Topic: Structural Health Monitoring using Artificial Intelligence techniques

for Aerospace systems.

Introduction: Aerospace structural systems experience a broad spectrum of
environmental and operational loads. Severe and/or prolonged load
exposures may trigger the damage accumulation process even in recently
deployed structures. The process of implementing a strategy of damage
detection for engineering structures is referred to as structural health
monitoring (SHM). SHM is normally based on non-destructive measurement
of responses that change due to damage in the structure. Comparisons of
measurements made in a damaged and healthy (reference) state allow
inference of the type, location, and severity of the damage. Vibration
measurements have been used in many ways for this purpose.
Objectives: This research will focus on the application of Artificial
Intelligence (AI) methods in the sensing, detection, analysis and classification
of ill-health status of Aerospace structures. Thus the objectives are:
To develop intelligent methods for in-situ health monitoring and damage
detection of aerospace structures using low frequency vibration.
To analyse these anomalous signals using AI techniques (Artificial neural
networks, Fuzzy Logic).
To classify the fault patterns observed.
Methodology: There are several approaches that can be employed in
exploring ways of performing failure prognostics requiring immediate
attention in aerospace structures. Such approaches go from simple historical
failure rate models to complex physics-based models. Information required
for such models include failure history, engineering model and data, past
operating conditions, observed fault patterns, system degradation and failure
modes. Artificial Intelligence (neural network based-approach) will be
employed. The proposed approach involves two steps. The first step, system
identification, uses NARX (Non-Linear Auto-Regressive with eXogenous)
neural networks to identify the undamaged and damaged states of the
structural system. The second step, structural damage detection, uses the
aforementioned trained NARX neural networks to generate free vibration
responses with the same initial condition or impulsive force. Feature
extraction techniques will be applied to obtain the desired structural
properties states in the signal. Comparing the periods, amplitudes and
spectrums of the free vibration responses of the damaged and undamaged
states allows the extent of changes to be assessed. This method is known to
have successfully supported the detection, diagnostic and forecasting
processes due to its flexibility, adaptability and robustness especially for
prediction purposes.
Scope of work: The scope of this work will comprise the treatment of SHM
as a comprehensive, multi-scale phenomenon in which damage detection

may be needed over a spectrum of length scales from the microscopic to the
macroscopic level. The strategies proposed will be intended for various
aerospace structures, including aircraft, launch vehicles, space vehicles and
robotic devices that patrol these structures for SHM.