论文部分内容阅读
采用粒子滤波算法,设计了一种无需涡扇发动机线性化模型的故障诊断方法。通过含有高斯白噪声的转速测量信号准确估计出相应转速值,构造残差,并设定合适的阈值,实现了故障诊断。压气机健康诊断仿真结果表明:高压压气机(HPC)效率突变30%会导致残差信号1 000倍以上的变化,可以明显检测到故障的发生及发生时刻;当HPC效率突变量为2%,残差信号将变成正常时的6倍,残差信号还可以反映退化程度。因此基于非线性粒子滤波的发动机健康诊断方法可较好地辨识发动机性能退化、故障及失效。
Particle filter algorithm is used to design a fault diagnosis method without turbofan engine linearization model. The rotational speed measurement signal containing Gaussian white noise is used to accurately estimate the corresponding rotational speed value, construct the residual, and set the appropriate threshold value to achieve the fault diagnosis. Compressor health diagnosis simulation results show that: high-pressure compressor (HPC) mutation efficiency of 30% will lead to more than 1000 times the residual signal changes, you can clearly detect the occurrence of the fault and the moment of occurrence; when the HPC efficiency of 2% The residual signal will become 6 times normal, and the residual signal may also reflect the degree of degradation. Therefore, the non-linear particle filter-based engine health diagnosis method can better identify engine performance degradation, failure and failure.