基于未知输入观测器的拾放机器人传感器故障检测

Sensor Fault Detection of Pick-and-place Robot Based on Unknown Input Observer

针对基于智能制造工业中广泛使用的拾放机器人,提出了一种基于未知输入观测器的传感器故障检测方法,以提高工业拾放机器人的可靠性。拾放机器人故障检测的挑战在于被拾放工件的类型等变化而带来工件质量变化,造成系统质量改变,使得系统模型不再准确。首先建立系统在竖直平面内运动的状态空间模型,将不同类型工件质量变化引起的未建模动态作为未知扰动。然后通过构建未知输入观测器将未知工件质量对系统的扰动从残差中解耦,增强残差对故障的敏感性,有效提高系统的故障检测性能。仿真实验表明,该算法能够有效地检测位置传感器故障。

In order to enhance the reliability of pick-and-place robots which have been widely used in modern intelligent manufacturing,this paper proposes an unknown input observer(UIO)-based sensor fault detection method.The main challenge in pick-and-place robot fault detection is the variation of whole system’s mass due to the changes of workpiece’s mass caused by the changes of workpiece’s type etc,because of which the mathematical model cannot represent the system dynamics correctly.Firstly,state space model of the pick-andplace robot is established when it moves in the vertical plane and the unmodeled dynamics caused by the mass variation of different types of workpieces are represented as unknown disturbances to the robot system.Then,a fault detection method is proposed,in which an UIO is adopted to decouple the unknown disturbances from the UIO’s fault detection residual information.The proposed method can improve the sensitivity to sensor faults and has higher performance of fault detection.Simulation experiments show that the algorithm can detect the fault of position sensors effectively.