Fault Detection Based on the Estimated Parameter of System Model

Most of the observer-based fault diagnosis methods, which are widely used techniques in many fields, require a nominal mathematical model of the systems. However, it is difficult and impractical to establish system mathematical model because it is hard to obtain accurate parameters of the plant in practical occasion. This paper presents a new fault detection technique using a innovative observer, which is named as Extended State Observer (ESO), can provide vital information for fault diagnosis with only partial information of the plant based on the input and output data. The ESO specifically identifies the un-modeled dynamics, a feat no other disturbance observer has ever attempted, and it cannot be easily accomplished with any existing methods. It is shown in this paper that faults can be detected with only one estimated parameter b₀, which related to system structure. In this paper, a strongly coupled nonlinear dynamic system - three tanks system is introduced as a case study. The simulation results demonstrate that the presented ESO based techniques effectively detected this nonlinear system´s multiple faults.