We present a method for synthesizing chemical process models that comb
ines prior knowledge and artificial neural networks. The inclusion of
prior knowledge is investigated as a means of improving the neural net
work predictions when trained on sparse and noisy process data. Prior
knowledge enters the hybrid model as a simple process model and first
principle equations. The simple model controls the extrapolation of th
e hybrid in the regions of input space that lack training data. The fi
rst principle equations, such as mass and component balances, enforce
equality constraints. The neural network compensates for inaccuracy in
the prior model. In addition, in equality constraints are imposed dur
ing parameter estimation. For illustration, the approach is applied in
predicting cell biomass and secondary metabolite in a fed-batch penic
illin fermentation. Our results show that prior knowledge enhances the
generalization capabilities of a pure neural network model. The appro
ach is shown to require less data for parameter estimation, produce mo
re accurate and consistent predictions, and provide more reliable extr
apolation.