Fault detection and isolation for robot manipulator using statistics

Abstract

Robotic systems are widely used in the manufacturing industries to improve high quality products. When a fault occurs in manipulator, the faulty robot operates in abnormal condition, which causes insecurity for devices and human. In general, the dynamic of residual vector consists of fault information. In order to evaluate the residual vector, data-driven technique based on statistical theory is used. This paper presents a method by using the combination between the residual vector and data-driven solution for detecting and isolating actuator, sensor faults in robot manipulator. To achieve the accurate and reliable detection and isolation of sensor and actuator faults, the threshold is proposed to combine with generalized momenta. Simulation results are realized to show the performance of proposed method.