UCLA

Critical infrastructures are complex engineering systems that play a fundamental role in the operation of modern societies. Due to the large costs associated with their design, operation, and maintenance, the application of an effective Prognosis and Health Management (PHM) strategy is of the utmost importance. By incorporating expert knowledge and data-driven approaches, PHM strategies can maximize the utilization of critical infrastructure while minimizing their maintenance expenditures. However, due to the scale, interconnectivity, and high levels of uncertainty under which these systems operate, the application of PHM strategies often requires solving computational challenges that strain the current technological capabilities.

While Artificial Intelligence methods have been used to overcome these challenges with great success, the rising scale, complexity, and cost of critical infrastructure systems motivate the exploration of novel computing paradigms. Quantum Computation has been heralded as a promising alternative to achieve speed or scalability advantages for certain operations when compared to traditional computational approaches. However, the application of quantum computation in the field of Civil Engineering, particularly for the PHM of critical infrastructures, remains heavily underexplored.

To address this relevant gap, this dissertation focuses on the development of quantum-based solutions to address challenges related to PHM of critical infrastructures. For this, three key focus areas within Quantum Computation are identified: Quantum Machine Learning, Quantum-based Combinatorial Optimization, and Quantum-Enhanced Sampling. For each of these areas, algorithms with a theoretical potential of producing a computational advantage for the PHM field are proposed. Relevant case studies of increasing size are designed to validate the proposed solutions and compare their performance and scaling against classical counterparts. Finally, the outputs of these comparisons are carefully examined to assess two primary results. First, whether an advantage is achievable nowadays when the algorithms are applied towards PHM applications. Second, the main limitations that still need to be overcome for Quantum Computation to be useful in practice.

The general contributions of this dissertation are listed below. A detailed list of contributions can be found in the respective introduction of each chapter.• Contribution #1: Assessing the viability of Quantum Kernel Functions towards fault diagnosis tasks in energy generation systems. • Contribution #2: Developing and assessing a general quantum-based framework for optimal sensor placement in civil infrastructure. • Contribution #3: Developing and assessing a quantum-based framework for the translation of Fault Trees as quantum algorithms and the enhanced identification of Minimal Cut Set configurations.