Abstract: Embodiments of the present disclosure provides a fine-grained image recognition method and apparatus using graph structure represented high-order relation discovery, wherein the method includes: inputting an image to be classified into a convolutional neural network feature extractor with multiple stages, extracting two layers of network feature graphs in the last stage, constructing a hybrid high-order attention module according to the network feature graphs, and forming a high-order feature vector pool according to the hybrid high-order attention module, using each vector in the vector pool as a node, and utilizing semantic similarity among high-order features to form representative vector nodes in groups, and performing global pooling on the representative vector nodes to obtain classification vectors, and obtaining a fine-grained classification result through a fully connected layer and a classifier based on the classification vectors.

Abstract: A holographic 3D display system based on virtual array splicing of a spatial light modulator includes a laser configured to generate a coherent light beam, first, second and third beam splitters, first and second reflectors, a shutter array, a spatial filter array, a solid lens, first and second light beam deflection elements and a spatial light modulator. The first and second beam splitters and the first reflector are configured to split the light beam generated by the laser into three parallel light beams to irradiate the shutter array. The shutter array is configured to control the three parallel light beams to sequentially pass therethrough according to a set time sequence. The three parallel light beams passing through the shutter array are expanded and collimated by the spatial filter array and the solid lens to form three parallel light beams with the same size and uniform intensity.

Abstract: The present disclosure provides a self-regulating electric heating film, including: an insulating isolation layer, an interdigital electrode arranged on the surface of the insulating isolation layer, a positive temperature coefficient coating covering the surface of a secondary electrode of the interdigital electrode, and an insulating protective layer covering the surface of a primary electrode of the interdigital electrode, wherein the positive temperature coefficient coating is not in contact with the primary electrode of the interdigital electrode, and the insulating protective layer overlaps the positive temperature coefficient coating.

Abstract: The present disclosure relates to the field of biomedicine, in particular to a drug delivery system for delivering a drug to cells. In an example of the present disclosure, the drug delivery system includes a drug delivery carrier, where a nucleic acid drug and a photothermal nanoparticle are encapsulated in the drug delivery carrier. The drug delivery system opens up an unprecedented means of intracellular drug delivery. Moreover, all materials of the system are desirable in biocompatibility, friendly to the environment without pollution, and harmless during use. Furthermore, compared with other operating methods, a use method of the system is simpler and more convenient to operate. The system shows low production and application costs and is economical-friendly.

Abstract: This disclosure relates to an immersive ankle rehabilitation training method and electronic apparatus based on upper limb motion signals. The innovation of this technology lies in that: firstly, this disclosure proposes a hybrid model of Regression LSTM corrected by a sliding window based voting algorithm, which predicts the changing angle of the user's wrist joint through wearable EMG sensors; secondly, this disclosure designs and develops two different VR training scenes for users of dorsiflexion and plantar flexion of foot; during the training, users wear VR headwear displays and EMG monitoring devices, to continuously perform wrist dorsiflexion or wrist palm flexion movements, predict the change in wrist joint angles in real-time through forearm sEMG signals, and map it to the contralateral ankle of the avatar in the virtual scene for rotation.

Abstract: The present invention discloses a coevolution-based multi-beamforming optimizing method for space-based ADS-B, which belongs to the field of aviation technologies. According to this method, after initializing population and acquiring an initial optimal solution, optimization parameters are firstly grouped randomly so that all beams form several significant uncovered regions. Then, grouping is performed by an adaptive grouping strategy based on the relative orientation between the mean center of the largest uncovered region and the direction of each beam, so that the parameters quickly converge to a full-coverage state within the half angle of the satellite. Finally, grouping is performed by a grouping strategy based on the aircraft density covered by each beam to obtain optimally the minimal update time interval under a full-coverage constraint.

Abstract: A magnetic particle imaging (MPI) device based on an asymmetric bilateral structure includes a fixed section and a movable section. The fixed section includes a pair of magnetic field generation coils, and the magnetic field generation coils are configured to generate a uniform and variable magnetic field and drive a field-free point (FFP) to move along a Z-axis direction. The movable section includes a permanent magnet, excitation coils and a reception coil. The permanent magnet is configured to cooperate with the magnetic field generation coils to generate the FFP. The excitation coils are configured to move the FFP on a two-dimensional plane and excite magnetic particles to generate a nonlinear response signal. The reception coil is configured to receive an MPI signal. The MPI device is simple in structure and convenient to operate, and can perform targeted local imaging or covered overall imaging on the target object.

Abstract: Embodiments of the present application provide a MEMS solenoid transformer, comprising a silicon substrate, a soft magnetic core, a first solenoid and a second solenoid; wherein, the soft magnetic core is wrapped inside the silicon substrate, the silicon substrate is provided with a first spiral channel and a second spiral channel, and two opposite sides of the soft magnetic core respectively pass through a center of the first spiral channel and a center of the second spiral channel; the first solenoid and the second solenoid are respectively disposed in the first spiral channel and the second spiral channel. By disposing the soft magnetic core, the first solenoid and the second solenoid of the transformer inside the silicon substrate completely, the thickness of the silicon substrate is fully utilized, and the obtained transformer has a larger winding cross-sectional area and a higher magnetic flux, increasing the inductance value of the transformer.

Abstract: The present invention discloses a method for UAV path planning in urban airspace based on a safe reinforcement learning (RL) algorithm called shield-DDPG, which combines a shield model with a DDPG algorithm and pertains to the field of UAV technologies. The method takes an attractive force from the destination point into account when an action is selected, which improves the convergence speed of the algorithm and also improve the efficiency of UAV path planning. More importantly, the method provided by the present invention can effectively verify safety of an action in terms of the air collision risk and the ground impact risk, and ensure that a final output of the algorithm is a safe optimal solution. Therefore, the present invention can effectively solve the problem that when the RL algorithm is used for UAV path planning, it is difficult to ensure the safety of the learning or execution process due to the lack of hard constraints.

Abstract: A global energy management optimization method in multi-task cross-core deployment for hybrid electric vehicles, belonging to the technical field of hybrid electric vehicle energy management, which is capable of solving the problem that conventional hybrid electric vehicles cannot realize global energy management at a vehicle end; through adopting the present invention, the problems relating to the failure of timely achieving global energy management for hybrid electric vehicles, poor vehicle speed performance and driving safety caused by the high difficulty of effectively connecting and deploying different computing tasks of global energy management are solved; a multi-core heterogeneous controller is used to perform multi-task deployment, which solves the problems caused by one type of tasks requiring high computing power but low real-time response while another type of tasks requiring low computing power but high real-time response, ensures the efficient utilization of control chips of the hybrid transmiss

Abstract: The present disclosure provides a cell segmentation and typing method, device, apparatus and medium based on machine learning. The method includes: acquiring at least one cell metabolism image of a target object; performing single cell image segmentation on the at least one cell metabolism image by using a machine learning segmentation model to obtain a plurality of single cell metabolism images; performing single cell feature extraction on each single cell metabolism image of the plurality of single cell metabolism images to obtain a single cell image feature map corresponding to the single cell metabolism image; combining the single cell image feature map corresponding to each single cell metabolism image of the plurality of single cell metabolism images to obtain an image feature map of the target object; performing typing of the cell by clustering the image feature map of the target object.

Abstract: The present disclosure provides a coordinated multiple access method for multi-cell ground-to-air data transmission, and belongs to the technical field of wireless communication. The method comprises the following steps: S1, constructing a model of a coordinated multiple access system for multi-cell ground-to-air data transmission; S2, calculating a transmission rate of an aircraft, and constructing a multi-cell airspace and power domain resource allocation optimization problem by taking maximizing a system transmission rate as an optimization objective; S3, constructing a Markov decision process model; S4, solving the optimization problem using a multi-agent deep reinforcement learning algorithm. The system transmission rate is maximized under the condition that the aircraft can decode the signal correctly and satisfy the constraint of minimum transmission rate, thereby realizing ground-to-air high-speed data transmission.

Abstract: The present application provides a CAD model analysis method and apparatus, a device and a storage medium, which relates to the technical field of industrial manufacturing. The CAD model based on a BRep representation is transformed into a graph structure. The graph structure is input into an analytical model to obtain a modeling sequence of the CAD model.

Abstract: An ultrasonic non-destructive testing method and system for detecting lithium precipitation in lithium-ion batteries are provided. The method includes the following steps: acquiring full matrix capture data of the lithium-ion battery using an ultrasonic phased array transducer; inputting the measured ultrasonic data into the initial fully convolutional network (FCN), which outputs a predicted ultrasonic velocity model; using the predicted model as the initial input for an acoustic wave equation that outputs reconstructed ultrasonic data; constructing a loss function using the deviation between the measured and reconstructed ultrasonic data, and iterating until the loss function converges. This process completes the training of the unsupervised FCN. The trained FCN then outputs an accurate reconstructed ultrasonic velocity model that represents the lithium precipitation in the battery.

Abstract: An electromagnetic leakage red and black signal identification method based on information entropy characteristics includes: acquiring electromagnetic information leakage signals of an information device in an idle state and an information processing state respectively, and calculating information entropy of the electromagnetic leakage signals in the idle state and the information processing state; and calculating an information entropy change of the electromagnetic leakage signals caused by information processing, and determining whether an electromagnetic leak contains red signals when the device processes information according to the information entropy change, so as to give an electromagnetic leakage red and black signal identification result.

Abstract: A three-dimensional (3D) magnetic particle imaging (MPI) method and system without rotating a field-free line (FFL) are disclosed. In accordance with the method, an FFL is generated such that magnetic particles in a region far from the FFL enter a magnetic saturation state. Further, a non-uniform mixed-frequency excitation magnetic field parallel to the FFL is applied to generate intermodulation response signals. In addition, the intermodulation response signals are acquired, amplified and filtered. Moreover, the intermodulation response signals are transmitted to a digital signal processing unit and an image reconstruction unit, and an encoding matrix is constructed. Additionally, a one-dimensional (1D) concentration distribution of the magnetic particles is reconstructed based on the encoding matrix and an actually measured voltage signal. The FFL is driven for line-by-line scanning along a vertical plane of the FFL, thereby achieving 3D imaging.

Abstract: An MPI device based on parallel scanning of multiple magnetic field free lines (FFLs) is provided. The MPI device includes a first coil, a second coil, an excitation coil, and a receiving coil; the first coil is configured to receive a first-frequency current and generate a plurality of FFLs; the second coil is configured to receive the first-frequency current and drive the FFLs to rotate under a joint action with the first coil; the excitation coil is configured to receive a second-frequency current and drive the FFLs to scan fast; the receiving coil receives a particle signal scanned by the FFLs at each time point; and the first coil, the second coil, the excitation coil and the receiving coil are provided in longitudinal symmetry along the center of the imaging field of view.

Abstract: An optoelectronic synaptic memristor includes: a bottom electrode layer, a porous structure layer modified with quantum dots, a two-dimensional material layer, a transparent top electrode layer, and a waveguide layer, which are arranged in sequence from top to bottom, wherein the waveguide is ridge shaped for light conduction, comprising a wedge-shaped output terminal, wherein: through the wedge-shaped output terminal of the waveguide, light is vertically injected into the two-dimensional material layer and the porous structure layer modified with the quantum dots. By integrating the waveguide and the optoelectronic memristor, the present invention obtains the highly controlled characteristics with high alignment and confinement for light effect on the device and has advantages in realizing optoelectronic synergy in the optoelectronic synaptic memristors.

Abstract: A steering wheel suspension structure of an omnidirectional mobile robot includes a steering wheel connecting ring, a plurality of leaf spring connecting beams and a chassis mounting section. The steering wheel suspension structure is an integrated plate structure. The steering wheel connecting ring is connected with a steering wheel, and the chassis mounting section is connected with a robot chassis. The leaf spring connecting beam is connected with the steering wheel connecting ring and the chassis mounting section, and the leaf spring connecting beam is used as the main component of suspension deformation. The suspension structure is simple, easy to install, light in weight, less in space, and highly integrated with the steering wheel assembly. When the suspension is deformed by the force, it weakens the radial force and only has the displacement and force in the vertical direction, thus ensuring the handling performance of the chassis.

Abstract: A large-scale UAV mission planning method includes constructing an objective function and its constraints; initializing a mission sequence and inserting customers into the mission sequence with the smallest objective function value to obtain a plurality of initial mission sequences; iteratively performing the update operations on the plurality of initial mission sequences. Fast acquisition of optimal missions is achieved by the invention. This invention solves the existing problem of not being able to obtain the optimal mission due to the complex and time-consuming calculation method and the difficulty of convergence of the algorithm.