基于LMD与AO-PNN的中介轴承故障诊断方法

针对航空发动机中介轴承受噪声干扰大、传递路径复杂导致采用传统方法难以进行故障诊断的问题，提出了一种基于局部均值分解（LMD）与相关系数-能量比-峭度准则、结合天鹰座优化算法（AO）优化概率神经网络（PNN）的中介轴承故障诊断方法。使用LMD对传感器采集的振动信号进行分解；利用相关系数-能量比-峭度准则判决筛选分解得到的PF分量，重构筛选后的信号；计算重构信号的多尺度排列熵（MPE），以构建特征向量；通过AO优化的PNN的平滑因子，将优化后的神经网络用于中介轴承的故障诊断。基于中介轴承故障试验数据对诊断结果进行了分析，结果表明：提出的方法可以有效诊断高背景噪声、复杂路径干扰下的航空发动机中介轴承的典型故障，与粒子群优化的概率神经网络方法（PSO-PNN）和传统的PNN方法相比，其诊断准确率分别提高了3.875%和8.125%，具有较好的全局收敛性和计算鲁棒性。

Inter-shaft Bearing Fault Diagnosis Method Based on LMD and AO-PNN

Aiming at the difficulty of the fault diagnosis of inter-shaft bearings of aeroengine due to large noise interference and complex transmission paths by traditional methods, a fault diagnosis method of inter-shaft bearings based on Local Mean Decomposition (LMD) and correlation-coefficient, energy-ratio, kurtosis criterion, combined with the Aquila Optimizer (AO) optimized Probabilistic Neural Network (PNN) was proposed. The vibration signals acquired by the sensors are decomposed by using LMD; The PF components obtained from the decomposition are screened and determined using the correlation-coefficient, energy-ratio, kurtosis criterion, and the screened signals are reconstructed; The Multiscale Permutation Entropy (MPE) of the reconstructed signals is calculated to construct feature vectors; By optimizing the smoothing factor of PNN through AO, the optimized neural network is used for the fault diagnosis of inter- shaft bearings. The analysis results based on the inter-shaft bearing fault test data show that the proposed method can effectively diagnose the typical faults of inter-shaft bearings of aeroengines with high background noise and complex path interference, compared with the Particle Swarm Optimized Probabilistic Neural Network (PSO-PNN)method and the traditional PNN method, the diagnostic accuracy is improved by 3.875% and 8.125%, respectively, with better global convergence and computational robustness.