The kinetics of thermogravimetric decomposition reactions are often difficult to model explicitly, because model parameters can depend on process conditions in an ill-defined way. Implicit kinetic models, such as those based on neural nets, often require extensive data and are not usually suitable for extrapolation of experimental data. In this paper it is shown that by combining explicit phenomenological models with artificial neural nets, more accurate modelling and extrapolation of these types of processes can be achieved.