为解决卡尔曼滤波算法难以实现燃气轮机多传感器故障诊断的难题，提出一种基于混合算法的燃气轮机多传感器故障诊断方法。首先，基于平方根容积卡尔曼滤波（SRCKF）算法构建了一组滤波器，每个滤波器对状态的最优估计被定义为故障检测因子用于传感器故障的特征提取；然后，利用基于密度的聚类算法对故障检测因子进行聚类以实现故障传感器的检测和隔离；最后，利用极大似然估计方法（MLE）实现故障传感器故障严重程度的估计。所提出的方法在GT25000三轴燃气轮机模拟机上进行了仿真验证，仿真结果表明：所提方法有效，多传感器故障诊断的准确率高于95%。

In order to solve the problem that it is difficult to realize gas turbine multisensor fault diagnosis based on Kalman filter, proposes a gas turbine multisensor fault diagnosis method based on a hybrid method.Firstly,based on the square root cubature Kalman filter (SRCKF) algorithm,a set of filters are constructed.The optimal state estimation of each filter is defined as a fault detection factor for feature extraction of sensor faults.Then,the density based clustering algorithm is proposed to cluster the fault detection factors to realize the detection and isolation of fault sensors.Finally,the maximum likelihood estimation (MLE) method is used to estimate the severity of the fault sensor.The proposed method is verified on a GT25000 threeaxis gas turbine simulator. The simulation results show that the proposed method is effective,and the accuracy of multisensor fault diagnosis is higher than 95%.

TK478

国家科技重大专项(2017-I-0007-0008)；国家自然科学基金(51976042)