基于NN-ELM的航空发动机燃油系统执行机构故障诊断

发布时间：2018-04-30 00:21

本文选题：航空发动机 + 执行机构　；参考：《航空动力学报》2016年02期

【摘要】：提出了一种航空发动机执行机构及其传感器单一故障诊断及定位方法.首先通过执行机构模型判断是否发生故障,然后运用发动机逆模型对故障进行定位.基于离线训练BP(back propagation)神经网络建立执行机构模型,根据某半物理仿真试验台的测试数据训练网络参数.提出离线训练和在线训练相结合的极端学习机(ELM)算法建立发动机逆模型,使网络在初始时刻就具有诊断能力,工作过程中具有适应能力,且在线训练过程采用阈值判别法筛选训练样本,减小了在线训练时间,提高了逆模型的实时性.以某型发动机燃油系统执行机构为例的设计和仿真结果表明:该诊断系统能够准确地对发动机在稳态和动态工况以及蜕化状态下的执行机构及其传感器单一故障进行准确诊断和定位,具有很好的实时性.
[Abstract]:A single fault diagnosis and location method for aeroengine actuator and its sensor is presented. First, the fault is judged by the actuator model, and then the engine inverse model is used to locate the fault. Based on the offline training BP(back propagation neural network, the executive mechanism model is established, and the network parameters are trained according to the test data of a semi-physical simulation test rig. An extreme learning machine (ELM) algorithm combining off-line training and on-line training is proposed to establish the engine inverse model, which makes the network have the ability to diagnose at the beginning and adapt to the working process. In the online training process, threshold discriminant method is used to screen the training samples, which reduces the online training time and improves the real-time performance of the inverse model. The design and simulation results of the actuator of a certain engine fuel system show that the diagnosis system can accurately diagnose the single fault of the actuator and its sensor in steady and dynamic conditions and in the degenerated state of the engine. For accurate diagnosis and localization, Has very good real-time.
【作者单位】：南京航空航天大学能源与动力学院江苏省航空动力系统重点实验室;
【基金】：航空科学基金(20110652003) 中央高校基本科研业务专项基金(NN2012033) 江苏高校优势学科建设工程资助项目
【分类号】：V263.6

【相似文献】