深度自编码观测器飞机操纵面快速故障诊断

Fast fault diagnostic method for aircraft actuators with deep auto-encoder observer

为了避免扩展多模型自适应估计故障诊断方法中的雅克比矩阵计算,解决飞机精确模型难以获得的问题,降低在线故障诊断的计算量,提出了一种基于深度自编码观测器的飞机操纵面快速故障诊断方法。基于离线训练、在线估计的思想,采用量测的飞行数据训练得到不同故障下的飞机模型,代替扩展多模型自适应估计方法的卡尔曼滤波器进行状态估计;基于基础自编码器的隐层节点数选取经验公式,推导了两种深度自编码器的隐层节点数选取的递推公式。仿真结果表明,该方法无需精确的飞机模型,故障诊断速度快、精度高。

To avoid the calculation of Jacobi matrix in traditional multiple model adaptive estimation method,solve the problem of difficult to obtain accurate plane model and reduce the amount of calculation for online fault diagnosis,a deep auto-encoder observer multiple-model fault diagnosis algorithm for aircraft actuator fault was proposed. Based on the thought of off-line training and online estimation,the method replaced Kalman filters in traditional multiple model adaptive estimation with different fault aircraft models obtained by training measured flight data. Based on the empirical formula of the basic auto-encoder hidden layer node number selection,two recursive formulas for deep auto-encoder hidden layer node number selection were derived. The simulation results show that the method does not require accurate aircraft models and has fast speed and high accuracy for fault diagnosis.