Joint Estimation of Additive and Parametric Faults: A Model-Based Fault Diagnosis Approach towards Predictive Maintenance

Abstract

The condition of systems, such as production equipment, typically deteriorates over time, increasing the risk of failure and associated unscheduled downtime. Predictive maintenance is a strategy to prevent failure, while maximizing the life cycle of equipment within a system. This paper contributes in this context to the theory of real-time fault diagnosis with an approach that can jointly estimate additive and parametric faults. The proposed fault diagnosis system consists of detection filters which are complemented with a residual evaluator, enabling effective fault isolation and fault estimation for open-loop and closed-loop controlled systems. The effectiveness of this unified approach is illustrated on a mass-spring-damper system.