27results about How to "Poor resolution" patented technology

The invention provides a big language model merging method and device, electronic equipment and a storage medium, and the method comprises the steps: determining a density hyper-parameter and a model merging coefficient hyper-parameter used for model merging based on a task target of a merging model in a downstream task; generating a mask matrix based on the density hyper-parameter, and determining task vectors of the two to-be-merged models based on the mask matrix, the respective model parameters of the two to-be-merged models and the model parameters of the basic model; based on the parameter vector angle of the corresponding neuron between the two to-be-merged models, the task vectors of the two to-be-merged models and the model merging coefficient hyper-parameter, determining the task vector of the merged model; and obtaining the merging model based on the task vector of the merging model and the model parameters of the basic model. By adopting the technical scheme, the problem that the performance of a merging model obtained by an existing model merging method is poor is solved.

The invention relates to the technical field of image processing and computer vision, in particular to a text-guided multi-view Mars topographic feature image segmentation method based on SAM, and the method comprises the steps: constructing and training an MTSNet model, inputting a to-be-segmented Mars image and a topographic prompt text into the trained MTSNet model, and obtaining a segmentation result; the MTSNet model comprises an MTG-SAM branch, an LTE (Long Term Evolution Network) branch, a feature fusion module, a text prompt encoder and a Mask Decoder; the MTG-SAM branch is formed by cascading a plurality of Transform Blocks, and a TCAM (Ternary Content Addressable Memory) module is introduced into each Transform Block; the text prompt encoder comprises a CLIP text encoder, a text affine layer and a prompt encoder; according to the method, the accuracy and convenience of terrain segmentation of the Mars image are effectively improved.

The embodiment of the invention provides a multi-domain cluster collaborative networking simulation test method, electronic equipment and a storage medium. Simulation scene information and simulation node information are obtained; the simulation node information represents virtual objects deployed in the virtual scene and the number of the virtual objects; deploying virtual objects in the virtual scene according to the object quantity to obtain object cluster information; obtaining multi-domain cluster information according to the simulation node information and the object cluster information; generating initial data transmission information according to the multi-domain cluster information; acquiring initial interference information set by a user; performing global synchronization on the initial interference information through the multi-domain cluster information, and determining global interference information; performing adaptive reconstruction on the initial data transmission information according to the global interference information to obtain reconstructed data transmission information; multi-domain cluster information, initial data transmission information and reconstructed data transmission information are displayed in a simulation mode, and the problem that simulation testing is poor in effectiveness is solved.

The invention relates to the technical field of battery negative electrode materials, in particular to a silicon-carbon composite material precursor and a preparation method and application thereof. The silicon-carbon composite material precursor comprises modified starch particles and a plurality of silicon-based material particles, the silicon-based material particles are at least embedded in the modified starch particles, the modified starch particles comprise carbon nanomaterials and starch, and the carbon nanomaterials and the starch are bonded and connected through oxygen-containing functional groups. The bonding connection of oxygen-containing functional groups limits the free movement of starch molecules and inhibits starch foaming. Therefore, the thickness of the carbon wall around the silicon particles in the silicon-carbon composite material prepared from the silicon-carbon composite material precursor is uniform. The preparation method comprises the following steps: carrying out surface modification treatment on the carbon nanomaterial, then mixing the carbon nanomaterial with starch, mixing the carbon nanomaterial with silicon-based material particles, and finally carrying out granulation treatment, so that the modified carbon nanomaterial and oxygen-containing functional groups in the starch form bonding connection. When the precursor is used for preparing the silicon-carbon composite material, the comprehensive performance such as conductivity, capacity and stability can be improved.

The heat pump water system comprises indoor heat exchange equipment, a heat pump device, auxiliary heat exchange equipment and a valve assembly, and the heat pump device communicates with a water inlet of the indoor heat exchange equipment through a first water supply pipeline and communicates with a water outlet of the indoor heat exchange equipment through a first water return pipeline; the auxiliary heat exchange equipment is communicated to the first water supply pipeline through a second water return pipeline and a second water supply pipeline; the valve assembly is arranged on the second water return pipeline and / or the second water supply pipeline and used for adjusting the flow of water flowing from the second water return pipeline to the second water supply pipeline through the auxiliary heat exchange equipment. According to the heat pump water system, the water flow flowing to the auxiliary heat exchange equipment can be adjusted by controlling the valve assembly according to the series complementary structure of the heat pump device and the auxiliary heat exchange equipment, the problem that the effect of an independent heat pump device is poor is solved, and therefore the heating effect of the heat pump water system is improved.

The utility model discloses an assembly type anti-slurry-leakage steel formwork for concrete structure pouring. The steel formwork is composed of a combined panel unit, a reinforcing unit and an anti-slurry-leakage structure. The combined panel unit comprises a plurality of panel basic frames and panel pieces, and the panel basic frames and the panel pieces are connected through concave-convex clamping structures. The reinforcing unit comprises a plurality of fixing rods which are connected in a clamped mode, the fixing rods are arranged on the outer side face of the combined panel unit in a surrounding mode to form a panel reinforcing frame, and the panel reinforcing frame is used for tensioning the distance between the panel basic frame and the panel pieces and locking and fixing the panel basic frame and the panel pieces. The device is simple in structure, and undesirable conditions such as mold expansion and slurry leakage are practically prevented; meanwhile, stability and reliability are achieved, the engineering construction quality and efficiency are powerfully guaranteed, circulation is convenient, the device can be used for pouring construction of concrete structures of different sizes, the space needed by supporting is small, the requirement for the narrow construction environment is met, and very good practical value and popularization prospects are achieved.

The application relates to a synergistic liquid antioxidant containing chicken-derived active peptides and phospholipids, and relates to the technical field of food additives or health products. The antioxidant comprises the following components in parts by weight: chicken-derived active peptides 2-10%; phospholipids 0.5-5%; stabilizers 0.1-2%; water 60-90%; and the weight ratio of the chicken-derived active peptides to the phospholipids is (2-6):1. By compounding the chicken-derived active peptides and the phospholipids in a specific ratio, the two can form a stable emulsification system and produce a significant antioxidant synergistic effect, effectively eliminating free radicals and inhibiting lipid peroxidation. Meanwhile, the liquid antioxidant has good stability and high bioavailability, can be widely applied in oils, pet foods, oil-containing foods, feed products and cosmetics, and can replace traditional synthetic antioxidants.

The invention discloses a self-supervised real image denoising method based on sampling difference as disturbance. The method mainly comprises the following steps: proposing and using a random sub-sample generation strategy to obtain more sub-samples; providing a sampling difference loss function according to a random subsample generation strategy; introducing a loop strategy into construction of a loss function, and proposing a loop sampling difference loss function; the denoising network is trained through the above steps to obtain the denoising model, then the image with noise is used as input, and a high-quality denoised image is obtained through the trained network model. The method does not need clean images for training, is superior to an existing pseudo-supervision / self-supervision method, and is an effective self-supervision real image denoising method.

The invention discloses a student grouping method based on reinforcement learning and an improved genetic algorithm, and relates to the field of teaching education. The method specifically comprises the following steps: S1, constructing a data set; s2, generating an initial solution; s3, genetic algorithm operation; and S4, dynamically adjusting parameters. The method has the beneficial effects that the problem of poor effect caused by single standard grouping in the prior art is solved by simultaneously considering multi-dimensional factors such as homogeneity, heterogeneity, gender, leaderness and estrus sharing, key parameters of the genetic algorithm are dynamically adjusted, and the method is suitable for large-scale popularization and application. And a high-quality initial solution is constructed by designing a minimum cost insertion strategy, so that the student grouping quality and the overall effect of collaborative learning are remarkably improved.

The invention discloses a hydraulic station remote flow control method and system based on the Internet of Things, and relates to the technical field of remote flow control, and the hydraulic station remote flow control method based on the Internet of Things comprises the following steps: collecting operation data of a hydraulic station in real time by using a preset sensor group; analyzing the oil mass change of an oil reservoir in the hydraulic station based on the preprocessed operation data, and starting cross-regional oil mass appointment scheduling based on a prediction result of the oil mass change; establishing a dynamic balance model among the oil reservoir, the hydraulic pump and the control valve; the fault monitoring model is used for recognizing an oil mass conveying fault in the oil mass conveying process; and the oil mass conveying state among the oil storage device, the hydraulic pump and the control valve is adjusted according to the recognized oil mass conveying fault, so that remote control over the oil mass is achieved. According to the method, the oil reservoir, the hydraulic pump and the control valve are regarded as participants of the dynamic balance model, the oil mass conveying state among the participants can be effectively analyzed and coordinated, and coordinated operation of the system is guaranteed.

The application provides a rapid surface defect detection method and system based on a GAN and an attention fusion mechanism. The detection method is as follows: a surface defect image sample set is acquired; a rapid scratch detection network is built, the network comprising: an information extraction module for extracting underlying detailed information; a rapid sampling module for extracting high-level semantic level information; an attention fusion module for fusing the high-level semantic level information and the underlying detailed information, and combining the underlying detailed information extracted by the information extraction module to position scratches, and using the high-level semantic level information extracted by the rapid sampling module to position and segment surface defects; the rapid surface defect detection network is trained using the surface defect image sample set; and the trained rapid surface defect detection network is used for surface defect detection. The rapid surface defect detection method accurately detects surface defects through rapid positioning and segmentation of the surface defects, and has the advantage of high precision.

The present invention discloses a chlorophyll content estimation method based on optical characteristic parameter inversion, which comprises the following steps: (1) constructing a leaf model based on a surface light source to simulate photon transmission and obtain a simulated spectral image; (2) inputting the simulated spectral image into a convolutional neural network for training and obtaining a pre-trained model; (3) based on the pre-trained model, performing maximum mean difference (MMD) migration on measured spectral data and obtaining an optical parameter inversion model of all training layers; (4) using the optical parameter inversion model obtained by MMD migration to estimate the chlorophyll content of the leaf. The optical parameter inversion method of green radish leaf based on MMD migration has better performance than the common model migration method.

The utility model discloses a multi-module fuel cell air tail gas treatment system and a fuel cell system, and the multi-module fuel cell air tail gas treatment system comprises at least two single-module fuel cell systems, or the liquid water flows to the heat exchanger of at least one other single-module fuel cell system through the conversion valve to adjust the temperature of the liquid water passing through the heat exchanger, and the water pump is used for conveying the temperature of the liquid water passing through the heat exchanger to the electric pile to adjust the temperature of the electric pile. According to the utility model, the tail gas exhausted by the electric pile of one single-module fuel cell system is selectively conveyed to the heat exchanger or the tail gas exhaust, and heat exchange is carried out on liquid water of other single-module fuel cell systems so as to improve the temperature of the liquid water, so that the temperature of the electric pile of other single-module fuel cell systems is close to the optimal working temperature; the problem that the working voltage of the electric pile is too high due to low required power of the whole vehicle is solved, and the problem that the performance of the electric pile which does not work is poor due to too low temperature when the electric pile is started next time is solved.

The application discloses a rail transit semantic segmentation method based on boundary calibration and receptive field evaluation, and steps are as follows: constructing an all-weather rail transit semantic segmentation dataset; constructing an all-weather rail transit semantic segmentation network composed of a semantic branch, a boundary branch, a detail branch, a feature aggregation layer, a convolution layer and a semantic segmentation head; adopting an uncertainty perception training strategy combined with a multi-task loss supervision mode to complete network training, so as to perform rail transit semantic segmentation on acquired rail transit scene images; and adopting the method to perform rail transit semantic segmentation on the rail transit scene images, which has the negative influence of the environmental noise region automatically inhibited, ensures the accuracy of the semantic segmentation structure, greatly enhances the perception stability of the processing target in the rail transit scene under the all-weather environment, and can also show good robustness even in the case of a sharp increase in environmental noise.

This invention discloses a UV-curable coating composition, its preparation method, and its application. The UV-curable coating composition includes a UV-curable primer and a UV-curable topcoat. The UV-curable primer is composed of polyurethane acrylate resin, a first acrylate monomer, a photoinitiator, an adhesion promoter, and a wetting and leveling agent in a specific ratio, and after curing, it forms a coating with strong adhesion on a nickel-plated substrate. Furthermore, by optimizing the formulation of the UV-curable topcoat, it promotes efficient bonding between the topcoat and primer when used in conjunction with the primer, imparting excellent insulation properties to the coating. The UV-curable coating composition provided by this invention can form a composite coating with strong adhesion, excellent insulation properties, and high stability on the surface of a nickel-plated substrate. Its adhesion performance is grade 1 or higher, and it passes voltage resistance and electrolyte resistance tests. Even after aging under double 85 conditions for 1000 hours, it still maintains excellent adhesion and insulation stability.

The invention relates to the technical field of power system photovoltaic prediction, in particular to a photovoltaic power short-term probability prediction method, device and equipment and a medium, and the method comprises the steps: processing original data, and carrying out the feature extraction of the preprocessing data based on EMD and KPCA models; performing deterministic prediction by adopting a Stacking model; evaluating the prediction result; correcting the deterministic prediction result by adopting an adaptive boundary optimization strategy to generate a short-term probability prediction result; the problem of low data resolution is solved through data interpolation, time sequence correlation of data is considered in correlation analysis, input feature redundancy is reduced through EMD and KPCA methods, model generalization is improved through a Stacking model, and prediction interval performance is improved through an adaptive boundary optimization strategy. The method has the advantages of efficiently processing multi-source data, considering data time sequence correlation, accurately capturing feature changes, having excellent model generalization and high-quality prediction error analysis and the like.

The utility model discloses a corrosion-resistant environment-friendly steel structure which comprises a bottom plate, supporting steel is arranged on the bottom plate, a reinforcing device is installed above the supporting steel, a fast clamping device is arranged on the supporting steel, and the fast clamping device comprises a fixing rod, a releasing sleeve, a fixing plate, a fixing sleeve, an elastic plate, a limiting block, a threaded groove, a limiting groove and a releasing groove. The fixing plate is connected to one side of the fixing sleeve and one side of the releasing sleeve, the elastic plates are arranged on the outer side of the fixing rod along threads, the limiting block is arranged in the releasing sleeve, the limiting groove is formed in the side wall of the fixing rod, the releasing groove is formed in the side wall of the releasing sleeve, the thread groove is formed in the fixing sleeve, and the limiting mechanism is arranged on the outer side of the fixing rod. The limiting mechanism comprises a receding hole, an ejector rod, a limiting sleeve and a top plate, the top plate is connected to one side of the fixing rod, the receding hole is formed in the top plate, the fixing rod is sleeved with the limiting sleeve, and the ejector rod is connected to one side of the limiting sleeve.

The utility model provides an electroplating pool waste gas filtering device which comprises a filtering box, an exhaust pipe is fixed to the upper end face of the filtering box, a bearing plate is fixed to the bottom of the front side wall of the filtering box, a motor is installed on the upper end face of the bearing plate, and the driving end of the motor penetrates through the filtering box and is fixedly connected with a stirring shaft. And the other end of the stirring shaft is rotationally connected with the filtering box. Through the arrangement of the motor, the stirring shaft, the stirring paddle, the first belt wheel, the rotating shaft, the first bevel gear, the supporting seat, the rotating rod, the second bevel gear, the second belt wheel, the synchronous belt, the poke rod, the mounting seat, the reciprocating frame, the sliding block, the bracket, the fixed pipe and the hose, waste gas introduced into purification liquid is fully fused with the purification liquid, and the waste gas filtering and purifying effect is improved; the waste gas filled with the purification liquid is input in a swinging manner, so that the use effect of the purification liquid is improved, and the problems of inconvenience in full contact of the waste gas with the purification liquid and reduction of the filtering effect due to fixed-point input in the prior art are solved.

The present application relates to the technical field of SAR image processing and self-supervised learning. A SAR image self-supervised pre-training method based on a mask autoencoder. In view of the problem that existing SAR target detection relies on natural image pre-training weights and there is a significant domain gap, resulting in low detection accuracy in a small sample scene, a multi-source SAR pre-training dataset is constructed, an asymmetric encoding and decoding architecture is adopted, ResNet50 is used as a feature extraction encoder to adapt to the local scattering characteristics of SAR, a lightweight convolutional decoder is constructed, a Gamma distribution speckle noise modeling mechanism is introduced, the input image is divided into non-overlapping image blocks and a random mask matrix is generated, the encoder only encodes the visible area, the decoder maps the latent representation back to the image space to complete the reconstruction, and the learning rate scheduling and gradient clipping mechanisms are combined to complete unsupervised pre-training, and the encoder weight is extracted and migrated to a downstream rotating target detection model to complete small sample fine tuning.

The invention belongs to the technical field of air conditioners, and particularly relates to a low-noise central air conditioner which comprises a fan coil, a noise reduction box is arranged outside the fan coil and used for isolating noise generated by the fan coil, the fan coil comprises a box body, an air inlet is formed in one end of the box body, and an air outlet is formed in the other end of the box body. An air blowing assembly, a first noise reduction mechanism, a second noise reduction mechanism and a heat exchanger are sequentially arranged in the box body, and the air blowing assembly is configured to form directional airflow; the directional airflow is subjected to two-stage noise reduction treatment through the first noise reduction mechanism and the second noise reduction mechanism in sequence, then exchanges heat with the heat exchanger and is finally blown out from the air outlet; the first noise reduction mechanism comprises a first shell, two spoiler sets are arranged in the first shell, the spoiler sets are in transmission fit with driving parts, the driving parts are electrically connected with a microcontroller, and the microcontroller controls the inclination angles of the spoiler sets according to the air volume of the directional airflow so as to reduce the air noise. And two-stage progressive processing of noise is achieved, and the noise reduction effect is good.

The invention provides a preparation method of an electrolyte-resistant adhesive tape for a lithium battery, which comprises the following four steps: preparing materials, preparing a paper cup adhesive, preparing a film and forming the adhesive tape. An epoxy resin curing agent, a nitrite accelerant, a hydrogen peroxide catalyst, an ozone catalyst, a magnesium oxide filling agent, a calcium carbonate filling agent, an aluminum oxide filling agent, a polyurethane sealant, a zinc oxide antioxidant, a ferric oxide antioxidant and a nano-silver preservative are added, so that the prepared adhesive has excellent electrolyte resistance; the adhesive tape has the advantages that the adhesive tape can meet the bonding requirement of an electronic product in an electrolyte environment, the adhesive tape is matched with a film, electrolyte permeation can be effectively prevented, the safety and performance of a battery are improved, the film is subjected to surface treatment, the bonding performance of the film and the adhesive is improved, and the stability of the adhesive tape in the using process is guaranteed.

The invention relates to an antibiotic-free biological fermentation compound feed for poultry and a preparation method thereof, and belongs to the technical field of feeds, the feed takes a fermentation basic raw material as a carrier, plant extracts and chemical addition components are added, and the synergistic effects of resisting bacteria, diminishing inflammation, regulating intestinal flora, promoting nutrient absorption and improving the immunity of poultry are achieved in a biological fermentation synergistic manner; and antibiotic residues are avoided, so that the safety of poultry products is guaranteed.

According to the scheme, the building environment monitoring device for civil building engineering comprises a bottom plate, electric telescopic rods, a lifting plate, a rotating assembly, a monitoring machine body, a camera, moving wheels and a supporting assembly, the electric telescopic rods are fixedly connected to the two sides of the top of the bottom plate, and the lifting plate is fixedly connected to the tops of the electric telescopic rods; the rotating assembly is rotationally connected to the top of the lifting plate, the monitoring machine body is fixedly connected to the rotating assembly, and the camera is fixedly connected to the monitoring machine body; the utility model belongs to the technical field of building environment monitoring, and particularly relates to a monitoring device which is convenient to move through moving wheels, can monitor most of places, is high in practicability, is convenient to stably place through a supporting assembly during monitoring of the monitoring device, avoids poor effect caused by movement of the device during monitoring, and is high in practicability. The electric telescopic rod drives the lifting plate to adjust the height, and the rotating assembly drives the monitoring machine body and the camera to rotate to adjust the monitoring angle.

The invention relates to the technical field of traditional Chinese medicine, in particular to a traditional Chinese medicine preparation for treating diabetes mellitus and a preparation method thereof.The preparation method comprises the following steps that biond magnolia flower, borneol, sandalwood, asarum, rhizoma atractylodis, astragalus membranaceus, Chinese yam and radix trichosanthis are weighed based on a preset proportion; pretreating the medicinal materials; the pretreated medicinal materials are treated; uniformly mixing the treated medicinal materials, putting into a decocting pot, adding a proper amount of clear water, boiling with big fire, and decocting with small fire; the decocted liquid medicine is filtered through a gauze net, and the traditional Chinese medicine preparation is obtained. Compared with side effects possibly accompanied by taking western medicines for a long time, the traditional Chinese medicine preparation possibly has lower side effect risk. The traditional Chinese medicines are generally known by natural components and mild action modes of the traditional Chinese medicines, so that adverse reactions such as hypoglycemia risk, weight gain and gastrointestinal discomfort are reduced, the compliance and life quality of patients are improved, and the problem of poor effect of an existing medicament treatment method is solved.

The invention provides a liquid metal battery series module rapid charging device, and relates to the technical field of secondary battery application, the device comprises a charger, a switch module, a liquid metal battery series module, a load, a voltage detection module, a central processing unit and a switch control module; the liquid metal battery series module is formed by connecting N sections of liquid metal batteries in series; the voltage detection module is used for detecting voltage information of the N liquid metal batteries; the central processing unit is used for receiving and processing the voltage information of the voltage detection module, generating a control instruction and inputting the control instruction to the switch control module; the switch control module is used for controlling the switch module according to the received control instruction; the switch module is used for changing a circuit connection mode, so that the liquid metal battery is connected with the charger, and the liquid metal battery is quickly charged; and the charger is used for quickly charging the liquid metal battery series module.

The invention discloses a seed treatment agent for preventing and treating garlic bulb rot, and belongs to the technical field of agricultural biology. The active ingredients of the seed treatment agent comprise metalaxyl-M, fludioxonil and clothianidin. Metalaxyl-M and fludioxonil are compounded to respectively show extremely high inhibitory activity on two main pathogenic bacteria, namely phytophthora and fusarium, causing garlic bulb rot, and after compounding, antagonism is avoided, so that the metalaxyl-M and fludioxonil can synergistically exert prevention and treatment effects. The treatment agent can be prepared into dosage forms such as a suspended seed coating and the like, and is used for seed dressing or coating treatment of garlic seeds. Field experiments show that the seedling death rate and the bulb rotting incidence rate can be remarkably reduced, yield increase is promoted, and the method is safe to seeds. The invention fills the blank of the special prevention and treatment medicament for garlic bulb rot, and has important application value.

The invention discloses a method, a device and equipment for removing rain and snow in an image, a medium and a product. The method for removing rain and snow in the image comprises the following steps: acquiring an image frame to be compared in railway monitoring video data; determining a reconstruction pixel value of each pixel point according to a pixel minimum value of multiple frames of images of the to-be-compared image frame at the same pixel point; and reconstructing the rain and snow filtered image according to the reconstructed pixel value of each pixel point. According to the technical scheme of the embodiment of the invention, rain and snow interference in the image can be quickly and well filtered out, the calculated amount is small, and a large number of paired rain and snow images in an actual scene are not needed to serve as training data.