基于多层感知机的航空发动机压气机盘应力和温度预测

Stress and temperature prediction of aero-engine compressor disk based on multilayer perceptron

将发动机可测参数作为初始特征,利用人工神经网络技术建立航空发动机压气机盘应力和温度预测的MLP(multilayer perceptron)模型,采用BP(back propagation)神经网络算法进行训练。结果表明:该方法预测结果与传统有限元计算结果吻合较好,相对偏差均在1%以内,判定系数达到0.95以上,方均根误差均在5以内,且计算速度由小时级提升为分秒级,可为后续工程应用提供依据。

Taking the measures parameters of the engine as the initial characteristics,the MLP(multilayer perceptron)model of aero-engine compressor disk stress and temperature prediction was established by using artificial neural network technology,and BP(back propagation)neural network algorithm was used for training.The results showed that the prediction results of this method were in good agreement with the traditional finite element calculation results.The relative deviations were all within 1%,the determination coefficients were above 0.95,and the root mean squared error was within 5.Moreover,the calculation speed increased from hour level to minute second level,providing a basis for subsequent engineering applications.