Abstract: An evaluation method of green product is set up, whi- ch is based on function approximation. It is to select product samples, then to use methods of function approximation and set up model. According to condition on the function approxi- mation of artificial neural network, an evaluation model of green product based on artificial neural network is set up. An instance of refrigerator is given. After establishing the structure of it’s evaluating index and collecting related evaluating samples, BP and RBF neural networks are respectively trained and simu- lated. Those models, which don’t need weighting, can fulfill approximation to ordinal-number utility function and cardinal- number utility function, and realize evaluation of the non-linear utility function of green product.

Key words: Function approximation, Artificial neural network, Green product, Utility function