State-Space (Kalman) Estimator in the Reconstruction of Thermal Diffusivity from Noisy Temperature Measurements

The identification of thermal diffusivity from transient temperature measurements is strongly influenced by various noises and disturbances affecting measurement instrumentation. Some relatively new statistical techniques, like 'Kalman filtering', provide the theory to take account of these disturbances, improving the methodology for solving the nonlinear inverse heatconduction problem. The Kalman algorithm is developed in order to identify both the thermal· diffusivity of materials and some experimental errors, like the uncertainty in the sensor locations inside the specimen, which can produce large variations in the reconstructed thermophysical properties. The results of numerical simulations provide estimates of diffusivity coefficients with a relative accuracy of ± 0.1%, in accordance with the variance estimates given by the filter itself.