Grey-box identification

In the modelling of dynamic systems several approaches exist. For the so-called black box approach the modelling is based solely on measurements obtained for the actual system, and any prior information about the system is not used. The opposite approach is used in the white box or physical approach where the model is formulated solely on prior knowledge. In the project methods for combining the two approaches are developed, and such methods are called grey-box methods. The methods are based on Bayesian statistics. Typically the structure of the models are formulated as stochastic differential equations and the parameters are estimated using available data. The modelling approach has been tested with great success on many physical and technical systems.