Optimization of a procedure for model calibration using a method based on scattered data interpolation

Abstract

In order to develop advanced control strategies for complex thermal systems such as in the case of refrigeration and air-conditioning systems, calibrated models against experimental measurements are necessary to reproduce with accuracy the behavior of the real plant, under static and dynamic conditions. Even if accurate information (e.g.:look-up tables) for each component (e.g.:heat exchangers, compressors, etc.) is known, and incorporated to the model plant, the effect of integration of the components inthe whole plant has to be taken into account. In a context of fast industrial development, a procedure of minimization of nº tests for the whole plant oriented to model plant calibration has been developed, suitable for Simulink or Modelica environments. This procedure is based on the optimized generation of the nº of tests to be done, according to the error against the available previous data. Once the experimental is doing the generated matrix of new data is used to calibrate the model, using scattered interpolation strategies.