本文研究了基于连续型Hopfield神经网络的模糊属性关系图(ARG)模型匹配算法。首先指出模糊ARG比通常的ARG模型更真实地反映了客观事物,具有良好的容错性,能够更好地完成建模和匹配过程。然后研究了连续型Hopfield神经网络进行模糊ARG模型匹配算法。采用由ARG模型直接映射到网络组态的形式,直接、准确地实现了ARG模型的匹配。仿真结果表明模糊ARG模型建立的合理性和神经网络匹配算法的有效性。 This paper studies the fuzzy attribute relation graph (ARG) model matching algorithm based on continuous Hopfield neural network. First of all, it points out that fuzzy ARG reflects objective things more truthfully than the usual ARG model, has good fault tolerance and can better complete the modeling and matching process. Then, the continuous Hopfield neural network is studied for fuzzy ARG model matching. The ARG model is matched directly and accurately using the form that the ARG model maps directly to the network configuration. Simulation results show that the fuzzy ARG model is reasonable and the neural network matching algorithm is effective.