基于粒子滤波器的移动机器人惯导传感器故障诊断

Particle filter-based fault diagnosis for inertial navigation system of mobile robot

提出一种粒子滤波器方法用于诊断移动机器人惯性导航系统传感器故障。该方法将基于规则的推理与多粒子滤波器结合,利用规则推理确定机器人运动状态,每一种运动状态用一个粒子滤波器监视。该方法有效地解决了单个粒子滤波器难以表示复杂逻辑的问题,降低了每个粒子滤波器的粒子数,从而提高了诊断效率和精度。对移动机器人在5种平面运动状态下(静止、直线运动、转动等)的8种工作模式(包括1种正常工作模式和7种故障模式)进行监视的仿真结果表明,采用所提出的方法可以有效地识别惯导系统的1个或多个硬故障。

A particle filter-based approach for fault diagnosis of inertial sensor of wheeled mobile robots was proposed, in which the rule-based inference and multiple particle filters were integrated. The rule-based inference method was employed to determine the movement states of the robot. Each movement state was monitored by a particle filter. This approach overcomes some shortcomings of particle filter such as weak logic inference capability, decreases particle number, increases efficiency and accuracy of each particle filter. The results of monitoring 8 kinds of operation mode of inertial navigation system （INS） of 5 kinds of movement states of mobile robot in a plane show that the approach presented can diagnose one or more hard faults of intertial navigation system sensors of mobile robots.