188results about How to "Reduced characteristics" patented technology

The invention discloses an attention aggregation and cross-modal graph fused unsupervised multi-modal emotion recognition method. The method specifically comprises the steps of 1, acquiring audio modal features, image modal features and text modal features of a video sample; step 2, acquiring an audio vector, an image vector and a text vector corresponding to the audio modal feature, the image modal feature and the text modal feature; step 3, constructing an unsupervised multi-modal emotion recognition model; 4, inputting the audio vector, the image vector and the text vector into an unsupervised multi-modal emotion recognition model for training, and obtaining the optimal output of the model as a final feature vector for prediction and classification; and 5, performing emotion recognition by using the trained unsupervised multi-modal emotion recognition model. The problem that in the prior art, the long-distance spatial dependency relationship of a tree or graph network cannot be processed, the syntactic dependency relationship in sentences cannot be captured, and consequently the model cannot make full use of multi-modal data is solved.

The invention relates to multi-mode signal processing, and provides a signal processing method of a computer system. The method comprises the following steps: acquiring different modal signal data by using a plurality of sensors and performing timestamp synchronization; a depth time sequence alignment network and a multi-layer dynamic time warping algorithm are utilized to ensure signal time consistency; multi-modal signal features are extracted and fused through a multi-stream depth feature extraction network and a cross-modal attention mechanism; the system adopts a dynamic convolution kernel generation and meta-learning technology to carry out adaptive signal processing; noise suppression and signal optimization are carried out by using a multi-mode noise reduction auto-encoder and a generative adversarial network, and parameters are optimized and processed through closed-loop feedback control and reinforcement learning strategies. According to the method, synchronization, fusion, adaptive processing, noise suppression and real-time feedback control of multi-mode signals are realized, and the method has the advantages of high precision, high robustness and real-time performance.

The invention provides a textile defect type identification method, and relates to the technical field of textile defect identification, and the method specifically comprises the steps: collecting original images of a plurality of textile samples, and carrying out the detailed marking; adjusting the original image to 224 * 224 pixels, constructing a CNN model, extracting feature maps of different convolutional layers, and generating a fusion feature vector in combination with the weights of the feature layers; performing pooling processing on the fused feature map to generate a single-channel space attention map, and obtaining a weighted feature map in combination with a channel attention weight; constructing a segmentation model, outputting a mask after training, and identifying a normal region and a defect region; analyzing and identifying a defect area and extracting related features through a connected component to form a feature set; and a comprehensive scoring coefficient is constructed based on the features, so that the textile defect type can be accurately identified. According to the method, the recognition capability on fine defects can be effectively enhanced, and finer deviations can be captured, so that the recognition rate is remarkably improved.

The invention provides a comprehensive energy multi-element load short-term prediction method based on secondary decoupling, and the method comprises the steps: collecting a multi-element load data set, reconstructing a winter data set and a summer data set in the multi-element load data set, and carrying out the primary decoupling; decomposing the reconstructed winter data set and summer data set; dividing the decomposed data into a periodic term and a trend term so as to perform secondary decoupling; screening the trend items, and predicting the screened trend items to obtain a trend item prediction result; and performing multivariate load prediction on the periodic term to obtain a periodic term prediction result, and adding the trend term prediction result and the periodic term prediction result to obtain a final prediction result. According to the method, the coupling of the multi-element load piece can be effectively captured, accurate decomposition of nonlinear and non-stationary signals can be realized, the anti-noise capability and the time frequency resolution are improved, and the balance is considered while the prediction accuracy is effectively improved.

The invention provides a traditional Chinese medicine abnormal state recognition method, system and device and a storage medium, and relates to the field of abnormal recognition. The method comprises the following steps: acquiring product information of a to-be-detected traditional Chinese medicine, and determining a data acquisition standard of the to-be-detected traditional Chinese medicine based on the product information of the to-be-detected traditional Chinese medicine and a preset detection rule; based on a data acquisition standard of the traditional Chinese medicine to be detected, inputting the image data into a preset image anomaly detection model to obtain a preliminary image anomaly region; performing feature fusion on the visual feature data and the spectral feature data to obtain an abnormal feature vector; and inputting the abnormal feature vector into a preset abnormal state judgment model, outputting abnormal state data, and performing abnormal state judgment based on the abnormal state data and a preset abnormal state judgment threshold to obtain an abnormal state detection result of the to-be-detected traditional Chinese medicine. Through the scheme, the abnormal state of the traditional Chinese medicine can be identified timely and accurately.

The invention provides a power load prediction method and device for an extreme weather event, computing equipment and a storage medium. The method comprises the steps of collecting power load data and weather forecast data; determining the correlation between the power load data and the weather forecast data based on a correlation coefficient method, and removing redundant features to obtain a load data sequence; processing the load data sequence through variable mode decomposition optimized by a particle swarm algorithm to obtain a high-frequency component, a low-frequency component and a residual component; processing the high-frequency component through a first model to obtain a first sequence, and processing the low-frequency component through a second model to obtain a second sequence; and processing the residual component through a third model, and combining a processing result with the first sequence and the second sequence to determine power load prediction data. On the premise that feature redundancy and information calculation amount are reduced, data processing complexity is weakened through high-reliability signal decomposition, and a complex extreme weather prediction task is completed.

The invention discloses core-shell type hydrophobic modified activated carbon and a preparation method thereof. The obtained hydrophobic activated carbon is composed of a core (activated carbon) and a shell (hydrophobic breathable silicon dioxide shell). The method comprises the following three steps: (1) constructing a silicon dioxide shell layer: dispersing pretreated activated carbon in a solvent, adding tetraethyl orthosilicate into the solvent, firstly adjusting the pH value of a reaction solution to be weakly acidic, stirring for 20-30 minutes under a certain stirring condition, then adjusting the pH value of the reaction solution to be weakly alkaline, continuously stirring for 0.5-1 hour to obtain a suspension A, and carrying out centrifugal drying treatment on the suspension A to obtain the silicon dioxide shell layer; an intermediate product coated with a hydrophilic silicon dioxide shell layer is obtained; (2) hydrophobic group grafting: dispersing the obtained intermediate product in a solvent, adding siloxane with different alkyl chain lengths into the solvent, fully stirring to obtain a suspension B, and performing centrifugal drying treatment to obtain hydrophobic modified activated carbon; cleaning and drying the hydrophobic modified activated carbon; and (3) adding a hydrophilic group: grafting the hydrophilic group on the silicon dioxide shell layer in an acid environment.

The invention relates to the technical field of target detection, and discloses a multi-stage refined camouflage target detection method based on wavelet enhanced recognition, which overcomes the defect that detection details are omitted due to the fact that global information is not effectively utilized in an existing method and solves the problem of feature redundancy caused by the fact that multi-layer information is not processed. And a position target detection task can be better completed. Firstly, a visible light image is preprocessed through a pre-training model, a feature quantity is expanded through pyramid operation, the calculated amount is controlled while the information quantity is guaranteed, part of initial features are selected for feature enhancement, and discrimination features are screened through a wavelet discrimination enhancement module; selecting the highest-layer features and sending the highest-layer features to an asymmetric convolution Mangbar module to obtain features with global information; processing global information features and distinguishing features through a feature inversion decoder composed of a plurality of guide units to obtain a prediction map, and training the model; when a specific detection task is executed, the trained model is adopted to obtain the prediction map, and the detection task is completed.

The present specification relates to a compound of Chemical Formula 1 and an organic light-emitting device comprising the same.

The invention relates to an unmanned aerial vehicle and bird identification method based on confidence coding super-dimensional calculation. The method is mainly suitable for classification and identification of the unmanned aerial vehicle and the flying bird by a low-resolution radar with high repetition frequency detection capability. The method mainly comprises the following steps: in a training stage, firstly, preprocessing track information and micro-Doppler spectrum data, and constructing a fusion feature vector; estimating the confidence coefficient of each element of the feature vector, and carrying out vector weighted coding on the feature vector; constructing an unmanned aerial vehicle and bird training super-dimensional vector set; and in a test stage, performing similarity measurement on the test sample super-dimensional vector and the training super-dimensional vector set, and performing unmanned aerial vehicle and bird classification identification according to a calculation result. The method provided by the invention has the characteristics of simple engineering implementation, low system resource consumption, sufficient theoretical basis of the used method and the like, and can quickly and effectively solve the problem of identifying the unmanned aerial vehicle and the flying bird.

The invention is applicable to the technical field of knowledge migration, and provides a graph domain adaptive node classification method based on instance learning, which comprises the following steps of: acquiring a source domain data set with a mark and a target domain data set without a mark; the representation of the source domain data set and the representation of the target domain data set are learned through a GNN model, modeling is conducted on the source domain data set and the target domain data set, and embedded vectors of the source domain data set and the target domain data set are obtained; the embedded vectors of the target domain data set are classified through KNN, and target domain category pseudo labels are obtained; performing fine-grained instance contrast learning on embedded vectors of the source domain graph data set and the target domain graph data set; and after classification is finished, displaying a final classification result of the target domain graph data set. According to the method, the defects of an existing graph domain adaptation technology in cross-domain tasks are overcome, and the method has wide applicability and can be applied to graph data analysis tasks in various cross-domain scenes, such as social network analysis, recommendation systems, knowledge graphs and biological networks.

The invention belongs to the technical field of artificial intelligence, discloses a human action understanding method based on a large language model and self-adaptive contrast decoding, and provides an FLARE framework, which enables actions and language semantics to be aligned by introducing a pre-trained frozen large language model LLM and a frozen action encoder, and utilizes a self-adaptive contrast decoding AdaCD method to understand actions and language semantics. According to the method, excessive dependence on language priori is reduced, so that the understanding ability of the model on actions is improved, the method can process various action understanding tasks, the generalization ability of the human action understanding method is improved, and meanwhile, the problems of task specificity and computing resource consumption in the prior art are reduced.

The invention relates to the technical field of biology, in particular to a variant AAV capsid protein and a recombinant adeno-associated virus rAAV. The variant AAV capsid protein provided by the invention comprises: 1) an amino acid sequence having at least 80%, 83%, 85%, 88%, 90%, 93%, 95% and 98% of identity with an amino acid sequence as shown in SEQ ID NO.1; and 2) an amino acid sequence shown as SEQ ID NO.1, wherein the amino acid sequence comprises amino acid substituted A587D, Q588G, A589N and Q590P, and an amino acid sequence TLAVPFK is inserted after the 588th site and before the 589th site; related products constructed by the variant AAV capsid protein can be used for cross-blood-brain-barrier efficient transduction to realize drug delivery to a central nervous system to treat related diseases of the central nervous system, and meanwhile, the related products have the characteristic of remarkably reducing liver transduction.

The invention discloses a bipolar plate with a composite carbon coating, a preparation method, an electrode and a battery, the bipolar plate comprises a substrate layer, a transition layer and a composite functional layer, the substrate layer is a metal layer formed by depositing metal Me on the surface of a substrate; the transition layer is formed on the substrate layer and is a metal Me / C layer; the composite functional layer is formed on the transition layer and is far away from the transition layer and comprises a first GLC layer with internal stress relieving and electron conduction functions, a tetrahedral amorphous carbon ta-C layer with a corrosion ion blocking function and a conductive and hydrophobic second GLC layer which are sequentially arranged. The coating provided by the invention has relatively low contact resistance, good ion corrosion resistance and hydrophobic property, and meets the requirement of long-term service of a fuel cell under a durable working condition.

The invention discloses an automatic fracture scanning electron microscope image recognition method based on a convolutional neural network, and the method comprises the steps: collecting a fracture scanning electron microscope image, and manually calibrating the type of the fracture scanning electron microscope image; secondly, performing image preprocessing to form a data set; then, based on a deep learning convolutional neural network, an improved CNN network model is adopted, and based on a plurality of convolution kernels, a maximum pooling layer and a global average pooling algorithm, the metal fracture scanning image is analyzed, and the type of the metal fracture image is determined; and finally, inputting a to-be-recognized image, and performing statistics on a classification model of the image by using a global average pooling algorithm. According to the method, the risk that the accuracy is reduced due to experience or long-time work of personnel can be avoided, the time cost of analysis work is reduced, the type of metal fracture analysis is obtained, and the analysis result is reliable and efficient.

The present invention discloses a planetary gear reducer with a large-caliber hollow structure, which solves the problem that when the existing planetary gear reducer is applied to a robot, the robot's cables cannot be built-in, resulting in a large device volume, and the existing reducer has a large axial size and weak rigidity. The reducer uses an end cover to axially position the planetary wheel of the first-stage planetary gear reduction mechanism, the current-stage planetary carrier axially positions the planetary wheel bearing, and the upper-stage planetary carrier axially positions the first external tooth of the next-stage sun wheel. At the same time, the same-stage planetary wheel and the planetary carrier directly contact each other to achieve mutual axial positioning; the reducer realizes power input from one side and output from the other side through the cooperation between the hollow sleeve, the input gear and the sun wheel of the first-stage planetary gear reduction mechanism, and the internal cavity of the hollow sleeve, the central through hole of each stage sun wheel and the through hole set in the center of the planetary carrier are connected in sequence, thereby forming a hollow structure for passing the line.

The invention relates to an enterprise credit evaluation system and method, the system comprises a data acquisition module, a feature engineering module and an evaluation module which are connected in sequence, and the method comprises the steps: obtaining the local data of an enterprise and the data of a regional big data center, and converting the data of the regional big data center into the data of an enterprise big data center; carrying out entropy calculation, IV value calculation and correlation coefficient analysis on the obtained data to determine in-mold multi-dimensional information, carrying out feature screening, screening out important features from the in-mold multi-dimensional information, and constructing a feature set; training an XGBoost model by using the feature set to obtain an enterprise credit evaluation model; and inputting the in-mold multi-dimensional information and the unified social credit code information of the to-be-assessed enterprise into the enterprise credit assessment model, and outputting a credit score and a suggested credit line corresponding to the to-be-assessed enterprise. Compared with the prior art, the method has the advantages that the problems of dispersion and inconsistency of multi-channel data sources in the prior art can be solved, timely updating of the data is ensured, and the accuracy of enterprise credit assessment results is improved.

The invention relates to a sea-land-air triphibian aircraft which is composed of an aircraft head, an aircraft body and foldable wings, each wing comprises a fixing part, a folding part and a driving part, and unfolding and folding of the wings are controlled through hydraulic jacks. The aircraft is provided with a propeller engine and a propeller thruster which provide lift force and propulsive force respectively, and air flight, underwater navigation and land advancing are achieved. The driving system assembly comprises a submarine engine and an aircraft star-shaped engine, and the submarine engine is connected with and drives propellers to rotate through gears to provide power; the aircraft star-shaped engine is mainly used for air flight. The triphibian aircraft has multifunctionality, high flexibility and wide applicability, and is suitable for various scenes such as emergency rescue, military reconnaissance and environment monitoring. Special materials and design endow the robot with concealment and maneuverability, the robot can quickly respond and avoid potential threats during task execution, and the robot shows powerful comprehensive performance and wide application potential.

The present invention provides a novel compound and an organic light emitting device comprising the same.

The invention provides a method for regulating and controlling the friction coefficient of a carbon-based thin film through friction-induced sliding interface reconstruction, and belongs to the technical field of carbon-based thin film materials. The preparation method comprises the following steps: firstly, depositing a transition metal intermediate layer on the surface of a pretreated substrate as a bonding layer, respectively taking carbon and transition metal carbide as sputtering targets, adjusting the condition of second sputtering to obtain a carbon-based film containing transition metal carbide nanocrystals with different sizes, and then performing friction treatment to obtain the carbon-based film containing the transition metal carbide nanocrystals with different sizes. The transition metal carbide nanocrystals are excited to be subjected to an oxidation reaction to form oxidation products, meanwhile, the oxidation products are driven to be screened and migrated, a transition metal oxide transfer film is spontaneously constructed, and the sliding interface with the low shearing force and weak adhesion characteristics is constructed; therefore, the friction coefficient of the carbon-based thin film is regulated and controlled through friction-induced sliding interface reconstruction.

The invention discloses a device and method for simultaneously determining the detonation velocity and detonation pressure of micro-scale charge, and belongs to the technical field of detonation performance testing of energetic materials. The device comprises a laser, a charging system, a detonation velocity testing system and a detonation pressure testing system, the charging system comprises a charging clamp, a sample containing groove and a sample ring, the sample containing groove is fixed in the charging clamp, and the sample ring is assembled in the sample containing groove; the detonation velocity testing system comprises a probe, a direct-current power supply and an oscilloscope; the detonation pressure testing system comprises an optical fiber and a photon Doppler velocity tester. During use, a to-be-tested sample is pressed into the sample rings, the probe is embedded between the two sample rings, the to-be-tested sample is detonated by using energy provided by the laser, the detonation velocity and detonation pressure of the to-be-tested sample can be measured at the same time, the sample amount required by a single shot experiment is within 200 mg, and compared with a traditional test method, the sample consumption is greatly reduced, and meanwhile, the test safety is improved.

The invention belongs to the technical field of electric vehicle speed reducers, and relates to a PSO-MCKD-based gear early fault diagnosis method, which comprises the following steps: firstly, adaptively optimizing parameters L and T in an MCKD algorithm by applying PSO for an acquired gear early fault signal containing a relatively weak gear early fault signal; based on the MCKD algorithm after parameter optimization, impact component enhancement is carried out on the early fault signal; thirdly, analyzing the enhanced signal by using an envelope spectrum analysis method, and comparing the enhanced signal with a theoretical fault feature; and finally, outputting a diagnosis result. According to the method, the problem that the parameters L and T of the MCKD cannot be optimal at the same time is solved, adaptive optimization is carried out on the parameters of the MCKD by adopting PSO, the problem that early fault features of the gear cannot be accurately extracted is solved, noise reduction is carried out on an original signal by using the optimized MCKD, and envelope spectrum demodulation is carried out to extract the fault features, so that the accuracy of the fault feature extraction is improved. And the efficiency and accuracy of early weak fault feature diagnosis of the gear are improved.

The utility model discloses a wide dual-passband filter topological structure and a filter. The topological structure comprises a first transmission line and a second transmission line which are arranged in parallel. The first end of the first transmission line is sequentially connected with a first microstrip line and an input end, and the second end of the first transmission line is sequentially connected with a second microstrip line, a third microstrip line and a first open circuit branch knot; the first end of the second transmission line is sequentially connected with a sixth microstrip line and an output end, and the second end of the second transmission line is sequentially connected with a fourth microstrip line, a fifth microstrip line and a second open-circuit branch knot; the first transmission line, the second transmission line, the third microstrip line, the fifth microstrip line and the sixth microstrip line are arranged along a first direction; the first microstrip line, the second microstrip line, the fourth microstrip line, the first open-circuit branch knot and the second open-circuit branch knot are arranged along a second direction, and the first direction is perpendicular to the second direction. According to the utility model, the defect that the existing dual-passband filter is large in size is overcome.

The invention discloses a foldable unmanned aerial vehicle rapid deployment system and method, and belongs to the technical field of unmanned aerial vehicle deployment. According to the system disclosed by the invention, the carrying system and the unmanned aerial vehicle storage device are arranged, the high-speed carrying system provides thrust and energy in the rapid deployment process, and the unmanned aerial vehicle storage device provides a protection function in the deployment process, so that rapid response and long-distance accurate and reliable deployment of the unmanned aerial vehicle are realized; the time for the unmanned aerial vehicle to arrive at the designated area is greatly shortened, the technical problem that an existing unmanned aerial vehicle deployment mode is poor in timeliness is solved, the time for the unmanned aerial vehicle to arrive at the designated area is greatly shortened through the high-speed and accurate deployment process, the burden that the unmanned aerial vehicle carries energy needed by take-off and climbing is avoided, and the unmanned aerial vehicle deployment efficiency is improved. And the task execution time of the unmanned aerial vehicle is prolonged.

The invention discloses a thin-film capacitor electrode defect detection method, device and equipment and a medium. The method comprises the following steps: acquiring a thin-film capacitor electrode image, and cutting the thin-film capacitor electrode image; performing manual defect labeling on the cut image, and generating an XML labeling file in a VOC format; performing first processing on the film capacitor electrode image with the defect manually marked and the XML marking file in the VOC format to obtain a first processing image and the XML marking file in the VOC format; correspondingly combining the thin-film capacitor electrode image of which the defect is manually labeled, the XML labeling file in the VOC format, the first processing image and the XML labeling file in the VOC format, performing data set division to generate a training set and a verification set of a detection model, and completing training of the detection model through the training set and the verification set; and through the training model, completing defect detection of a newly collected film capacitor electrode image. According to the invention, the complex image preprocessing and feature extraction steps are reduced, and the feature capture of the small target is more accurate.

The invention relates to the field of blood products, in particular to a preparation method of human activated C1 esterase. The preparation method comprises the following steps: carrying out contact adsorption on cryoprecipitate-removed human plasma and anion exchange resin, washing and eluting to obtain an eluent containing C1 proesterase, carrying out membrane filtration, and carrying out ultrafiltration desalination treatment; adding an anionic substance into the eluent, incubating in vitro to effectively activate the C1 esterase, and filtering by a membrane to obtain a feed liquid containing the activated C1 esterase; and carrying out purification treatment on the feed liquid to obtain the human activated C1 esterase, and carrying out cryopreservation. According to the method, an anion substance is introduced into an eluent system obtained through anion exchange, mild and effective activation of the C1 proesterase is achieved under the condition of not depending on foreign proteolytic enzyme, the risk of non-specific hydrolysis and self-cutting inactivation is remarkably reduced, good synergy is formed with a follow-up purification process, and the method is suitable for industrial production. The human activated C1 esterase product with relatively high activity and good stability is obtained.

The application discloses an article recommendation method and related device, and relates to the field of article recommendation. According to the article ID contained in the article recommendation request, a positive query result is obtained. If the positive query meets the first missing level, and the incremental benefit evaluated based on the article recommendation scene data is not less than the first benefit threshold, a semantic vector is generated according to the article key field to recall a first group of candidate articles similar to the current article from a vector library, a second group of candidate articles corresponding to the current article is recalled from other multiple different databases, multi-dimensional feature fusion evaluation and sorting are performed on each candidate article, and the top N articles are selected as recommended articles. The application reduces invalid filling of low-value recommendation requests through information missing level and incremental benefit evaluation. When the positive feature is missing, candidate articles are recalled from the vector library and multiple different databases outside the vector library, multi-dimensional feature fusion evaluation and sorting are performed on each candidate article, and article recommendation is realized.

The invention relates to the technical field of computer vision and coal mine intelligent mining, and discloses a lightweight working face three-dimensional reconstruction method and device fusing geometric prior constraints, and the method comprises the steps: firstly screening a dynamic effective region based on a projection relation and texture features, and removing static redundant scenes; performing sparse visual reasoning and depth regression only for the effective area to generate an initial depth map; carrying out physical structure correction on the depth map by utilizing geometric priori such as vertical upright posts and parallel top plates of the hydraulic support; and finally, constructing an optimization objective function containing deviation constraints of the vertical column perpendicularity and the top plate parallelism, and completing high-precision fusion of the incremental point cloud. According to the method, inference delay is reduced by reducing invalid calculation, structural distortion and registration drift caused by an underground severe environment are effectively inhibited by using a geometric law, and real-time and high-fidelity reconstruction of a working face scene is realized.

The application discloses a feature placement method and system for multi-GPU sampling graph neural network training, uniformly models a sampling graph neural network training process, and obtains hardware topology information; in a pre-sampling process, the number of times of accessing node features is counted as node heat, the node features are sorted according to the node heat, the node features are divided into a plurality of feature blocks according to a single-block capacity, and the feature blocks are divided into hot blocks or cold blocks according to block heat; based on the feature block set obtained by the division and the obtained hardware topology information, a mixed integer programming model is constructed, with the minimum bottleneck link time in a feature extraction stage and the memory load imbalance degree as targets, the mixed integer programming model is solved, and the placement position, access path and link flow of the globally jointly optimized hot block are obtained. The application can obtain a high-quality feature placement scheme under a given hardware constraint, minimize the extraction time, and improve the resource utilization rate of a multi-GPU system.

This utility model relates to a PE foam car cargo box mat, comprising: a base plate, an insert groove inside the base plate, multiple sets of mounting holes on both sides of the bottom of the insert groove, six sets of positioning slots on both sides of the top of the base plate, a set of snap-fit ​​components on both sides of the inner wall of the positioning slots, a set of limiting slots on both sides of the inner wall of the insert groove between adjacent sets of positioning slots, multiple mat body parts inserted into the insert groove at positions corresponding to the limiting slots, pressure plate grooves on both sides of the top of the mat body and at the center of the mat body, multiple sets of positioning insertion holes at the bottom of the pressure plate grooves on both sides, and a pressure plate bracket inserted into the top of the base plate above the mat body. The beneficial effects of this utility model are: the design facilitates the disassembly and replacement of damaged areas of the mat, enhancing overall practicality.