Academic Journals Database
Disseminating quality controlled scientific knowledge

Hierarchical Fault Diagnosis Using Sensor Data Fusion for Robotic System

ADD TO MY LIST
 
Author(s): Mohammad A. Nekoui | Amir R. Kashanipour | Karim Salahshoor

Journal: Electronics
ISSN 1450-5843

Volume: 14;
Issue: 1;
Start page: 55;
Date: 2010;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Fault diagnosis | Signal processing | Process control | Neural networks | Multi-agent system | Hierarchical system | Data fusion

ABSTRACT
In this paper, we present a sensor data fusion based approach in a multi-layered, multi-agent architecture toward more reliable fault detection and isolation. This paper describes a sensor fusion approach to the detection of faults in robotic systems, based on multi-agent distributed diagnosis. Each of the agents uses an Adaptive Neuro-Fuzzy Inference System (ANFIS) and decision fusion is achieved by a Dempster-Shafer supervisor. In this work sensor level fusion used for generating residual signal more accurate and by considering uncertainties in sensors tried to compensate innate lack of model-based approach in detecting abrupt change. Based on distributing the diagnosis task by dedicating intelligent diagnosis agents for different parts of robot, decision fusion used in supervisory level to consider reliability of each agent, considering contingency faults scenario and failure symptoms. Adaptive Neuro-Fuzzy Inference System has been used as inference architecture for distributed sensor networks. Two types of ANFIS networks are used. An ANFIS based learning and adaptation is used for modeling the jointstoward residual generation, while for residual evaluation another ANFIS is used. In this work online fault diagnosis based on sensordata fusion techniques for online fault diagnosis targeting robotic systems has proposed. Considering possibility of diagnosis one specific fault by more than one agent, practiced by using multiple ANFIS for different type of faults in each agent. Simulation results of controlled 3-Link Rigid Planar Manipulator controlled demonstrate capability of the proposed method toward achieving a fault tolerant controller.
Affiliate Program      Why do you need a reservation system?