盐酸伊达比星杂质A的合成工艺优化及其高效液相分析

通过对盐酸伊达比星脱羟基杂质A的合成研究，为盐酸伊达比星的质控标准提供参考。以盐酸伊达比星为起始原料，经过4步反应得到脱羟基杂质A。本文提供了盐酸伊达比星杂质A的制备方法以及高效液相色谱分析方法，为杂质含量控制提供了依据。该工艺可以稳定、快速制备高纯度脱羟基杂质A，制备总收率为57.20%，纯度为57.20%，其结构经¹H NMR， ¹³C NMR和MS(ESI)表征。     

基于深度学习的ELM实时识别研究

基于深度学习的方法，在HL-2A装置上开发出了一套边缘局域模(ELM)实时识别算法。算法使用5200次放电数据(约24.19万数据切片)进行学习，得到一个深度为22层的卷积神经网络。为衡量算法的识别能力，识别了HL-2A装置自2009年实现稳定ELMy H模放电以来所有历史数据(约26000次放电数据)，共识别出1665次H模放电，其中误识别35次，误报率为2.10%。在实际的1634次H模放电中，漏识别4次，漏识别率为0.24%。该误报率和漏报率可以满足ELM实时识别的精度要求。识别算法在实时控制环境下，对单个时间点的平均计算时间为0.46ms，可以满足实时控制的计算速度要求。     

Recent Experimental Results of HL-1M Tokamak and Progress of HL-2A Project

Recent experimental results of HL-1M tokamak and progress of HL-2A project are presented. Strong fishbone instability was observed during off-axis ECRH. This is first observation of the fishbone instability purely driven by energetic electrons produced by ECRH. The MBI was first proposed and demonstrated on HL-1M. Recently new results of MBI experiment were obtained by increasing the pressure of gas. A stair-shape density increment was obtained with high-pressure multi-pulse MBI just like the density evolution behavior during the multipellet injection. It is shown     

Banach空间上有界可逆线性变换逆变换的结构

Fillmore在[1]中得到一个定理:设A,T是Banach空间X上的线性变换,A有界,若Lat(A) Lat(T)且AT=TA,则T是A的多项式.在本文里,以此作为引理,讨论了Banach空间上可逆线性变换A在什么情况下,A-1可表示为A的多项式.本文最主要的结论是定理3.4:设X是Banach空间,A是X上的有界线性变换,且可逆,则A-1是A的多项式当且仅当A-1是A的局部多项式.     

Development of Vacuum Monitoring System in the HL-2A tokamak

HL-2A is the first divertor tokamak in China. The vacuum system is one of the important parts of the HL-2A tokamak, which has to be work well during the physics campaign in 2004. As the tokamak machine is separated from the operation staff during discharges, to guarantee the safe and reliable operation of the vacuum system, it is necessary to watch the vacuum system outside the machine hall with an automatic monitoring system. The paper describes the design and manufacture of the monitoring system.     

The Design for Automation of Pellet-Forming Process in HL-2A Pellet Fueling System

In the ever fusion experiments in SWIP, pellet forming process was carried out through adjusting relative devices by staff member in site, which will make every pellet-forming process slight distinction and will result in pellet difference in shape, size and intensity. In the intervals of HL-2A discharges, staff member have to go site to accomplish the pellet-forming process, this wastes human power and increase the potential danger. So it is necessary to develop a remote control system to perform the pellet-forming process. The control system needs have the features of real-time, reliability and be easy to operate and maintain.     

Vacuum System for HL-2A Tokamak

The vacuum system of today＇s tokamak devices is designed to meet the operational requirements of the experiments. The operation can be divided into five modes, （1） pumping down and leak detecting of the vacuum vessel, （2） baking, （3） plasma-facing component （PFC） conditioning, （4）evacuating and controlling of the particles at plasma edge, （5） plasma discharge experiments.     

Control of LHCD on the HL-2A Tokamak

in order to study non-indutive plasma current production, the lower hybrid current drive（LHCD） experiment on the HL-2A tokamak is carried out. Simultaneously a microcomputer has been used to control the whole LHCD system. During the experiment this year, we can monitor and protect the LHCD system by use of the microcomputer control system, which will imediately switch off the microwave power to be launched into the tokamak if the plasma is disrupted. All this ensure that the microwave is injected into the equipment correctly.     

基于BDF的无约束优化方法的收敛性分析

1.介 绍 在上个世纪的七十年代末、八十年代初,基于常微分方程的优化方法或者说同伦方法是一类与拟牛顿法和共轭梯度法等我们所熟知的优化方法相竞争的重要方法[1-6,8,13,14,16].由于这类方法只是简单地利用现成的数值求解常微分方程的软件包,如CVODE[7]、LSODE[12],对同伦方程(一般是一个常微分方程的初值问题)进行计算,除了一些特殊的病态问题     

覆盖亚临界到临界区的制冷工质物性快速算法

为了在亚临界到临界点这个大范围内保证制冷剂物性计算的快速、稳定、可逆,提出将整个数据区间划分成若干个子区间分别进行隐式拟合,并且保证相邻子区间连接点处的连续与光滑,以及通过求分析解从隐式方程中得到显式物性计算式的方法.以美国国家标准局开发的程序REFPROP6.01的计算结果作为数据源,以R410A为例对该方法做了验证.得出快速物性计算公式的应用范围为饱和温度-60℃至临界温度,与NIST REFPROP 6.01的计算结果进行比较表明,计算速度快1500倍,平均相对偏差小于0.01%.