5 SIMULATION EXPERIMENTS The stack which theoretical maximum output power is 480W consists of 50 pieces of single fuel cells in series. Every single fuel cell has ʹͷܿ݉ଶ polarization area . The anode of the stick is recycling and the Hydrogen flow is constant at ͷͲͲ݉ܮȀ݉݅݊ . The data collection time interval is 2 seconds, and through our experiments, the maximum time for all input become stable is 1.5 seconds. There are three input variables in this model so each group of data will activate eight rules. Corresponding to these eight rules, every input variable has a separated fuzzy subspace. The three input variables make up a three dimensional fuzzy subspace. There are three consequence parameters for this fuzzy subspace. According to the principle of least square estimation, the amount of data must be much more than the number of parameters to be identified, so the input data size to activate each rules should be equal to or greater than 4. Totally 2000 groups response value of the stack temperature under different air flow, different Hydrogen humidity and different output power were collected, then we separate these data to training set and test set. The training set was used for model training process and the test set was used to test the validity of the model. Finally the temperature model for the fuel cell stack are: