确定采样型强跟踪滤波飞机舵面故障诊断与隔离

Deterministic sampling strong tracking filtering algorithms: fast detection and isolation for aircraft actuator fault

中文关键词  故障诊断  多模型自适应估计  多重渐消因子  无迹卡尔曼滤波  中心差分卡尔曼滤波  容积卡尔曼滤波

英文关键词  fault detection  multiple model adaptive estimation  multiple fading factors  unscented Kalman filter  central difference Kalman filter  cubature Kalman filter

中文摘要

      为了克服扩展多模型自适应估计中扩展卡尔曼滤波的理论局限性, 多重渐消因子强跟踪改进引起的滤波发散现象以及多维高斯故障概率计算量大等问题, 本文将一类基于确定解析采样近似方法的非线性次优高斯滤波与多模型自适应估计相结合, 提出了改进的多重渐消因子强跟踪非线性滤波快速故障诊断方法. 确定采样型滤波克服了扩展卡尔曼滤波的理论局限性; 推导了等效多重渐消因子计算方法, 避免了非线性系统雅克比矩阵的计算, 提高了故障突变时的跟踪性能; 提出了基于平方根分解的改进的一步预测协方差更新方程, 保证了滤波稳定性; 提出了基于欧几里得范数简化的故障概率计算方法, 降低了计算量. 通过对比仿真验证了3种不同非线性滤波算法及其强跟踪改进算法的有效性, 故障诊断方法跟踪性强、\!\!速度快、\!\!精度高, 具有较好的鲁棒性和稳定性.

英文摘要

      The extended Kalman filter used in the extended multi-model adaptive estimation method has theoretical limitations; multiple fading factors may result in divergence of the strong tracking filter. Moreover, the fault probability calculation cost is large. In this paper, a class of deterministic sampling nonlinear suboptimal Gauss filters is combined with the multi-model adaptive estimation, and an improved nonlinear filter algorithm based on equivalent multiple fading factors is proposed for fast fault diagnosis. The theoretical limitation of the extended Kalman filter is eliminated by deterministic sampling filters. The equivalent multiple fading factor equations are derived, which avoid the calculation of the Jacobian matrix of the target nonlinear system. An improved update equation of the prediction covariance matrix based on the square-root decomposition is proposed, which ensures the stability of filter. Furthermore, based on the Euclidean norm, a fast fault isolation method that reduces the fault probability calculation cost is proposed. The simulation results show that the algorithm is more efficient and has better performance than the standard method.