Structural Health Monitoring using Pattern Recognition

There are two main approaches to the diagnostic phase of Structural Health Monitoring (SHM): the first, is based on the solution of inverse problems, and the second, is based on pattern recognition or machine learning. The first approach usually depends on the construction of a model of the structure based on physical principles, while the second relies on building a model based on measured data. The complexity of many modern structures and their environments sometimes makes the second option an attractive proposition. While many engineers are familiar with the process of building physics-based models e.g. finite element models, familiarity with the principles of pattern recognition is less common. The objective of this chapter is to provide an introduction to the concepts of data-based modelling and pattern recognition in the context of the SHM problem.