一种Sigmoid为激活函数的高精度ANN电路的设计  

Design of a High-Precision ANN Circuit with Sigmoid Activation Function

作　　者：何从蓉 邢世威 吴晨健[1] HE Congrong;XING Shiwei;WU Chenjian(School of Electronic and Information Engineering,Soochow University,Suzhou Jiangsu 215006,China)

机构地区：[1]苏州大学电子信息学院,江苏苏州215006

出　　处：《电子器件》2024年第3期617-622,共6页Chinese Journal of Electron Devices

基　　金：国家自然科学基金项目(61801321)。

摘　　要：提出了一款以Sigmoid为激活函数的人工神经网络(ANN)电路,实现对目标数据的分类,基于TSMC 180 nm CMOS工艺完成了电路设计、版图设计、仿真与流片测试,芯片面积为0.3375 mm^(2),1.8 V电源电压下电路功耗为216μW。3块芯片的测试结果显示,电路对目标数据的分类准确率为97.5%,误差在(-3.2%,3.27%)之间,平均误差为-1.51%。人工神经网络电路含有3个网络层,由15个Sigmoid激活函数神经元电路组成,两输入Sigmoid激活函数神经元电路由两个加权电路与一个Sigmoid激活函数电路组成。后仿真结果显示,在1.8 V电源电压下,神经元电路的功耗为8.02μW,最大噪声为√Hz,版图面积为0.001 mm^(2),最大相对误差为1.76%。An artificial neural network(ANN)circuit with Sigmoid activation function(AF)is proposed to realize the classification of target data.Based on the TSMC 180 nm CMOS process,the circuit design,layout design,simulation and chip test are completed.The chip area is 0.3375 mm2,and the circuit power consumption is 216μW at 1.8 V supply voltage.The test results of three chips show that the classification accuracy of the circuit for target data is 97.5%,the error is between-3.2% and 3.27%,and the average error is-1.51%.The ANN circuit consists of 3 network layers and is composed of 15 Sigmoid AF neuron circuits.The two-inputs Sigmoid AF neuron circuit is composed of two weighted circuits and a Sigmoid AF circuit.The post-simulation results show that the power consumption of neuron circuit is 8.02μW at 1.8 V supply voltage,the maximum noise is 80 pA/Hz,the layout area is 0.001 mm2,and the maximum relative error is 1.76%.

关 键 词：Sigmoid神经元电路 CMOS 人工神经网络电路 

分 类 号：TN702[电子电信—电路与系统] TP183[自动化与计算机技术—控制理论与控制工程]