UC San Diego

The desire to push aging civil, mechanical, and aerospace structures beyond their intended design lives has highlighted the need for structural health monitoring (SHM) strategies that are able to detect, locate, and quantify various forms of damage within them. SHM strategies may also be tailored for newly-deployed structures in an attempt to optimize their performance and maintenance over an entire life cycle so that total ownership costs are reduced. Specifically within the aerospace industry, standard non-destructive evaluation (NDE) techniques have been used for decades for inspection of components and systems. One of the most common and widely-accepted NDE domains is ultrasonic inspection, where components are imaged with the component out of service. Recent advances in sensor technology, distributed networks, and advanced signal processing techniques have begun to be exploited for in situ ultrasonic (and other forms of) SHM systems that are being deployed in a wide variety of real-world structures. In most cases, however, the ultrasonic excitation signals and feature extraction techniques being employed are the same as the standard NDE methods that have been in use for decades and are only applicable to relatively simple component geometries. This dissertation contributes to the body of knowledge in this field by introducing a new class of excitation signals and pattern recognition algorithms that, when paired with novel sensor networks, improve on the ability of standard SHM techniques to locate and identify damage on more complex geometry systems, including bolted joints and composite materials. This dissertation describes a methodology whereby chaotic guided waves are created and optimized (in a detection sense) and used as probes to perform damage assessment by building both time- and state -space domain models (rooted in pattern recognition) and using statistical modeling for performing damage classification under Type I/II error control. Multiple chaotic ultrasonic excitation formats are explored, including short-time chaotic wave packets and long-time chaotic bulk insonification, in which the diffuse, reverberant wave field is examined to identify structural changes. This method of insonification, in addition to enhanced pattern recognition techniques, allows this damage detection scheme to be employed on complex structural geometries with which standard ultrasonic-based SHM methods cannot be used. The outlined SHM method is applied to various test structures with different forms of induced damage including an aluminum plate with corrosion damage, bolted connections on several aluminum test structures (single and multiple-bolt configurations) and several adhesively-bonded composite wing-to-spar structures. Chaotic signal creation parameters are optimized for the composite structures and attempts are made to examine and, where possible, compensate for several sources of variability, such as temperature, within-unit variability and unit-to-unit variability