针对燃气轮机故障诊断过程中诊断精度不足，卡尔曼滤波易出现“弥散”现象的问题，提出了一种基于滤波阵列的燃气轮机气路故障诊断方法。该方法首先构建了一组基于健康参数调度的平衡流形展开模型阵列，然后结合平方根容积卡尔曼滤波对气路部件健康参数进行了实时估计，最后通过隔离因子实现了对故障部件的检测与隔离。仿真表明：该方法有效解决了卡尔曼滤波在故障诊断过程中出现“弥散”现象的问题，针对燃气轮机气路部件突变故障，能够有效实现故障的检测、隔离与估计。

Aiming at the problem that Kalman filter is prone to "dispersion" phenomenon due to insufficient diagnosis accuracy in gas turbine fault diagnosis process, a gas path fault diagnosis method based on filter array was proposed.Firstly, an equilibrium manifold expansion (EME) model array based on health parameter scheduling was constructed.Then, health parameters of gas path components were estimated in real time by combining square root cubature Kalman filter (SRCKF).Finally, an isolation factor was proposed to detect and isolate the fault components.The simulation results show that the proposed method effectively solves the problem of "dispersion" phenomenon of Kalman filter in the fault diagnosis process,and the abrupt fault of gas turbine gas path components can be detected, isolated and estimated effectively.

TK471

国家自然科学基金(51976042)；国家重大科技专项(2017-I-0007-0008)