Dissipativity based fault detection and diagnosis for linear systems

The increasing complexity of modern industrial processes can make the plants susceptible to faults (e.g. sensor and actuator failures). In this paper, the dissipativity theory is used to develop a fault detection and diagnosis scheme for process systems. The dissipativity properties represent the features of process dynamics, which will change when certain faults occur. Moreover, the dissipativity property of a system is not unique, and can be shaped to capture the process dynamic features that are sensitive to the faults. The proposed fault diagnosis scheme is constructed based on a number of dissipativity inequalities, which are quadratic functions of the input and output trajectories of the process (in the Quadratic Difference Form). These inequalities are checked in real time for online fault detection and diagnosis. The proposed fault detection and diagnosis approach is illustrated using a heat exchanger case study.