Model acceptance criteria for the time-domain technique explainability

3 Model acceptance criteria for the time-domain technique explainability 

Unlike the transfer-function-based technique, the time-domain technique does not require the central values of the nominally constant parameters to be determined from a minimization exercise. Nevertheless, we will expect the modeller to use sensible estimates, which may be expressed as a condition similar to inequality (24.52)  

Using the time-domain method described above, it is possible to calculate the time-behaviour of the modelmatching parameter variations, and hence it is a simple matter to determine the contributory variance (Soj) of each parameter that is necessary to match model to all the transient measurements recorded from the plant. To satisfy the explainability criterion, the variance of each parameter should be within the variance anticipated in advance, i.e.  

Provided that this condition is satisfied for each of the p parameters, the model is acceptable in the first instance in the sense that the deviations of the model outputs from the plant records are explainable in terms of known approximations. 

The same caveat as uttered in Section 24.4.4 applies when the model is to be used for subsequent prediction purposes, namely that the recorded plant transients against which the model has been tested should be sufficiently wide-ranging to traverse all the regions of the mt el utilized by the later model calculations if we are to have a high confidence in the model s predictions. 

---