| 摘要: |
| 为了解决在基于解析冗余关系的故障诊断应用中难以实现故障隔离的问题,提出了一种基于双重改进粒子滤波器的故障隔离方法。该方法利用状态和参数估计粒子滤波器组成的联合估计模型,对系统状态和潜在故障参数值进行联合估计,通过对比潜在故障参数估计值与其标称值实现故障隔离。在联合估计模型中,一方面,在传统的随机扰动法的基础上,利用最大似然估计法获得参数时间更新梯度,使用一种改进随机扰动法实现参数时间更新；另一方面,在采样过程中考虑当前量测值,并引入粒子群和模拟退火优化思想,使用一种采样粒子质量改进方法实现粒子采样,以提升其估计性能。仿真结果表明:在假设的两类参数型故障下,基于双重粒子滤波器的联合估计模型在鲁棒性、计算速度和估计精度上均优于基于扩展状态空间的粒子滤波器联合估计模型,在基于双重粒子滤波器的联合估计模型上,使用所提出的改进方法能显著提升其估计性能。所提出的方法基本满足参数型故障隔离对计算效率和估计精度的要求,可作为基于解析冗余关系故障诊断中的故障隔离方法。 |
| 关键词: 粒子滤波器 故障隔离 联合估计方法 粒子群优化 模拟退火优化 最大似然估计 |
| DOI：10.11918/202108109 |
| 分类号:TN713 |
| 文献标识码:A |
| 基金项目:航空科学基金(2020033052001) |
|
| Fault isolation method based on a dual improved particle filter |
|
MO Haobin,LI Yanjun
|
|
(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China)
|
| Abstract: |
| In view of the problem that it is difficult to realize fault isolation when using the analytical redundancy relations-based fault diagnosis (ARRBFD) method, this paper presents a dual improved particle filter method for fault isolation. The method employs the joint estimation model composed of state and parameter estimation particle filters to jointly estimate the values of system state and potential fault parameter, and it achieves fault isolation by comparing the estimated value of the potential fault parameter with its nominal value. In the joint estimation model, on the basis of the traditional random perturbation method, an improved random perturbation method was developed to realize the parameter time update, which uses the maximum likelihood estimation method to obtain the parameter time update gradient. Then, a sampling method was proposed to improve the sampling particle quality, which takes into account the current measured values in the sampling process and introduces the idea of particle swarm optimization and simulated annealing optimization. Simulation results show that under the two types of parametric faults assumed in this paper, the joint estimation model based on dual particle filter outperformed the joint estimation model based on extended state space in terms of robustness, calculation speed, and estimation accuracy. The proposed method significantly improved the estimation performance in the joint estimation model based on the dual particle filter. In conclusion, the proposed method meets the requirements of computational efficiency and estimation accuracy for parametric fault isolation, and it can realize the fault isolation when applying the ARRBFD method. |
| Key words: particle filter fault isolation joint estimation method particle swarm optimization simulated annealing optimization maximum likelihood estimation |
