Sensor Fusion and Inferential Sensing for Fault Detection and Isolation in Uncertain Systems

When attempting to meticulously conduct fault detection and isolation, inferential sensing approaches can prove to be valuable tools, as they constitute cost-effective and reliable alternatives to expensive and often impractical measuring devices. In this work, we combine advanced model-based sensor selection and inferential sensing techniques to yield accurate fault detection outcomes, in the presence of system noise and uncertainty. In this regard, the most informative sensor set is chosen, by comparing all possible sensor combinations, based on various optimality criteria that extract the maximum knowledge from the system in functions of the Fisher Information Matrix. Subsequently, optimal inferential sensors are derived, by means of genetic and mathematical programming that incorporate symbolic regression and optimization techniques. For built-in test deployment, k-Nearest Neighbors (k-NN) classification is employed to assess the accuracy of each sensor network, as well as the performance enhancement due to the inclusion of inferential sensor(s). The proposed methods are applied in steady state and dynamic models of a cross-flow plate-fin heat exchanger system for various levels of measurement noise and uncertainty. The results illustrate that the augmented system of composite sensors (i.e., inferential and hardware) provide more accurate information on system fault(s), while reducing the evidence of uncertainty and system noise.