• 学术搜索
  • 科研智能体
    • Research Labs
    • AI 阅读
    • AI 文库
    • 深度研究
    • 学者亮点
  • 学术资源
    • AI2000
    • 期刊/会议
    • 学者库
    • 学术API
    • 溯源树
    • 数据集
  • 知识沉淀
    • 学术空间
订阅小程序
旧版功能
aminer vip
开通会员低至0.73元/天
一次搞定AI科研
立即登录
  • English
    • 联系方式
    专

    海上风电漂浮式基础C115~C140超高性能自密实混凝土

    1262005039452D0
    发明人
    王军, 刘晓峰, 邹荔兵, 张启应, 陈发桥, 余辉
    受让人
    MING YANG SMART ENERGY GROUP CO LTD
    申请人
    SIGNIFY HOLDING B.V.
    申请号
    136704
    申请日
    1994-08-30
    公开(公告)号
    1262005039452D0
    公开(公告)日
    1980-05-28
    IPC分类号
    G01M001/04G01M001/16F16C032/06
    CPC分类号
    -
    优先权号
    2363370
    优先权日
    2001-11-20
    摘要

    Process comprises filling a thermoplastic resin cpd. in a slush mould, or resin filling process in which resin film is formed on the inside surface of mould, or the resin filling process in which excess resin cpd. is discharged from the mould are carried out under vacuum. In the device, a slush mould is mounted on a frame through bearings so as to be rotatable; a reserve tank is provided in the frame; the slush mould is provided with temp. adjusting piping and with a vacuum valve for evacuation; and the reserve tank can be connected to the slush mould by a lifting table.

    ADVANTAGE -Skin materials without defects such as pinholes, etc. can be obtd. by slush moulding.

    权利要求
    1 . A semantic segmentation method for aircraft point cloud based on voxelization and three views, comprising: (S1) collecting, by a three-dimensional (3D) scanner, a plurality of point clouds of an aircraft; assigning a label to each point in the plurality of point clouds; and filtering each of the plurality of point clouds followed by centralization to obtain a centralized point cloud; (S2) inputting the centralized point cloud into a 3×3 T-Net rotation matrix network until a rotation loss function converges; and rotating the centralized point cloud to a front side; (S3) adjusting a voxelization resolution of the centralized point cloud according to an adaptive scale to obtain a point cloud having the same voxel size; and subjecting the point cloud having the same voxel size to voxelization to obtain a voxelized point cloud; (S4) subjecting an unoccupied voxel in the voxelized point cloud to voxel filling to obtain a voxel-filled point cloud; (S5) respectively calculating a thickness map of three views of the voxel-filled point cloud; (S6) building a point cloud semantic segmentation network; sequentially stitching thickness maps of the three views of the voxel-filled point cloud; and inputting a stitched thickness map to the point cloud semantic segmentation network to train the point cloud semantic segmentation network until a cross-entropy loss function converges; and (S7) inputting the plurality of point clouds into the point cloud semantic segmentation network; and predicting a 3D point cloud semantic segmentation result of the aircraft; wherein the step (S5) comprises: (S501) for a front view of the voxel-filled point cloud, calculating the number of occupied voxels in z-direction for each voxel position on an xy-plane; and taking the number of occupied voxels in the z-direction as a thickness map of the front view; (S502) for a left side view of the voxel-filled point cloud, calculating the number of occupied voxels in x-direction for each voxel position on a yz-plane; and taking the number of occupied voxels in the x-direction as a thickness map of the left side view; and (S503) for a top view of the voxel-filled point cloud, calculating the number of occupied voxels in y-direction for each voxel position on an xz-plane; and taking the number of occupied voxels in the y-direction as a thickness map of the top view.
    2 . The semantic segmentation method of claim 1 , wherein the step (S1) comprises: (S101) collecting, by the 3D scanner, the plurality of point clouds of the aircraft; assigning a label to each point in the plurality of point clouds; filtering each of the plurality of point clouds according to a filter radius r; and filtering out a point with a point distance less than the filter radius r from each of the plurality of point clouds; and (S102) calculating a coordinate (X, Y, Z) of a center point of the plurality of point clouds after filtering; and subjecting all points in the plurality of point clouds to centralization by taking the center point as an original point to obtain the centralized point cloud.
    3 . The semantic segmentation method of claim 2 , wherein the coordinate (X, Y, Z) of the center point is calculated as follows: X = ∑ p i ∈ P x p i / n ; Y = ∑ p i ∈ P y p i / n ; and Z = ∑ p i ∈ P z p i / n ; wherein n is the number of points in the plurality of point clouds after filtering; i is an index of a point in the plurality of point clouds after filtering; p i represents an i th point in the plurality of point clouds after filtering; P represents the plurality of point clouds after filtering; x pi is a horizontal coordinate of the p i ; y pi is a vertical coordinate of the p i ; and z pi is a z-direction coordinate of the p i .
    4 . The semantic segmentation method of claim 1 , wherein the rotation loss function L reg is expressed as: L r e g = I − A A T F 2 ; wherein I represents a unit matrix; A is a predicted rotation matrix; A T is a transposed matrix of the A; and || ​ F 2 represents Frobenius norm (F-norm).
    5 . The semantic segmentation method of claim 1 , wherein the step (S3) comprises: (S301) selecting a maximum horizontal coordinate, a minimum horizontal coordinate, a maximum vertical coordinate, a minimum vertical coordinate, a maximum z-direction coordinate and a minimum z-direction coordinate in the centralized point cloud rotated to the front side; calculating a difference Δ× between the maximum horizontal coordinate and the minimum horizontal coordinate; calculating a difference Δy between the maximum vertical coordinate and the minimum vertical coordinate; and calculating a difference Δz between the maximum z-direction coordinate and the minimum z-direction coordinate; (S302) setting a voxel size of each of the plurality of point clouds as d×d×d such that a horizontal coordinate resolution of each of the plurality of point clouds is α x =Δx/d, a vertical coordinate resolution of each of the plurality of point clouds is α y =Δy/d, and a z-direction coordinate resolution of each of the plurality of point clouds is α z =Δz/d; (S303) performing steps (S301)-(S302) for each centralized point cloud rotated to the front side to obtain the point cloud having the same voxel size; and (S304) subjecting points in the point cloud having the same voxel size to voxelization to obtain the voxelized point cloud, wherein a coordinate of a voxel is expressed as (x t i ,y t i ,z t i ); the voxel is labeled as occupied; x t i = x ′ p i α x ; y t i = y ′ p i α y ; z t i = z ′ p i α y ; indicates an operation of rounding down; x ′ p i is a horizontal coordinate of an i th point in the point cloud having the same voxel size; y ′ p i is a vertical coordinate of the i th point in the point cloud having the same voxel size; and z ′ p i is a z-direction coordinate of the i th point in the point cloud having the same voxel size.
    6 . The semantic segmentation method of claim 1 , wherein the step (S4) comprises: if the unoccupied voxel in the voxelized point cloud has at least one occupied voxel in each normal direction of a cubic surface of the unoccupied voxel, labeling the unoccupied voxel as occupied; and obtaining the voxel-filled point cloud.
    7 . The semantic segmentation method of claim 1 , wherein the point cloud semantic segmentation network comprises a first convolutional neural network (CNN), a second CNN, a third CNN, a fusion network, a first fully connected layer, a max-pooling layer, a second fully connected layer and a softmax layer; and the step (S6) comprises: (S601) connecting the first CNN, the second CNN and the third CNN to the fusion network; connecting the fusion network to the first fully connected layer and the max-pooling layer; stitching the max-pooling layer with an output of the fusion network; and connecting the max-pooling layer successively to the second fully connected layer and the softmax layer; (S602) for each voxel of the voxel-filled point cloud, inputting a thickness map of a front view into the first CNN to extract a feature vector of an xy-plane; inputting a thickness map of a left side view into the second CNN to extract a feature vector of a yz-plane; inputting a thickness map of a top view into the third CNN to extract a feature vector of an xz-plane; and inputting the feature vector of the xy-plane, the feature vector of the yz-plane and the feature vector of the xz-plane into the fusion network for concatenation to obtain a first concatenated vector; (S603) inputting the first concatenated vector successively into the first fully connected layer and the max-pooling layer to extract a global feature vector; concatenating the global feature vector with the first concatenated vector to obtain a second concatenated vector; inputting the second concatenated vector successively into the second fully connected layer and the softmax layer; and predicting a category of each voxel after division; and (S604) performing steps (S602)-(S603) for each voxel-filled point cloud until the cross-entropy loss function converges to complete the training of the point cloud semantic segmentation network.
    8 . The semantic segmentation method of claim 7 , wherein the cross-entropy loss function L cls is expressed as: L c l s = − ∑ i = 1 N y i l o g y i ′ ; wherein i is an index of a point in the plurality of point clouds after filtering; y i is a category of a label of an i th point; and y i ’ is a predicted category of the i th point.
    说明书
    [0001]技术领域
    [0002]本发明涉及制冷技术领域，尤其涉及一种制冷电器。
    [0003]背景技术
    [0004]目前，冰箱、冷柜等制冷电器是人们日常生活中的常用电器，制冷电器的内胆通常配置有排水管，相对应的，制冷电器的外壳中还设置有接水盒，排水管将水排入到接水盒中。但是，当接水盒中的水积满后，用户需要从内胆内部采用排水塞将排水管封堵住，然后，才可以将接水盒从外壳中取出。由上可知，现有技术中的制冷电器，如果内部用排水塞塞住排水管，内部积水会浸泡货物，排水时需要人为的控制排水塞，如果排水塞不安装，接水盒内积水满了，无人倒水，则积水则会从接水盒中溢出，导致用户使用不方便，操作繁琐。如何设计一种自动控制排水管开关、方便用户使用以提高用户体验性的制冷电器是本发明所要解决的技术问题。
    [0005]发明内容
    [0006]本发明提供了一种制冷电器，实现制冷电器自动控制排水管开关、方便用户使用以提高用户体验性。
    [0007]为达到上述技术目的，本发明采用以下技术方案实现：
    [0008]一种制冷电器，包括外壳、设置在所述外壳中的内胆、连接所述内胆的排水管，所述外壳中设置有安装腔体，所述排水管的排水口位于所述安装腔体中，所述排水管中设置有用于开关所述排水口的电磁阀，所述排水管位于所述排水口上部的管道为蓄水管段，所述安装腔体中还设置有用于触发所述电磁阀打开的第一开关；所述制冷电器还包括接水组件，所述接水组件包括接水容器，所述接水容器上设置有用于触发所述第一开关动作的触发部，所述接水容器可拆卸的安装在所述安装腔体中；所述电磁阀、所述第一开关与所述制冷电器的供电部串联形成控制电路。
    [0009]进一步的，所述安装腔体中设置有承载座，所述承载座上设置有弹簧、第二开关和垫块，所述弹簧固定在所述承载座上，所述垫块设置在所述弹簧的上部，所述第二开关位于所述垫块的下方，所述第二开关串联在所述控制电路中；所述接水容器的侧壁设置支撑杆，所述支撑杆的自由端部搭接在所述垫块上。
    [0010]进一步的，所述垫块上形成有用于导向所述支撑杆的自由端部移动的倾斜面。
    [0011]进一步的，所述支撑杆的自由端部为球形结构。
    [0012]进一步的，所述安装腔体的顶部第二开关，所述第二开关串联在所述控制电路中；所述接水容器中设置有浮球，所述浮球上设置有用于在所述接水容器中的水达到设定值后触发所述第二开关动作的顶杆。
    [0013]进一步的，所述接水容器中设置有用于导向所述浮球上下移动的导槽。
    [0014]进一步的，所述第一开关为常开开关，所述第二开关为常闭开关。
    [0015]进一步的，所述电磁阀上并联有指示灯，所述第一开关为常开开关，所述第二开关为二位置选择开关，所述二位置选择开关连接所述电磁阀和所述指示灯，所述二位置选择开关连接所述电磁阀的电路为常闭电路。
    [0016]进一步的，所述第一开关设置在所述安装腔体的开口处，所述接水容器的前面板设置有向外延伸的延长边；在所述接水容器装入到所述安装腔体中后，所述延长边抵靠在所述第一开关上，所述延长边为所述触发部。
    [0017]进一步的，所述接水容器设置有水位计或水位观察窗。
    [0018]与现有技术相比，本发明的优点和积极效果是：通过在排水管下部的排水口中设置电磁阀，并通过安装接水容器以触发第一开关动作实现电磁阀打开，这在实际使用过程中，当接水容器接满水后，用户可以直接将接水容器从安装腔体中取出，第一开关复位后使得电磁阀断电，利用电磁阀将自动关闭排水口，以避免内胆中的水从排水口排出，与此同时，排水管上部的蓄水管段还能够继续收集内胆中的水，避免内胆积水造成货物浸水，实现制冷电器自动控制排水管开关、方便用户使用以提高用户体验性。
    [0019]附图说明
    [0020]为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
    [0021]图1为本发明制冷电器实施例的剖视图一;
    [0022]图2为本发明制冷电器实施例的剖视图二；
    [0023]图3为图1中A区域的局部放大示意图；
    [0024]图4为图2中B区域的局部放大示意图；
    [0025]图5为本发明制冷电器实施例的局部结构示意图；
    [0026]图6为本发明制冷电器实施例的电路原理图一；
    [0027]图7为本发明制冷电器实施例的电路原理图二；
    [0028]图8为本发明制冷电器实施例中接水组件的结构示意图；。
    [0029]具体实施方式
    [0030]为使本发明实施例的目的、技术方案和优点更加清楚，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
    [0031]如图1-图7所示，本实施例制冷电器，包括外壳1、设置在所述外壳1中的内胆11、连接所述内胆11的排水管2，所述外壳1中设置有安装腔体12，所述排水管2的排水口位于所述安装腔体12中，所述排水管2中设置有用于开关所述排水口的电磁阀41，所述排水管2位于所述排水口上部的管道为蓄水管段20，所述安装腔体12中还设置有用于触发所述电磁阀41打开的第一开关42；所述制冷电器还包括接水组件3，所述接水组件包括接水容器31，所述接水容器31上设置有用于触发所述第一开关42动作的触发部312，所述接水容器31可拆卸的安装在所述安装腔体12中；所述电磁阀41、所述第一开关42与所述制冷电器的供电部10串联形成控制电路。
    [0032]具体而言，本实施例制冷电器在排水管2的排水口中设置电磁阀41，电磁阀41在断电状态下处于关闭状态，实现自动封堵关闭排水口，而接水组件3利用接水容器31收集排水管2排出的水，在接水容器31安装到安装腔体12中后，接水容器31上的触发部312触动第一开关42使得控制电路导通，供电部10向电磁阀41供电，使得电磁阀41通电打开，使得打开排水管2的排水口，实现内胆11中的水从排水管2排入到接水容器31中；而在接水容器31接满水从安装腔体12中取出后，第一开关42复位使得控制电路断开，电磁阀41断电后自动关闭排水口，从而可以有效的避免在接水容器31取出后排水管2继续排水而污染地面的问题；另外，由于排水管2上部形成的蓄水管段20可以继续收集内胆11中的水，避免在接水容器31取出后，内胆11发生积水而导致货物浸水。在接水容器31拆卸后，排水管2能够通过电磁阀41自动关闭，而无需用户从内胆11手工放置排水塞，方便用户使用。而接水容器31可以设置有水位计或水位观察窗，以便于用户查看接水容器31内的水量。本实施例制冷装置包括两种接水模式：主接水模式和辅助接水模式，具体的，主接水模式下，电磁阀41通电以打开排水管2，内胆11中的水从排水管2排入到接水容器31中；辅助接水模式下，接水容器31从安装腔体12中取出，电磁阀41断电关闭排水管2，内胆11中的水暂存在排水管2的蓄水管段20。
    [0033]其中，为了在接水容器31内的水量达到设定值后自动关闭电磁阀41以提高使用可靠性和用户体验性，所述安装腔体12中设置有承载座（未标记），所述承载座上设置有弹簧45、第二开关43和垫块44，所述弹簧45固定在所述承载座上，所述垫块44设置在所述弹簧45的上部，所述第二开关43位于所述垫块44的下方，所述第二开关43串联在所述控制电路中；所述接水容器31的侧壁设置支撑杆310，所述支撑杆310的自由端部搭接在所述垫块44上。具体的，接水容器31装入到安装腔体12中后，接水容器31的前端部通过安装腔体12的开口进行支撑，而接水容器31的后端部通过支撑杆310支撑在垫块44上，垫块44受弹簧45施加的弹力能够有效的支撑柱支撑杆310，而在接水容器31中收集水的过程中，垫块44受到的压力逐渐增大，使得垫块44向下移动靠近第二开关43，而当接水容器31中收集的水达到设定值后，垫块44将触动第二开关43，通过第二开关43控制电磁阀41断电，从而可以避免接水容器31中的水溢出。而为了便于安装接水容器31，垫块44上形成有用于导向所述支撑杆310的自由端部移动的倾斜面（未标记），同样的，支撑杆310的自由端部为球形结构。
    [0034]同样的，为了实现上述接水容器31水满自动切断电磁阀41供电的目的，还可以采用下述方案，如图6和图8所示，所述安装腔体的顶部第二开关43，所述第二开关43串联在所述控制电路中；接水容器31中设置有浮球32，所述浮球32上设置有用于在所述接水容器31中的水达到设定值后触发所述第二开关43动作的顶杆321。具体的，为了避免接水容器31接水过多发生溢水的现象，通过在接水容器31中设置浮球32，浮球32能够根据接水容器31中的水位高低自动升降，而在接水容器31中的水位达到设定值后，浮球32上升到特定的高度位置，使得顶杆321触动第二开关43动作，通过第二开关43控制电磁阀41断电，从而可以避免接水容器31中的水溢出。优选的，为了确保顶杆321准确的触动第二开挂43，接水容器31中设置有用于导向所述浮球32上下移动的导槽311。
    [0035]另外，针对上述记载的第一开关42和第二开关43，可以配置为第一开关42为常开开关，所述第二开关43为常闭开关。或者，为了达到对用户进行水满提醒的目的，所述电磁阀43上并联有指示灯13，所述第一开关42为常开开关，所述第二开关43为二位置选择开关，所述二位置选择开关连接所述电磁阀41和所述指示灯13，所述二位置选择开关连接所述电磁阀41的电路为常闭电路。
    [0036]与现有技术相比，本发明的优点和积极效果是：通过在排水管下部的排水口中设置电磁阀，并通过安装接水容器以触发第一开关动作实现电磁阀打开，这在实际使用过程中，当接水容器接满水后，用户可以直接将接水容器从安装腔体中取出，第一开关复位后使得电磁阀断电，利用电磁阀将自动关闭排水口，以避免内胆中的水从排水口排出，与此同时，排水管上部的蓄水管段还能够继续收集内胆中的水，避免内胆积水造成货物浸水，实现制冷电器自动控制排水管开关、方便用户使用以提高用户体验性。
    [0037]最后应说明的是：以上实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明个实施例技术方案的精神和范围。
    同族专利
    暂无同族专利