BP与RBF神经网络在航空发动机气路故障诊断中的应用

Application of BP and RBF Neural Networks in Airway Fault Diagnosis of Aero-engines

采用模拟样本与实际故障样本建立网络训练所需样本库,基于BP与RBF神经网络建立故障诊断模型,用PW4000航空发动机的故障样本进行训练,对训练好的网络进行测试仿真。仿真结果显示BP与RBF两种神经网络具有较高的诊断正确率,在航空发动机故障诊断中具有应用价值;与实际故障数据对比表明BP与RBF神经网络在航空发动机故障诊断中具有实用性。

Using a combination of simulated samples and actual fault samples, a sample library for network training and a fault diagnosis model are established based on back-propagation （BP） and radial basis function （RBF） neural networks. The operating principle of the neural network models is analyzed and the neural networks are trained with failure samples from the PW4000 aero-engine. The networks are tested and the simulation results show that the BP and RBF neural networks have high diagnostic ac-curacy and application value. Verification of the practicality of BP and RBF neural network in aero-engine fault diagnosis is performed by using actual fault data.