Analytical redundancy, fault detection and health monitoring for structures

In today's society there is an increasing concern regarding the safety of structures and structural damage associated with impact, fatigue loading, and high load excursions which can eventually result in catastrophic failure of the structure if not detected at the earliest stage of development. The detection of damage can be accomplished by either routine inspection of the structure or with an in-situ health monitoring system. Continuously monitoring the structure's health with an in-situ system, reduces in-service maintenance time and operating costs, and increases structural reliability.; While the potential payoffs are high, developing a reliable technique to monitor damage evolution in a structure is a difficult task to achieve. In past decades, many different approaches have been proposed and among them, the investigation of changes in vibration characteristics caused by damage are the most prominent. However, to date these methods have relatively poor sensitivity which is problematic for practical implementation.; The approach for the health monitoring system proposed here utilizes the structural vibration changes associated with the presence of damage. This is done in two different methods: (1) a method that utilizes the changes that damage produces in the dynamic response energy and (2) a method that utilizes the Douglas-Speyer fault detection filter. Both the methods are implemented in a simply supported beam with piezoelectric transducers utilized as sensor and actuator.; The first method utilizes the power spectral density (PSD) that can be regarded as a measure of the structure power output as a function of frequency. For this method, two different types of investigations are proposed: (1) a broad band investigation and (2) a narrow band investigation. The Broad band investigation shows that, if a sufficiently large bandwidth is considered, the energy changes associated with damage is approximately independent from its locations and rather is a function of the damage size. This is shown with experimental tests performed on a simply supported beam with simulated damage. This investigation, although it allows to infer the presence of damage, gives no information on damage location. For this reason, a narrow band investigation is proposed that allows the detection and location of damage by means of a Damage Location ( DL) function defined in our analysis. Results are predicted with the analytical model and verified experimentally with good agreement obtained.; The second method implemented utilizes the Douglas-Speyer fault detection filter. To date, in practical applications, this filter is utilized mostly for faults that occur in communication systems and has never been implemented for a structure. The filter allows a structure to be continuously monitored and as soon as the fault occurs, the output of the filter changes. These changes can be utilized to infer dam age as well as to locate the damaged region. The filter is implemented numerically with a simply supported beam model for two different cases (1) a simply supported beam with as many measurements as half of the dimension of the state vector and (2) a simply supported beam with a reduced number of measurements. The first case adopts the ordinary Douglas-Speyer theory for the filter design while the second case uses a modified theory that has been developed.