摘要：【中文】本发明属于制氮机技术领域，具体涉及一种改进的制氮机角座阀，包括阀杆、螺纹外活套、阀芯和软阀座；所述阀杆的末端设置凸台，凸台与螺纹外活套一端的内径相配合，螺纹外活套另一端的内径带有内螺纹，所述阀芯上设有外螺纹，阀芯通过外螺纹与螺纹外活套内径的内螺纹连接，所述阀芯内嵌入软阀座。本发明简化了阀杆与阀芯的连接结构，强化了阀杆与阀芯的连接强度，避免造成阀门故障，导致制氮机因阀门故障停机维修，保障制氮机安全平稳运行。 【EN】The invention belongs to the technical field of nitrogen making machines, and particularly relates to an improved nitrogen making machine angle seat valve which comprises a valve rod, a thread outer loop, a valve core and a soft valve seat; the end of the valve rod is provided with a boss which is matched with the inner diameter of one end of the thread outer loop, the inner diameter of the other end of the thread outer loop is provided with an internal thread, the valve core is provided with an external thread, the valve core is connected with the internal thread of the inner diameter of the thread outer loop through the external thread, and the valve core is embedded into a soft valve seat. The invention simplifies the connecting structure of the valve rod and the valve core, strengthens the connecting strength of the valve rod and the valve core, avoids causing valve failure, and ensures the nitrogen making machine to be stopped and maintained due to the valve failure, thereby ensuring the safe and stable operation of the nitrogen making machine.

摘要：【中文】本发明提供了一种结合BERT模型的图像描述生成方法，首先提取图像的特征向量，对特征向量进行压缩、维度扩充，其次，用外部语料数据扩充词典，然后，将特征向量和词典输入基于端到端加入注意力机制的图像描述生成模型，生成弱语义描述语句A，最后，通过BERT模型对弱语义描述语句A进行语义调整，获取完整的图像描述语句。本发明通过对特征向量进行压缩与维度扩充，增强图像数据的特征表达含义；利用基于端到端并加入注意力机制的图像描述生成模型生成弱语义图像描述语句，同时对于词汇不足的问题采用应用外部语料数据扩充词典的方式，增强语义含义，使所生成的图像描述能更准确的表征图像的内容，具有更丰富的语义。 【EN】The invention provides an image description generation method combined with a BERT model, which comprises the steps of firstly extracting a feature vector of an image, compressing and dimension expanding the feature vector, secondly expanding a dictionary by using external corpus data, then inputting the feature vector and the dictionary into an image description generation model based on an end-to-end attention adding mechanism to generate a weak semantic description statement A, and finally, carrying out semantic adjustment on the weak semantic description statement A through the BERT model to obtain a complete image description statement. The feature expression meaning of the image data is enhanced by compressing and dimension expanding the feature vector; an image description generation model based on end-to-end and with an attention mechanism is used for generating weak semantic image description sentences, meanwhile, a dictionary is expanded by applying external corpus data for solving the problem of insufficient vocabulary, semantic meaning is enhanced, and the generated image description can accurately represent the content of the image and has richer semantics.

摘要：【中文】本发明针对跨媒体检索问题，提出了一种基于跨媒体统一表征模型的跨媒体检索方法，包括以下步骤：(1)跨媒体数据库构建，建立面向政务新闻领域大跨媒体数据库；(2)跨媒体数据预处理，文本、图像、视频和音频等数据的输入预处理；(3)跨媒体数据原域特征提取，跨媒体数据的原域特征向量提取；(4)跨媒体数据统一表征，跨媒体数据在共同表示空间的特征向量提取；(5)数据检索语义相似度计算与排序，检索目标数据与跨媒体数据库中数据语义相似度计算，并排序输出结果。本发明不仅提出了一种支持四种媒体数据的相互检索方法，同时提出多种媒体数据的统一表征模型，提高了跨媒体语义检索精度，具有关阔的应用前景。 【EN】The invention provides a cross-media retrieval method based on a cross-media uniform representation model aiming at the problem of cross-media retrieval, which comprises the following steps: (1) constructing a cross-media database, and establishing a large-cross-media database facing the government affair news field; (2) cross-media data preprocessing, input preprocessing of data such as texts, images, videos and audios; (3) extracting original domain features of cross-media data, and extracting original domain feature vectors of the cross-media data; (4) uniformly representing the cross-media data, and extracting feature vectors of the cross-media data in a common representation space; (5) and calculating and sequencing the semantic similarity of the data, calculating the semantic similarity of the data of the retrieval target and the data in the cross-media database, and sequencing to output results. The invention not only provides a mutual retrieval method supporting four media data, but also provides a unified representation model of multiple media data, improves the cross-media semantic retrieval precision, and has a broad application prospect.

摘要：【中文】本发明提供一种螺栓状态的诊断方法及系统，所述方法包括：采集待诊断螺栓的第一振动信号和第二振动信号；根据第一振动信号和第二振动信号的信号差，获得信号差的自相关函数；对信号差的自相关函数进行小波包N层分解，提取频率成分信号；根据每个频率成分信号，获得每个频率信号对应的信号能量；根据各个频率信号对应的信号能量，获得诊断特征向量；将诊断特征向量输入至螺栓状态诊断模型，输出待诊断螺栓的状态诊断结果，螺栓状态诊断模型是基于螺栓训练样本的样本特征向量以及对应的状态标签进行训练后获得的。所述装置用于执行上述方法。本发明实施例提供的螺栓状态的诊断方法及系统，提高了螺栓状态诊断的可靠性。 【EN】The invention provides a method and a system for diagnosing bolt states, wherein the method comprises the following steps: acquiring a first vibration signal and a second vibration signal of a bolt to be diagnosed; obtaining an autocorrelation function of the signal difference according to the signal difference between the first vibration signal and the second vibration signal; carrying out wavelet packet N-layer decomposition on the autocorrelation function of the signal difference, and extracting frequency component signals; obtaining signal energy corresponding to each frequency signal according to each frequency component signal; obtaining a diagnosis characteristic vector according to the signal energy corresponding to each frequency signal; and inputting the diagnosis characteristic vector into a bolt state diagnosis model, and outputting a state diagnosis result of the bolt to be diagnosed, wherein the bolt state diagnosis model is obtained after training based on the sample characteristic vector of the bolt training sample and the corresponding state label. The device is used for executing the method. The bolt state diagnosis method and the bolt state diagnosis system provided by the embodiment of the invention improve the reliability of bolt state diagnosis.