37results about How to "Solve balance problems" patented technology

The utility model discloses a kind of mine dump truck antenna support, comprising: support assembly, vertical pole, support base, transition base, buffer base, buffer pad, antenna installation support;Support assembly is frame structure, support assembly is provided with assembly connecting end head in bottom, vertical pole lower end is hinged with support base, vertical pole upper end is hinged with assembly connecting end head;Transition base is hinged in support assembly lower part, transition base, assembly connecting end head are located at the both sides of support assembly respectively;Antenna installation support is connected in support assembly top, buffer base is connected in support assembly, buffer base is located between antenna installation support, transition base, buffer pad is installed between support assembly, buffer base. The utility model can effectively solve the balance and stability problem of antenna in driving and unloading process, provide reliable guarantee for high-precision positioning of unmanned mine dump truck.

The present disclosure relates to the technical field of intelligent robots, and specifically provides a robot, a control method and device thereof, and a storage medium. A robot control method includes obtaining first attitude information of a robot and determining second attitude information of a standing surface according to the first attitude information, determining a target dynamic equation according to a dynamic equation of the robot and a dynamic equation of the standing surface, establishing a control optimization task of the robot based on the target dynamic equation, obtaining target torques of each joint of the robot based on the first attitude information, the second attitude information and the control optimization task, and controlling movements of each joint of the robot according to the target torques. In the present disclosure, through dynamic analysis and optimization control of the attitude of the robot and the attitude of the standing surface, the balance problem of the robot on a dynamic standing surface is solved, the balance control of the robot on the dynamic standing surface is realized, the application scenario of the robot is expanded, and the control stability of the robot is improved.

The utility model relates to a high-strength light-weight vehicle cellular board which comprises a top layer cellular hole, a bottom layer cellular hole and a middle cellular hole, the top layer cellular hole, the bottom layer cellular hole and the middle cellular hole form a cellular hole unit, and peripheral holes incline towards a central hole at the same curvature. Through the unique honeycomb hole structure design, stress is effectively dispersed, the strength is improved, meanwhile, the weight is reduced, and the problem that the strength and the light weight of an existing vehicle honeycomb plate are difficult to balance is solved. Honeycomb holes are of a hexagonal structure and are tightly arranged to form a honeycomb shape, material consumption is reduced, and light weight is achieved. Connecting edges are arranged at the tops and the bottoms of the top-layer honeycomb holes and the bottom-layer honeycomb holes and used for being connected with other parts, the connecting edges and the honeycomb holes are connected through reinforcing ribs, and the edge strength is enhanced. Edge sealing structures are arranged on the side faces of the cellular board, internal cellular holes are prevented from being damaged, and connecting holes are formed in the edge sealing structures and used for being fixedly connected with other parts. The cellular board is made of high-strength aluminum alloy.

The invention provides a lead halide perovskite seed crystal and a preparation method thereof, and a lead halide perovskite microcrystal and a preparation method thereof. The preparation method of the lead halide perovskite seed crystal comprises the following steps: mixing a pseudo-halogen additive, first AX, first BX2 and a first solvent to form first sol, and heating the first sol to separate out crystals to obtain the lead halide perovskite seed crystal. The microcrystal is prepared by using the seed crystal as a raw material through slow heating crystallization. According to the method provided by the invention, the crystallization quality can be improved while the crystallization speed is increased.

The invention discloses a thoracolumbar vertebra fracture detection method and system based on sequence modeling and a prior graph, and the method comprises the steps: obtaining thoracolumbar vertebra CT volume data and fracture type truth value label data, carrying out the preprocessing, and constructing a training set and a test set; and constructing a fracture detection multi-branch classification network model, training and testing the constructed fracture detection multi-branch classification network model by using the training set and the test set, preprocessing thoracolumbar vertebra CT body data of a to-be-detected patient, inputting the preprocessed thoracolumbar vertebra CT body data into the trained fracture detection multi-branch classification network model, and outputting a fracture type classification prediction result. By designing an end-to-end framework of an image feature branch, a label relation branch and a branch fusion module, the problems that in the prior art, continuity of a vertebral body structure is deficient, long-range dependence is insufficient, feature fusion is insufficient and the like are solved, and accurate detection and refined classification of thoracolumbar vertebral fractures are achieved.

The application belongs to the technical field of robot control, and particularly relates to a master-slave operating arm with electromagnetic force feedback for precision assembly and a bidirectional transparent control method, which comprises a master-slave robot unit, an information transmission unit, a master-slave control unit, an electromagnetic force feedback unit, a bidirectional transparent control unit, a singular point processing unit, an adaptive path planning unit and a master-slave proportion matching unit. The electromagnetic force feedback unit obtains environmental force information through a magnetic coupling component and transmits the information to the master control unit; the bidirectional transparent control unit constructs a master-slave control closed loop, balances the system transparency and stability under communication delay, detects and avoids joint singular points through a Jacobian matrix, and constructs a potential field function to generate an optimal motion path. The application realizes high-precision control and force feedback under a communication delay environment, and improves the accuracy and safety of remote operation.

This invention provides an amino-functionalized transparent impact-resistant isophthalic resin and its preparation method, belonging to the field of functional polymer material synthesis. The amino-functionalized transparent impact-resistant isophthalic resin is a multi-component copolymer of styrene, isoprene, and a p-chloromethylstyrene derivative monomer containing tertiary amine functional groups. The preparation method employs a classic anionic polymerization method. First, an amino-functionalized polystyrene active segment SN1 is prepared. Then, it is added to a styrene / isoprene mixed monomer for a two-stage reaction to obtain a block copolymer segment SN1-SP with alternating styrene / isoprene sequences. Finally, styrene and the amino-functionalized monomer are added for a three-stage reaction or directly coupled with a coupling agent to obtain a special block copolymer SN1-SP-SN2, successfully yielding the amino-functionalized transparent impact-resistant isophthalic resin. By controlling the length of the special alternating sequence molecular chain and the block composition ratio, the mechanical properties of the product are controlled. Simultaneously, the tertiary amine polar groups are quantitatively introduced into the molecular chains, successfully achieving high impact resistance and transparency, and also enhancing the resin polarity. Using a lithium-based polymerization method, the resin particles are free of monomer and harmful impurities.

The embodiment of the invention relates to the technical field of intelligent assistants, and provides an intelligent agent task execution method and system, a medium and electronic equipment. The method comprises the steps that a first end sends a target task to a second end; the second end starts from the unified starting point state of the target task, parallel exploration is carried out through the multiple branch agents, and a target path corresponding to the target task is obtained; the second end sends the target path to the first end; and the first end executes the target task based on the target path by using the main agent. According to the method provided by the invention, the branch agent is utilized to perform path exploration at the second end to obtain the target path, and the main agent executes the target task at the first end, so that the task execution success rate of the main agent can be improved.

The application discloses a knowledge-based temporal convolution neural network (KB-TGCN) blast furnace fault diagnosis method. The method can solve the space-time dependence and sample imbalance problem of the blast furnace ironmaking process. First, the calculated variables that can further reflect the process state are calculated by using the variables directly detected by the sensor. Then, the knowledge-based graph structure is constructed according to the spatial position and calculation relationship of the variables. Subsequently, a one-dimensional time information extraction module is embedded in the graph convolutional neural network. Therefore, the TGCN can capture time information while maintaining the original spatial relationship. By using the knowledge-based graph structure, the KB-TGCN can fully mine the spatial information of different blast furnace positions. In addition, the method uses a focal loss function with an adaptive balance factor instead of a traditional cross-entropy loss function to overcome the sample imbalance problem.

This invention belongs to the field of materials science, specifically relating to a crystal structure generation method based on variational autoencoders and Cartesian coordinate derivation. The aim is to construct a crystal structure design method driven by target properties. The method includes: acquiring real crystal structure data and performing data augmentation processing; training a variational autoencoder using the crystal structure dataset as the real sample and the target material properties as the input conditions; converting the real crystal structure data in the crystal structure dataset into crystal diagrams, performing feature derivation on the edges of the crystal diagrams based on the geometric information of atoms in the Cartesian coordinate system, and training a graph neural network based on the edge-derived features; inputting the target material properties into the fully trained variational autoencoder to generate candidate crystal structures; inputting the candidate crystal structures into the fully trained graph neural network for property prediction; comparing the predicted property values ​​with the target material properties within a preset target tolerance range and selecting the desired crystal structure.

The utility model belongs to the technical field of power capacitor manufacturing, and particularly relates to a composite capacitor core shaft and a capacitor element. The composite capacitor core shaft comprises a core shaft body provided with a regular hexagon through hole, the core shaft body is composed of insulators, annular long holes are formed in the wall thickness of the two ends of the core shaft body in the length direction of the core shaft body, and an insulator interval is arranged between the two ends, close to each other, of the two annular long holes to form isolation. Heat conduction metal bodies are inserted into the two annular long holes respectively, the heat conduction metal bodies and the mandrel body are integrally formed, and the ends, away from each other, of the two heat conduction metal bodies make contact with metal spraying layers at the two ends of a metallized film on the capacitor element so that heat transfer of the capacitor element can be achieved. The internal hot spot temperature of the capacitor element is effectively reduced, the transverse heat conduction efficiency of the capacitor element is improved, the heat transfer effect is further improved, and the problem that heat dissipation and heat balance in the capacitor element with the enlarged geometric size become poor is solved.

The present application belongs to the field of reusable launch vehicle trajectory planning and intelligent control, and relates to a reusable launch vehicle cruise phase intelligent trajectory planning method. By constructing a three-degree-of-freedom mass center motion dynamics model of the cruise phase launch vehicle, an offline optimal trajectory sample library covering a wide range of working conditions is generated based on the Gauss pseudospectral method, a deep neural network is designed to learn the mapping relationship between task parameters and trajectory features, and high-quality trajectory planning initial values are output online. In combination with the adaptive collocation method, the collocation distribution is dynamically adjusted, and the nonlinear programming problem is iteratively solved with the high-quality initial values as the starting point, thereby realizing efficient and accurate optimization of the trajectory. Simulation results show that the method significantly reduces the sensitivity of online optimization to initial guesses, still has fast response capability and high robustness under complex constraints and uncertain conditions, effectively improves the autonomous trajectory planning capability of the reusable launch vehicle cruise phase, and guarantees flight reliability and control accuracy.

This invention provides a large-image-area, long-working-distance objective lens system for micro-projection, comprising: a first projection lens, a first lens group (1), a first aperture stop (3), a second lens group (2), a first beam splitter (4), a first protective glass (5), and a first imaging surface (6) arranged sequentially; light emitted from the first projection lens passes through the first lens group (1), the second lens group (2), the first beam splitter (4), the first protective glass (5), and the first imaging surface (6), and forms an image on the first imaging surface (6). This invention, by setting up the lens group and specifying the material and number of lenses, solves the problems of low image quality, imbalance between lens size and mass, and excessive cost caused by projection over long distances.

The invention relates to a power distribution network fault identification method based on a whale optimization algorithm LSSVM, and belongs to the technical field of power system fault diagnosis, the method uses an improved TS-GAN model to solve the problem of insufficient fault samples, and provides sufficient data support for subsequent diagnosis; the fault starting moment is efficiently determined by using a break variable monitoring algorithm, and complex mathematical transformation and model establishment do not need to be carried out; according to the DBO-VMD algorithm, the optimization efficiency and precision of key parameters are improved, complex fault signals can be effectively decomposed, IMF with abundant features can be separated out, envelope demodulation analysis is carried out by selecting a maximum kurtosis IMF component, and features such as envelope information are extracted; according to the WOA-LSSVM algorithm, kernel parameters and penalty factors of the LSSVM are optimized, local extreme values are avoided, various short-circuit faults of the power distribution network are recognized, more accurate and more comprehensive decision support is provided for operation and maintenance of the power distribution network, and finally the operation efficiency and reliability of the power distribution network are improved.

The invention belongs to the technical field of video quality evaluation, and particularly relates to a neural typing method based on a multi-modal brain map and semi-supervised deep clustering. Comprising the following steps: constructing a multi-modal brain network according to multi-modal neural image data; the brain network of each mode comprises a normal brain network and an abnormal brain network; performing pre-training on the MBVAE model by adopting a multi-modal brain network to obtain a pre-trained MBVAE model; performing fine tuning training on the pre-trained MBVAE model in combination with a clustering module to obtain a trained deep embedded clustering model; obtaining multi-modal neural image data of a user, constructing a multi-modal brain network, and inputting the multi-modal brain network into the trained deep embedded clustering model for processing to obtain a neural typing result; according to the method, the accuracy and reliability of ASD neural typing results are improved, the problems of gender imbalance and multi-site heterogeneity existing in ASD data are effectively solved, and the robustness and generalization ability of the model are enhanced.

This invention provides an intelligent prediction method for low-resistivity oil reservoirs based on hierarchical ensemble learning. This method uses small-layer data as a benchmark, integrates heterogeneous data from multiple sources such as well logging, well logging, and production data, and constructs a labeled dataset after standardized preprocessing. Multi-dimensional features are automatically extracted using the Tsfresh framework, and key feature subsets are selected using random forest. Subsequently, generative adversarial networks that integrate noise filtering, adaptive clustering, and residual connections are used to optimize sample distribution, addressing data imbalance and insufficient sample problems. A model cluster containing traditional machine learning models and convolutional hybrid neural networks is constructed, adapting to static and temporal feature learning respectively. A hierarchical stacking ensemble strategy is adopted, using the unbiased prediction results of five types of base learners as meta-features, and integrating the advantages of each model through meta-learners. Furthermore, this invention also provides an intelligent prediction device for low-resistivity oil reservoirs based on hierarchical ensemble learning. The technical solution provided by this invention can improve the prediction accuracy and generalization ability of low-resistivity oil reservoirs under complex geological conditions, thereby increasing the recovery rate.

The invention belongs to the technical field of high polymer materials, and particularly relates to a sulfonated porous aromatic framework doped sulfonated polybenzimidazolyl polymer electrolyte membrane and a preparation method thereof. The preparation method comprises the following steps: preparing sPAF and sPBI, dispersing the sPAF in dimethyl sulfoxide, dissolving the sPBI in DMSO, uniformly blending and stirring the sPAF and the sPBI, pouring the mixed solution on a clean glass plate, and drying to prepare a uniform and transparent composite film; soaking the composite membrane in sulfuric acid to remove inorganic salt in the membrane, and then washing; obtaining a film for low temperature use; and soaking the composite membrane in phosphoric acid to obtain the phosphoric acid doped polymer electrolyte membrane used at high temperature. The composite membrane disclosed by the invention can absorb water and efficiently conduct protons by virtue of a sulfonate hydrophilic group at a low temperature, and can retain phosphoric acid by virtue of a siphoning effect of a porous structure and an imidazole group on PBI in a high-temperature environment, so that the equilibrium effect of conductivity and temperature of a current proton exchange membrane material is solved.

This invention belongs to the field of deep learning clustering, specifically relating to an ASD neural typing method based on synthetic brain map data augmentation. The method includes: acquiring brain map data for training and inputting it into a pre-trained dual-decoder graph autoencoder to obtain latent embeddings; inputting the latent embeddings into a pre-trained latent space conditional diffusion model to obtain an augmented brain map; training a deep embedding clustering model based on MBVAE or a deep graph clustering model based on HGDAE using the augmented brain map to obtain a trained deep embedding clustering model based on MBVAE or a deep graph clustering model based on HGDAE; acquiring brain map data to be detected and inputting it into the trained deep embedding clustering model based on MBVAE or a deep graph clustering model based on HGDAE to obtain the ASD neural typing result. This invention effectively solves the gender imbalance and multi-site heterogeneity problems existing in ASD data, enhancing the robustness and generalization ability of the model.

This invention provides a method and apparatus for UAV target detection based on infrared and visible light image fusion. The method includes: simultaneously acquiring images of the same scene in two modalities using sensors mounted on the UAV; extracting global illumination information from the visible light image to generate illumination weights; calculating weighting coefficients for the two modalities based on the illumination weights; weighting the visible light feature map and the infrared feature map using the corresponding weighting coefficients to obtain weighted features; determining the frequency domain difference between the weighted features of the two modalities; enhancing the weighted features based on the frequency domain difference to obtain enhanced features; concatenating the enhanced features of the two modalities and inputting them into a global branch and a local branch respectively to obtain a global offset field and a local offset field, which are then superimposed to obtain a final offset field; implicitly aligning the infrared enhanced features using the final offset field; and concatenating the aligned enhanced features of the two modalities and inputting them into a target detection head to obtain the target detection result. This invention improves the accuracy and robustness of target detection.

The invention discloses a keratoconus element learning diagnosis method under an unbalanced small sample condition. The method comprises two parts of model training and auxiliary reasoning. According to the model training part, firstly, an acquired corneal topographic map is preprocessed and labeled, and then a meta-learning task ensuring complete categories is constructed by adopting a balance task sampler. And a pre-trained Swin Transform is used as a feature extractor to acquire the multi-scale depth features of the image. A plurality of prototypes are generated for early keratoconus categories with high diagnosis difficulty by using K-means clustering, and other categories are uniformly grouped. Classification is completed by calculating the distance between a query sample and a prototype and carrying out probability aggregation, and meta-learning optimization is carried out by adopting a joint loss function combining enhanced focus loss and contrast learning loss. And the auxiliary reasoning part is used for extracting input image features by using the trained model, comparing the input image features with the learned prototype, and outputting a diagnosis category and confidence. According to the method, the auxiliary diagnosis accuracy and generalization ability of the keratoconus are remarkably improved.

The invention relates to the technical field of permanent magnet synchronous motors, and particularly discloses a structure capable of comprehensively weakening cogging torque and electromagnetic vibration of a permanent magnet synchronous motor, which comprises a rotating shaft, a rotor and a stator which are coaxially sleeved from inside to outside, the rotor comprises a rotor iron core which coaxially sleeves the rotating shaft, and a rotor section I, a rotor section II, a rotor section III and a rotor section IV which are arranged on the rotor iron core at equal intervals along the axial direction, and the rotor section I and the rotor section IV comprise a plurality of permanent magnet magnetic steels I which are circumferentially arranged at equal intervals; the second rotor section and the third rotor section comprise second permanent magnet magnetic steel which is the same in number and arranged in the circumferential direction at equal intervals, magnetic poles on the two sides of the first permanent magnet magnetic steel and the two sides of the second permanent magnet magnetic steel offset in the circumferential direction at equal angles, and the third rotor section and the fourth rotor section have the same axial dislocation offset relative to the first rotor section and the second rotor section. The problem that synchronous suppression of cogging torque and radial electromagnetic force is difficult to realize by an optimized structure of a traditional permanent magnet synchronous motor is solved.

The application provides a multi-dimensional identity recognition method and device, electronic equipment and program product, and relates to the technical field of artificial intelligence. The method comprises the following steps: collecting video stream data containing a to-be-recognized object, extracting multi-dimensional feature information of the to-be-recognized object based on a first target image in the video stream data, recognizing the identity of the to-be-recognized object according to face information, if the recognition fails, performing target tracking on the to-be-recognized object based on a second target image before the first target image in the video stream data to determine whether a third target image containing the identity recognition result of the to-be-recognized object exists in the second target image, and if not, recognizing the identity of the to-be-recognized object according to body key point information. Through face recognition, target tracking and identity recognition based on body key points, the high accuracy of face recognition is retained, and when face recognition fails, identity recognition can be accurately performed based on body key points, so that the balance between the high accuracy and the easy failure of face recognition is achieved.

The invention discloses a vehicle-mounted-cloud road time health multi-dimensional analysis and risk assessment system, and belongs to a vehicle-mounted analysis and assessment technology. Capacitance, pressure and radar signals are fused, the defect of a single sensor can be overcome, a lightweight self-adaptive filtering model generated based on hardware sensing meta-neural architecture search is adapted to a vehicle-mounted edge computing power constraint and interference feature real-time sensing mechanism, so that the model can adjust a filtering strategy according to vehicle vibration and illumination changes, and the vehicle-mounted edge computing power constraint and interference feature real-time sensing mechanism is optimized. The dynamic scene self-adaptive capability is enhanced, the delay of the generation of the health state snapshot of the edge end can be effectively reduced through model quantification and hardware optimization, and the occupancy rate of vehicle-mounted computing resources can be effectively reduced by unloading a complex model training task to the cloud and only executing lightweight reasoning by the edge end. The parameter updating strategy is dynamically adjusted based on the communication quality, the data transmission amount can be effectively reduced in a weak communication environment, and the model updating failure caused by communication fluctuation is avoided.

The invention discloses an intelligent identification and classification method for a TAP condensate microscopic image based on deep learning, and relates to the technical field of biomedical image processing and artificial intelligence. A TAP condensate microscopic image data set is constructed and preprocessed, and a deep learning identification model based on a convolutional neural network is constructed; the deep learning recognition model comprises a feature extraction backbone network, an attention mechanism module and a classification output layer, training the deep learning recognition model by using the training set, optimizing model parameters through a loss function, inputting a to-be-detected microscopic image into the trained deep learning recognition model, and outputting the to-be-detected microscopic image. The Label-free TAP condensate recognition method has the advantages that dependence on physical color marking in the prior art is abandoned, the high-dimensional features of the image are automatically extracted through deep learning, and accurate recognition of the Label-free TAP condensate is achieved.

This application relates to the fields of polymer synthesis and coating technology, specifically providing a reactive polyether silane and its stabilized alkali metal silicate coating; the reactive polyether silane is prepared by an addition reaction of an unsaturated silane and a polyether amine; the structure of the polyether amine is: R'-[OCH2-CHR]. x -[OCH2-CH(CH3)] y The coating contains -NH2, with an x / y ratio of (1-55):(1-36), an HLB value of 0.5-20 for the polyetheramine, and a weight-average molecular weight of 500-6000; the mass ratio of unsaturated silane to polyetheramine is 100:(2-70); the coating composition includes 10-50% by mass of reactive polyether silane-stabilized alkali metal silicates. This application demonstrates that a small amount of polyether silane can effectively control the thermal stability of alkali metal silicate coatings, solving the problem of performance degradation after thermal storage, and possesses significant application value.

The application discloses a kind of sludge solidifying agent using carbon fixation desulfurization ash, by cement clinker 12~15%, mineral powder 50~55%, steel slag powder 15~20%, carbon fixation desulfurization ash 10~15%, solid additive 1~2%, liquid additive 1~2% composition by mass percentage.The application uses desulfurization ash to fix CO2 in industrial tail gas to improve the activity of desulfurization ash, then carbon fixation desulfurization ash is compounded with cement clinker, mineral powder, steel slag powder, solid additive, liquid additive to obtain sludge solidifying agent.The application does not need to use various chemical means of flotation agent to enrich the carbon fixation component in desulfurization ash, and the cementation activity of desulfurization ash after carbon fixation is greatly improved, can better play sulfate activation effect, and with additive play superposition coupling effect, and with cement clinker, a kind of cementing material system, can jointly prepare sludge solidifying agent, to waste treat waste at the same time, improve the solidification strength of the sludge solidifying agent to sludge and the solidification effect of heavy metals in sludge.

The invention provides a terminal positioning method based on a CIS and electronic equipment, and relates to the technical field of communication. The method comprises the following steps: constructing a channel state information (CSI) fingerprint database of an area to be positioned, and constructing a CSI fingerprint map based on the CSI fingerprint database; wherein the CSI fingerprint database comprises CSI amplitude information of a plurality of reference points; training a feature extractor by adopting a contrast learning strategy based on the CSI fingerprint map, and training a positioning network based on the trained feature extractor; wherein the positioning network comprises a feature extractor, a predictor and a multi-layer sensor; and inputting a real-time CSI measurement value acquired by a to-be-positioned terminal into the trained positioning network, and outputting a predicted coordinate of the to-be-positioned terminal. According to the invention, the problem of cost and precision balance of indoor positioning in a small sample scene can be solved, and the accuracy and robustness of positioning are improved.

A deep learning-based method and system for detecting rail damage belongs to the field of rail transit monitoring and damage identification technology. The method includes: acquiring an ultrasonic two-dimensional image of the rail to be detected; inputting the two-dimensional image into a pre-trained neural network model, outputting detection results including damage category and location coordinates; neural network model training includes: acquiring a training sample set containing labeled information; inputting the training sample set into the neural network model, and having the model output predicted bounding boxes representing the predicted location and size of the damage; calculating the error between the predicted bounding box and the ground truth bounding box using a bounding box regression loss function, and updating the model parameters through backpropagation; the bounding box regression loss function includes an overlap loss term, a center distance loss term, and a size penalty term, and is adjusted by a dynamic weighting factor. This invention solves the problems of imbalanced positive and negative samples and uneven distribution of easy and difficult samples during training, improving the detection accuracy and robustness for small targets and hidden damage.

The application discloses a kind of construction method and application of intelligent signal detection model, belong to mobile communication technical field;The application constructs a new type of graph neural network containing aggregation update module, the global feature of each node is aggregated with its neighbor node information, so as to extract the hidden feature state of data according to the feature information of neighbor node and itself, and then realize the effective capture of the feature of communication signal such non-euclidean space data;In addition, after each aggregation is completed, the intermediate feature obtained by aggregation is further extracted in the application Feature, to improve the expression ability of intermediate feature, so that the designed graph neural network has more efficient information interaction ability and more powerful relationship reasoning ability, greatly improve the precision of signal detection.The application effectively solves the problem that the performance and computational complexity of traditional signal detection method are difficult to balance, and has strong robustness to channel estimation error.