Gas-liquid separator modelling and simulation with Gaussian-process models

The Gaussian process model is an example of a flexible, probabilistic, nonparametric model with uncertainty predictions. It can be used for the modelling of complex nonlinear systems and recently it has also been used for a dynamic systems identification. Gaussian process models have become an emerging, complementary method for a nonlinear system identification. The output of the GP model is a normal distribution, expressed in terms of mean and variance. The mean value represents the most likely output and the variance can be interpreted as the measure of its confidence. The modelling case study of gas-liquid separator is presented in the paper. The emphasis of the paper is on the comparison of three methods for dynamic model simulation based on Gaussian processes in the phase of model validation. All three presented methods are approximations. The ‘naive’ simulation is feeding back only the mean values of model predicted values. The Taylor approximation approach and ‘exact’ approach approximate the model predicted distribution with Gaussian distribution, but in different ways. The level of computational burden associated with each approach rises with the complexity of computation necessary for approximation of uncertainty propagation. Therefore a trade off between the level of uncertainty propagation approximation and computational burden is present.