433 results about "Reference vector" patented technology

The invention relates to the technical field of industrial intelligent monitoring, in particular to a remote online monitoring method and system based on machine vision and artificial intelligence, and the method comprises the following steps: synchronously collecting a visible light video stream, infrared thermal imaging data and three-dimensional vibration spectrum data through an edge calculation node, and forming a multi-modal sensing data set; performing cross-domain feature alignment processing on the data set to generate a space-time synchronous composite feature matrix; inputting the composite feature matrix into a cascaded deep learning model, and outputting a three-dimensional diagnosis vector which comprises an equipment health state, an abnormal region and a fault probability; and matching the three-dimensional diagnosis vector with a historical reference vector through a DTW and cosine similarity combination algorithm to generate a deviation index, and outputting a self-adaptive alarm threshold and a maintenance suggestion scheme according to the deviation index. The method has the advantages of high multi-modal fusion degree, light diagnosis model and intelligent and adjustable alarm mechanism, and is suitable for real-time remote operation and maintenance in a high-risk industrial scene.

A machine learning system configured to map an input data instance to an output according to a system mapping. The system mapping may be composed from multiple functions. A latent input vector may be determined in a second latent space for the input data instance, from latent reference vectors related to a number of reference data instances identified as parents.

The invention discloses a retrieval response method and device, a storage medium, electronic equipment and a program product, and relates to the technical field of big data, and the retrieval response method comprises the following steps: under the condition that a prompt instruction for processing an original document is received, performing segmentation processing on the original document according to sequence termination probabilities corresponding to different data sub-blocks in the original document, n parent text blocks are obtained; paragraph splitting processing is conducted on each parent text block in the N parent text blocks, N * M sub text blocks are generated, and N and M are positive integers; vectorizing the N * M sub-text blocks to obtain N * M query reference vectors; according to the method, retrieval response is performed on the query problem issued by the target object based on the N * M query reference vectors, the technical problems of semantic fracture and low retrieval accuracy caused by traditional fixed blocking are solved, and the technical effect of improving the retrieval accuracy after document blocking is achieved.

The invention relates to the technical field of time synchronization, and discloses a hybrid time service method based on Beidou satellite navigation and a TSN, which comprises the following steps: acquiring network topology structure information and historical time service error data of each synchronization node in a target distributed system, and constructing a dynamic time delay state set for jitter identification; based on the dynamic time delay state set, a current global time reference is extracted through a time calibration algorithm, a time difference reference vector sequence is generated, and initial synchronization adjustment is carried out on the TSN main node; constructing an inter-node clock convergence function curve according to the slave node response stability, and judging whether the synchronous network is in a low jitter interval or not based on the time offset rate and the time drift trend change; in combination with the node time state after directional offset suppression, a multi-channel redundant time synchronization mechanism in a time window is applied, a weight adjustment coefficient is extracted, and an unbalanced calibration factor is introduced; and according to a regression correction result, evaluating a time service stability index in real time. The method has the advantage of improving the time service stability.

The invention discloses an ML-based rear subframe structure lightweight and modal optimization target design method, which comprises the following steps: (1) through three-dimensional modeling, statics analysis and modal analysis, constructing a mathematical model corresponding to the weight of a rear subframe and first-order inherent frequency simulation; (2) generating a population based on a Maximin criterion and a Latin hypercube, and establishing a radial basis function machine learning model; (3) generating candidate offspring individual vectors by adopting a Pbest-driven DPM evolutionary strategy based on an Eplison function; (4) constructing a minimum and maximum Pareto front lifting function to screen real offspring individual vectors; and (5) performing simulation evaluation on a real offspring individual vector, adaptively switching the reference vector type based on a reverse generation distance updating condition, returning to the step (3) until an optimization target meets a design requirement, and outputting an optimal optimization design parameter value. According to the method, the evolution direction is adaptively adjusted according to the population simulation result, and the optimization design effect on the two targets of lightweight and modal optimization is good.

Currently, technical expertise is required to construct data maps between data sources and targets within integration processes. Disclosed embodiments enable automated generation of data maps using semantic comparisons and artificial intelligence. In particular, data may be converted into reference vector embeddings that define a semantic location of the data within a multi-dimensional vector space. User requests may be converted into input vector embeddings, based on source and target information. The input vector embeddings may be semantically compared to the reference vector embeddings, potentially using a graph database, to identify semantically similar data. This semantically similar data may then be used to generate a prompt, which may be input to a generative language model to automatically produce a data map for use in an integration process.

The invention discloses a noise signal recognition method based on deep learning. The method comprises the steps of original signal collection, signal preprocessing, noise recognition feature extraction, noise signal recognition model establishment and noise signal intelligent recognition. The invention relates to the technical field of digital signal data processing, in particular to a noise signal identification method based on deep learning, and innovatively provides a self-adaptive signal mode decomposition method based on a combined judgment mechanism of energy threshold value judgment and energy range value judgment. Decoupling and reservation of different frequency bandwidth information in an original signal can be effectively completed, and the recognition accuracy of non-stationary communication noise is improved; a frequency band filter bank constructed based on a Gaussian function is introduced, so that the accuracy of communication signal identification is improved; a method for constructing a noise recognition model by fusing graph neural network structure modeling and an original type reference vector matching mechanism is innovatively proposed, and the recognition robustness of the model to non-stationary communication interference signals is improved.

The various implementations described herein include methods and systems for coding video. In one aspect, a method includes receiving a video bitstream comprising a current picture composed of a plurality of blocks, including a current block; identifying a reference block by applying a template-matching technique to the current block; identifying a reference vector for the reference block, where the reference vector is a motion vector or a block vector; populating a merge candidate list for the current block using the reference vector; and reconstructing the current block using information from the merge candidate list.

The invention discloses a multi-modal collaborative recommendation method, system and device, a medium and a program product, and relates to the technical field of multi-modal recommendation, and the method comprises the steps: carrying out the feature enhancement of a user portrait and a commodity attribute through a large-scale language model, and obtaining a more comprehensive and precise text modal representation through generating more abundant semantic description; a modal reference vector is introduced, and the similarity between the modal features and the modal reference vector is calculated to evaluate the importance of different modals, so that modal noise is filtered, a user-commodity bipartite graph and a commodity-commodity similarity graph based on the modal features are constructed, and user preferences are captured from user-commodity interaction and commodity-commodity semantic relationships; by integrating a difference perception attention mechanism and a dynamic modal preference gating mechanism, fine-grained fusion of multi-modal features is realized, so that the model can adaptively adjust the importance of each modal according to different scenes.

The invention relates to the technical field of crop disease and insect pest monitoring, and particularly provides a crop disease and insect pest intelligent monitoring system based on image recognition and machine learning, and the system comprises a reference model generation subsystem which obtains a continuous state flow in a crop growth period through a multispectral imaging array; performing micro-variation field comparison on the real-time image of the continuous state flow and the ideal form reference vector, and outputting a biological stress difference chart; the mutual exclusion association engine subsystem purifies a pathological response map, inputs the pathological response map into a pathological characteristic dissociation network, and outputs an environmental immune type disease mark set; and the disease source trajectory deduction subsystem diffuses a power chain through organisms on the basis of the initial infection focus coordinates in combination with a real-time wind direction vector field and a plant density topological graph. According to the method, the problem of high false positive rate in a complex farmland scene is solved through a dual denoising mechanism of growth rhythm decoupling and environmental interference stripping; and finally, monitoring dimension jump from single-point identification to group prevention and control is realized.

The invention relates to the technical field of intelligent monitoring and fault diagnosis of a photovoltaic power generation system, in particular to a photovoltaic grid-connected box electric energy quality monitoring system based on data acquisition of the Internet of Things. Comprising an environment perception and time domain synchronization unit which is used for triggering sampling by using a unified clock source and generating a full-dimensional synchronous original data stream; the ideal reference physical reconstruction unit is used for calculating a theoretical output amplitude envelope and generating an ideal state reference vector; the fault mechanism parameterization evolution unit is used for converting the fault mechanism into a fault operator, selecting a high-risk fault mode according to the current operation condition, and superposing the corresponding fault operator onto the ideal state reference vector to generate a theoretical damaged state waveform; the double-rail-difference manifold extraction unit is used for obtaining a real residual vector and a theoretical fault residual vector; and the isomorphic feature coupling judgment unit is used for calculating the similarity between the real residual vector and the theoretical fault residual vector. According to the invention, the robustness of power quality monitoring under complex meteorological conditions is significantly improved.

An approach is provided for computer-vision-based object motion detection. The approach involves receiving a video sequence captured from a perspective of a vehicle / device traveling at street level. The approach also involves processing frames of the videos sequence using computer vision to determine semantic localization feature(s) associated with object(s) detected in the frames. The approach further involves clustering the object(s) into object cluster(s) across the frames based on the semantic localization feature(s). The approach further involves creating an object vector connecting the object(s) in a given cluster of the object cluster(s) across the frames. The approach further involves computing an angle between the object vector and a reference vector associated with a movement of the vehicle / device across the frames. The approach further involves classifying a motion of the object(s) relative to the vehicle / device based on the angle.

Currently, technical expertise is required to construct data objects within integration processes. Disclosed embodiments enable automated generation of data objects using semantic comparisons and artificial intelligence. In particular, data may be converted into reference vector embeddings that define a semantic location of the data within a multi-dimensional vector space. User requests may be converted into input vector embeddings. The input vector embeddings may be semantically compared to the reference vector embeddings to identify semantically similar data, and related data may be identified using a graph database. This semantically similar and related data may then be used to generate a prompt, which may be input to a generative language model to automatically produce a data object for use in an integration process.

The invention discloses a data fusion method. The method comprises the following steps: acquiring an image and radar point cloud data collected in a to-be-detected area; performing feature extraction on the image data, mapping an obtained feature map to a preset three-dimensional coordinate system to generate a first visual point cloud feature, mapping the radar point cloud data to the three-dimensional coordinate system to generate a radar point cloud feature, and fusing the first visual point cloud feature and the radar point cloud feature to generate a point cloud fusion feature; performing feature extraction on the point cloud fusion features to generate aerial view features at the current moment; based on a plurality of preset speed reference vectors, performing feature aggregation on the aerial view features at the current moment and aerial view features at a plurality of historical moments to obtain speed-separated aerial view features; and carrying out feature fusion on the bird's-eye view features subjected to speed separation and the bird's-eye view features at the current moment to obtain time sequence fusion features, and carrying out target detection based on the time sequence fusion features. The smear problem of the moving target can be effectively relieved, and the target detection precision is improved.

The invention relates to the technical field of image analysis, in particular to a battery cell assembly polarity detection system based on machine vision, which comprises a reference acquisition module, a brightness analysis module, a feature extraction module, a response identification module and a classification judgment module. According to the invention, the reference containing the reference vector direction angle and the initial gray average value is established, so that the direction and brightness deviation caused by illumination change in the current detection period can be quantified, and the compensation instruction is generated according to the direction and brightness deviation to dynamically adjust the exposure time length and the light source voltage, thereby realizing the real-time correction of the acquired image; meanwhile, the local direction entropy of the tab area is calculated in the corrected image, and weighting processing is carried out in combination with the neighborhood brightness mean value difference, so that the distinction degree of surface textures and gloss characteristics of materials with different polarities is enhanced, and the accuracy of the surface texture and gloss characteristics of the materials with different polarities is improved. And finally, performing comprehensive judgment according to multi-dimensional indexes such as a brightness reflectivity mean value and a local direction entropy strength ratio.

A method and apparatus for generating checking data, an electronic device and a non-volatile computer-readable storage medium. The method includes: acquiring a store-status matrix corresponding to a redundant array of independent disks; determining non-reference vectors among the magnetic-disk vectors, and performing data-vector-element evenly dividing processing to an initial data region formed by the non-reference vectors, to obtain first and second data regions; performing data-exchange processing to the first and second data regions in the store-status matrix, to obtain a first matrix; determining a plurality of second-order matrixes in the first and second data regions of the first matrix, and performing position-exchange processing to target magnetic-disk elements in the second-order matrixes, to obtain a second matrix; based on the second matrix, performing single-error-correction encoding generating processing, to obtain first checking data; and based on the second matrix, performing double-error-correction encoding generating processing, to obtain second checking data.

The invention discloses a lightweight surface electromyogram signal classification method and system. The method comprises the steps that original surface electromyogram signals are acquired; maintaining a dynamic reference vector capable of tracking the change of the physiological baseline of the signal; standardizing the original signal by using the dynamic reference vector, and generating a channel hybrid matrix according to the vector; extracting a local time-frequency characteristic pattern from the standardized signal; transforming the local time-frequency feature map by using a channel hybrid matrix so as to output a collaborative adaptive feature map capable of adapting to muscle collaborative mode change; and through a parallel sequence modeling module fusing a state space model and a multi-head self-attention mechanism, processing the collaborative adaptive feature map to obtain a final action category. The method improves the robustness of the model to individual difference and muscle fatigue, has the advantages of light weight and low delay while maintaining high classification precision, and is suitable for mobile or wearable equipment.

Examples in the present disclosure relate to hand chirality estimation. Tracking data captured by one or more sensors associated with an extended reality (XR) device is processed to determine positions of a plurality of joints of a hand of a person. A reference vector is generated based on a first subset of the positions. The first subset of the positions includes positions of at least two metacarpophalangeal joints. A plurality of bending angles is determined based on at least a second subset of the positions. Each bending angle represents an angle between a respective pair of articulating bones that is measured in relation to the reference vector. An estimated chirality of the hand is identified based on the plurality of bending angles. Operation of the XR device is controlled using the estimated chirality of the hand.

The invention belongs to the field of attitude estimation, and discloses a SigKAN network assistance-based Kalman filtering attitude estimation method, which comprises the following steps of: establishing a discrete time state equation and a measurement equation, and modeling a disturbance component and a reference component; a SigKAN model is trained; a SigKAN model is used to predict the dynamic characteristics of the noise; removing a disturbance component by using a Kalman filtering algorithm, and initializing state variable priori estimation and a covariance matrix thereof; calculating a residual error and a covariance thereof by using the measurement information, calculating a Kalman gain, and obtaining a state variable estimation value and a covariance matrix thereof; the gravitational acceleration and the geomagnetic field serve as observation vectors, a reference vector is obtained, attitude calculation is conducted through a QUEST algorithm, and an attitude estimation quaternion is obtained; and obtaining attitude estimation at all moments. According to the method, the dynamic characteristics of disturbance are learned by adopting the SigKAN network, disturbance estimation is carried out by assisting the filter in a mode of dynamically adjusting process noise covariance, and the anti-interference capability of a traditional Kalman filter is enhanced.

The invention provides a wear coefficient correction method for a shield cutter, and relates to the technical field of wear monitoring, and the method comprises the steps: collecting a time sequence magnetic field signal obtained by a magnetic sensor array disposed on a non-rotating part of a shield tunneling machine, and synchronously obtaining a cutter angle coding signal; performing time-frequency conversion and filtering on the magnetic field signal to obtain a noise reduction time sequence signal; determining a geometric position sequence of each shield cutter in combination with the angle coding signals; based on the magnetic sensor pose, the tool position and the magnetic dipole model, theoretical magnetic field mapping between the tool and the sensor is constructed, and a system response matrix is generated; reconstructing an independent magnetic feature vector of each cutter to obtain an optimal feature set; generating a vector difference through comparison with a reference vector, inputting the vector difference into an abrasion identification model, obtaining the abrasion loss and type identification of each cutter, and correcting an abrasion coefficient; the method improves the wear recognition precision and reliability, and is suitable for tunnel construction under complex geological conditions.

The invention relates to the technical field of textile bio-manufacturing and industrial process automation control, in particular to a bio-based fabric production temperature and humidity intelligent regulation and control system which comprises a data acquisition unit used for acquiring temperature and humidity sensor values in a fermentation curing workshop in real time and reading current heating power of an actuator and a fan rotating speed instruction; the reference state reconstruction unit is used for receiving the input state vector and generating an interference-free reference vector; the disturbance simulation unit is used for generating a theoretical damaged state vector representing a specific fault mode; the double-track difference unit is used for generating a real difference feature vector and a theoretical difference feature vector; the coupling regulation and control unit is used for calculating a similarity value of the real difference feature vector and the theoretical difference feature vector in a number domain space; according to the method, non-ideal factor interference is effectively eliminated, and the physical interpretability and numerical traceability of the reference data are ensured.

The invention discloses a model illusion detection method and system and a related device. The method comprises the steps that a first reply text obtained by inputting a question text into a trained first language model and a second reply text obtained by inputting the question text into an untrained second language model are obtained; obtaining key information based on the question text and the first reply text, and obtaining feature vectors of multiple output hierarchies corresponding to the key information based on reference vectors of multiple output hierarchies obtained in the output process of the key information and the second reply text; and obtaining an information vector corresponding to the key information based on the feature vectors of the plurality of output hierarchies, and determining an illusion detection result of the first reply text based on the information vector. In this way, the accuracy of text illusion detection can be improved.

The invention relates to a BIM model-based maintenance method and system for a split-phase vertical laminated GIS bus cylinder. The method comprises the following steps: collecting point cloud data through laser scanning to construct a three-dimensional model; key feature points are extracted, and displacement reference vectors are determined; constructing a displacement constraint matrix and optimizing a path; generating an interactive three-dimensional space model; simulating a dismounting sequence and evaluating the risk; generating an emergency instruction and a tool track; and the maintenance process is optimized by feeding back the cycle correction model. Visual simulation, precise control and dynamic optimization of the whole overhaul process are achieved, and the overhaul operation precision, efficiency and safety are remarkably improved.

The embodiment of the invention discloses a database query optimization method and device based on a large model and a medium, and relates to the technical field of databases. The method comprises the steps that under triggering of a current query request, a query SQL text of the current query request is collected in real time, and real-time environment data of a current execution engine is obtained; vectorizing the query SQL text and the real-time environment data to determine a query reference vector; the query reference vector is input into a pre-trained space-time attention model, performance bottleneck positioning data for the current query is determined, a query optimization strategy is determined based on the performance bottleneck positioning data, and the query optimization strategy comprises an index optimization strategy and a query rewriting strategy; and optimizing the current query request according to the query optimization strategy, determining optimized query information, performing rewriting security verification on the optimized query information in a preset database copy environment, and executing database query based on the optimized query information after the rewriting security verification is passed.

The various implementations described herein include methods and systems for coding video. In one aspect, a method includes receiving a video bitstream comprising a current picture composed of a plurality of blocks, including a current block. The method also includes identifying a merge candidate for the current block and identifying a refined merge candidate by refining a position of the merge candidate using a template-matching technique. The method further includes identifying a reference vector for the refined merge candidate and populating a merge candidate list for the current block using the reference vector. The method also includes reconstructing the current block using information from the merge candidate list.

The invention provides a multi-source heterogeneous data unified semantic representation method based on a large language model. The method comprises the following steps: extracting semantic embedding vectors of modal data through a shared trunk model, realizing cross-modal fusion by virtue of a dynamic modal adapter, generating a uniform representation vector, performing semantic alignment by taking a text modal as a reference, and finally obtaining a text reference vector capable of comprehensively reflecting semantic information of multi-source heterogeneous data. According to the method, the strong representation capability of a large language model is fully utilized, the multi-modal data is encoded in a unified semantic space, the problems of feature space splitting and difficulty in semantic alignment caused by modal difference in a traditional method are effectively solved, the flexibility and semantic consistency of multi-modal data processing are improved, and the method is suitable for large-scale popularization and application. A high-quality semantic representation basis is provided for subsequent tasks such as cross-modal retrieval and semantic reasoning, and the generalization ability and efficiency of the model in practical application are enhanced.

A method for classifying a digital certificate as malicious or non-malicious includes receiving the digital certificate from a network source and extracting textual fields from the certificate. The extracted text is embedded into a high-dimensional vector using a pretrained transformer-based encoder. The resulting test vector is queried against a vector data structure populated with reference vectors derived from known benign and malicious certificates. A similarity search is performed to identify a set of nearest reference vectors. A classification decision is made based on the labels of the most similar / nearest neighbors, using a voting mechanism. If a given set or number of them are labeled as malicious, the certificate is classified as malicious. If not, it is classified as benign. The classification result may trigger a network security action, such as blacklisting the associated IP address or identifying a botnet command and control server. The system may use various embedding techniques, including concatenating subject and issuer fields or embedding individual certificate attributes separately.

The invention discloses an image generation method and device, equipment and a storage medium, and relates to the technical field of image generation, and the method comprises the steps: determining intermediate vector data used for generating a target image set according to a pre-constructed text prompt word set and a noise image set, and the intermediate vector data comprises reference vector data and target vector data; generating a first target image based on the reference vector data; the reference vector data and the target vector data are fused to generate a second target image, and the first target image and the second target image form a target image set corresponding to the style cue words in the text cue word set. According to the image generation method and device, the data of the first target image used as the reference in the generation process is shared to the generation process of the second target image, so that the image set which is consistent in style, diversified and high in quality is generated, the requirement for generating the images consistent in style is effectively met, and the image generation effect and efficiency of the diffusion model are improved.

The invention discloses a wind field forecast data processing method and system based on deep learning, and belongs to the technical field of weather forecast big data analysis. According to the invention, various meteorological data resources can be fully utilized and can be effectively processed according to the integration indication. A previous observation element embedding vector is obtained through observation element mining, and meteorological observation information features can be deeply extracted through feature pooling and vector mining. Meteorological element derivation and trend deduction are carried out based on past cross-modal space mapping vectors, and the connotation of meteorological information is expanded. The subsequently generated weather forecast reference vector improves the vector accuracy. A wind field forecasting training result is obtained through discriminant analysis, forecasting confidence features are obtained, and the credibility of the forecasting result is improved. And the data assimilation thermal feature set locates the target interaction variable to determine the target wind field forecasting processing network, finally, the wind field forecasting result can be accurately generated according to the current information, and the accuracy, reliability and practicability of wind field forecasting are integrally improved.

The invention belongs to the technical field of electric appliance insulation testing, and particularly relates to a switch cabinet insulation state evaluation method based on transient overvoltage signal analysis, which comprises the following steps: acquiring a transient overvoltage signal of a switch cabinet bus, and obtaining a first-order differential sequence and a phase space trajectory through differential filtering; by calculating the vertical deviation degree of a phase space trajectory motion vector relative to a reference vector, the trajectory deviation degree is obtained, and a divergence accumulation index is calculated so as to represent the geometric distortion of the waveform; estimating the oscillation frequency and the voltage amplitude of the signal, and obtaining the frequency amplitude synchronism based on the correlation of the variation of the oscillation frequency and the voltage amplitude so as to represent the nonlinear physical characteristics of the insulating medium; and the divergence accumulation index and the frequency amplitude synchronism are input into a preset two-dimensional characteristic plane, and the insulation state of the switch cabinet is output, so that the evaluation accuracy is effectively improved.