System Identification for Structural Health Monitoring is the first text on smart techniques of mechanical system identification using records from limited locations. The techniques explained in the book are based on rich content published in international journal papers by the authors, to which have been added introductory explanations to make the material accessible for a broad class of readers.
System identification (SI) techniques play an important role in investigating and reducing gaps between constructed structural systems and their structural design models, as well as in structural health monitoring for damage detection. A great amount of research has been conducted on SI.
The two major branches of SI are modal-parameter SI and physical-parameter SI. The former is appropriate for identifying the overall mechanical properties of a structural system and exhibits stable characteristics in implementation. While the latter is important from different viewpoints, e.g. enhancement of reliability in active controlled structures or base-isolated structures, its development is limited due to the requirement for multiple measurements and the necessity of complicated manipulation. A mixed approach is often used in which physical parameters are identified from the modal parameters obtained by the modal-parameter SI. However, a sufficient number of modal parameters must be obtained in order for the unique and accurate identification of the physical parameters to take place. This requirement is usually hard to satisfy.
The authors explain a unique system identification theory for a shear building model that overcomes some of the difficulties with SI methods. They show that unique identification of story stiffness and viscous damping coefficients is possible when acceleration records at the floors just above and below a specific story are available; this eliminates the need for acceleration records for all the floors above a specific story, an unrealistic possibility within multi-storied buildings because of the instrumentation that would be required. It also makes it possible to identify both stiffness and damping simultaneously and it requires only simple manipulation of Fourier transforms.