摘要
大型氦低温系统广泛应用于各类大科学装置中,运行中往往会产生热脉冲,通过负载端传导给制冷系统,对制冷系统产生热冲击。为了研究和应对热冲击,建立了一种多变量控制策略并得到了相关仿真和实验结果。首先以真实系统为基础建立了氦低温系统的动态仿真模型,同时建立了一个基于模糊神经网络的多变量协同控制策略,并将其应用在仿真液化器模型和一个真实的氦透平制冷系统上,得到了低温系统降温过程和控制过程的仿真和实验数据。仿真和实验结果显示本策略的偏差积分量为0.016 5,下降时间为102 s,上升时间为112 s。普通PID的的偏差积分量为0.026 9,下降时间为154 s,上升时间为170 s。通过仿真和实验过程的比较,验证了本文建立的动态仿真模型具有可用的精度,证明了本策略具有较好的控制效果。
Helium large-scare cryogenic system is widely used in the fields of superconducting, nuclear fusion energy and high-energy physics. These systems generate heat pulse load, and helium refrigerators and liquefiers have to handle such heat loads. Thus, a multi-strategy controller is proposed in this paper, and some simulation and experiment results are obtained. At first, a dynamic simulation model of a helium liq- uefier is developed. Then, a multi-strategy controller based on fuzzy neural network is proposed, and is ap- plied on the simulation model and a real helium refrigerator. To study the control effect, an operation process including cool-down, steady-state and pulse of heat is described. The simulation and experiment re- suits show that the integrated quantity of deviation is 0. 016 5, the fall time is 102 s, the rise time is 112 s for the FNN controlling system, and the integrated quantity of deviation is 0. 026 9, the fall time is 154 s, the rise time is 170 s for the PID controlling system. The simulation and experiment results indicate that the simulation model has acceptable validity and accuracy, and the fuzzy neural network control system has bet- ter control effect than traditional PID control.
出处
《低温工程》
CAS
CSCD
北大核心
2017年第1期26-30,58,共6页
Cryogenics
基金
国家自然科学基金项目(51106169)资助
关键词
大型低温系统
氦制冷机
氦液化器
模糊控制
神经网络
large-scare cryogenic system
helium refrigerator
helium quefier
fuzzy control
neuralnetwork