Abstract: A method for diagnosing transformer fault based on a deep coupled dense convolutional neural network, includes: obtaining datasets of dissolved gas in oil of a transformer in normal and fault states; expanding the datasets by using an adaptive synthetic oversampling method; performing, in a form of a two-dimensional matrix, feature reconstruction on characteristic gas dissolved in the oil; building a transformer fault diagnosis model based on a deep coupled dense convolutional neural network; and dividing an expanded dataset into a training set and a test set, and taking the two-dimensional matrix as an input of the deep coupled dense convolutional neural network and a set label as an output to train the network to obtain a fault diagnosis model. The present disclosure can resolve a problem that a fault diagnosis accuracy rate of the transformer is low due to insufficient and unbalanced fault samples in the dissolved gas in the oil.

Abstract: A DC/DC converter fault diagnosis method based on an improved sparrow search algorithm, includes: establishing an simulation module of the converter, selecting a leakage inductance current of a transformer as a diagnosis signal, and collecting diagnosis signal samples under OC faults of different power switching devices of the converter as a sample set; improving a global search ability of a sparrow search algorithm by using a Levy flight strategy; dividing the sample set into a training set and a test set, preliminarily establishing an architecture of a deep belief network, and initializing network parameters; optimizing a quantity of hidden-layer units of the deep belief network by using an improved sparrow search algorithm, to obtain a best quantity of hidden-layer units of the deep belief network; and training an optimized deep belief network obtained based on the improved sparrow search algorithm, and obtaining a fault diagnosis result based on a trained network.

Abstract: The disclosure relates to a method for fabricating a speckle for high temperature deformation measurement of a shaped refractory material. A technical solution includes mixing a hercynite micropowder and a liquid mixing agent in a mass ratio of (3-6):1, and ultrasonically treating to obtain a speckle mixture; polishing a surface of a shaped refractory material to be measured, removing impurities, and spraying the speckle mixture on the surface of the shaped refractory material to be measured with a pneumatic airbrush in a time hood to obtain an uncured speckle; heating the uncured speckle to 60-80° C., keeping for 1-3 h, then heating to 100-120° C., and keeping for 1-3 h to obtain a speckle for high temperature deformation measurement of a shaped refractory material. The fabricated product is suitable for high temperature deformation measurement of a shaped refractory material at 1,600° C.

Abstract: This invention proposes a pedestrian re-identification method based on virtual samples, comprising following steps: s1) obtaining virtual persons generated by game engine, and generating the virtual samples with person labels by fusing a background of a target dataset and a pose of real persons through a multi-factor variational generation network; s2) rendering the generated virtual samples according to lighting conditions; s3) sampling the rendered virtual samples according to person attributes of target dataset; s4) constructing a training dataset according to virtual samples obtained by sampling to train a pedestrian re-identification model, and verifying identification effect of the trained model.

Abstract: A person re-identification method based on a perspective-guided multi-adversarial attention is provided. The deep convolutional neural network includes a feature learning module, a multi-adversarial module, and a perspective-guided attention mechanism module. The multi-adversarial module is followed by a global pooling layer and a perspective discriminator after each stage of a basic network of the feature learning module. The perspective-guided attention mechanism module is an attention map generator and the perspective discriminator. The training of the deep convolutional neural network includes learning of the feature learning module, learning of the multi-adversarial module, and learning of the perspective-guided attention mechanism module. The proposed method uses the trained deep convolutional neural network to extract features of the testing images, and using an Euclidean distance to perform feature matching on images in a query set and images in a gallery set.

Abstract: An automatic and intelligent spraying device for traffic marking intelligent. The spraying device for traffic marking includes a fixed bottom plate, a coating storage tank, a control box provided with a controller, an armrest, a row of nozzles, a speed sensor, and a driving motor. The coating storage tank is connected with coating pipes equipped with an electromagnetic valve. The fixed bottom plate is provided with a driving motor and rotating wheels. The driving motor drives the rotating wheels to move. The speed sensor is electrically connected with the controller and the driving motor. The controller controls a speed of the driving motor according to speed information detected by the speed sensor. The controller controls the driving motor, and controls the row of nozzles to spray according to a preset spray marking and spray pattern through the electromagnetic valve.

Abstract: Disclosed in the present invention is hyperspectral image classification method based on context-rich networks. The method comprises a training stage and a prediction stage, wherein the training stage comprises image pre-processing, sample selection and network training. Firstly, performing normalization on a hyperspectral image, and then randomly selecting an appropriate proportion of marked samples from each category to generate a label map, and performing training by using the designed network; in the prediction stage, directly inputting the whole image into the trained network and obtaining a final classification result. By means of the present invention, data pre-processing, feature extraction, the process of context-rich information capturing, and classification are taken into comprehensive consideration in the whole flow; and the classification of a hyperspectral image is realized by means of constructing an end-to-end network.

Abstract: This invention proposes a pedestrian re-identification method based on virtual samples, comprising following steps: sl) obtaining virtual persons generated by game engine, and generating the virtual samples with person labels by fusing a background of a target dataset and a pose of real persons through a multi-factor variational generation network; s2) rendering the generated virtual samples according to lighting conditions; s3) sampling the rendered virtual samples according to person attributes of target dataset; s4) constructing a training dataset according to virtual samples obtained by sampling to train a pedestrian re-identification model, and verifying identification effect of the trained model.

Abstract: The present invention relates to a polyvinyl alcohol/sodium alginate/hydroxyapatite composite fibrous membrane, and a preparation method and an application thereof. The preparation method of the composite fibrous membrane includes the following steps: firstly, reacting a diammonium hydrogen phosphate with a calcium nitrate to prepare a hydroxyapatite, and performing ultrasonic dispersion on the hydroxyapatite with a sodium alginate to form a stable hydroxyapatite suspension; separately preparing a sodium alginate solution of which the mass fraction is 2% and a polyvinyl alcohol solution of which the mass fraction is 18% using the above stable hydroxyapatite suspension; and finally, proportionally and uniformly mixing the two solutions, and performing electrospinning.

Abstract: The present disclosure discloses a molten aluminum erosion resistant casting tip paint, the casting tip paint is formed by mixing magnesia-alumina spinel, alumina, magnesia, water, sodium tripolyphosphate and sodium carboxymethyl cellulose, wherein the proportions of various raw materials for the casting tip paint are as follows: magnesia-alumina spinel of 40-50 wt %, alumina of 10-20 wt %, magnesia of 5-10 wt %, water of 20-40 wt %, sodium tripolyphosphate of 0-10 wt %, and sodium carboxymethyl cellulose of 0-10 wt %. The present disclosure also discloses a preparation method of a molten aluminum erosion resistant casting tip. The molten aluminum erosion resistant casting tip paint provided by the present disclosure can be evenly coated onto the inner wall of the casting tip, so as to form a compact molten aluminum erosion resistant coating, thereby preventing infiltration and erosion of the molten aluminum.

Abstract: The invention discloses an attention-based joint image and feature adaptive semantic segmentation method. First, the image adaptation procedure is used to transform the source domain image Xs to a target-domain-like image Xs-t with an appearance similar with the target domain image Xt, to reduce the domain gap between the source domain and the target domain at the image appearance level; then using the feature adaptation procedure to align the features between Xs-t and Xt in the semantic prediction space and the image generation space, respectively, to extract the domain-invariant features, to reduce the domain difference between Xs-t and Xt. In addition, the present invention introduces an attention module in the feature adaptation procedure to help the feature adaptation procedure pay more attention to image regions worthy of attention. Finally, combining the image adaptation procedure and the feature adaptation procedure in the end-to-end manner.

Abstract: A hyperspectral remote sensing image classification method based on a self-attention context network is provided. The method constructs a spatial dependency between pixels in a hyperspectral remote sensing image by self-attention learning and context encoding, and learns global context features. For adversarial attacks in the hyperspectral remote sensing data, the proposed method has higher security and reliability to better meet the requirements of safe, reliable, and high-precision object recognition in Earth observation.

Abstract: Disclosed is a method for classifying hyperspectral images on the basis of an adaptive multi-scale feature extraction model, the method comprising: establishing a framework comprising the two parts of a scale reference network and a feature extraction network, introducing a condition gate mechanism into the scale reference network, performing determination step-by-step by means of three groups of modules, inputting features into a corresponding scale extraction network, deep mining rich information contained in a hyperspectral remote sensing image, effectively combining features of different scales, improving a classification effect, and generating a fine classification result map.

Abstract: The invention discloses high-strength glass-ceramic-based lightweight aggregates and the preparation method thereof. The mass ratio of raw material components is 50-70 parts of engineering muck, 20-40 parts of glass, 3-7 parts of calcium carbonate, 3-7 parts of magnesium oxide, and 2-10 parts of a nucleating agent; the nucleating agent is at least one of calcium fluoride, titanium dioxide, and chromium oxide. After crushing, mixing, and granulating, spherical particles with a particle size of 10-12 mm are formed; and then the product can be obtained after drying, sintering, and cooling. The obtained lightweight aggregate from the invention has a diopside matrix which provides high strength and a low water absorption rate at low densities. Moreover, waste glass and engineering muck could be utilized with high value.

Abstract: A method for evaluating a damage power of confined explosion, includes: determining a size of a confined case and a thickness of a target plate according to a requirement of evaluation on damage effectiveness of a warhead to establish an explosive damage effectiveness evaluation platform; calibrating parameters of the platform with a series of confined explosion tests of trinitrotoluene (TNT) bare charges in the case to form a relational map, in which a deflection of a mid-point of the target plate changes with a charge mass; obtaining a deflection of the mid-point of the target plate by detonating different warheads, forming different detonation atmosphere and forming different venting configuration in the case, and performing interpolation through the map to obtain an equivalent bare charge as an evaluation index. The method quantitatively evaluates the damage effectiveness of the warhead, and is not limited by the damage component.

Abstract: A method for evaluating a damage power of confined explosion, includes: determining a size of a confined case and a thickness of a target plate according to a requirement of evaluation on damage effectiveness of a warhead to establish an explosive damage effectiveness evaluation platform; calibrating parameters of the platform with a series of confined explosion tests of trinitrotoluene (TNT) bare charges in the case to form a relational map, in which a deflection of a mid-point of the target plate changes with a charge mass; obtaining a deflection of the mid-point of the target plate by detonating different warheads, forming different detonation atmosphere and forming different venting configuration in the case, and performing interpolation through the map to obtain an equivalent bare charge as an evaluation index. The method quantitatively evaluates the damage effectiveness of the warhead, and is not limited by the damage component.

Abstract: The invention discloses an attention-based joint image and feature adaptive semantic segmentation method. First, the image adaptation procedure is used to transform the source domain image Xs to a target-domain-like image Xs-t with an appearance similar with the target domain image Xt, to reduce the domain gap between the source domain and the target domain at the image appearance level; then using the feature adaptation procedure to align the features between Xs-t and Xt in the semantic prediction space and the image generation space, respectively, to extract the domain-invariant features, to reduce the domain difference between Xs-t and Xt. In addition, the present invention introduces an attention module in the feature adaptation procedure to help the feature adaptation procedure pay more attention to image regions worthy of attention. Finally, combining the image adaptation procedure and the feature adaptation procedure in the end-to-end manner.

Abstract: Disclosed are a method and equipment for reinforcing a concrete-filled steel tubular column with an outer concrete-filled steel tube. The method includes following steps: step 1, grinding and derusting an outer wall of a concrete-filled steel tubular column to be reinforced by an auxiliary device; step 2, welding two prefabricated semi-rectangular steel plates in advance, forming a complete rectangular steel tube with the two semi-rectangular steel plates and sleeving the rectangular steel tube outside the concrete-filled steel tubular column to be reinforced; and step 3, filling self-compacting concrete between the outer wall of the concrete-filled steel tubular column to be reinforced and an inner wall of the rectangular steel tube to complete a reinforcement.

Abstract: A person re-identification method based on a perspective-guided multi-adversarial attention is provided. The deep convolutional neural network includes a feature learning module, a multi-adversarial module, and a perspective-guided attention mechanism module. The multi-adversarial module is followed by a global pooling layer and a perspective discriminator after each stage of a basic network of the feature learning module. The perspective-guided attention mechanism module is an attention map generator and the perspective discriminator. The training of the deep convolutional neural network includes learning of the feature learning module, learning of the multi-adversarial module, and learning of the perspective-guided attention mechanism module. The proposed method uses the trained deep convolutional neural network to extract features of the testing images, and using an Euclidean distance to perform feature matching on images in a query set and images in a gallery set.

Abstract: A non-blind area real-time monitoring and alarming system for an accident on a freeway is provided, which belongs to the field of photoelectric technology and can solve the existing problems in whole journey information monitoring for freeways, such as failure to cover all freeway sections and lack of all-weather and prompt monitoring, and being subject to severe environments such as rain, fog and snow, or conditions such as poor visibility at night. The system includes a distributed sound wave detection fiber-optic cable, a sound wave signal demodulator, a network switch, a workstation, and a monitoring terminal. A fiber-optic sensing network composed of a series of reflection nodes distributed at equal distances is utilized to monitor sound wave signals from traffic accidents efficiently in real time without blind areas, accurately locate a traffic incident by analyzing frequency components, and transmit alarming information in time.