提出了一种利用FPGA对生物神经元网络进行硬件仿真的方法。该方法充分考虑了多进程流水线模型中各神经元状态输出的时序问题,设计了节点选择器完成对流水线数据通路输出数据的保存和选择功能,实现了多状态耦合情况下精确的仿真方法。最后,利用该方法对Morris-Lecar神经元网络模型进行了FPGA硬件仿真,再现了Morris-Lecar神经元网络的非线性动力学特性。
This paper proposed a method of biological neural network FPGA hardware simulation.The method fully considered the timing of each neuron state output in the multi-process pipeline model,designed the node selector to complete the pipeline data path output data saving and selection,and implemented the precise simulation of the multi-state coupling model.Finally,the methods implemented the FPGA hardware simulation of Morris-Lecar neural network model,and achieved the Morris-Lecar neural network's nonlinear dynamics.