论文部分内容阅读
针对发动机气路部件故障,提出了一种基于模型和基于数据驱动的融合诊断方法。采用极端学习机(ELM)实现基于数据驱动的故障诊断。针对ELM随机选择输入层权值和隐含层偏置带来的缺点,采用改进微分进化(IDE)算法以训练样本的均方根误差(RMSE)和输出层权值的范数为评价标准对其进行优化,减少了ELM的隐含层节点数,提高了网络的泛化能力。同时,由于传感器数目的不足,采用基于奇异值分解(SVD)的Kalman(SVD-Kalman)滤波器实现基于模型的部件故障诊断。为了提高航空发动机部件故障诊断的精度,利用改进的迭代约简最小二乘支持向量回归机(IRR-LSSVR)算法对两种算法的估计结果在特征层进行定量融合。仿真结果表明,在发动机稳态状态下,与单独使用基于模型和数据驱动的诊断方法相比,采用特征层融合有效地提高了部件故障诊断的精度和准确率。
Aiming at the fault of engine air line components, a model-based and data-driven fusion diagnosis method is proposed. Data-driven troubleshooting with Extreme Learning Machine (ELM). Aiming at the shortcomings of ELM random selection of input layer weight and hidden layer bias, the modified differential evolution (IDE) algorithm is used to evaluate the root mean square error (RMSE) of training samples and the norm of output layer weights It is optimized to reduce the number of hidden layer nodes in the ELM and improve the generalization ability of the network. At the same time, due to the shortage of sensors, a model-based fault diagnosis of components is implemented based on singular value decomposition (SVD) Kalman (SVD-Kalman) filter. In order to improve the accuracy of fault diagnosis of aeroengine components, the improved iterative reduction least squares support vector regression (IRR-LSSVR) algorithm was used to quantitatively fuse the estimation results of the two algorithms at the feature level. The simulation results show that, compared with the single diagnosis method based on model and data driven, the fusion of feature layer effectively improves the accuracy and accuracy of component fault diagnosis under the steady state of engine.