基于数字孪生的红外光学设备数据模型混合驱动故障预测

Hybrid-driven Fault Prediction of Infrared Optical Equipment Data Model Based on Digital Twin

摘要:

传统的数据驱动预测建模理论支撑不足，导致红外导引头设备预测建模中存在传统的建模方法置信度精度不高、预测算法工程化水平不足等问题。基于数据驱动和模型融合的故障预测技术为解决上述问题提供了重要技术途径。本研究建立了红外导引头设备光电探测系统性能退化预测的数字孪生框架，根据物理实体的工作机理将其抽象为能量域上的物理模型，结合物理模型和历史检测数据进行仿真分析得到系统的初始性能退化模型，利用实时监测数据建立性能退化的时序贝叶斯网络图（TBN）演化模型以实现与物理实体之间的同步演化，通过无迹卡尔曼滤波（UKF）算法实现了对物理实体的状态估计和剩余使用寿命预测并进行了仿真验证，克服了单一的数据驱动和基于模型预测建模方法存在的适应面较窄等不足，并较典型数据驱动方法提升100倍量级的精度，为开展红外导引头设备预测评估研究提供了技术基础。

Abstract:

The traditional data-driven prediction modeling theory lacks sufficient support, resulting in problems such as low confidence accuracy of traditional modeling methods and insufficient engineering level of prediction algorithms in the prediction modeling of infrared guidance head equipment. The fault prediction technology based on data-driven and model fusion provides an important technical approach to solve these problems. This study established a digital twin framework for the performance degradation prediction of the optical-electrical detection system of infrared guidance head equipment. According to the working mechanism of the physical entity, it abstracts it into a physical model in the energy domain. Combining the physical model with historical detection data, simulation analysis is conducted to obtain the initial performance degradation model of the system. The Temporal Bayesian network diagram (TBN) evolution model for performance degradation is established using real-time monitoring data to achieve synchronous evolution with the physical entity. The state estimation and remaining useful life prediction of the physical entity are realized using the unscented Kalman filter (UKF) algorithm and verified through simulation. It overcomes the shortcomings of narrow adaptability of single data-driven and model-based prediction modeling methods and improves the accuracy by 100 times compared to the data-driven method, providing a technical foundation for the research on prediction and evaluation of infrared guidance head equipment.