基于神经网络的航空发动机滑油监测分析

Aeroengine Lube Monitoring Analysis Based on Neutral Net

提出了一种基于BP神经网络的航空发动机滑油金属含量预测方法,给出了运用自回归模型(AR模型)预测模型和神经网络进行预测的一般公式。将其应用于某型发动机滑油的铁金属含量预测,结果表明,与传统的AR预测模型相比,神经网络表现出优秀的推广能力。经过数值仿真得出AR模型仅能预测出序列的变化趋势;神经网络预测推广能力强、具有较强的鲁棒性和容错性,可以为发动机的监控提供重要的依据。

A method was proposed which can be used to forecast lube oil metal content of aviation motor based on BP neural net,and the ordinary application formula of auto-regressive（AR） forecasting model and neural net model was given out. The two methods were used to forecast lube oil metal content of a kind of motor,the result shows that,in contrast to AR forecasting model,the neural net model has excellent popularization capability. The numeric value imitation show that AR model can only forecast the array＇ s alternation tendency, but the neural net has powerful rousting and fault-tolerance,it can be used in the motor supervision.