aerospace relay as a case study. Comparative experiments show that the proposed model maintains high accuracy and achieves advantages in interpretability.
健康状态评估是维护航天继电器安全性的一项重要措施。由于继电器系统内部存在不确定性，模型评估的准确性受到挑战。此外，过程的不透明性和结果的不可理解性往往会导致对模型失去信任，尤其是在高安全领域，因此保持模型的可解释性至关重要。因此，本文提出了一种用于航天继电器健康状态评估的具有步长收敛策略的新型可解释信念规则库模型（IBRB-Sc）。首先，考虑了BRB的一般可解释性标准，并设计了保持模型可解释性的策略。其次，使用证据推理（ER）方法作为推理机。然后，基于可解释投影协方差矩阵自适应进化策略（IP-CMA-ES）进行优化。最后，以JRC-7M航天继电器为例验证了该模型的有效性。对比实验表明，所提出的模型保持了较高的准确性，并在可解释性方面具有优势。
