晶体生长炉的PID神经网络温度控制算法  

Study on PID Neural Network Temperature Control Algorithm for Crystal Growth Furnace

作　　者：仲元昌[1] 郭耿涛[1] 贾年龙[1] 赵贞贞[1] 

机构地区：[1]重庆大学通信工程学院通信与测控中心,重庆400044

出　　处：《人工晶体学报》2010年第5期1302-1307,共6页Journal of Synthetic Crystals

基　　金：国家自然科学基金(No.50875272);重庆市自然科学基金资助项目(CSTC;2008BB2340)

摘　　要：本文采用了数字PID与神经网络相融合的晶体生长炉温度测控方法,设计了生长晶体的PID神经网络温度控制器,并用FPGA(Field Programmable Gate Array)实现了PID神经网络的并行结构;神经网络采用符合32位IEEE754的单精度浮点数运算单元。理论分析和仿真结果表明:该PID神经网络温度控制器具有很高的控制精度和响应速度,有利于提高晶体生长炉的控温精度和加热效率。A PID neural network controller of crystal growth temperature was designed using a digital PID and neural network fusion method of crystal growth furnace temperature measurement and control.A parallel architecture of the PID neural network was implemented using FPGA.The 32-bit IEEE754 single-precision floating-point computing unit is used in the neural network.Theoretical analysis and simulation results show that the neural network PID temperature controller with high control accuracy and response speed,helping to make crystal growth furnace temperature control accuracy and improve heating efficiency.

关 键 词：现场可编程门阵列 晶体生长炉 神经网络 浮点 

分 类 号：O78[理学—晶体学]