Abstract: The present invention discloses deep learning based robot target recognition and motion detection methods, storage media and devices, the method consists of the following steps: Step S1. adding masks to regions where potentially dynamic objects are located through instance segmentation networks incorporating attention mechanisms and positional coding; Step S2, estimation of the camera pose using static feature points outside the instance segmentation mask in the scene; Step S3, estimation of the object pose transformation matrix from the camera pose; Step S4, determining the state of motion of the object's characteristic points from the relationship between motion parallax and differential entropy, and thus the state of motion of the object as a whole; Step S5, rejects the dynamic objects therein and repairs the static background of the rejected area for positional estimation and map construction.

Abstract: This disclosure provides a perovskite light-emitting diode with an adjustable light field, including a glass layer, an anode, a hole transport layer, an emission layer, an electron transport layer and a cathode in sequence from top to bottom. The electron transport layer is provided with a periodic nano-grating structure.

Abstract: Disclosed is a ceramic piezoelectric underwater detection and 5th generation (5G) mobile phone antenna, including: a ceramic cavity, a thin film, electrode feed holes and a coating, where the thin film is embedded inside the ceramic cavity; the electrode feed holes are drilled on both sides of the thin film at one end of a bottom of the ceramic cavity; the coating is coated outside the ceramic cavity for realizing omnidirectional radiation; the thin film drives the ceramic cavity to make the coating on an outer surface of the ceramic cavity generate radiation vibrations and generate electromagnetic wave radiation for communication and underwater detection.

Abstract: A dynamic obstacle avoidance method based on real-time local grid map construction includes: acquiring and inputting Red-Green-Blue-RGBD image data of a real indoor scene into a trained obstacle detection and semantic segmentation network to extract obstacles of different types and semantic segmentation results in the real indoor scene and generate 3D point cloud data with semantic information; according to the 3D point cloud data, extracting and inputting state information of a dynamic obstacle to a trained dynamic obstacle trajectory prediction model, and predicting a dynamic obstacle trajectory in the real indoor scene to build a local grid map; and based on a dynamic obstacle avoidance model, sending a speed instruction in real time to the mobile robot to avoid various obstacles during the navigation process.

Abstract: Disclosed is a dual-band very low frequency antenna, which comprises positive and negative electrodes, silicon substrates, piezoelectric material units, stress-electromagnetic conversion material units and an insulator. The piezoelectric material units are arranged between the positive and negative electrodes; the stress-electromagnetic conversion material units and the silicon substrates are respectively arranged at both ends of the positive and negative electrodes. The positive and negative electrodes are used for driving the piezoelectric material units; the piezoelectric material units are used for generating flutter and conducting the flutter to the stress-electromagnetic conversion material units; the stress-electromagnetic conversion material units are used for converting vibration waves generated by flutter into electromagnetic wave radiation.

Abstract: Disclosed is a multi-layer low-profile four-arm spiral antenna, including a metal grounding plate, and a top surface of the metal grounding plate is fixedly connected with three layers of cylindrical dielectric substrates; the three layers of dielectric substrates are coaxially arranged, and the three layers of dielectric substrates are coaxially arranged with the metal grounding plate, and a gap is set between two adjacent layers of dielectric substrates; four spiral metal strips with a same rotation direction are respectively arranged on each layer of dielectric substrate, the four spiral metal strips on each layer of dielectric substrate form spiral radial arms, and a phase difference of the four spiral metal strips on each layer of dielectric substrate is 90°; the corresponding spiral metal strips on the two adjacent layers of dielectric substrates are connected by connecting bridges, the connecting bridges are arranged in the gap.

Abstract: Disclosed are a hydro-mechanical coupling experimental device with CT real-time scanning and a use method thereof. The hydro-mechanical coupling experimental device with the CT real-time scanning includes a CT scanning room and further includes a support frame, a hydro-mechanical coupling mechanism and a jack that are arranged in the CT scanning room. The support frame includes a base, a top plate, a plurality of columns for arranging the top plate and the base at intervals, and a movable plate that is arranged between the top plate and the base and can slide on the plurality of columns. The hydro-mechanical coupling mechanism includes an experimental box, a pressure box arranged inside the experimental box and a compression leg slidingly worn on a top of the experimental box; and the experimental box is arranged on the movable plate, and the jack is arranged on the base.

Abstract: The metal ceramic powder with a particle size of 15-45 µm and suitable for thermal spraying is prepared through a combination of mechanical ball milling, spray granulation, and vacuum sintering. The metal ceramic powder is sprayed on a surface of a steel substrate adopting the high velocity oxygen fuel (HVOF) technology with oxygen-propane as fuel and taking oxygen as a combustion improver, propane as fuel, nitrogen as powder feeding carrier gas, and air as a cooling medium to prepare and form the NiCrBSi—ZrB2 composite coating. The present disclosure solves the problem that ZrB2 ceramic is difficult to compact during sintering and improves powder bonding strength and fluidity. The preparation method is simple, has advantages of high coating deposition efficiency and convenient equipment operation, and is cost-effective. The preparation method can improve thermal corrosion resistance and high-temperature wear resistance of a surface of boiler, and prolonging lifetime of the boiler.

Abstract: The present disclosure provides a method for fine-grained detection of driver distraction based on unsupervised learning, belonging to the field of driving behavior analysis. The method includes: acquiring distracted driving image data; and inputting the acquired distracted driving image data into an unsupervised learning detection model, analyzing the distracted driving image data by using the unsupervised learning detection model, and determining a driver distraction state according to an analysis result. The unsupervised learning detection model includes a backbone network, projection heads, and a loss function; the backbone network is a RepMLP network structure incorporating a multilayer perceptron (MLP); the projection heads are each an MLP incorporating a residual structure; and the loss function is a loss function based on contrastive learning and a stop-gradient strategy.

Abstract: The present disclosure provides a transformer-based driver distraction detection method and apparatus, belonging to the field of driving behavior analysis. The method includes: acquiring districted driving image data; building a driver distraction detection model FPT; inputting the acquired distracted driving image data into the driver distraction detection model FPT, analyzing the distracted driving image data by using the driver distraction detection model FPT, and determining a driver distraction state according to an analysis result. The present disclosure proposes a new network model, i.e., a driver distraction detection model FPT, based on Swin, Twins, and other models. Compared with a deep learning model, the FPT model compensates for the drawback that the deep learning model can only extract local features; the FPT model improves the classification accuracy and reduces the parameter quantity and calculation amount compared with the transformer model.

Abstract: The present disclosure provides a fire source detection method and device under the condition of a small sample size, and a storage medium, and belongs to the field of target detection and industrial deployment. The method includes the steps of acquiring fire source image data from an industrial site; constructing a fire source detection model; inputting the fire source image data to the fire source detection model, and analyzing the fire source image data via the fire source detection model to obtain a detection result, where the detection result includes a specific location, precision and type of a fire source. By means of the method, the problems of insufficient sample capacity and difficulty in training under the condition of a small sample size are solved, and different enhancement methods are used to greatly increase the number and quality of samples and improve the over-fitting ability of models.

Abstract: A contextual visual-based synthetic-aperture radar (SAR) target detection method and apparatus, and a storage medium, belonging to the field of target detection is described. The method includes: obtaining an SAR image; and inputting the SAR image into a target detection model, and positioning and recognizing a target in the SAR image by using the target detection model, to obtain a detection result. In the present disclosure, a two-way multi-scale connection operation is enhanced through top-down and bottom-up attention, to guide learning of dynamic attention matrices and enhance feature interaction under different resolutions. The model can extract the multi-scale target feature information with higher accuracy, for bounding box regression and classification, to suppress interfering background information, thereby enhancing the visual expressiveness.

Abstract: Disclosed are a mine dust real-time detection system based on double-photoacoustic spectrometry and a detection method. The mine dust real-time detection system based on double-photoacoustic spectrometry includes a first sampling unit, a first photoacoustic detection cavity, a second sampling unit, a second photoacoustic detection cavity, a signal unit and a processing unit; the first sampling unit is used for sampling in respective, the first photoacoustic detection cavity provides a photoacoustic effect field to substances sampled by the first sampling unit, the second sampling unit is used for sampling in respective, the second photoacoustic detection cavity provides the photoacoustic effect field to substances sampled by the second sampling unit, the signal unit is used for providing a laser signal, and the processing unit is used for collecting and processing a photoacoustic signal.

Abstract: A high sensitivity flexible three-dimensional force tactile sensor includes a hemispherical contact, wherein the hemispherical contact includes a tray with a groove on the surface and a hemispherical protrusion arranged in the groove. A flexible inverted cone component connected to the lower surface of the hemispherical contact, wherein a plurality of flexible triangular excitation electrode is arranged on the side surface of the flexible inverted cone component. A flexible common electrode surrounding part of the flexible triangular excitation electrode, wherein a first cavity with an opening is opened inside the flexible common electrode, parts of the flexible triangular excitation electrode and the flexible inverted cone component are arranged in the first cavity of the flexible common electrode. The flexible triangular excitation electrode and the flexible inverted cone component have no contact with the inner wall of the first cavity of the flexible common electrode to form an air cavity.

Abstract: The present invention proposes a coal and gangue separation device, including a first supporting seat, a guide part and multiple material bins. The guide part is hinged with the first supporting seat; the guide part rotates in a vertical direction; and inlets of the plurality of material bins are sequentially distributed on a rotation track of a movable end of the guide part. In the coal and gangue separation device of the present invention, the high-pressure airflow of a high-pressure nozzle does not directly act on the materials, but on grate bar driving mechanisms; and cantilever grate bars are driven to rotate by the grate bar driving mechanisms so as to directionally discharge the mineral aggregate. Therefore, the gangue can accurately enter a gangue collection bin; and the clean coal can accurately enter a clean coal collection bin.

Abstract: Disclosed is a rake-free thickening device including driving area. The device includes a feed assembly, a diversion assembly and a clean coal collection assembly. The clean coal collection assembly includes a driving area. The diversion assembly includes a central tank. Slime water passes through the feed assembly and flows with a medicament from an upper part of the central tank to a middle of the central tank, and then diffuses around. Bubbles carry the fine slime up after reacting. The driving zone drives the dispersed bubbles to a defoaming zone located in the middle of the central tank. The slime water in the central tank flows through the central tank after defoaming. With the continuously filling of slime water, the slime water above the central tank overflows the central tank to the clean coal collection assembly within the diversion and settlement area.

Abstract: A pitch control method and system of a symmetrical-airfoil vertical axis wind turbine collects data by an anemometer, an anemoscope and an angle sensor, outputs an optimum pitch angle based on a control law of a pitch angle and controls the pitch angle to be the optimum pitch angle through a pitch control actuator. In addition to input variables of the control law such as a wind velocity vin and a blade azimuth angle ?, constants such as a rotation radius R, a rotation velocity ? of the blade and aerodynamic coefficients c1, c2 and c3 are also related. A Reynolds number has little influence on three aerodynamic coefficients c1, c2 and c3. The pitch actuator controls the adjustment rods to realize the automatic pitch control of the blades. An expression of the control law of the pitch is concise, the calculation time is short, and a response speed is fast.

Abstract: The present disclosure discloses a method for preparing 2,5-dimethylphenol by selective catalytic conversion of lignin, relates to the technical field of chemistry, and includes the following steps: mixing lignin, a catalyst, and ethanol, and then carrying out a catalytic conversion reaction of lignin under the gaseous supercritical conditions of ethanol; and cooling the reaction product by quenching after the completion of reaction, and then subjecting it to separation and extraction to obtain 2,5-dimethylphenol. The catalyst comprises a modified sepiolite carrier, an active metal Mo, and auxiliary agents Zr and Fe. The process of the present disclosure is simple, and the prepared catalyst is a solid catalyst, which avoids problems of difficult recovery, serious environmental pollution and equipment corrosion caused by the use of homogeneous organic acid-base catalysts.

Abstract: A flocculating, grading, and dewatering device includes a tank body, a mixing zone, a flocculent settling zone, a centrifugal dewatering zone, and a particle screening zone. An initial material is mixed with a chemical agent for a flocculation reaction, a lower sediment material settles and an upper liquid overflows. In the centrifugal dewatering zone, a screen basket is allowed to rotate to facilitate a centrifugal movement of the sediment material, water in the sediment material moves into a guide cavity and then flows to a centrifugate outlet, and then a dewatered material in the screen basket is removed out of the screen basket. In the particle screening zone, a screening mechanism screens particles of different size grades in the dewatered material, and water generated flows downward and is discharged; liquids discharged from the centrifugate outlet and the overflow port are combined, and then discharged through a mixed liquid outlet.

Abstract: A dosimetry assessment method for an organ at risk (OAR) in an esophageal radiotherapy plan includes: acquiring volumetric-modulated arc therapy (VMAT) plan data of a patient with esophageal cancer, and obtaining a distance-target histogram (DTH); subjecting a geometric feature vector of the DTH to dimension reduction to obtain a dimension-reduced geometric feature vector; establishing a deep belief network (DBN) model, and completing training the DBN model; non-linearly fitting a correlation between a dose feature vector and the dimension-reduced geometric feature vector to obtain a dose feature vector with a dimension being identical to a dimension of the dimension-reduced geometric feature vector; and reconstructing the dose feature vector through a decoding layer of an autoencoder structure to obtain a dose feature vector with a dimension being identical to a dimension of the geometric feature vector before the dimension reduction, and finally obtaining a dose-volume histogram (DVH) for predicting the OAR