28results about How to "Solve redundancy" patented technology

The present application relates to the technical field of medium and shallow coalbed methane development, in particular to a kind of medium and shallow coalbed methane well production dynamic prediction method, comprising the following steps: S1, target well data acquisition and preprocessing;S2, random forest algorithm filters production main control factor;S3, fusion attention mechanism's CNN-LSTM multimodal time series feature extraction;S4, physical constraint deep learning model construction;S5, mixed loss function construction and model training;S6, production dynamic prediction and result output.The present application fuses physical constraint and deep learning technology, both utilize the efficient feature extraction capability of deep learning, and also guarantee the physical consistency of prediction result by physical constraint, solve the problem that pure data driven model generalization ability is poor, long-term prediction error is big.

The application discloses a system and method for accelerating reading compressed files based on A Virtual File System, and belongs to the technical field of time series databases; the technical problem to be solved by the application is how to realize column format compressed storage and reading of time series data types, reduce IO, memory and CPU consumption caused by decompression, and improve the ability of the system to process time series data; the technical scheme adopted is as follows: in the time series engine of the open database ZNBase, an execution engine generates an execution plan according to SQL semantic analysis, calls a storage layer interface of ZNBase, and queries accurate time series history partition compressed data; the storage layer completes fast execution of the query in the storage layer based on various compressed query interfaces provided by AVFS, and accelerates the query of the database; the specific steps are as follows: pre-analysis; compressed file pre-reading; reading compressed data.

The invention provides an image-driven three-dimensional animatable asset generation method and device based on unified field representation and a program product. The method comprises the following steps: carrying out feature extraction on an obtained single RGB image to obtain a multi-view feature and a global semantic feature; generating sparse three-dimensional voxel representation based on the multi-view features and the global semantic features; generating, by a structured encoder, a unified field representation based on the sparse three-dimensional voxel representation, the unified field representation comprising a shape field, a bone field, and a skin field; generating three-dimensional animatable assets with geometric shapes, skeleton structures and skin weights based on the unified field representation; wherein the skeleton field adopts a confidence attenuation mechanism to process the fuzziness of skeleton connection, and the skin field adopts double skin feature fields to respectively associate geometric features and skeleton features. The effect of directly generating animatable three-dimensional assets with high-fidelity geometry, reasonable skeleton structure and accurate skin weight from a single image is achieved.

The application relates to an aluminum template installation precision control method, system, device and medium. The method comprises the following steps: extracting the actual local characteristics of the installed aluminum template, matching the actual local characteristics with the predefined key local characteristics, and giving corresponding semantic labels; calculating six-degree-of-freedom deviation parameters and weight coefficients to construct a multi-objective optimization function; based on the function, based on the physical connection relationship and the physical stroke limit value, a mathematical optimization problem is constructed; the problem is numerically solved to obtain the optimal adjustment amount of each aluminum template unit, generate and execute the adjustment instruction, obtain the verification point cloud data of the adjusted aluminum template, and verify and iteratively optimize until the new structured deviation data set meets the preset precision requirement. The method constructs a multi-objective optimization model, combines the physical connection relationship and the stroke limit value, realizes high-precision and intelligent adjustment of the aluminum template installation, significantly improves the precision and efficiency of the aluminum template installation, and enhances the global optimization capability and intelligent level of the adjustment process.

The invention belongs to the field of artificial intelligence, and relates to a video key frame extraction method and device based on self-supervision, equipment and a medium, and the method comprises the steps: obtaining a frame sequence of a target video, extracting the feature vector of each frame in the frame sequence through a pre-training feature extractor, and obtaining a feature vector sequence; determining graph nodes according to the frame sequence, calculating edge weights according to the feature vector sequence, and constructing an initial directed graph; calculating a graph Laplacian matrix, screening positive and negative sample frame pairs from the frame sequence by using a target sampler, and calculating time sequence perception comparison loss; circularly using parameters of the loss updating feature extractor to finely adjust the feature vector, update the edge weight, construct the graph and the like until the loss converges; reconstructing a directed graph, and calculating a Laplacian matrix of the graph to obtain a Laplacian matrix of a target graph; and performing Laplacian matrix spectral clustering on the target image to obtain a target video key frame set. The method can be applied to the business fields of financial science and technology, insurance, medical treatment and the like, and can improve the extraction accuracy of the video key frame.

The invention provides a landslide area extraction method and device, electronic equipment and a storage medium, and belongs to the technical field of image data processing. The method comprises the following steps: respectively inputting a first remote sensing image of a time phase before occurrence of a landslide, a second remote sensing image of a time phase after occurrence of the landslide and a slope map into a first feature extraction network, a second feature extraction network and a third feature extraction network of a landslide extraction depth model, a first feature map, a second feature map and a third feature map which are correspondingly output are obtained; inputting the first feature map, the second feature map and the third feature map into a feature fusion module of a landslide extraction depth model to obtain a fused feature map output by the feature fusion module; and inputting the fused feature map into a decoder of the landslide extraction depth model to obtain a landslide extraction result output by the decoder. According to the method, the accuracy, robustness and efficiency of landslide automatic detection and mapping are improved, the slope map is introduced as a supplement of topographic information, and the identification capability of the model for the potential landslide area is enhanced.

This invention provides a method and system for cross-variable long-term time series prediction that integrates linear and enhanced Transformer methods. The method includes: extracting an input sequence using a sliding window based on historical data and calculating the temporal statistical features of the sequence; generating scaling and bias factors for the window through a shared fully connected network to perform affine normalization on the input and eliminate distribution bias; performing a Fast Fourier Transform on the normalized tensor to extract the main frequency domain components and map them back to the time domain, then fusing positional encoding to form a time-frequency hybrid embedding; constructing an adjacency matrix based on the correlation between variables, using this matrix as a mask in cross-variable attention calculation to suppress interference between weakly correlated variables and enhance variable co-representation; mapping the representation to the prediction length through a time projection layer, and performing an inverse transform using previously generated normalization parameters to restore the original dimensions of the data and obtain the prediction result.

The application relates to a multi-source data fusion man-machine interaction task online acceleration method and system, and belongs to the field of man-machine interaction. The application provides multiple interaction modes through the arrangement of multiple sensors, improves the self-adaptive capability and high reliability; through a task self-adaptive multi-source data fusion method, feature parameters of acquired multi-source sensing signals are extracted, and data fusion is carried out according to a fusion discrimination criterion, the time for processing multiple information of multi-mode man-machine interaction tasks under uncertain conditions is reduced, interaction instruction redundancy is avoided, and the response speed is improved; in combination with a fusion parameter online updating method, the fusion parameters are predicted in multiple steps in advance, the fusion parameters are corrected and updated in real time, and the problems that multi-mode man-machine interaction data cannot be fused online and the prediction accuracy is low are solved. The application is suitable for the fields of man-machine interaction, automatic driving, medical diagnosis and the like, and is used for improving the accuracy of multi-source sensor data prediction results and the reliability of man-machine interaction under various uncertain conditions.

The application discloses a hyperspectral image classification method based on spectral attention and enhanced second-order pooling. First, the acquired hyperspectral original data is subjected to unsupervised dimension reduction by using a self-encoder to reduce information redundancy. Then, the dimension-reduced data is input into a deep learning network, a two-dimensional convolution layer is used to extract spatial information, and a learnable nonlinear spectral attention mechanism is designed to further utilize the spectral information in the data. Next, a diagonal line enhanced second-order pooling method is introduced to capture high-order discriminative features. Finally, a full connection layer is used to obtain a classification result. The application effectively improves the classification accuracy, especially in the case of uneven sample class distribution, significantly improves the recognition performance of the minority class, and enhances the robustness and generalization ability of the model.

The application provides a construction worker unsafe behavior intelligent identification method and early warning system, the method comprises the following steps: obtaining a to-be-detected image of a construction site; inputting the to-be-detected image into a construction worker unsafe behavior identification model to obtain an unsafe behavior identification result of a construction worker; wherein the unsafe behavior at least includes an unworn state, a wrong wearing state or a blocked state of personal protective equipment; and the construction worker unsafe behavior identification model is a neural network model for target detection. The detection accuracy is significantly improved, and the fine-grained identification capability is enhanced. Since a learnable position coding component is introduced into the backbone network, the model can explicitly model the spatial position relationship of the feature map, enhance the spatial position perception capability of the personnel (especially small targets and blocked targets) in the construction scene under complex background interference, solve the problem that the traditional convolutional network is not sensitive to absolute position information, and improve the feature extraction accuracy in a complex background.

The application provides a haptic object recognition method based on gradient adaptive sampling and a 3D neural network, and is used for solving the technical problems of information redundancy / loss caused by the use of a uniform sampling strategy by an existing haptic object recognition model and the technical problem of insufficient generalization ability to process haptic data under different grasping speeds. The steps of the application are as follows: the original haptic frame is sent into a gradient adaptive sampling strategy for adaptive selection of the haptic frame, and a haptic frame set with rapid gradient change is obtained; the haptic frame set is down-sampled under multiple time scales; an MR3D-18 network is used to extract features of the down-sampled haptic frame to obtain features under different time scales; the features under different time scales are fused, and the category of the object is recognized according to the fused features to obtain a predicted classification result. The application is based on a gradient adaptive sampling strategy and a multi-time scale 3D convolutional neural network, and can effectively improve the recognition accuracy of the haptic object recognition task.

The utility model relates to microfiltration membrane detection technical field, concretely is a kind of microfiltration membrane test bubble point clamping device.The utility model, including: holder base, the arc surface of holder base lower end is equipped with bracket seat;Three support rods, three support rods are located below bracket seat;And adjusting structure;Three support rods are connected with bracket seat by adjusting structure;Wherein, the inside of holder base is equipped with porous support plate, the upper surface of porous support plate is placed with porous support screen, the upper surface of porous support screen is placed with test diaphragm, the upper surface of holder base is placed with holder upper cover, the inside of holder upper cover is equipped with several bolt bodies, bolt body is connected with holder base screw thread, the inner wall of holder upper cover is equipped with inlet, the lower surface of holder base is equipped with outlet, the inside of holder upper cover is equipped with exhaust bolt.The problem that existing clamping device is prone to extrude deformation of filter screen is solved.

The application relates to a multilingual news abstract extraction method based on reinforcement subgraph mining, and belongs to the technical field of natural language processing and artificial intelligence. The application is proposed in view of the problems of the traditional extractive abstract method, such as insufficient coherence between sentences, much redundant information, low abstract quality and the like in long document processing, and comprises the following steps: document multi-level semantic heterogeneous graph construction and initialization; feature enhancement is performed on the sentence nodes through a heterogeneous graph hierarchical structure encoder, a global context encoder and an extraction history encoder; key subgraph mining is strengthened; the abstract extraction process is modeled as a multi-step Markov decision process, key sentence nodes are dynamically selected based on a sentence-level feature graph to form an abstract subgraph, and the selection process is optimized through a reinforcement learning strategy. The application is suitable for automatic abstract generation tasks of long texts such as scientific literature, academic papers and long news reports, and has good interpretability, adaptability and abstract quality.

This disclosure provides a method for multicast transmission of routing messages in proximity service relays, relating to the field of communication technology. The method includes configuring a multicast Internet Protocol (IP) address; establishing a connection between UEs supporting the same relay service code and UEs relaying the same UE; and uploading mobile ad hoc network (MAN) routing messages to the multicast IP address. This disclosure, by configuring a multicast IP address in a UE-to-UE relay and establishing a connection between UEs supporting the same relay service code and UEs relaying the same UE, achieves the uploading of MAN routing messages via multicast IP addresses, solving the problem of radio traffic redundancy during routing table generation / maintenance in proximity service multi-hop UE-to-UE relay communication.

The invention discloses a method for determining a test scene and a related device, and aims to obtain a plurality of candidate scenes by combining element values of different scene elements so as to determine a potential scene which may occur but does not occur in a real road and improve the scene coverage rate. Thirdly, clustering a plurality of candidate scenes so as to cluster similar scenes into one scene cluster and cluster scenes with relatively large differences into different scene clusters; according to the method and the device, scene clusters are obtained, candidate scenes included in the scene clusters are sampled, target scenes as representatives are obtained, and test scenes are determined according to the target scenes corresponding to the multiple scene clusters respectively, so that the test scenes are prevented from including excessive similar scenes, and the problem of test scene redundancy is solved. In addition, due to the fact that sampling is conducted from all the scene clusters, the difference between all the scene clusters can be reserved, and therefore various driving scenes can be covered. On the basis, the scene coverage rate and the test efficiency can be considered, more comprehensive simulation test can be completed more quickly, and the simulation test effect is improved.

This invention belongs to the interdisciplinary field of biomedical engineering and artificial intelligence-assisted diagnosis. It discloses a training method for a SERS spectral classification prediction model and its application. The disclosed SERS spectral classification prediction model training method combines CatBoost feature selection and deep learning. First, surface-enhanced Raman spectroscopy (SERS) data of serum from different categories of subjects is collected. After baseline removal, filtering, and normalization preprocessing, CatBoost gradient boosting algorithm is used to evaluate feature importance, considering the high-dimensional redundancy of the spectral data, and to select a subset of discrete feature bands containing key biomarker information. This feature subset is then input into a one-dimensional convolutional neural network model for training. The model constructed by this invention possesses deep feature mining capabilities, and the biological interpretability of the decisions is verified through SHAP analysis. It effectively solves the problems of large spectral noise interference and difficulty in extracting weak pathological features in traditional methods. For example, it can be used for the auxiliary diagnosis of coronary heart disease, achieving non-invasive, rapid, and high-precision classification and diagnosis of coronary heart disease and its subtypes.

This invention discloses a modular dynamic modeling method for complex assembled systems that incorporates origin-based dynamic stiffness correction. This method decomposes the complex system into secondary structural units (including supports and multi-excitation device-vibration isolator combinations) and primary structural units (including excitation device-vibration isolator combinations), establishing modular dynamic equations. This enables rapid adaptive adjustment of the dynamic equations when parameters such as the number of vibration isolators and motors change. Furthermore, to improve the response prediction accuracy of the vibration isolator stiffness model, an origin-based dynamic stiffness parameter is introduced to quantify the influence of component dynamic deformation characteristics on vibration transmission, resulting in a corrected dynamic equation. This invention effectively avoids the redundancy problems caused by repetitive derivations in traditional modeling and significantly improves the prediction accuracy of system vibration response, providing a solid theoretical framework for rapid response to structural changes in the system.

The application discloses a kind of federal learning communication methods for cloud radio access network, its characteristics are the method includes: S1, constructs the federal learning framework for cloud radio access network, the federal learning system of cloud radio access network and quantization neural network, training, cloud radio access network communication model;S2, convergence analysis is carried out to federal learning framework;S3, construct federal learning system energy model, and carry out convergence analysis, according to the convergence result and energy consumption model constructs an optimization problem of system energy minimization, and solve the problem by alternating optimization joint optimization.This application compared with prior art has by quantization neural network using low bit data reduces the energy consumption when training, by using the characteristics of high scalability, throughput and coverage of cloud radio access network, realize the federal learning framework of low energy consumption, solve the problem of large energy consumption of wireless federal learning.

The invention provides a data processing method and device based on a special optical fiber, equipment, a medium and a product. The method comprises the following steps: acquiring special line request information of a user branch through an optical fiber special line; wherein the private line request information represents that the user branch requests to obtain the cloud data; determining a target cloud resource pool according to the private line request information; and sending the cloud data in the target cloud resource pool to the user branch through the optical fiber private line according to the private line request information. According to the method, during data processing based on the optical fiber special line, the effects of reducing redundancy of the optical fiber special line and reducing the cost of the optical fiber special line are achieved.

The invention discloses a polarization spectrum-based unfrozen water content prediction method and system, a medium and a terminal, and relates to the technical field of soil detection. The method mainly aims at solving the problem that the prediction precision of an unfrozen water content prediction model constructed based on reflectivity is reduced due to the change of soil surface reflection characteristics in the ice crystal structure forming process. Comprising the following steps: collecting a multi-angle polarization spectrum of a to-be-detected soil sample at a target temperature, and performing decomposition treatment to obtain a polarization component spectrum set; determining a target freezing and thawing stage based on the target temperature; generating a target input feature combination based on the stage input feature subset corresponding to the target freeze thawing stage; and based on a stage unfrozen water content prediction model matched with the target freezing and thawing stage, generating predicted unfrozen water content according to the target input feature combination.

The application discloses a multi-dimensional tensor-based test literacy cognitive level intelligent labeling method and system, in order to solve the problem of multi-dimensional literacy rating performance collapse caused by expert labeling scarcity, loss of mathematical symbol semantics and long tail distribution in the prior art, a three-dimensional tensor space is constructed by fusing course standards; the first data set is used as a few-sample example, and a diagnostic thinking chain is combined to drive a large language model to generate structured pseudo-labels for unlabeled test questions in the second data set; through the incremental pre-training of the mask language model by injecting mathematical symbol prior and combining the cost-sensitive weighting mechanism based on real slot statistics, the lightweight training of the high-dimensional multi-task joint rating network is realized; finally, the active backflow closed loop of low confidence samples is executed through confidence evaluation. The application greatly improves the detection rate and rating accuracy of high-order long tail literacy under very small samples, significantly reduces the labeling cost, and realizes the engineering intelligent evaluation of cross-version massive question banks.

The invention provides a multi-level mesh knowledge graph end-to-end construction method for a plurality of long documents, which is suitable for the plurality of long documents and comprises the steps of document slicing and preprocessing; a label layer construction and combination step: integrating the initial keywords of all the effective fragments to form an original keyword set; the original keyword set is input into a dynamic spectral clustering double-judgment merging module to be processed, and a tag set is generated by merging keywords; wherein the processing of the dynamic spectral clustering double-judgment merging module comprises dynamic capacity spectral clustering and double-judgment merging; a node layer merging and map generation step: inputting the initial node set after self-supervision optimization into the dynamic spectral clustering double-judgment merging module for processing, and generating a final node set through merging nodes; and integrating the final node set, the final relationship set and the label layer graph, and constructing a multi-stage mesh knowledge graph.

The application relates to a path planning method based on improved bidirectional A* algorithm through opposite node query, which improves the traditional A* algorithm through the mode of bidirectional search and opposite node query, takes the starting point and the ending point as the algorithm origin in the global map, simultaneously carries out bidirectional path search, calculates a global evaluation function to determine the best path. Although the mode of bidirectional search improves the search efficiency, in the search process, the two paths are misaligned and do not intersect in the middle. In view of the situation, when the global evaluation function is the same, the opposite node is taken as the query information to further determine the search node, and a cost function of the current node and the latest node of the opposite search is constructed to ensure that the paths intersect in the middle point. Compared with the traditional A* algorithm, the method provided by the application reduces the number of search nodes and improves the search efficiency.

The invention relates to the technical field of image processing, in particular to a feedforward type large-scale scene reconstruction method based on three-dimensional Gaussian splashing. The method comprises the following steps: acquiring a multi-view image set and constructing an initial camera token carrying view angle identification information; carrying out feature interaction between the intra-frame and the cross-view through an alternating attention geometry Transform, and outputting an updated camera token and an updated image feature token; the pose, the depth map and the depth confidence map of each frame are predicted through decoding branches; the pixels are projected back to a three-dimensional space by using the prediction parameters, and Gaussian attributes and confidence coefficients are regressed by combining image features; and finally, realizing redundancy elimination of Gaussian primitives through voxelization aggregation, and outputting a target view. According to the method, high-efficiency and high-precision three-dimensional reconstruction of a large scene can be realized without the prior of internal and external parameters of a camera and a complex sparse reconstruction process, the calculation overhead is remarkably reduced, and the problems of geometric drift and video memory overflow which are easy to occur in the large scene in a traditional method are solved.

The invention discloses a data acquisition method based on TCP communication and related equipment, and relates to the field of intelligent transmission and acquisition of sensor data. According to solar azimuth information and future weather prediction, sunny areas are identified, and dynamic illumination partition labels and environment sudden change risk indexes are given to equipment in the areas. Differentiated scheduling is carried out based on the real-time illumination influence and the future weather risk. According to the invention, the problem of low-value repeated polling of a non-key area caused by a fixed polling period in the related technology is effectively solved, and the environment monitoring efficiency of the whole operation resource is improved.

The invention relates to the technical field of intelligent monitoring, in particular to a monitoring data processing method and system for degraded rock mass of a hydro-fluctuation belt in a hydroelectric reservoir area. The method comprises the following steps: determining standard data based on original monitoring data and example data of degraded rock mass of the hydro-fluctuation belt, and obtaining difference data between the standard data and the example data; based on the standard data, the paradigm data and the difference data, extracting corresponding standard features, paradigm features and difference data features by adopting a local splicing thread, and splicing the standard features, the paradigm features and the difference data features to obtain final splicing features; performing independent numerical analysis on the difference data to obtain a positioning difference queue; preprocessing the positioning difference queue and the final splicing features by adopting a local important feature splicing thread to obtain abnormal risk features; and based on the abnormal risk features, determining the risk range label of each monitoring item and the rock mass abnormity monitoring result in the standard data. And early-stage accurate monitoring of the rock mass abnormity of the hydro-fluctuation zone in the hydroelectric reservoir area is realized.