数据驱动的红外导引头伺服预测仿真验证平台

Data-Driven Predictive Simulation and Verification Platform for Infrared Seeker Servo Systems

摘要:

针对红外导引头伺服系统在“光-机-电-算”耦合退化下难以获得可重复全寿命数据、传统仿真与标定难以支撑寿命预测的问题，构建了基于物理同构映射的半实物预测评估仿真验证平台。平台复现感知-控制-执行闭环信号链路，实现多域关键参数的毫秒级同步采集与统一数据封装；基于多维特征工程，采用GMM连续健康评分与HSMM退化阶段识别的混合诊断框架，并结合残差约束实现RUL预测。加速老化与分级故障注入实验复现8类演化故障，系统级预测准确率90.45%，相对误差稳定小于15%。结果可为视情维护与全寿命周期管理提供高置信度验证载体。

Abstract:

To address the lack of repeatable full-life degradation data and the limited capability of conventional simulation and calibration for infrared seeker servo systems under coupled opto-mechanical-electrical-computational effects, a hardware-in-the-loop simulation and verification platform based on physical isomorphism mapping is developed. The platform reconstructs the sensing–control–actuation closed-loop signal chain and enables millisecond-level synchronized acquisition and unified packaging of multi-domain parameters. A hybrid diagnostic scheme is implemented by combining GMM-based continuous health scoring with HSMM-based degradation-stage recognition, together with residual-constrained remaining useful life (RUL) prediction. Accelerated aging tests with graded fault injection reproduce eight evolving fault patterns; the overall prediction accuracy reaches 90.45%, with relative errors stably within 15%. The platform provides a high-confidence testbed for condition-based maintenance and life-cycle management of electro-optical equipment.