面向高效微分求解电解质晶体管基储备池计算系统

杜明阳; 韩孟涛; 邢千叶; 崔航源; 万昌锦

doi:10.12263/DZXB.20260257

您当前的位置：

首页 >

文章列表页 >

面向高效微分求解电解质晶体管基储备池计算系统

更新时间：2026-06-09

- 面向高效微分求解电解质晶体管基储备池计算系统
- Electrolyte Transistor-Based Reservoir Computing for High-Efficiency Differential Solving System
- 电子学报 2026年页码：1-11
- 作者机构：
  
  南京大学电子科学与工程学院，江苏南京 210023
- 作者简介：
  
  [ "杜明阳男，2005年1月出生于四川省遂宁市。现为南京大学电子科学与工程学院本科生。中国电子学会会员编号：E190199051A。E-mail: dumingyang@smail.nju.edu.cn" ]
  [ "韩孟涛男，现为南京大学电子科学与工程学院研究生。主要研究方向为微电子器件。E-mail: nymph@smail.nju.edu.cn" ]
  [ "邢千叶女，现为南京大学电子科学与工程学院研究生。E-mail: 652025230023@smail.nju.edu.cn" ]
  [ "崔航源男，现为南京大学电子科学与工程学院研究生。E-mail: cuihy2017@163.com" ]
  [ "万昌锦男，现为南京大学电子科学与工程学院副教授。主要研究方向为神经形态电子器件。E-mail: cjwan@nju.edu.cn" ]
- 基金信息：
  
  国家自然科学基金(92364106;U25A20500);江苏省研究生科研与实践创新计划项目（No.KYCX25_0278,No.KYCX25_0284）
- DOI：10.12263/DZXB.20260257
  中图分类号： TN386;TP183
- 收稿：2026-04-09，
  
  录用：2026-04-19，
  
  网络首发：2026-06-09，
- 稿件说明：
移动端阅览
杜明阳, 韩孟涛, 邢千叶, 等. 面向高效微分求解电解质晶体管基储备池计算系统[J/OL]. 电子学报, 2026,1-11.

DU Mingyang, HAN Mengtao, XING Qianye, et al. Electrolyte Transistor-Based Reservoir Computing for High-Efficiency Differential Solving System[J/OL]. ACTA ELECTRONICA SINICA, 2026, 1-11.
杜明阳, 韩孟涛, 邢千叶, 等. 面向高效微分求解电解质晶体管基储备池计算系统[J/OL]. 电子学报, 2026,1-11. DOI： 10.12263/DZXB.20260257.

DU Mingyang, HAN Mengtao, XING Qianye, et al. Electrolyte Transistor-Based Reservoir Computing for High-Efficiency Differential Solving System[J/OL]. ACTA ELECTRONICA SINICA, 2026, 1-11. DOI： 10.12263/DZXB.20260257.

摘要

物理储备池计算（Physical Reservoir Computing，PRC）凭借其低训练成本与高并行效率，已成为时序信号处理与复杂动力学系统建模的新兴范式。该方法主要依赖于器件物理特性中的非线性与短时记忆动力学特性，实现对时序信号的高维映射，获得丰富的储备池状态，从而实现对复杂状态方程较为精确的求解及预测。PRC仅需求解输出层权重、推理功耗低、少样本数据驱动等优势，为在端侧硬件上实现混沌微分方程与偏微分方程的实时求解提供了全新路径。然而，如何将高阶复杂时序任务的处理算法与物理器件的本征调控机制有效适配，仍是制约其实际部署的关键瓶颈。本文报道了一种基于铟镓锌氧化物（Indium Gallium Zinc Oxide，IGZO）电解质晶体管的高效物理储备池计算系统，成功实现了混沌含时微分方程序列的求解预测，并首次报道了基于物理计算系统的偏微分方程求解。借助IGZO器件的低功耗特性、连续非线性映射能力，设计了轻量化的储备池训练推理与方程求解协同算法，对IGZO晶体管的电学响应与离子迁移动力学行为进行了精确建模表征，并在现场可编程门阵列（Field Programmable Gate Array，FPGA）平台上完成了系统级仿真验证。资源评估表明，相较于传统四阶Runge-Kutta算法实现，本方案在保持同等精度的前提下，数字信号处理单元（Digital Signal Processor，DSP）与触发器（Flip-Flop，FF）资源占用分别降低约64%与50%，整体逻辑开销显著优化。求解性能方面，实验选取Mackey-Glass、Lorenz这两类混沌微分方程及一维热扩散方程为基准，与主流数值方法（Runge-Kutta）及现有物理计算平台进行对比。结果表明，该系统在混沌微分方程求解中展现出约11 μs的延迟，相比Runge-Kutta方法功耗降低15%，归一化均方根误差（Normalized Root Mean Square Error，NRMSE≈0.04）与现有大多数高效物理计算系统相当；针对一维热扩散偏微分方程，计算时间较二阶Runge-Kutta缩短至1/400，且精度方面NRMSE≈0.008也与之相当。本研究不仅验证了IGZO基PRC系统在通用微分方程求解中的高效性与准确性，更为下一代轻量化、低功耗的端侧科学计算与神经形态硬件平台提供了可拓展的架构基础与工程实践参考。

Abstract

Physical reservoir computing (PRC)

characterized by its low training cost and high parallel efficiency

has emerged as a promising paradigm for temporal signal processing and complex dynamical system modeling. By leveraging the intrinsic nonlinearity and short-term memory dynamics of physical devices

PRC maps temporal inputs into a high-dimensional state space

generating rich reservoir states that enable accurate solving and prediction of complex state equations. With advantages such as readout-layer-only training

low inference power consumption

and few-shot data driven

PRC offers a novel pathway for real-time solving of chaotic differential and partial differential equations on edge hardware. However

effectively co-designing high-order temporal processing algorithms with the intrinsic modulation mechanisms of physical devices remains a key bottleneck hindering practical deployment. Here

we report a highly efficient PRC system based on indium gallium zinc oxide (IGZO) electrolyte transistors

achieving the solving and prediction of time-dependent chaotic differential equations and partial differential equations. Leveraging the low-power characteristics and continuous nonlinear mapping capability of IGZO devices

we designed a lightweight algorithm co-optimizing reservoir training

inference

and equation solving. The electrical response and ion migration dynamics of the IGZO transistors were accurately modeled and characterized

followed by system-level validation on the field programmable gate array (FPGA). Resource evaluation demonstrates that

compared to a conventional fourth-order Runge-Kutta (RK4) implementation

our approach reduces digital signal processor (DSP) and flip-flop utilization by approximately 64% and 50%

respectively

while maintaining equivalent accuracy. In performance benchmarks using the Mackey-Glass equation

Lorenz chaotic sequence

and one-dimensional heat diffusion equation

our system achieves an inference latency of 11 μs and reduces power consumption by 15%

with a normalized root mean square error (NRMSE≈0.04) comparable to the Runge-Kutta method. For the partial differential equation (PDE) task

computation time is reduced to 1/400 of that required by the second-order Runge-Kutta (RK2) method

while maintaining a comparable NRMSE≈0.008. This work not only validates the efficiency and accuracy of IGZO-based PRC in solving general differential equations

but also provides a scalable architectural foundation and practical engineering reference for next-generation lightweight

low-power edge scientific computing and neuromorphic hardware platforms.

关键词

Keywords

references

Walter W . Ordinary differential equations [M ] . New York : Springer , 1998 . DOI: 10.1007/978-1-4612-0601-9_5 http://dx.doi.org/10.1007/978-1-4612-0601-9_5

Driver R D . Ordinary and delay differential equations [M ] . New York : Springer , 1977 . DOI: 10.1007/978-1-4684-9467-9_5 http://dx.doi.org/10.1007/978-1-4684-9467-9_5

唐卷 , 罗凯 , 曾晓青 , 等 . 用于求解约束机械多体系统高指标微分代数方程的神经网络构建方法及系统 : 中国 , CN119558357A [P/OL ] . 2025-12-05 [ 2025-12-20 ] . http://epub.cnipa.gov.cn/patent/CN119558357A http://epub.cnipa.gov.cn/patent/CN119558357A .

Tang Juan , Luo Kai , Zeng Xiaoqing , et al . Neural network construction method and system for solving high-index differential-algebraic equations of constrained mechanical multibody systems : China , CN119558357A [P/OL ] . 2025-12-05 [ 2025-12-20 ] . http://epub.cnipa.gov.cn/patent/CN119558357A http://epub.cnipa.gov.cn/patent/CN119558357A .

王志刚 . 基于差异演化算法的非线性方程组求解 [J ] . 计算机工程与应用 , 2010 , 46 ( 4 ): 54 - 55 .

Wang Zhigang . Solving nonlinear systems of equations based on differential evolution [J ] . Computer Engineering and Applications , 2010 , 46 ( 4 ): 54 - 55 . (in Chinese)

Schilling T E , Raddum H . Solving equation systems by agreeing and learning [M ] // Arithmetic of Finite Fields . Berlin, Heidelberg : Springer , 2010 : 151 - 165 . DOI: 10.1007/978-3-642-13797-6_11 http://dx.doi.org/10.1007/978-3-642-13797-6_11

Hamming R W . Hybrid computers: Electronic analog and hybrid computers . granino A. Korn and Theresa M. Korn. McGraw-hill, New York, 1964. xxiv + 584 pp. Illus . $17.50 [J ] . Science , 1965 , 148 ( 3668 ): 356 - 361 . DOI: 10.1126/science.148.3668.356-b http://dx.doi.org/10.1126/science.148.3668.356-b

Tomović R , Karplus W J . High-speed analog computers [M ] . New York : Wiley , 1962 .

Bodenstedt E . Analog computer for the differential equation y ″+ f ( x ) y + g ( x )=0 [J ] . Review of Scientific Instruments , 1956 , 27 ( 4 ): 218 - 221 . DOI: 10.1063/1.1715525 http://dx.doi.org/10.1063/1.1715525

张庆平 . 一种流水线结构的分数阶微分方程组求解电路和方法 : 中国 , CN119474623A [P ] . 2025-02-18 .

Zhang Qingping . Pipeline-structured circuit and method for solving fractional-order differential equation systems : China , CN119474623A [P ] . 2025-02-18 .

Awaad T A , Elbehery A M , Abdelhamid A , et al . An ultrafast neural network-based hardware acceleration for nonlinear systems' simulators [J ] . Computers & Electrical Engineering , 2019 , 79 : 106452 . DOI: 10.1016/j.compeleceng.2019.106452 http://dx.doi.org/10.1016/j.compeleceng.2019.106452

YuLianfeng , ZhangTeng , HanYang , et al . High-concurrency tri-mode memristor-based ordinary differential equation solver [J ] . Nature Communications , 2026 , 17 ( 1 ): 1373 . DOI: 10.1038/s41467-025-68122-4 http://dx.doi.org/10.1038/s41467-025-68122-4

Zhong Yanan , Tang Jianshi , Li Xinyi , et al . Dynamic memristor-based reservoir computing for high-efficiency temporal signal processing [J ] . Nature Communications , 2021 , 12 ( 1 ): 408 . DO I: 10.1038/s41467-020-20692-1 http://dx.doi.org/10.1038/s41467-020-20692-1

Liang Xiangpeng , Zhong Yanan , Tang Jianshi , et al . Rotating neurons for all-analog implementation of cyclic reservoir computing [J ] . Nature Communications , 2022 , 13 ( 1 ): 1549 . DOI: 10.1038/s41467-022-29260-1 http://dx.doi.org/10.1038/s41467-022-29260-1

Yang Yang , Cui Hangyuan , Ke Shuo , et al . Reservoir computing based on electric-double-layer coupled InGaZnO artificial synapse [J ] . Applied Physics Letters , 2023 , 122 ( 4 ): 043508 . DOI: 10.1063/5.0137647 http://dx.doi.org/10.1063/5.0137647

Cui Hangyuan , Xiao Yu , Yang Yang , et al . A bioinspired in-materia analog photoelectronic reservoir computing for human action processing [J ] . Nature Communications , 2025 , 16 ( 1 ): 2263 . DOI: 10.1038/s41467-025-56899-3 http://dx.doi.org/10.1038/s41467-025-56899-3

Tanaka G , Yamane T , Héroux J , et al . Recent advances in physical reservoir computing: A review [J ] . Neural Networks , 2019 , 115 : 100 - 123 . DOI: 10.1016/j.neunet.2019.03.005 http://dx.doi.org/10.1016/j.neunet.2019.03.005

Moon J , Ma Wen , Shin J H , et al . Temporal data classification and forecasting using a memristor-based reservoir computing system [J ] . Nature Electronics , 2019 , 2 ( 10 ): 480 - 487 . DOI: 10.1038/s41928-019-0313-3 http://dx.doi.org/10.1038/s41928-019-0313-3

Guo Dongkai , Kapur O , Dai Peng , et al . Reservoir computing using back-end-of-line SiC-based memristors [J ] . Materials Advances , 2023 , 4 ( 21 ): 5305 - 5313 .. DOI: 10.1039/d3ma00141e http://dx.doi.org/10.1039/d3ma00141e

Chen Mingze , An Xiaoqiu , Ki S J , et al . Nanoelectronics-enabled reservoir computing hardware for real-time robotic controls [J ] . Science Advances , 2025 , 11 ( 13 ): eadu2663 . DOI: 10.1126/sciadv.adu2663 http://dx.doi.org/10.1126/sciadv.adu2663

Kim S , Yoon J , Kim H D , et al . Carbon nanotube synaptic transistor network for pattern recognition [J ] . ACS Applied Materials & Interfaces , 2015 , 7 ( 45 ): 25479 - 25486 . DOI: 10.1021/acsami.5b08541 http://dx.doi.org/10.1021/acsami.5b08541

Milano G , Pedretti G , Montano K , et al . In materia reservoir computing with a fully memristive architecture based on self-organizing nanowire networks [J ] . Nature Materials , 2022 , 21 ( 2 ): 195 - 202 . DOI: 10.1038/s41563-021-01099-9 http://dx.doi.org/10.1038/s41563-021-01099-9

Yu Jie , Li Yi , Sun Wenxuan , et al . Energy efficient and robust reservoir computing system using ultrathin (3.5 nm) ferroelectric tunneling junctions for temporal data learning [C ] // 2021 Symposium on VLSI Technology . Kyoto, Japan : IEEE , 2021 : 1 - 2 .

Chen R T Q , Rubanova Y , Bettencourt J , et al . Neural ordinary differential equations [C ] // Proceedings of the 32nd International Conference on Neural Information Processing Systems . Red Hook : Curran Associates Inc. , 2018 : 6572 - 6583 .

Backus J . Can programming be liberated from the von Neumann style : A functional style and its algebra of programs [J ] . Communications of the ACM , 1978 , 21 ( 8 ): 613 - 641 . DOI: 10.1145/359576.359579 http://dx.doi.org/10.1145/359576.359579

Marković D , Mizrahi A , Querlioz D , et al . Physics for neuromorphic computing [J ] . Nature Reviews Physics , 2020 , 2 ( 9 ): 499 - 510 . DOI: 10.1038/s42254-020-0208-2 http://dx.doi.org/10.1038/s42254-020-0208-2

Furber S . To build a brain [J ] . IEEE Spectrum , 2012 , 49 ( 8 ): 44 - 49 . DOI: 10.1109/mspec.2012.6247562 http://dx.doi.org/10.1109/mspec.2012.6247562

Indiveri G , Liu S C . Memory and information processing in neuromorphic systems [J ] . Proceedings of the IEEE , 2015 , 103 ( 8 ): 1379 - 1397 . DOI: 10.1109/jproc.2015.2444094 http://dx.doi.org/10.1109/jproc.2015.2444094

Du Chao , Cai Fuxi , Zidan M A , et al . Reservoir computing using dynamic memristors for temporal information processing [J ] . Nature Communications , 2017 , 8 ( 1 ): 2204 . DOI: 10.1038/s41467-017-02337-y http://dx.doi.org/10.1038/s41467-017-02337-y

Xu Yinhao , Gottwald G A , Kuncic Z . Dynamic reservoir computing with physical neuromorphic networks [C ] // 2025 International Joint Conference on Neural Networks (IJCNN) . Piscataway : IEEE , 2025 : 1 - 8 . DOI: 10.1109/ijcnn64981.2025.11229247 http://dx.doi.org/10.1109/ijcnn64981.2025.11229247

Liu Jia , Wu Qiuhao , Sui Xiubao , et al . Research progress in optical neural networks: Theory, applications and developments [J ] . PhotoniX , 2021 , 2 ( 1 ): 5 . DOI: 10.1186/s43074-021-00026-0 http://dx.doi.org/10.1186/s43074-021-00026-0

Smerieri A , Duport F , Paquot Y , et al . Towards fully analog hardware reservoir computing for speech recognition [J ] . AIP Conference Proceedings , 2012 , 1479 ( 1 ): 1892 - 1895 .

Appeltant L , Soriano M C , Van der Sande G , et al . Information processing using a single dynamical node as complex system [J ] . Nature Communications , 2011 , 2 : 468 . DOI: 10.1038/ncomms1476 http://dx.doi.org/10.1038/ncomms1476

Ganguly S , Ghosh A W . Building reservoir computing hardware using low energy-barrier magnetics [C ] // International Conference on Neuromorphic Systems 2020 . New York : ACM , 2020 : 17 . DOI: 10.1145/3407197.3407217 http://dx.doi.org/10.1145/3407197.3407217

Zhu Yixin , Peng Baocheng , Zhu Li , et al . IGZO nanofiber photoelectric neuromorphic transistors with indium ratio tuned synaptic plasticity [J ] . Applied Physics Letters , 2022 , 121 ( 13 ): 133502 . DOI: 10.1063/5.0109772 http://dx.doi.org/10.1063/5.0109772

Wan Changjin , Zhu Liqiang , Wan Xiang , et al . Organic/inorganic hybrid synaptic transistors gated by proton conducting methylcellulose films [J ] . Applied Physics Letters , 2016 , 108 ( 4 ): 043508 . DOI: 10.1063/1.4941080 http://dx.doi.org/10.1063/1.4941080

Chua K W , Loy K C , Idris R . Adaptive time-stepping for Runge-Kutta methods for ordinary differential equations [J ] . Universiti Malaysia Terengganu Journal of Undergraduate Research , 2021 , 3 ( 1 ): 25 - 36 . DOI: 10.46754/umtjur.2021.01.004 http://dx.doi.org/10.46754/umtjur.2021.01.004

Bhattacharya S , Chakraborty D . Design and analysis of a hardware accelerator with FPU-based Runge-Kutta solvers [C ] // 2023 International Conference on Electrical, Communication and Computer Engineering (ICECCE) . Dubai, United Arab Emirates : IEEE , 2023 : 1 - 6 . DOI: 10.1109/icecce61019.2023.10442325 http://dx.doi.org/10.1109/icecce61019.2023.10442325

Liu Keqin , Dang Bingjie , Zhang Teng , et al . Multilayer reservoir computing based on ferroelectric α-In(2) Se(3) for hierarchical information processing [J ] . Advanced Materials , 2022 , 34 ( 48 ): e2108826 . DOI: 10.1002/adma.202108826 http://dx.doi.org/10.1002/adma.202108826

Choi S , Shin J , Park G , et al . 3D-integrated multilayered physical reservoir array for learning and forecasting time-series information [J ] . Nature Communications , 2024 , 15 ( 1 ): 2044 . DOI: 10.1038/s41467-024-46323-7 http://dx.doi.org/10.1038/s41467-024-46323-7

Gaurav A , Song Xiaoyao , Manhas S K , et al . Nano-ionic solid electrolyte FET-based reservoir computing for efficient temporal data classification and forecasting [J ] . ACS Applied Materials & Interfaces , 2025 , 17 ( 11 ): 17590 - 17598 . DOI: 10.1021/acsami.5c00092 http://dx.doi.org/10.1021/acsami.5c00092

Yang Biao , Zhang Weifeng , Li Pengfei , et al . Ultra-robust Y-doped hafnium oxide ferroelectric memristors for intelligent edge computing [J ] . Science China Materials , 2026 , 69 ( 5 ): 2904 - 2912 . DOI: 10.1007/s40843-025-3817-3 http://dx.doi.org/10.1007/s40843-025-3817-3

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

储备池计算概述

一种信号外推和层级学习SAR超分辨率方法

基于国密SM9的分层标识签名方案

基于超表面的低副瓣高口径效率反射阵列天线

基于自适应群体掩码图卷积网络的行人轨迹预测