59results about How to "Accurate expression" patented technology

The invention discloses a semantic understanding-based technical literature multi-dimensional analysis method, which comprises the following steps of: performing text vectorization processing on text data by utilizing a Sension-BERT model to generate dense vector representation of a text; vector dimensionality reduction is carried out by adopting UMAP, and redundant features are removed; performing unsupervised clustering analysis by using HDBSCAN to generate a clustering result; extracting a subject term from a clustering result by adopting a TF-ICF method; and carrying out multi-dimensional analysis on the clustering result. Compared with the prior art, a comprehensive and efficient patent analysis tool is provided through an advanced semantic understanding patent clustering method, so that deeper interpretation and analysis of technical literatures are realized while the cost benefit is ensured.

The invention relates to an emotion recognition method and system based on brain map features, and belongs to the field of signal processing. Comprising the following steps: constructing a video material database, and collecting an electroencephalogram signal when a subject watches a video in the video material database; preprocessing the electroencephalogram signal, performing time domain segmentation, frequency domain frequency division, feature extraction and spatial domain mapping on the preprocessed electroencephalogram signal to obtain a two-dimensional map, and performing frequency domain superposition and time domain superposition on the two-dimensional map to obtain a four-dimensional brain map; and inputting the four-dimensional brain map into a trained emotion recognition model to obtain a predicted emotion. According to the method, by collecting the electroencephalogram signals when the testee watches the video in the constructed video material database and converting the collected electroencephalogram signals into the four-dimensional data block containing the frequency domain information, the space domain information and the time domain information through the four-dimensional brain map, the emotion recognition accuracy is improved; the problem that fine-grained accurate recognition of corresponding emotions is achieved for materials with small emotional spans based on electroencephalogram signals is solved.

The invention discloses a power cable heating state monitoring and fault prediction method based on artificial intelligence, and relates to the technical field of cables, and the method comprises the steps: collecting power cable data in an operation state, and marking the collected power cable data; a training framework of the state monitoring and fault prediction model is a meta learning and transfer learning fusion framework; a neural network based on feature dynamic influence is adopted; and an auto-encoder based on feature path decoupling is adopted. According to the method, the feature independence is optimized through a feature path decoupling mechanism, the independence among the power cable data features is evaluated and optimized through a regularization loss function, in addition, linear dependence among the power cable data features is calculated by adopting covariance, the expression ability of the independent power cable data features is enhanced through a penalty mechanism, and the power cable data features are more accurately extracted. The processing capability of the model to the power cable data is improved, and the model can still keep accurate expression of the power cable data structure even under the condition that the power cable data features are missing.

The invention discloses a multi-unmanned-vehicle autonomous exploration method and system for a real large-scale complex scene, and the method comprises the steps: carrying out the voxelization of a three-dimensional environment, and carrying out the detection of a leading edge point; constructing an initialized undirected passable topological graph for sharing of multiple vehicles and calculation of path distances from the multiple vehicles to a target point; obtaining the pose of the radar relative to the world coordinate system and the pose of the robot relative to the world coordinate system by using observation data of the radar and the IMU; ground radar data extracted through ground segmentation is combined with obstacle point cloud data to guarantee that the robot operates in a complex environment. A brand-new distributed task allocation method for multiple unmanned vehicles is provided, and the dispersion rate and the exploration efficiency between robots are considered; and obtaining a reachable path from the robot to the target point based on the cost function distribution. The problem that in the field of multi-unmanned-vehicle exploration, multi-unmanned-vehicle exploration based on 3D laser radar is mostly applied to a simulation environment and is difficult to deploy to a mobile robot to explore a real environment is solved.

The invention discloses an AFDX (Avionics Full Duplex Switched Ethernet) data processing method, and relates to the technical field of data processing. Comprising the following steps: step 1, data acquisition; step 2, data storage; 3, data preprocessing; and 4, data processing. According to the AFDX data processing method, an initial distillation sample is obtained by sampling an initial data set, and a first expression loss value of each initial sample is calculated according to the similarity between the initial distillation sample and each initial sample in the initial data set; therefore, whether the initial distillation sample can completely express each initial sample in the initial data set or not can be determined, so that the sample data can express each initial sample in the initial data set more completely and more accurately.

The invention discloses a meteorological monitoring method and system for low-altitude navigation and the Internet of Things. The meteorological monitoring method comprises the steps that a historical time sequence of meteorological monitoring data and historical observation records of meteorological events are obtained; preprocessing the historical time sequence of the meteorological monitoring data to generate standard sequence data; carrying out correlation analysis between the standard sequence data and meteorological event historical observation records to generate a meteorological event correlation; inputting the standard sequence data and the meteorological event association into an improved meteorological monitoring model, and executing model training; obtaining meteorological monitoring data in real time, inputting the meteorological monitoring data into the improved meteorological monitoring model, and generating real-time meteorological event prediction; and the real-time meteorological event prediction is sent to a low-altitude meteorological monitoring guarantee platform for real-time publishing, and if the real-time meteorological event prediction shows an extreme meteorological event, an early warning signal is broadcasted. The problem that a meteorological monitoring method and system specially oriented to low-altitude navigation and low-altitude internet-of-things scenes are lacked at the present stage is solved.

The invention relates to a pre-drilling stratum depth prediction method, and belongs to the technical field of oil and gas field exploration and development. According to the seismic interpretation result and the geological data, determining well position coordinates of a design well or an intention well in the target area, converting the well position coordinates of the design well or the intention well into geographic latitude and longitude coordinates, and obtaining well point elevation data values of the design well or the intention well according to the current latitude and longitude coordinates and the topographic data of the target area; performing well-seismic joint time-depth conversion work, determining sand body top and bottom positions and a horizontal section track of a well section of the design well according to the horizon after depth conversion and a transverse change trend of seismic inversion, and predicting a horizontal section track depth value of the design well or the intention well according to a seismic datum plane and an elevation data value of the design well or the intention well; according to the method, well point elevation data can be rapidly and accurately obtained, and the well point depth value can be rapidly and accurately determined.

The embodiment of the invention discloses a method for providing commodity information and electronic equipment, which are applied to a client of a commodity information service system, and the method comprises the following steps: in a target interface of the client of the commodity information service system, providing an operation option for performing artificial intelligence (AI) design based on a commodity image of a target commodity; after an operation request of a user is received, obtaining design demand information of commodity design based on first main body image content of the commodity image input by the user; obtaining a target image according to the design demand information, the target image being generated by the following steps: generating prompt information for dialogue with an AI large model, and calling the AI large model according to the commodity image and the prompt information, the AI large model processes first main body image content in the commodity image according to the prompt information, and a target image is generated; and returning the target image to the user. According to the embodiment of the invention, a user can conveniently purchase commodities meeting creative requirements of the user.

The invention discloses a method for carrying out simultaneous transmission call translation by using an earphone, relates to the field of call translation, and solves the problem that in the process of simultaneous transmission call translation, if only the paraphrase of original voice is rigidly transmitted in a target language, the connotation and context of an original text cannot be fully captured and expressed, and the interpretation of the original text cannot be fully captured and expressed. The problems of serious information distortion and connotation loss are often caused; by simulating the mood of a speaker, the quality of simultaneous transmission call translation can be remarkably improved, the emotion expression is enhanced, the information accuracy is improved, the communication effect is enhanced, and the user experience is improved; in the aspect of technology implementation, the speech synthesis and sentiment analysis technology can provide powerful support, so that tone simulation is more natural and real; according to the method, voice data of different accent, different speakers, different background noise and different speech speeds are collected, a more robust and accurate voice recognition model can be trained, the method is suitable for diversified voice input, and the generalization ability of the method in different scenes is improved.

The invention relates to a training method of an emotion recognition model, and belongs to the field of signal processing. Comprising the following steps: constructing a discrete emotion set and a video material database, and collecting electroencephalogram signals when a subject watches a video in the video material database; preprocessing the electroencephalogram signal, obtaining a four-dimensional brain map based on the preprocessed electroencephalogram signal, labeling the four-dimensional brain map based on the discrete emotion set, and constructing a data set based on the labeled four-dimensional brain map; and constructing an emotion recognition model, and determining parameters of the emotion recognition model based on a genetic algorithm and the data set to obtain a trained emotion recognition model. According to the method, frequency domain, space domain and time domain information of electroencephalogram signals is integrated through the four-dimensional brain map, parameters of the emotion recognition model are optimized through the genetic algorithm, the emotion recognition accuracy is improved, and the problems that an existing emotion recognition model is low in recognition fine grit, and particularly the recognition accuracy of materials with small emotion spans is poor are solved.

The invention belongs to the field of artificial intelligence, is applied to the field of financial virtual customer service, and relates to a digital human generation method which comprises the following steps: receiving an audio signal and initial digital human image data, and converting the audio signal and the initial digital human image data into audio features and initial image implicit features; inputting the audio features and the initial image implicit features into an autoregressive Transform model, generating image implicit features corresponding to each time step of the audio features through the autoregressive Transform model, and taking the image implicit features as target image implicit features; decoding the implicit features of the target image to obtain a target digital human image; a digital human video sequence is generated based on the audio features and the target digital human image. The invention further provides a digital human generation device, computer equipment and a storage medium. In addition, the invention also relates to a block chain technology, and the digital human video sequence can be stored in a block chain. According to the invention, the trueness and fluency of the generated digital human are effectively improved.

The invention discloses an operation area electronic fence automatic generation and intrusion alarm method and a medium in the technical field of electronic fences. The method comprises the steps of obtaining a full-scene standard three-dimensional model based on three-dimensional data of a transformer substation building and primary equipment and shooting data of secondary equipment; a digital model is constructed based on key information of typical work tickets and secondary safety measure tickets, correlation mapping of the digital model and a full-scene standard three-dimensional model is established through a global unique identifier to generate digital expression, and intervals and screen cabinets of operation and inspection work are divided into a plurality of work areas. Automatically generating a virtual electronic fence based on PRM sampling path planning and collision early warning; different monitoring precision virtual electronic fences are set for indoor and outdoor environments according to the type of a working area, the position of an irregular obstacle is obtained in real time, and an alarm is triggered immediately when an intrusion behavior is found. According to the invention, automatic generation of the electronic fence and intrusion alarm are realized, the operation safety risk is reduced, and the human resource configuration and process management and control efficiency are optimized.

The present invention provides a method and device for classifying demand documents based on machine learning, which can be used in the field of financial technology. The present invention classifies demand documents through a machine learning model, which solves the problem that current demand documents need to be manually classified, meets the needs of architects and developers, and reduces the learning cost of architects and developers for flowchart design. The requirements are visualized, allowing the architecture to express the requirements more accurately, allowing developers to understand the requirements more clearly, increasing the effective coding time of developers, and improving the accuracy of classification through continuous learning.

The disclosed embodiments provide a method, device, equipment and readable storage medium for generating a video preview animation. The method comprises: obtaining all the image frames to be selected corresponding to the video to be processed; performing a screening operation on all the image frames to be selected to obtain a plurality of target image frames, wherein the similarity between any two target image frames is lower than a preset threshold; generating a preview animation based on the plurality of target image frames. Since the similarity between any two target image frames is lower than the preset threshold, the preview animation produced based on the plurality of target image frames can more accurately express the content of the video to be processed. The technical problem that the animation produced by the existing animation production method cannot capture the changes in the video is overcome. In addition, the quality of the generated preview animation can be improved, so that the user can understand the video content more intuitively.

The invention discloses a three-view depth vision observation space construction and calibration method. The method comprises the following steps: constructing a depth vision observation unit based on a depth camera and a range finder; the three depth vision observation units are arranged in a mutually perpendicular configuration to form a three-view-angle depth vision observation space, and a main coordinate system is set in the three-view-angle depth vision observation space; executing camera coordinate system X-axis, Y-axis and Z-axis calibration, so that the view boundary line of the depth camera is parallel to the corresponding calibration line on the reference calibration object, and the camera coordinate system Z-axis is perpendicular to the corresponding calibration surface; an external reference calibration object is placed, a transformation matrix between coordinate systems is calculated according to the three-dimensional coordinates of the surface center point of the external reference calibration object in the main coordinate system and the camera coordinate systems, and the mapping relation between the main coordinate system and the camera coordinate systems is established. According to the method, the three-view-angle depth vision observation space is established, calibration between each view angle and the main coordinate system is completed, target surface information is extracted from three directions, and observation comprehensiveness of shielded objects with changeable attitudes is improved.

The invention discloses an extensible three-dimensional vector plotting graph situation deduction method, and belongs to the field of situation deduction methods, and the method comprises the following steps: 1, formulating corresponding situation rules for various vector plotting graphs, including flicker, rotation, scaling, roaming and growth situation parameter setting; 2, designing and realizing vector plotting graph and situation file management, and storing vector plotting graph information and situation information; 3, editing a vector plotting graph situation; and 4, realizing vector plotting graph situation simulation. Compared with a traditional three-dimensional vector plotting graph situation deduction method, the three-dimensional vector plotting graph situation deduction method has good expansibility and flexibility, and is suitable for making various professional combined primitive animation schemes; the three-dimensional vector plotting graph situation deduction method has the advantages of being accurate in expression, high in intuition, easy to operate, capable of supporting complex scenes, capable of being applied to multiple fields and the like.

The invention provides a question query method, system and device, equipment, a storage medium and a program product. The method comprises the steps of determining a target question; splitting the target question to obtain a plurality of first sub-questions, and determining an association relationship among the plurality of first sub-questions; according to the incidence relation and the question theme types of the first sub-questions, determining query results of the first sub-questions; and determining a target query result of the target question according to the query results of the plurality of first sub-questions. And the accuracy of determining the target query result is improved.

The invention relates to an ammeter error solving method and device and a storage medium, and the method comprises the following steps: obtaining a measurement switch in a metering box and data of all sub-meters corresponding to the measurement switch, and carrying out the preprocessing of the data of all sub-meters; constructing a split-phase energy conservation equation set and a total energy conservation equation set based on the preprocessed data of all the sub-tables; establishing a simultaneous equation set of a split-phase energy conservation equation set and a total energy conservation equation set, and solving through a BFGS quasi-Newton method to obtain an initial fitting residual sequence; screening out the metering period data based on the data in the initial fitting residual error sequence to obtain an optimal data set, solving the optimal data set through the BFGS quasi-Newton method again, carrying out repeated iteration, and stopping iteration until an iteration stopping condition is met to obtain a final optimal data set; and calculating the error of each sub-table of the measurement switch based on the final optimal data set.

The invention discloses a non-pairing cross-modal retrieval method based on entity semantic guidance and relation fusion. The method comprises the following steps: extracting entity image features and entity text features from an entity data set; obtaining an entity public semantic space through negative sample selection and intra-modal neighbor relation constraint based on the entity image features and the entity text features; based on the entity public semantic space, a high-confidence image-text pseudo-pairing set is obtained through a pseudo-pairing generation and screening method; on the basis of the high-confidence image-text pseudo-pairing set, relation information is captured and aligned through a relation fusion module, and a fused public semantic space is obtained; and calculating the similarity between the images and texts based on the fused public semantic space, sorting according to the similarity, and outputting a cross-modal retrieval result. Compared with other unpaired cross-modal retrieval methods, the method has the advantages that the performance is obviously improved, a space obtained by learning can accurately express data from a deeper level, and the whole method is better in interpretability.

The invention provides an image semantic communication method and system based on a multi-modal large language model (MLLMs), and aims to improve the semantic accuracy and system robustness of image transmission. The method comprises the following steps: extracting semantic features from an image source, detecting through an expert AI model, processing an OOD problem in combination with a general AI model, optimizing recognition precision by using a Bayesian method, encoding the features and transmitting the encoded features through a wireless channel. And after decoding by a receiving end, reconstructing the image by using a generation-evaluation framework based on the MLLMs and performing iterative optimization so as to ensure accurate reconstruction of semantic information of the image. According to the method, the OOD data processing capability is particularly enhanced, the image reconstruction quality is improved through the generation-evaluation framework, the limitation of a traditional model on new data distribution is solved, and the reliability of the system and the accuracy of image reconstruction are enhanced.

The embodiment of the application provides a video generation method, system, electronic equipment, storage medium and program product, relates to the computer technical field, and the method comprises the following steps: identifying the skeleton structure information of a target role from a reference image containing the target role; acquiring a reference posture sequence used for representing reference posture changes; adjusting the skeleton structure information in the reference posture sequence by using the skeleton structure information of the target role, and generating a target posture sequence, wherein the skeleton structure information in the target posture sequence matches the skeleton structure information of the target role, and the posture change in the target posture sequence matches the reference posture change; and generating a target video according to the target posture sequence and the reference image, wherein the target video expresses the reference posture changes based on the target role. The technical scheme of the embodiment of the application can avoid the problem that the deviation of the target video and the target role in the reference image is large, and accurately expresses the reference posture changes in the reference posture sequence.

The embodiments of the present application disclose a method, apparatus, device and computer-readable storage medium for representing a word, wherein the method comprises: obtaining the root of the word to be represented; when there is no vector representation corresponding to the root of the word to be represented in the trained vector model, decomposing the word to be represented with simple words as the decomposition granularity to obtain a set of simple words; obtaining the vector representation of each simple word in the set of simple words according to the vector model; and determining the word vector representation of the word to be represented according to the vector representation of each simple word.

This invention discloses a robot interaction method, device, electronic device, and storage medium. The method includes: detecting first pose parameters of the first facial target in response to a first condition being triggered by a first facial target; detecting first feature information of the first facial target in response to a angular difference between the facial orientation indicated by the first pose parameters and the optical axis of the target robot's camera being less than a preset angular difference, wherein the first feature information is used to indicate facial feature information and gesture feature information; determining the task intent associated with the first gesture feature information from multiple sets of task intents associated with the first gesture feature information based on the first gesture feature information in response to a matching degree greater than a preset matching degree of the first feature information and a third feature information among multiple sets of second feature information; and controlling the target robot to execute a corresponding feedback action based on the task intent associated with the first gesture feature information. This invention can improve the environmental anti-interference capability and intent recognition accuracy of robot interaction.

The present application discloses an obstacle recognition method, device, equipment and storage medium based on an unmanned vehicle, wherein the method comprises: three-dimensionally rasterizing the body area of ​​the unmanned vehicle to obtain an initial body grid; obtaining the first point cloud data of the unmanned vehicle; matching the first point cloud data with the initial body grid to determine the target point cloud data falling into the initial body grid; optimizing the initial body grid based on the target point cloud data to generate a target body grid; obtaining the second point cloud data of the unmanned vehicle; matching the second point cloud data with the target body grid to determine the body point cloud data falling into the target body grid, as well as the vehicle point cloud data outside the target body grid; using the vehicle point cloud data outside the vehicle as obstacle point cloud data. The separation accuracy, efficiency and stability of the vehicle point cloud data and the vehicle point cloud data outside the vehicle can be improved, thereby improving the accuracy of obstacle recognition.

The invention discloses an intelligent question and answer method and system for enterprise safety and environmental protection management, and belongs to the technical field of intelligent question and answer. The method comprises the following steps: acquiring a security and environmental protection problem asked by a user, analyzing the security and environmental protection problem asked by the user by using a security and environmental protection semantic analysis model to obtain an analysis result, and judging whether the security and environmental protection problem asked by the current user belongs to the field of security and environmental protection systems or not according to the analysis result; if the method belongs to the field of security and environmental protection systems, rewriting a security and environmental protection problem proposed by a user by using a statement rewriting model to obtain an instruction statement; searching a rule unit in a security and environmental protection knowledge base based on the instruction statement to obtain a search result; and the recall summary model extracts content directly related to the user intention from the search result, and generates a safety and environmental protection answer. According to the method, the corresponding safety and environmental protection result can be generated in time according to the questions of the employees, the operation risk increase is reduced, and the safety and environmental protection system awareness rate of the employees is improved.

The invention relates to an oil-immersed CT reignition arc modeling method, device and equipment and a storage medium, and belongs to the technical field of arc modeling, and the method comprises the steps: taking instantaneous input power, arc loss power and arcing time as input to determine a differential equation of an arc universal model; establishing a differential equation of a first electric arc model based on a hypothesis condition of a Cassie model and the differential equation of the electric arc general model; establishing a differential equation of a second electric arc model based on a Schwarz model and the differential equation of the electric arc general model; and setting a transition function, and obtaining a reignition arc model according to the differential equation of the first arc model, the differential equation of the second arc model, the differential equation of the arc general model and the transition function. According to the method, accurate expression of the arc combustion process is realized through differential equations of a plurality of models.

The invention relates to the technical field of commercial data analysis, in particular to an intelligent demand prediction analysis method and system for commodity auction. The method comprises the following steps: acquiring auction item ontology data; analyzing auction item ontology feature data according to the auction item ontology data; performing dynamic association network analysis on the auction item ontology based on the auction item ontology feature data to generate a dynamic auction item ontology association network; detecting user interaction data; performing user interaction behavior feature analysis processing according to the user interaction data to generate user interaction behavior feature data; performing instant user demand auction probability prediction based on the user interaction behavior feature data to generate instant user demand auction probability prediction data; and transmitting the instant user demand auction probability prediction data to a dynamic auction body association network to carry out instant user association demand auction intelligent analysis, and generating instant user association demand auction prediction data. According to the invention, the accuracy and efficiency of demand prediction in the commodity auction platform are improved.

A neural network-based image processing method and apparatus according to an embodiment of the present invention may generate a motion vector field by using motion information used in motion prediction of a processing unit included in a current picture, and generating a tensor of the motion vector field by performing compression on the motion vector field based on a neural network including a plurality of neural network layers.

The present application provides a method and system for remote sensing assessment of surface anomaly intensity based on multi-scale features, which belongs to the field of remote sensing data processing technology. A conceptual model of surface anomaly intensity is constructed from four scales: pixel scale, structure scale, object scale, and scene scale; a mathematical model of surface anomaly intensity is constructed based on the conceptual model of surface anomaly intensity; a remote sensing assessment model of surface anomaly intensity is constructed based on the conceptual model and the mathematical model; remote sensing image data is used as basic data to measure each remote sensing feature in the remote sensing feature system of surface anomaly intensity, and the weight of each remote sensing feature is calculated using the entropy weight method, and then the value of the comprehensive surface anomaly intensity of the study area is calculated through the mathematical model of surface anomaly intensity according to the value of each remote sensing feature and its corresponding weight. This scheme can adaptively adjust the feature weights at each scale, improving its ability to be applicable to different types of surface anomaly intensity assessments.

The present invention relates to the technical field of information processing and three-dimensional modeling, and specifically to a method for multi-source marine geological information fusion and three-dimensional visualization modeling, comprising the following steps: S1: collecting marine geological data from multiple sources; S2: preprocessing the collected data; S3: extracting marine geological features from the preprocessed data; S4: multi-dimensionally fusing the features extracted in S3; S5: constructing a marine geological three-dimensional model based on the fused data in S4 using a three-dimensional grid construction algorithm; S6: optimizing the constructed three-dimensional model; S7: visualizing the optimized three-dimensional model to generate an interactive three-dimensional display interface. The present invention improves the accuracy, information integrity and user interactive analysis capabilities of the marine geological model through multi-source data fusion, refined three-dimensional model construction and efficient visualization processing, and provides reliable technical support for geological research and engineering applications.