提出了一种基于Σ-Δ调制的单比特Hopfield人工神经网络(HNN)的设计方法.网络中各神经元的输入输出均为基于Σ-Δ调制的单比特信号,这解决了多比特并行分布网络扇入扇出的难题.同时在设计12神经元全并行Hopfield网络过程中,通过数字后端设计,优化了关键模块,减小了系统的面积及功耗.最后采用同步扰动学习算法(SLA)实现了在片训练系统,通过此训练系统Hopfield网络实现了文字的联想记忆. A design method of single-bit Hopfield artificial neural network (HNN) based on Σ-Δ modulation is proposed, in which input and output of each neuron in the network are single-bit signals based on Σ-Δ modulation, which solves the problem of multi-bit parallel distribution Network fans into the fan out of the problem.At the same time in the design of 12 neuron full parallel Hopfield network process, through the digital back-end design, the key modules are optimized to reduce the system area and power consumption.At last, the synchronous perturbation learning algorithm (SLA ) Has realized the on-chip training system, and realized the associative memory of words through this training system Hopfield network.