19results about How to "Lightweight structure" patented technology

The application discloses a rice disease image segmentation method based on frequency domain modulation and topological constraint, which comprises the following steps: 1, frequency domain decomposition is carried out on rice disease image data to obtain amplitude spectrum and phase spectrum features; 2, a frequency domain adaptive modulation module is constructed to selectively enhance the frequency components of the structural key information; 3, the topological invariants of the image are extracted through a topological constraint module, and the reference topological structure is obtained by down-sampling the label to construct a topological consistency loss; 4, the topological consistency loss and the rice disease image segmentation loss are jointly optimized, the model training under multi-level structure constraint is realized, and a rice disease image segmentation model is constructed. The sensitivity of the model to the structure features of the rice disease lesion is enhanced through frequency domain modulation, and the morphological consistency of the segmentation result is maintained through topological constraint, so that the robustness of the model under natural conditions such as light change and background complexity difference is effectively improved, and the performance of the disease image segmentation in an open farmland scene is improved.

The invention relates to the field of photovoltaic technology, and discloses a photovoltaic power station power prediction method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining the operation data and environment data of a photovoltaic power station; calculating a target attenuation coefficient of the photovoltaic power station according to the operation data and the environment data; and correcting an initial power predicted value of the photovoltaic power station based on the target attenuation coefficient to obtain a target power predicted value, the initial power predicted value being determined according to meteorological data of the photovoltaic power station. The method solves the problems that in the prior art, the interaction effect between photovoltaic panel aging and dust deposition is not considered, and due to the fact that a complex model or newly-added hardware is depended on, the power prediction precision of an old photovoltaic power station is low, and the applicability is poor.

The invention discloses a bearing fault identification method based on multi-scale space-time synchronization attention and space-time alignment. The method comprises the following steps: carrying out sliding window segmentation and normalization preprocessing on an original vibration signal; constructing an adjacent matrix by using a K-nearest neighbor algorithm, and extracting local spatial features by using a graph convolutional neural network; bidirectional sequence features are extracted through a bidirectional gating loop unit network, and weighted fusion is carried out in combination with a global attention mechanism; and after space and time sequence features are spliced, fault prediction is realized through adaptive average pooling and a full-connection classifier. According to the method, graph modeling, GCN, BiGRU and an attention mechanism are fused, cooperative extraction of space-time double-path features is realized, the problems of weak space modeling and lack of time sequence dependence in a traditional method are effectively solved, and the bearing fault recognition precision and generalization ability in a multi-working-condition and strong-noise environment are remarkably improved.

The utility model discloses a kind of rotating devices, belong to X-ray automatic detection equipment technical field, specifically related to a kind of high-precision X-ray source intelligent rotating device, comprising: driving motor, installation bottom plate, hollow rotating platform and X-ray source assembly;The driving motor is installed in the rear side of the installation bottom plate, the hollow rotating platform is installed in the front side of the installation bottom plate, and the X-ray source assembly is installed on the hollow rotating platform;Arranged on the installation bottom plate is in-place inductor and limit block;The X-ray source assembly is provided with the inductive sheet matched with the in-place inductor;The utility model uses hollow rotating platform as core rotating element, and is directly driven by motor, and this design reduces the generation of cumulative error from the root, ensures the high-precision rotating performance;And simplify overall structure, under the premise of guaranteeing function, reduce volume, weight and cost, show significant technical advantage.

The invention discloses a web version train number drawing method, system and device and a medium, and the method comprises the steps: dividing a drawing area into continuous subareas, determining a corresponding initial graph area according to the text attribute of a target train number when the target train number is drawn or updated, and determining an initial subarea occupied by the initial graph area; indexing the graphic area coordinate information of other train numbers stored in the initial partition through a cache structure, and carrying out collision detection on the coordinate information of the initial graphic area and the graphic area coordinate information of other train numbers; under the condition that collision is detected, selecting a candidate graphic area without collision from the generated candidate graphic areas, and determining the selected candidate graphic area as a final drawing area; and drawing the target train number in the final drawing area, and updating the cache structure. The calculation process of collision detection is effectively reduced, the problem of overlapping and shielding of train numbers is solved, and the visual definition of a working diagram is improved.

The application discloses a Jerusalem unit double-layer FSS broadband wave-absorbing structure and a preparation method and application thereof, and belongs to the field of electromagnetic wave-absorbing technology design in electromagnetic stealth technology. The structure is sequentially arranged from bottom to top as a metal bottom layer, a dielectric layer 1, a patterned resistance layer 1, a dielectric layer 2 and a patterned resistance layer 2. Periodic array distributed multiple unit cell structures are formed on the surfaces of the patterned resistance layer 1 and the patterned resistance layer 2, and each unit cell structure adopts a Jerusalem cross pattern. The Jerusalem unit double-layer FSS broadband wave-absorbing structure provided by the application has a wide-frequency absorption characteristic. Through the regulation of geometric sizes on a macro scale and the structure design on a micro scale, the wave-absorbing structure can realize an absorption rate of more than 99% in a full frequency band of 8-12 GHz, and has a good absorption effect on low-frequency electromagnetic waves.

The present application relates to whip antenna technical field, specifically to carbon fiber whip antenna structure and forming method. The existing carbon fiber whip antenna conduction performance is poor, the resistance value is large, the radiation efficiency is low. The carbon fiber whip antenna structure includes first metal joint, carbon fiber tube and second metal joint and the conductive wire connecting two metal joints. Its forming method is that the first metal joint and the second metal joint are installed at both ends of the mold; carbon fiber prepreg is laid; the conductive wire is wound; carbon fiber prepreg is laid again; the antenna outer surface is wrapped with metal mesh; the metal mesh is wound and fixed with the conductive wire; the connecting part of the metal mesh and the first metal joint and the second metal joint is wound with carbon fiber prepreg respectively; compaction and solidification; demolding. The carbon fiber whip antenna has good conduction performance, small resistance value, high resistance radiation efficiency, light structure quality, excellent mechanical properties, fatigue resistance, corrosion resistance, large power bearing, high radiation efficiency, long service life, high efficiency of the workpiece.

This invention discloses an onboard multi-sensor active-passive collaborative stabilization device and method, relating to the field of stability control technology for autonomous vehicle perception systems. The device includes a vehicle mounting base, a passive vibration isolation module, an active compensation module, a multi-sensor common reference support platform, an attitude and vibration sensing module, and a collaborative control module. The passive vibration isolation module attenuates mid-to-high frequency vibrations and impact loads transmitted from the vehicle body to the support platform. The active compensation module compensates for the pitch, roll, and / or yaw attitudes of the support platform in real time. The collaborative control module preprocesses disturbance information, identifies disturbances, divides frequency bands, and allocates control, ensuring that low-frequency attitude disturbances are compensated by the active compensation module, mid-to-high frequency vibrations are attenuated by the passive vibration isolation module, and outputs stabilized multi-sensor data and / or residual attitude information. This invention improves the installation stability and relative pose stability of multiple sensors such as cameras, lidar, and IMUs during vehicle operation, reduces image blurring, point cloud distortion, and attitude drift, and enhances the stability and accuracy of multi-sensor fusion perception, localization mapping, and autonomous driving systems.

The application discloses a gas nozzle adjusting device and method for a solid rail control engine, and the gas nozzle adjusting device is used to adjust the angle of the gas nozzle based on the engine cabin body, improve the position precision of the thrust axis of the rail control engine, and realize the accurate control of the rail control engine. The structure comprises a nozzle, a connecting flange, a base, a spherical column, an end cover, a positioning tool, a positioning core rod, a cabin body, an O-shaped ring, a fastening screw, a fixing screw and a positioning pin. The layout form is shown in the drawings, the nozzle of the rail control engine is fixed with the connecting flange through threads, the positioning tool is positioned with the cabin body through two positioning pins, two fixing screws are used for fixing, the positioning tool and the positioning core rod ensure that the angle of the gas nozzle is adjusted in place, and the connecting flange, the base and the end cover are fixed with the spherical column through the fastening screw.

The utility model relates to the technical field of power transmission, and discloses an aluminum alloy bus duct shell, which comprises a first side plate, a second side plate, a first cover plate and a second cover plate, the first side plate and the second side plate are C-shaped plates, the first side plate and the second side plate are oppositely arranged, a gap is arranged between the first side plate and the second side plate, a first heat dissipation part is arranged outside the first side plate, and a second heat dissipation part is arranged outside the second side plate. A first heat dissipation part is arranged on the outer side of the first side plate, a second heat dissipation part is arranged on the outer side of the second side plate, the first cover plate is connected with one end of the first side plate and one end of the second side plate, the first cover plate is an I-shaped plate and comprises a first vertical plate, a first horizontal plate and a second vertical plate which are sequentially connected, and a first supporting plate is arranged on the outer side surface of the first horizontal plate; the second cover plate is connected with the other end of the first side plate and the other end of the second side plate, the second cover plate is an I-shaped plate, the second cover plate comprises a third vertical plate, a second horizontal plate and a fourth vertical plate which are connected in sequence, and a second supporting plate is arranged on the outer side surface of the second horizontal plate. The LED lamp is high in supporting capacity, not prone to deformation, even in heat dissipation and light in weight.

The utility model relates to the technical field of traffic facilities, and discloses a traffic isolation guardrail which comprises a plurality of connecting knots detachably and fixedly arranged on a road at intervals, protective plates detachably and fixedly installed on the two sides of the connecting knots and used for connecting every two adjacent connecting knots, and end columns fixedly arranged at the ends of the multiple mutually-connected protective plates. The bottoms of the end columns are detachably and fixedly connected with a road. A brand-new frame skin structure is adopted, the stability and safety of the guardrail can be considered at the same time, and meanwhile the overall structure is light in weight and convenient to transport; during installation, different structural parts are firmly connected through the fasteners, so that the overall stability and external force resistance of the guardrail are high.

The application provides a turbine disc ultra-light topology optimization method considering equal strength targets, which comprises the following steps: discretizing a turbine disc design domain and setting design variables, obtaining unit physical density through density filtering and Heaviside mapping, establishing a finite element model to solve structure displacement and von Mises stress, calculating enhanced stress standard deviation, dispersion factor and mass fraction after von Mises stress punishment, constructing an aggregate objective function and an approximate maximum stress constraint, performing sensitivity analysis and updating design variables, and iteratively obtaining a turbine disc topology configuration. The method can improve the stress utilization degree of the bearing material region while meeting the strength constraint.

The application belongs to the technical field of computer vision, and particularly relates to an image semantic segmentation method based on dynamic fusion and context perception, comprising the following steps: inputting an input image into a lightweight multi-scale feature extraction backbone network to output shallow, middle and deep layer features; inputting the shallow, middle and deep layer features into an MRFM module to output first, second and third enhanced semantic features; inputting the third enhanced semantic feature into an LACP to output an enhanced context feature; inputting the first, second enhanced semantic features and the enhanced context feature into a DWFM to output a fusion feature; inputting the enhanced context feature into a double-branch prediction structure to output a final segmentation prediction map; constructing a joint loss function to perform end-to-end training and optimization on the network formed by steps S1 to S5; and solving the problems of rigid fusion strategy, weak context perception capability and model parameter redundancy in the existing image semantic segmentation method.

The application relates to a combined air unit frame connecting structure, which comprises a plurality of frame rods, reinforcing rods, corner connecting pieces for connecting the frame rods and intermediate connecting pieces for connecting the frame rods and the reinforcing rods; the corner connecting piece comprises a bearing base, a first plug connected to the bearing base and used for transversely connecting the frame rods and a second plug used for longitudinally connecting the frame rods; outward first convex structures are arranged on two mutually-perpendicular side walls of the second plug; the intermediate connecting piece comprises a connecting plate and a plug-in block arranged on the connecting plate, the connecting plate is fixed on the frame rod, and the plug-in block is used for plugging in the reinforcing rod; the plug-in block has at least one partially concave side wall. The modular connecting structure designed in the application has high assembly efficiency, light structure quality, high connecting rigidity and is suitable for application places with large air volume and high static pressure.

The embodiment of the invention provides a multi-node UWB relative positioning method and system based on a liquid neural network. The method comprises the steps that observation data between known node pairs are collected in real time through UWB; obtaining a multi-frame time sequence feature sequence corresponding to a moment based on observation data between known node pairs acquired in real time; inputting the multi-frame time sequence feature sequence into a liquid neural network model based on an ordinary differential equation, and outputting a prediction distance result between known node pairs; based on the real distance result and the predicted distance result, optimizing the liquid neural network model by using a loss function to obtain a trained liquid neural network model; using the trained liquid neural network model to obtain a prediction distance result between the unknown node pairs; and obtaining the relative position of the unknown node at the current moment based on the predicted distance result between the unknown node pairs and the relative position of the unknown node at the previous moment. The method can give consideration to the real-time performance and robustness of positioning.

The invention discloses a multivariate time series deep federation anomaly detection method and system, and the method comprises the steps: carrying out the local feature coding and normal behavior paradigm learning at an edge device, constructing a local paradigm set, only uploading an embedded network parameter and a paradigm module to a central server, and achieving the collaborative modeling under a data localization processing condition; the central server executes federal optimization aggregation on model parameters uploaded by multiple devices, performs clustering alignment on local examples, generates a global embedded model and a global example set, and then synchronizes the global embedded model and the global example set to each edge device for iterative optimization; according to the method, a unified anomaly judgment mechanism based on cosine similarity is adopted, so that the deviation degree of the to-be-detected sample and the paradigm feature is taken as an anomaly score, and accurate recognition of unknown anomaly and fine anomaly is realized. According to the method, original time sequence data does not need to be transmitted, the communication overhead can be remarkably reduced, the privacy protection capability is enhanced, and the method has the adaptive capability to multiple types of equipment and dynamic environment changes.

The invention discloses an extremely few sample remote sensing scene classification method based on dynamic sparse feature enhancement. The method comprises the following steps: acquiring high-resolution remote sensing image data; extracting multilayer features of the image data through a trunk convolutional neural network; performing adaptive sparsification processing on the features extracted by the trunk convolutional neural network to obtain sparse features; carrying out confidence weighted fusion on the sparse features to obtain enhanced features; and performing classification based on the enhanced features, and outputting a scene category and a confidence result of the remote sensing image. According to the method, a dynamic sparse mapping and uncertainty perception feature fusion mechanism in a deep learning network is utilized to realize adaptive feature extraction and confidence coefficient weighted classification of high-resolution remote sensing images, and a high-precision and high-robustness classification result can be obtained under the condition of extremely few labeled samples; and carrying out scene category judgment and result output on the purified key representation. The remote sensing scene classification method is suitable for the condition of extremely few sample supervision.

The slope excavation supporting device for the water conservancy project comprises a retaining wall, blocking mechanisms and a plurality of sand filling grooves formed in the top end of the retaining wall, the multiple sets of blocking mechanisms are arranged at the bottom of the retaining wall in an arrayed mode, and the multiple sets of blocking mechanisms are used for sand discharging and blocking of the multiple sand filling grooves respectively; the top supporting structure is arranged at the bottom of the retaining wall and used for supporting the bottom of the retaining wall in an abutting mode. A plurality of through openings are formed in the middle of the retaining wall, and blocking mechanisms are detachably installed at the through openings and used for blocking the through openings. The retaining wall is provided with the through hole and the sand filling groove, so that the volume structure of the retaining wall is reduced, the prefabricated retaining wall structure is lighter in weight, the retaining wall is more convenient to transfer, and meanwhile, the blocking mechanism which can be detachably mounted is additionally arranged, so that the blocking mechanism can be additionally arranged according to actual support requirements; and the retaining wall is arranged more flexibly.

The invention relates to a high-efficiency flight welding positioning device and method.The positioning device comprises a rack, a left truss and a right truss are arranged on the rack, and the two ends of the left truss or the two ends of the right truss move in the length direction of the rack; the left truss and the right truss are respectively provided with two sets of copper nozzle pressing mechanisms; each set of copper nozzle pressing mechanism comprises a mounting frame, each mounting frame comprises a sliding plate, a lifting plate and a mounting plate, the sliding plates of two adjacent sets of copper nozzle pressing mechanisms are mounted on a supporting beam at the same time, and the supporting beams move in the length direction of the rack through a large-stroke moving mechanism; a lifting electric cylinder is arranged on the sliding plate and fixed to the sliding plate, and an output shaft of the lifting electric cylinder is connected with the lifting plate. The mounting plate is mounted at the bottom of the lifting plate, two rows of pressing copper nozzles are arranged on the mounting plate, and the pressing copper nozzles are in one-to-one correspondence with palladium sheets to be welded. The device has the effects of simplifying mechanism configuration, saving production takt, improving production efficiency and improving welding precision.