摘要：【中文】一种聚醚醚酮复合材料及其制备方法，由包括聚醚醚酮55～90质量份、锌铝合金5～30质量份，石墨5～15质量份、氧化石墨烯0.3～1质量份、以及加工助剂制得；锌铝合金由锌90.5～91.1质量份、铝8.5～8.6质量份、铜1.1～1.2质量份经熔炼‑氮气雾化法制成粉末，再经湿式球磨法制得片层状合金粉末；复合材料包括以下步骤制得：将锌铝合金放入季铵盐表面活性剂水溶液中，超声分散，过滤，洗涤后烘干；将氧化石墨烯溶解在去离子水中，锌铝合金分散在去离子水中，将氧化石墨烯溶液滴加至锌铝合金分散液中，制得氧化石墨烯/锌铝合金复合体；将聚醚醚酮、氧化石墨烯/锌铝合金复合体，石墨以及加工助剂混合后在100～120℃温度下干燥3～4小时；经混料机混料后在380～400℃温度下模压成型。 【EN】The polyether-ether-ketone composite material is prepared from 55-90 parts by mass of polyether-ether-ketone, 5-30 parts by mass of zinc-aluminum alloy, 5-15 parts by mass of graphite, 0.3-1 part by mass of graphene oxide and a processing aid; the zinc-aluminum alloy is prepared by preparing powder from 90.5-91.1 parts by mass of zinc, 8.5-8.6 parts by mass of aluminum and 1.1-1.2 parts by mass of copper by a smelting-nitrogen atomization method, and preparing lamellar alloy powder by a wet ball milling method; the composite material is prepared by the following steps: putting the zinc-aluminum alloy into a quaternary ammonium salt surfactant aqueous solution, performing ultrasonic dispersion, filtering, washing and drying; dissolving graphene oxide in deionized water, dispersing a zinc-aluminum alloy in the deionized water, and dropwise adding a graphene oxide solution into a zinc-aluminum alloy dispersion liquid to obtain a graphene oxide/zinc-aluminum alloy composite; mixing polyether-ether-ketone, graphene oxide/zinc-aluminum alloy composite, graphite and a processing aid, and drying at the temperature of 100-120 ℃ for 3-4 hours; and (3) mixing the materials by a mixer, and then performing compression molding at the temperature of 380-400 ℃.

摘要：【中文】本发明涉及一种用于涂料检测的连续式进料装置，以解决目前可实现连续进料时的检测设备存在检测效果较差的问题。该用于涂料检测的连续式进料装置，包括支撑架，支撑架上设有环体，环体为环状结构，其内表面光滑，环体底部设有与其密封连接的底板，底板为透明材料，环体内卡接有与底板平行设置的调节板，调节板外圈设有密封圈，调节板上表面的左右两侧设有拉杆，环体左侧壁底部设有进料管，环体右侧壁底部设有出料管；支撑架底部设有检测设备，检测设备的摄像头穿过支撑架上的通孔并对准底板的正下方。 【EN】The invention relates to a continuous feeding device for paint detection, which aims to solve the problem of poor detection effect of detection equipment in the process of realizing continuous feeding at present. The continuous feeding device for paint detection comprises a support frame, wherein a ring body is arranged on the support frame, the ring body is of an annular structure, the inner surface of the ring body is smooth, a bottom plate hermetically connected with the ring body is arranged at the bottom of the ring body, the bottom plate is made of transparent material, an adjusting plate arranged in parallel with the bottom plate is clamped in the ring body, a sealing ring is arranged on the outer ring of the adjusting plate, pull rods are arranged on the left side and the right side of the upper surface of the adjusting plate, a feeding pipe is arranged at the bottom of the; the bottom of the support frame is provided with a detection device, and a camera of the detection device penetrates through a through hole in the support frame and is aligned to the position right below the bottom plate.

摘要：【中文】本发明涉及一种图文序列匹配的跨媒体综合推理方法和系统。该方法包括以下步骤：1.建立跨媒体蕴涵推理数据集，其中前提包括图像和文本两种媒体类型，结论的媒体类型为文本。2.利用跨媒体蕴涵推理数据集训练图文序列匹配深度网络，在网络训练时使用了跨任务跨媒体迁移的训练方法。3.利用训练好的图文序列匹配网络结构进行推理，判断给定前提和结论的蕴涵关系。与现有方法相比，本发明能够实现基于图像和文本前提的蕴涵推理，并将检索任务蕴含的关联知识迁移至推理任务网络，提高推理准确率。 【EN】The invention relates to a cross-media comprehensive reasoning method and a system for matching image-text sequences. The method comprises the following steps: 1. and establishing a cross-media implication reasoning data set, wherein the premise comprises two media types of an image and a text, and the media type of the conclusion is the text. 2. The method is characterized in that a cross-media implication reasoning data set is used for training a graph and text sequence matching deep network, and a cross-task and cross-media migration training method is used in network training. 3. And reasoning by using the trained image-text sequence matching network structure to judge the implication relation of the given premise and the conclusion. Compared with the prior art, the method can realize implication reasoning based on image and text premises, and migrate the associated knowledge implied by the retrieval task to the reasoning task network, thereby improving the reasoning accuracy.

摘要：【中文】本发明涉及一种基于异构交互学习的跨媒体推理方法和系统。该方法包括以下步骤：1.建立跨媒体蕴涵推理数据集，其中前提包括两种不同媒体类型，结论包括一种媒体类型。2.利用跨媒体蕴涵推理数据集训练异构交互学习网络结构，主要包括跨媒体交互注意力学习和异构张量空间构建。3.利用训练好的异构交互学习网络进行推理，判断给定前提和结论的蕴涵关系。与现有方法相比，本发明能够实现基于不同媒体前提的蕴涵推理，并提高蕴涵推理的准确率。 【EN】The invention relates to a cross-media reasoning method and a system based on heterogeneous interactive learning. The method comprises the following steps: 1. a cross-media implication reasoning data set is established, wherein the premises includes two different media types and the conclusion includes one media type. 2. The heterogeneous interactive learning network structure is trained by using a cross-media implication reasoning data set, and the method mainly comprises cross-media interactive attention learning and heterogeneous tensor space construction. 3. And reasoning by using the trained heterogeneous interactive learning network, and judging the implication relation of the given premise and the conclusion. Compared with the prior art, the method can realize the implication reasoning based on different media premises and improve the accuracy of the implication reasoning.

摘要：【中文】本申请公开了一种源网荷储协同方法及系统，源网荷储协同服务系统根据网络控制器发送的客户侧源网荷储协同单元的预测功率协同能力生成并发送协同响应指令，使网络控制器在确定协同响应指令携带的功率调整目标值超过客户侧源网荷储协同单元的当前功率协同能力最大值时，将当前功率协同能力最大值作为功率调整计划值，以便源网荷储协同服务系统根据功率调整计划值确定响应收入预测金额；源网荷储协同服务系统将功率调整计划值和响应收入预测金额发送至客户侧源网荷储协同单元的客户端，并在接收到客户端发送的响应确认信息时，将功率调整计划值发送给网络控制器，使网络控制器根据功率调整计划值，对客户侧源网荷储协同单元中设备进行功率调整。 【EN】The source network load-storage cooperative service system generates and sends a cooperative response instruction according to the predicted power cooperative capacity of a client side source network load-storage cooperative unit sent by a network controller, so that when the network controller determines that a power adjustment target value carried by the cooperative response instruction exceeds the maximum value of the current power cooperative capacity of the client side source network load-storage cooperative unit, the maximum value of the current power cooperative capacity is used as a power adjustment plan value, and the source network load-storage cooperative service system determines a response income prediction amount according to the power adjustment plan value; and the source network charge-storage cooperative service system sends the power adjustment plan value and the response income prediction amount to a client of the client side source network charge-storage cooperative unit, and sends the power adjustment plan value to the network controller when receiving response confirmation information sent by the client, so that the network controller adjusts the power of equipment in the client side source network charge-storage cooperative unit according to the power adjustment plan value.

摘要：【中文】本发明实施例公开了一种自适应的定址寻路规划方法、装置、设备及存储介质，方法包括：获取初始路径规划结果并初始化全局最优规划结果和初始路径优化基础；为各个算子分配分数；基于初始路径优化基础先后进行下级优化和局部搜索优化得到第三路径优化结果；若第三路径优化结果的成本小于初始路径规划结果的成本，使用第三路径优化结果替代初始路径规划结果；若初始路径规划结果的成本小于全局最优规划结果的成本，使用初始路径规划结果替代全局最优规划结果；迭代进行优化步骤到达预设次数后输出全局最优规划结果。本发明实施例合理设置循环过程中各算子的使用概率提高了优化结果的质量，保障了路径规划结果的质量同时缩短了规划的用时。 【EN】The embodiment of the invention discloses a self-adaptive addressing route-finding planning method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring an initial path planning result and initializing a global optimal planning result and an initial path optimization basis; distributing scores for each operator; performing subordinate optimization and local search optimization in sequence based on the initial path optimization basis to obtain a third path optimization result; if the cost of the third path optimization result is less than the cost of the initial path planning result, replacing the initial path planning result with the third path optimization result; if the cost of the initial path planning result is less than that of the global optimal planning result, replacing the global optimal planning result with the initial path planning result; and after the iteration optimization step reaches the preset times, outputting a global optimal planning result. The embodiment of the invention reasonably sets the use probability of each operator in the circulation process, improves the quality of the optimized result, ensures the quality of the path planning result and shortens the planning time.

摘要：【中文】本发明公开了一种碳化硅涂层沉积炉流体场的数字孪生控制方法，包括以下步骤：（1）将沉积炉的炉内空间划分成多个控制区域，并在各控制区域内分别安装流量传感器；（2）采集各控制区域内的流量传感器检测到的各控制区域的实际流量值并传送给数字孪生设备；（3）计算出各控制区域的实际流量值与各控制区域的理论流量值之间的差值；（4）根据差值进行流量补偿。本发明利用数字孪生技术，对沉积炉内空间分区域建造数字孪生模型，达到生产时，分区域实时监控沉积炉内影响沉积的物质的流量，通过对采集的实际流量与理论数据融合，得到误差补偿结果，以此进行分区域流量补偿，解决了沉积炉内流量场分布均匀性差的问题，获得高质量的碳化硅涂层。 【EN】The invention discloses a digital twinning control method for a fluid field of a silicon carbide coating deposition furnace, which comprises the following steps of: (1) dividing the furnace space of the deposition furnace into a plurality of control areas, and respectively installing flow sensors in the control areas; (2) collecting actual flow values of the control areas detected by flow sensors in the control areas and transmitting the actual flow values to digital twin equipment; (3) calculating the difference between the actual flow value of each control area and the theoretical flow value of each control area; (4) and carrying out flow compensation according to the difference. According to the invention, a digital twin model is built in the space in the deposition furnace in different areas by using a digital twin technology, so that when production is achieved, the flow of substances affecting deposition in the deposition furnace is monitored in different areas in real time, and an error compensation result is obtained by fusing the acquired actual flow with theoretical data, so that the regional flow compensation is carried out, the problem of poor distribution uniformity of a flow field in the deposition furnace is solved, and the high-quality silicon carbide coating is obtained.

摘要：【中文】本发明针对neo4j图数据库查询语句传统的编写方法效率低、泛化能力差、成本高等缺点，提出了一种基于BERT(Bidirectional Encoder Representations from Transformers，BERT)和知识图谱感知的CQL(CQL代表Cypher查询语言)生成方法。首先采用基于BERT的知识图谱感知编码器对自然语言查询和整个图谱的标签以及属性进行编码，获得包含自然语言查询信息以及图谱结构信息的特征向量；然后采用本专利提出的SHALLOW‑LAYER模型，通过损失函数控制BERT有规则的信息流，最终将自然语言查询映射成CQL查询。 【EN】Aiming at the defects of low efficiency, poor generalization capability, high cost and the like of the traditional writing method of a neo4j graph database query statement, the invention provides a CQL (CQL stands for Cypher query language) generation method based on BERT (Bidirectional Encoder retrieval from transforms, BERT) and knowledge graph perception. Firstly, a knowledge graph perception encoder based on BERT is adopted to encode natural language query, labels and attributes of the whole graph, and feature vectors containing natural language query information and graph structure information are obtained; then, by adopting the SHALLOW-LAYER model provided by the patent, the regular information flow of BERT is controlled by a loss function, and finally, the natural language query is mapped into a CQL query.

摘要：【中文】本发明公开了一种优化的化学气相沉积工艺，包括以下步骤：（1）将沉积炉的炉内空间划分成多个控制空间；（2）采集各控制空间内的实际温度值和实际流量值；（3）计算出实际与理论的差值；（4）根据差值进行温度补偿和流量补偿。本发明利用数字孪生技术，对沉积炉的内腔分空间建造数字孪生模型，达到生产时，分空间实时监控沉积炉内温度和影响沉积的物质的流量，实现整个温度场及流体场控制过程的可视化，通过对采集的实际温度和影响沉积的物质的实际流量与理论数据融合，得到误差补偿结果，以此分空间进行温度和流量补偿，从而实现了保证沉积炉内温度场和流体场分布均匀。 【EN】The invention discloses an optimized chemical vapor deposition process, which comprises the following steps: (1) dividing the furnace inner space of the deposition furnace into a plurality of control spaces; (2) collecting actual temperature values and actual flow values in each control space; (3) calculating the difference between the actual value and the theoretical value; (4) and performing temperature compensation and flow compensation according to the difference. According to the invention, a digital twin model is built in a cavity of the deposition furnace in a spatial division manner by utilizing a digital twin technology, so that when production is achieved, the temperature in the deposition furnace and the flow of substances influencing deposition are monitored in real time in the spatial division manner, the visualization of the control process of the whole temperature field and the fluid field is realized, an error compensation result is obtained by fusing the acquired actual temperature and the actual flow of the substances influencing deposition with theoretical data, and temperature and flow compensation is carried out in the spatial division manner, so that the uniform distribution of the temperature field and the fluid in the deposition furnace is ensured.

摘要：【中文】本发明公开一种主动式毫米波实时三维成像安检系统的校准系统及方法，包括左通道墙体、右通道墙体、吸波壁、校准体、所述左通道墙体和右通道墙体之间形成被检通行通道，所述被检通行通道底部设置有底板；其中，所述左通道墙体和右通道墙体相对设置，且所述左通道墙体和右通道墙体均配置有毫米波天线阵列口面，用于收发宽带线性调频信号，完成对所述校准体回波信号采集；所述校准体配置在所述地板上，用于反射所述左通道墙体和右通道墙体上的毫米波天线阵列口面发射的宽带线性调频信号；所述吸波壁设置在所述左通道墙体或右通道墙体前，用于防止所述左通道墙体工作时，右通道墙体的信号干扰，或用于防止所述右通道墙体工作时，左通道墙体的信号干扰。 【EN】The invention discloses a calibration system and a method of an active millimeter wave real-time three-dimensional imaging security inspection system, wherein the calibration system comprises a left channel wall body, a right channel wall body, a wave absorbing wall, a calibration body, a detected passage channel formed between the left channel wall body and the right channel wall body, and a bottom plate arranged at the bottom of the detected passage channel; the left channel wall and the right channel wall are arranged oppositely, and both the left channel wall and the right channel wall are provided with millimeter wave antenna array port faces for receiving and transmitting broadband linear frequency modulation signals and finishing the acquisition of the calibration body echo signals; the calibration body is arranged on the floor and used for reflecting broadband linear frequency modulation signals emitted by the millimeter wave antenna array openings on the left channel wall and the right channel wall; the wave absorbing wall is arranged in front of the left channel wall body or the right channel wall body and used for preventing signal interference of the right channel wall body when the left channel wall body works or preventing signal interference of the left channel wall body when the right channel wall body works.