Unified Sensing, Diagnosis and Active Self-Healing

The application of self-healing material in industrial systems has the potential to improve reliability and save cost. This is because faults do occur in systems, and for life critical or remote or difficult to access ones, current maintenance practices may be insufficient. When such systems are self-healed, the materials making up the systems regain some or all of its lost physical properties to keep the systems functioning. However, a majority of self-healing approaches are not yet at the level of industrial integration. These self-healing approaches are passive and do not guarantee a match between the damage and healing rate. A significant step in their advancement is the development of an integrated sensing, fault diagnosis and active self-healing system, which is the focus of this thesis. A mathematical model based on a previously experimented electromechanical self-healing process, whose healing mechanism combines piezoelectricity and electrolysis is developed. The model demonstrates the poor match between the damage and healing rate due to the ineffectiveness of the healing process to counteract the onset of damage, the dominant effect of uncertainties and disturbances on the healing process, the dependence of healing on the location of the healing mechanism relative to the fault location, etc. In addition, nonlinearities, such as the inherent dead-zone of the chosen healing mechanism affect the response of the healing process. The model also provides a benchmark for the work in this thesis. The model is then the foundation for the development of a novel active self-healing system. This is a closed loop system that takes advantage of sensing and adaptive sliding mode feedback control with the modelled healing mechanism to achieve a desired response. Importantly, it is shown in simulation that adaptive feedback control (sliding mode control) can minimize the effect of uncertainties, regulate the healing rate of a self-healing system to meet user or environmental demands, such as the damage rate, and compensate for the non-linear dead-zone associated with this healing mechanism. Finally, a novel fault diagnosis method that combines the beam curvature, proportional orthogonal decomposition, Hölder exponent and supervised regression is presented as a step to define the environmental demands. This essentially captures the effect of damage of a beam structure. It is combined with the active self-healing system, leading to a novel framework for an integrated sensing, fault diagnosis and closed loop active self-healing system. It is shown through simulation that the proposed active system can potentially estimate the damage rate, provide adequate actuation to match the healing rate with the estimated damage rate and provide real time insight into the healing dynamics