基于迁移学习的电力缴费终端数据实时高速传输系统  

Real⁃time high⁃speed transmission system of power payment terminal data based on transfer learning

作　　者：王林信 杨鹏 杨波 台树杰 李俊业 WANG Linxin;YANG Peng;YANG Bo;TAI Shujie;LI Junye(State Grid Gansu Electric Power Company,Lanzhou 730050,China)

机构地区：[1]国网甘肃省电力公司,甘肃兰州730050

出　　处：《电子设计工程》2021年第18期35-38,43,共5页Electronic Design Engineering

基　　金：国网甘肃省电力公司信息通信公司科技项目资助(SGGSXT00SJJS1900134)。

摘　　要：为解决传统电力缴费终端数据传输系统存在的传输相似性数据较多,导致系统内存不够、数据传输不完整等问题,设计了基于迁移学习的电力缴费终端数据实时高速传输系统。通过设置FPGA核心控制器,调整数据采集系统,利用XC7K325T-2FFG900芯片采集信号,并交由DDR3 SDRAM处理器对数据进行处理。使用现场转发器为系统供电,确保电池容量。引入迁移学习算法迁移相似数据,并在STORM平台上设计传输流程。由实验结果可知,该系统数据传输完整度最高为0.98,具有良好的传输效果。In order to solve the problems of the traditional power payment terminal data transmission system,such as the lack of system memory and incomplete data transmission,a real⁃time high⁃speed data transmission system based on migration learning is designed.By setting the FPGA core controller and adjusting the data acquisition system,XC7K325T-2FFG900 chip is used to collect the signal,and the data is processed by DDR3SDRAM processor.Use field repeater to power the system and ensure battery capacity.Transfer learning algorithm is introduced to migrate similar data,and the transfer process is designed on STORM platform.The experimental results show that the highest data transmission integrity of the system is 0.98,which has good transmission effect.

关 键 词：迁移学习 电力缴费 终端数据 传输 

分 类 号：TM915[电气工程—电力电子与电力传动]