Fault Detection and Isolation of Automotive Air Conditioning Systems using First Principle Models

Abstract

Although model-based fault tolerant control (FTC) has become prevalent in various engineering fields, its application to air-conditioning systems is limited due to the lack of controloriented models to characterize the phase change of refrigerant in the vapor compression cycle. The emergence of moving boundary method (MBM) illuminates a promising way for FTC design. In this paper, we exploit a control-oriented nonlinear model comparable to MBM to design an FTC framework with a generalized internal model control (GIMC) approach. A fault detector and isolator (FDI) is developed to identify potential actuator and sensor faults. A fault compensator is employed to compensate these faults if detected. Comprehensive simulations are carried out to evaluate the developed FTC framework with promising results. Plant variations are explicitly considered to enhance the gain-scheduled FTC developments.

Review on Fault Detection and Diagnosis Feature Engineering in Building Heating, Ventilation, Air Conditioning and Refrigeration Systems

“…For HVAC&Rs, a simulation model that obeys thermo-physical equations and control logics can be used to describe fault operations in a relatively systematic way. Researchers can select FFs by comparing simulated fault data with normal data even under dynamic working conditions [195,[203][204][205].…”

A hidden-Gamma model-based filtering and prediction approach for monotonic health factors in manufacturing

“…HFs can also be the output of Soft Sensor modules [21,22], where the status health is impossible/costly to be monitored. Moreover, HFs can be the residual of first principle FDI models [23]. In fact, in many practical examples [1,17,23,24], residuals have a monotonic behaviour and thresholdbased policies to maintenance management are implemented on such quantities.…”

“…Moreover, HFs can be the residual of first principle FDI models [23]. In fact, in many practical examples [1,17,23,24], residuals have a monotonic behaviour and thresholdbased policies to maintenance management are implemented on such quantities. HFs are therefore relevant quantities in both model-based [24,25,26] and modelfree [1,3,27,28] prognostic approaches.…”