Estimation of material properties using thermography and data assimilation

Real-time monitoring of systems and processes based on the estimation of material
parameters is a common approach to enable an optimization of such systems
and processes in terms of cost, time and quality. However, for complex, non-linear
systems classical estimation methods based on linear models may fail. Therefore,
the aim of this thesis is to develop a method for parameter estimation based on
thermography, which overcomes these difficulties. The Ensemble Kalman Filter, a
data assimilation technique, is applied to transform the temperature data from the
thermography into the parameters of interest. Thereby, a dynamic model is combined
with measurement data to sequentially improve the state and parameters of
the dynamic model.
First, the proposed method is evaluated by means of a simulation study with artificial
data. The convective heat transfer coefficient as well as selected thermophysical
parameters such as thermal conductivity and heat capacity are to be estimated.
Building on the knowledge gained from the simulation study, the method is applied
to real thermographic measurement data. Thermal parameter estimation based on
the proposed method require heat transfer mechanisms to occur in or at the surface
of the material of interest. Thus, the test specimen is brought out of the thermal
equilibrium in the thermography experiments by external thermal excitations such
as laser or heat gun.
Data assimilation allows the determination of parameters based on thermography,
whereby no direct measurement of the parameters is necessary. The chosen measuring
method has the advantage of being contactless, which means that it can be used
during the process without disturbing it. It is recommended to adapt the method
to the material to be examined or the parameter to be estimated by means of a
simulation study.