Proton exchange membrane fuel cell hybrid modeling research

Support vector machine (SVM) is a general purpose algorithm that has superior generalization performance. It is suitable to model nonlinear multi-variable systems such as proton exchange membrane fuel cell (PEMFC). But building a high dimensional SVM model needs huge of training data, hence a hybrid modeling approach was reported: SVM model only concerns the input variables of temperature and current, and a mechanistic pressure-incremental model incorporates anode and cathode pressures. Thus, combining them together can do the prediction on any current, temperature, anode and cathode pressures. The predictions are highly consistent with experimental results. Data analyses show that the model can make the predictions with the errors less than 1.6%.