摘要：【中文】本发明涉及一种具有低残留铬离子的金属络合染料的制备方法，包括如下步骤：分别将偶氮化合物与铬络合剂以不同比例反应，得到产物A和产物B，然后将产物A、产物B和偶氮化合物混合并粉碎成浆料反应，过滤除去沉淀后，得到金属络合染料。铬离子在形成产物A和产物B的步骤中，通过过滤随滤液一同去除，从而避免了向最终的金属络合染料溶液中引入铬离子，通过本发明所述制备方法，得到的金属络合染料铬离子含量很低，在10 ppm以下。 【EN】The invention relates to a preparation method of a metal complex dye with low residual chromium ions, which comprises the following steps: respectively reacting azo compounds with chromium complexing agents in different proportions to obtain a product A and a product B, then mixing the product A, the product B and the azo compounds, crushing the mixture into slurry for reaction, and filtering to remove precipitates to obtain the metal complex dye. In the step of forming the product A and the product B, the chromium ions are removed together with the filtrate through filtration, so that the chromium ions are prevented from being introduced into the final metal complex dye solution, and the chromium ion content of the obtained metal complex dye is very low and is below 10 ppm by the preparation method.

摘要：【中文】本发明属于病毒检测领域，具体涉及一种适于快速直扩PCR检测的病毒裂解反应液。本发明所述病毒裂解反应液兼有拭子样品采样液，样品裂解液，扩增反应缓冲液的三种功能，避免了传统核酸检测体系需要核酸提取步骤，时间长，因样品处理量少检测敏感度低等问题；又避免了近几年建立的“一步裂解法”及“免提取直扩方法”中实际采样体积过低造成的敏感度低的缺陷。本发明所述病毒裂解反应液特别适用于人及动物传染病筛查，尤其是高风险的呼吸道病毒的安全筛查。 【EN】The invention belongs to the field of virus detection, and particularly relates to a virus lysis reaction solution suitable for rapid direct-amplification PCR detection. The virus lysis reaction solution has three functions of a swab sample sampling solution, a sample lysate and an amplification reaction buffer solution, and solves the problems that the traditional nucleic acid detection system needs a nucleic acid extraction step, the time is long, the detection sensitivity is low due to less sample processing amount and the like; and the defect of low sensitivity caused by too low actual sampling volume in a one-step cracking method and a hands-free straight-extending method established in recent years is avoided. The virus lysis reaction liquid is particularly suitable for screening infectious diseases of human and animals, in particular to the safety screening of high-risk respiratory viruses.

摘要：【中文】本发明公开了一种NBIOT的频域时频同步方法，包括：进行一次采样，并和本地的时域PSS做匹配滤波后寻找前3大峰值及其对应的3个位置；分别以3个位置为起始位置，从一次采样的数据中取三组数据，二次降采样并作16点FFT得到频域信号；按照5种不同的子载波偏移方式从频域信号中取5种频域数据；将5种频域数据和本地频域PSS信号互相关，得到互相关结果数据；将互相关结果数据做延时滑动自相关并加权累加，得到待判决结果；在待判决结果中找到最大值并和门限比较，超过门限则得到NB‑IoT同步信息；取过门限的峰值位置数据，计算频偏，完成小区时频同步。本发明能够在频域对低信噪比大频偏的场景下的小区进行初始同步，并且一次完成粗同步和精细同步。 【EN】The invention discloses a frequency domain time frequency synchronization method of NBIOT, which comprises the following steps: carrying out primary sampling, and searching the front 3 large peak values and 3 corresponding positions thereof after carrying out matched filtering with the local time domain PSS; taking 3 positions as initial positions respectively, taking three groups of data from the data sampled for the first time, performing secondary down-sampling and performing 16-point FFT to obtain frequency domain signals; taking 5 kinds of frequency domain data from the frequency domain signal according to 5 different subcarrier offset modes; performing cross correlation on the 5 kinds of frequency domain data and the local frequency domain PSS signal to obtain cross correlation result data; performing time-delay sliding autocorrelation on the cross-correlation result data and performing weighted accumulation to obtain a result to be judged; finding a maximum value in a result to be judged, comparing the maximum value with a threshold, and obtaining NB-IoT synchronization information if the maximum value exceeds the threshold; and taking peak position data which passes through a threshold, calculating frequency deviation and completing cell time-frequency synchronization. The invention can perform initial synchronization on the cell under the scene of low signal-to-noise ratio and large frequency offset in the frequency domain, and complete coarse synchronization and fine synchronization at one time.

摘要：【中文】本申请公开了一种光刻机对位方法，涉及半导体制造技术领域。该方法包括利用对位光斑周期性扫描晶圆上的对位标记，获取每条扫描路径对应的强度位置相关性图；根据强度位置相关性图，确定每条扫描路径对应的多重相关系数和晶圆对位信号强度；根据多重相关系数和晶圆对位信号强度，计算出每条扫描路径对应的评估值；将最大评估值对应的扫描路径确定为最佳扫描路径，最佳扫描路径用于定义对位标记的位置；评估值＝多重相关系数*多重相关系数*晶圆对位信号强度；解决了现有方式在选择最佳扫描路径时容易受到异常情况的影响，导致晶圆次品的问题；达到了减少对位信号受损或偏移的影响，提高最佳扫描路径选择的准确度，降低晶圆次品率的效果。 【EN】The application discloses a lithography machine alignment method, and relates to the technical field of semiconductor manufacturing. Periodically scanning alignment marks on a wafer by using alignment light spots to obtain an intensity position correlation diagram corresponding to each scanning path; determining multiple correlation coefficients and wafer alignment signal intensity corresponding to each scanning path according to the intensity position correlation diagram; calculating an evaluation value corresponding to each scanning path according to the multiple correlation coefficients and the wafer alignment signal intensity; determining the scanning path corresponding to the maximum evaluation value as an optimal scanning path, wherein the optimal scanning path is used for defining the position of the alignment mark; evaluating the wafer alignment signal intensity as multiple correlation coefficient; the problem that the existing method is easily influenced by abnormal conditions when the optimal scanning path is selected, so that defective wafers are caused is solved; the effects of reducing the influence of damage or deviation of alignment signals, improving the accuracy of selecting the optimal scanning path and reducing the defective rate of wafers are achieved.

摘要：【中文】本发明公开了一种自注意力深度学习端到端的行人再识别方法，涉及机器学习与模式识别技术领域，用于提升行人再识别的性能，包括以下步骤：1)以三元组图像作为深度学习网络的输入数据，采用预先训练的Resnet50深度网络去除最后一层分类层提取图像的深度特征；2)在深度特征的基础上，通过自注意力网络，进一步获取自注意力特征；3)融合自注意力特征与深度特征，产生更具辨识力的图像特征；4)以多分类任务和验证任务的损失函数共同监督网络的训练，经过多次迭代不断地优化网络模型参数，从而获取用于行人再识别的优选模型。 【EN】The invention discloses an end-to-end pedestrian re-identification method through self-attention deep learning, which relates to the technical field of machine learning and pattern recognition and is used for improving the performance of pedestrian re-identification, and the method comprises the following steps: 1) taking the triple images as input data of a deep learning network, and removing the depth features of the extracted images of the last classification layer by adopting a pre-trained Resnet50 deep network; 2) on the basis of the depth features, further obtaining self-attention features through a self-attention network; 3) fusing the self-attention feature and the depth feature to generate an image feature with better identification power; 4) and (3) supervising the training of the network by using the loss functions of the multi-classification task and the verification task together, and continuously optimizing the parameters of the network model through multiple iterations so as to obtain the optimal model for re-identifying the pedestrian.

摘要：【中文】本发明公开了一种凝胶型硬性接触镜润滑液，按重量份计，包括以下组分，凝胶成0.05~0.2份，保湿剂3.0~20份，油脂成分1.0~5.0份，渗透压调整剂0.1~10份，消毒剂0.05~1.0份，pH调节剂0.2~10份，所述润滑液的pH为5.0~8.0，渗透压为240~290 mosm/kg∙H₂O。本发明制备的凝胶型硬性接触镜润滑液可改善硬性接触镜材质引发的表面疏水性，提高夜间戴镜时的舒适性，又能作为生理泪液，缓解眼睛干涩症状，改善戴镜时的异物感。 【EN】The invention discloses a gel type hard contact lens lubricating liquid which comprises, by weight, 0.05-0.2 parts of gel, 3.0-20 parts of humectant, 1.0-5.0 parts of grease component, 0.1-10 parts of osmotic pressure regulator, 0.05-1.0 part of disinfectant and 0.2-10 parts of pH regulator, wherein the pH of the lubricating liquid is 5.0-8.0, and the osmotic pressure is 240-290 mosm/kg ∙ H₂And O. The gel type hard contact lens lubricating liquid prepared by the invention can improve the surface hydrophobicity caused by hard contact lens materials, improve the comfort when wearing the lens at night, and can be used as physiological tears to relieve the dry eye symptom and improve the foreign body sensation when wearing the lens.

摘要：【中文】本发明公开了一种窄带物联网联合小区搜索及同步方法。本发明一种窄带物联网联合小区搜索及同步方法，包括：接收到数字前端的输出数据，并对其进行数字增益控制；根据上层控制信息判定当前需要执行小区搜索还是小区恢复；若需要执行小区搜索，则根据上层控制信息判定需要执行初始小区搜索还是邻小区搜索。本发明的有益效果：适用于多种小区搜索或同步场景，复杂度降低并减少资源占用。 【EN】The invention discloses a narrowband Internet of things united cell searching and synchronizing method. The invention discloses a narrowband Internet of things united cell searching and synchronizing method, which comprises the following steps: receiving output data of a digital front end and performing digital gain control on the output data; judging whether cell search or cell recovery is required to be executed currently according to the upper control information; and if the cell search needs to be executed, judging whether the initial cell search or the adjacent cell search needs to be executed according to the upper control information. The invention has the beneficial effects that: the method is suitable for various cell searching or synchronous scenes, the complexity is reduced, and the resource occupation is reduced.

摘要：【中文】本发明涉及半导体制造技术领域，具体涉及一种晶圆背面固体颗粒侦测方法。包括以下步骤：对晶圆的参考层进行光刻，在参考层上形成参考图形标记；读取参考图形标记，获得参考图形标记的位置；对准参考图形标记的位置，对晶圆的当前层进行光刻，在当前层上形成测试图形标记；测量在参考层与当前层之间，晶圆的各区域的区域套刻误差；当一区域的区域套刻误差大于阈值范围，确定区域中有热点。本发明通过测量在不同层间，晶圆的套刻误差，根据套刻误差是否符合预定阈值，即可侦测定位即使粒径很小的固体颗粒所在的位置。 【EN】The invention relates to the technical field of semiconductor manufacturing, in particular to a method for detecting solid particles on the back surface of a wafer. The method comprises the following steps: photoetching a reference layer of the wafer, and forming a reference pattern mark on the reference layer; reading the reference pattern mark to obtain the position of the reference pattern mark; aligning the position of the reference pattern mark, photoetching the current layer of the wafer, and forming a test pattern mark on the current layer; measuring the area overlay error of each area of the wafer between the reference layer and the current layer; and when the area overlay error of one area is larger than the threshold range, determining that the area has hot spots. According to the invention, the alignment error of the wafer among different layers is measured, and the position of the solid particle with small particle size can be detected and positioned according to whether the alignment error meets the preset threshold value.

摘要：【中文】本发明公开了一种窄带物联网同频邻小区搜索方法及搜索系统。本发明一种窄带物联网同频邻小区搜索方法，包括：窄带物联网终端接收到无线信号并下变频到基带；数字前端给出一个帧长的时域基带信号，并与本地NPSS序列进行互相关，随后对互相关序列进行延迟自相关；从NPSS相关序列的结果中找到若干峰值，计算各峰值与相关序列均值的比值；计算记录的一组峰均比值的均值和标准差，并据此判定存在NPSS信号的峰值及其对应位置；利用检测到的NPSS序列实现频偏估计，并使用频偏估计结果判定误检的NPSS序列。本发明的有益效果：本发明可以有效降低由于同频邻小区窄带同步信号在时间上重叠引起的高虚警概率，进而提高邻小区搜索性能。 【EN】The invention discloses a method and a system for searching same-frequency adjacent cells of a narrow-band Internet of things. The invention discloses a method for searching same-frequency adjacent cells of a narrow-band Internet of things, which comprises the following steps: the narrowband Internet of things terminal receives the wireless signal and down-converts the wireless signal to a baseband; the digital front end gives a time domain baseband signal with a frame length, and performs cross correlation with a local NPSS sequence, and then performs delay auto-correlation on the cross correlation sequence; finding a plurality of peak values from the result of the NPSS correlation sequence, and calculating the ratio of each peak value to the mean value of the correlation sequence; calculating the mean value and standard deviation of a group of recorded peak-to-average ratio values, and judging that the peak value of the NPSS signal and the corresponding position thereof exist according to the mean value and the standard deviation; and the frequency offset estimation is realized by utilizing the detected NPSS sequence, and the error-detected NPSS sequence is judged by utilizing the frequency offset estimation result. The invention has the beneficial effects that: the invention can effectively reduce the high false alarm probability caused by the overlapping of the same-frequency adjacent cell narrow-band synchronous signals in time, thereby improving the adjacent cell searching performance.

摘要：【中文】本发明NBIOT系统多小区迭代干扰抵消检测方法包括：初始化迭代计数器λ＝0同时产生本地SSS信号集，并提取频域11个符号上12个子载波信号r为频域接收SSS信号；将频域接收SSS信号和本地SSS信号集作互相关；将得到的互相关值作峰值过门限检测；将过门限的频域接收SSS信号作LS信道估计，并进一步得到LS的信道响应；根据得到的信道响应和检测到的该ID本地SSS信号重新生成该ID接收SSS信号；将频域接收SSS信号中减去ID接收SSS信号后作为新的频域接收SSS信号重新进入步骤S20，迭代直到规定的邻小区都检测完成或者达到最大迭代检测次数。本发明能够在多个小区中准确检测到多个邻居小区并降低虚警和漏检概率，同时采用迭代方法能够显著降低邻小区检测SINR门限。 【EN】The invention discloses a NBIOT system multi-cell iterative interference cancellation detection method, which comprises the following steps: initializing an iteration counter lambda to be 0 and simultaneously generating a local SSS signal set, and extracting 12 subcarrier signals r on 11 symbols of a frequency domain as frequency domain receiving SSS signals; cross-correlating the frequency domain received SSS signals with a local SSS signal set; performing peak value threshold-crossing detection on the obtained cross-correlation value; performing LS channel estimation on the frequency domain receiving SSS signals which pass through the threshold, and further obtaining the channel response of LS; regenerating the ID receiving SSS signal according to the obtained channel response and the detected ID local SSS signal; and step S20 is re-entered as a new frequency domain received SSS signal after subtracting the ID received SSS signal from the frequency domain received SSS signal, and the iteration is performed until all the specified neighbor cells are detected completely or the maximum number of iterative detections is reached. The invention can accurately detect a plurality of neighbor cells in a plurality of cells and reduce false alarm and missed detection probability, and can obviously reduce the detection SINR threshold of the neighbor cells by adopting an iteration method.