Smart Fault Detection Approach Leveraging Soft Sensor and Model-Free Control: Application to Robot Manipulator

Abstract

Fault detection and diagnosis (FDD) in smart manufacturing (SM) has become a significant area of research due to its potential impact on safety, robustness, and overall durability. This paper presents a novel Model-Free Fault Detection (MFFD) method inspired by the Model-Free Controller (MFC), addressing the limitations of existing FDD approaches. Unlike traditional data-driven and model-based methods, the MFFD method eliminates the dependence on physical laws and reduces the need for large datasets and extensive computational resources. The proposed method is based on fault estimation leveraging soft sensors to compare them with a reference signal obtained under normal conditions. The validation and effectiveness of the proposed MFFD method is demonstrated through simulation results on a SCARA robot, focusing on actuator faults.