航空发动机气路故障诊断的SANNWA-PF算法  

SANNWA-PF algorithm of aero-engine gas path fault diagnosis

作　　者：许梦阳 黄金泉 鲁峰 

机构地区：[1]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,南京210016 [2]先进航空发动机协同创新中心,北京100191

出　　处：《航空动力学报》2017年第10期2516-2525,共10页Journal of Aerospace Power

基　　金：国家自然科学基金(51276087);国家自然科学青年基金(61304133);南京航空航天大学研究生创新基地(实验室)开放基金(kfjj20160211)

摘　　要：针对航空发动机非线性、非高斯的特点,提出一种用于航空发动机气路故障诊断的自适应神经网络权值调整粒子滤波(SANNWA-PF)算法。该算法根据粒子分布情况确定分裂和调整的粒子数目,进而根据粒子权重采用正态分布的方式进行分裂,采用反向传插(BP)神经网络进行权值调整,缓解了粒子的退化和贫化,具有更强的自适应性能和跟踪能力。通过一维非线性跟踪模型和航空发动机气路故障诊断仿真研究表明:SANNWA-PF算法具有良好的非高斯性能,相对粒子滤波一维非线性追踪模型估计精度提高约21%,航空发动机气路故障诊断在高斯噪声和非高斯噪声下分别提高约30%和26%,诊断速度分别提高约7倍和10倍。A self-adaptive neural network weight adjustment particle filter algorithm was proposed for aero-engine gas path fault diagnosis of the nonlinear and non-Gaussian properties of aero-engine.Number of particles split and adjusted was determined by the distribution of particles.Then particles were spilt by the way of normal distribution and adjusted by back propagation（BP）neural network,which avoided the degradation and impoverishment of particles and had stronger self-adaptive and tracking ability.The simulation results of one-dimensional nonlinear tracking model and aero-engine gas path fault diagnosis show that self-adaptive neural network weight adjustment-particle filter（SANNWA-PF）algorithm has a good non-Gaussian performance.Compared with normal particle filter,SANNWA-PF improved 21%in accuracy of one-dimensional nonlinear tracking model,30% withGaussian noise and 26% with non-Gaussian noise in aero-engine gas path fault diagnosis;and the diagnosis speed improved about 7times with Gaussian noise and 10 times with nonGaussian noise.

关 键 词：航空发动机 故障诊断 粒子滤波 自适应 神经网络 非高斯噪声 

分 类 号：V233.7[航空宇航科学与技术—航空宇航推进理论与工程]