基于试飞数据的航空发动机加力瞬态过程模型辨识

Identification of Aero-engine Afterburner Transient Process Based on Flight Test Data

为了对某航空发动机慢车至最大以及最大至慢车加力瞬态过程的工作状态进行监控,在该发动机实际飞行试验数据基础上,基于三层前向人工神经网络,辨识得到了该型发动机加力瞬态过程模型,对模型预测精度进行了分析讨论,利用额外的飞行试验数据对模型进行了检验。结果表明,辨识模型能够指示加力燃烧室工作状态,模型计算模拟参数与实际飞行试验数据吻合良好。该辨识模型可应用至该型发动机飞行试验的实时监控中,也可为其他型号发动机模型辨识提供参考。

For monitoring a type aero-engine＇s state during process of idle to maximum and maximum to idle, based on the actual flight test data, the afterburner transients model has been identified using three-layers forward artificial neural networks. The model precision was analyzed based model output error methods, and then the identified model was validated on out-of-sample flight test data. It shows that the state of afterburner indicator has been well predicted and the turbofan＇s state parameter agree well with actual flight test data during afterburner transients. The identified model could be applied in real-time monitoring during such type aero-engine＇s flight test, meanwhile could be a reference for other aero-engine＇s model identification.