Abstract: To study the problem of product performance evaluation from the perspective of system evolution, the non-equilibrium statistical mechanics method is introduced in the modeling process, which is successfully used to analyze the complex ecosystem in environmental field. The system constituted of indexes which reflect the product performance is taken as study object. The stochastic evolution equation of the system is set up on the basis of non-equilibrium statistical mechanics method, and the product performance evaluation is realized by conferring reasonable weight to the indexes to make the system become stable. The formation pattern of performance level and the dynamics law of evolution are revealed by the new model. Then, a specific evaluation method is presented by the combination of this model and self-organization feature map algorithm, and the effectiveness of this evaluation method is confirmed by examples in the end. Using this method to evaluate the product performance can exclude the effects of factors such as man-made weighting, and its has wide adaptability. This model also provides an approach to analyzing the index system architecture and the relationship between indexes, and can be further applied to the research on performance optimization of products..

Key words: Non-equilibrium statistical mechanics, Performance evaluation, Self-organization feature map