基于BiSNet的航天发动机转子智能健康监控

航天发动机某部件转子健康监控,对检测发动机能否正常工作进而影响航天发射任务的成败至关重要。传统健康监控方法需要依赖经验和专业知识,在对故障机理深入理解后才能构建相应的健康监控模型。针对传统转子健康监控手段效率低、适应性差、过度依赖专家经验等不足,结合海量转子振动健康监控历史数据,提出并设计采用神经网络BiSNet的人工智能方法,对航天领域运载火箭发动机某部件转子的健康状态进行智能建模并完成监控。通过与传统监控手段和主流基准神经网络方式对照实验得出结论,基于历史数据驱动的神经网络BiSNet可为航天转子智能健康监控提供便捷准确的建模预测。

Intelligent health monitoring of aerospace engine rotor based on BiSNet

Health state monitoring of certain rotor portion in the aerospace engine is critical to the aerospace engine working normally,which effects launch task success further.Traditionally,health state monitoring depends on subjective experience and professional knowledge,so the corresponding health monitoring model is built after studying and understanding the failure mechanism behind.In view of the low efficiency,poor adaptability and over-reliance on expert experience of traditional rotor health state monitoring means,combined with massive amounts of rotor vibration health state monitoring historical data,neural network BiSNet is proposed and designed as an artificial intelligent method to model the certain rotor portion in the aerospace engine health monitoring model and monitor the health status timely.Compared with the traditional monitoring method and the mainly general baseline intelligent method,the conclusion is drawn that the historical data-driven intelligent neural network BiSNet could provide convenient and accurate modeling and prediction for the health monitoring of aerospace rotors.