9 - Model Validation

Summary

Introduction

Model validation is the process of determining the degree to which a model is an accurate depiction of the real world from the perspective of the intended use of the model. Model validation has different purposes depending on the person's perspective. For example, for the modeling engineer the main purpose of model validation is to guide the development and refinement of a model, while for the user of the model, its validation provides a confidence level for the accuracy and limitations of the model. A solid and comprehensive model validation exercise results in increased confidence in assumptions behind the construction of the model and a higher level of assurance of its predictive capabilities outside the validation domain.

There is a subtle but important difference between the validation and verification of a model [1–3]. The verification of the model is the process by which the implementation of the model in the CAD package is demonstrated to be consistent with the equations and topology of the model, and to ensure that the model produces the expected results. In other words, the verification ensures that the model was properly implemented in the circuit simulator, as has been outlined in Section 8.4. On the other hand, model validation is the process by which the model simulation results are compared with an independent set of data not used during the model extraction. In essence, the model validation provides confidence and guidance on the predictive qualities of the model.