314 results about "Data slice" patented technology

The invention discloses an intelligent report generation method and system based on multi-source heterogeneous data fusion, and particularly relates to the technical field of data fusion, and the method comprises the steps: constructing a fact anchor point containing a stable identifier in a fusion layer, and solidifying an aperture version signature and a time interval; packaging a screening condition, a sorting rule, an access plan, time window limitation and the like into a controlled context packet and freezing the controlled context packet into a unique fact source; in the range, generating sentences by using a limited template and writing sentence-level reference marks, so that each sentence can be traced back to a data slice and an evidence fragment; performing consistency verification according to a source, an aperture and time, performing conservative degradation on the attribution expression according to a consistency threshold value, and outputting a degradation strategy description; and finally, page two-way pointer and cross-sentence consistency recording and audit playback are realized through an evidence chain pointer table and a published version signature, and the problems of caliber drift, cross-window access and difficulty in evidence tracing are solved.

The invention relates to a large model data protection method and device based on a trusted environment, equipment and a medium. In the scheme, after a server side obtains original large model data, the original large model data needs to be encrypted and signed, and then the original large model data and an encrypted information table are issued to a target application side; when the target application end calls the large model data, performing exception check according to a credible strategy; if the check is passed, performing signature verification on each encrypted data slice; if the signature verification succeeds, decrypting the encrypted slice data by using the encrypted information table to obtain original large model data; wherein the server side and the target application side are both in a trusted environment. It can be seen that the trusted environment and the key system are combined, and a dual protection mechanism for large model data is achieved; moreover, before the large model data is called, the operation environment of the target application end needs to be subjected to exception check, so that an attacker is prevented from acquiring or tampering the sensitive data, and all-around data protection is provided for the large model data.

The invention discloses a railway tunnel foreign matter detection method based on radar point cloud and RGB image fusion. The method comprises the following steps: acquiring a laser radar point cloud and an RGB image in a railway tunnel; obtaining an RGB image by using an adversarial network model GAN to judge the abnormal occurrence state of the foreign matter in the railway tunnel, and if the abnormal state of the foreign matter occurs, projecting the obtained laser radar point cloud to an image plane to obtain a dense depth map; processing through a sliding window to obtain data slices which are approximately square; and carrying out data feature extraction, fusion and foreign object target detection on the data slices of the RGB image and the dense depth map by using a fusion target detection network based on the OfficientNet-FPN, and outputting a foreign object target information detection result containing a target category, a position, a confidence coefficient and a distance. According to the railway tunnel foreign matter detection method based on radar point cloud and RGB image fusion, high precision, high efficiency and high reliability of railway tunnel foreign matter detection are realized.

The invention discloses an archive information processing method and system based on digital security, and relates to the technical field of computer application, and the method comprises the following steps: S1, constructing a time sequence consistency profile, and collecting the actual arrival time and the corresponding position index of each archive data fragment, a network jitter fingerprint model reflecting instantaneous delay fluctuation characteristics is generated based on multi-round transmission data; s2, in a data inbound process, injecting a unique position label for each archive data fragment, and enabling the position labels and fragment contents to jointly participate in Hash calculation to generate a content Hash and position Hash parallel dual verification baseline; according to the method, sequence verification and correction are realized by constructing a time sequence consistency profile and a dual Hash mechanism, the Hash collision risk is prevented in combination with disturbance verification, result write-back and model updating are introduced, a closed-loop control link is formed, and the safety, accuracy and traceability of file synchronization are remarkably improved.

The invention relates to the technical field of communication quality, and particularly discloses an acquisition terminal for communication quality monitoring, which comprises an electric energy meter interface module for reading an instrument value from an electric energy meter; the time synchronization module is used for aligning the terminal clock to the coordinated world time and generating a time frame with 1s as granularity; the communication quality monitoring module is used for measuring a communication quality index and calculating a value at the 95% position within a preset statistical time period as a statistical result preset statistical time period; the data slicing and counting module is used for forming 10T0 fragments according to 10 fundamental wave periods in each time frame; the caching and out-of-order correction module is used for performing grouping, out-of-order rearrangement and missing compensation on instrument values and communication quality indexes according to the same time frame before uploading and completing framing in a waiting window; and the alarm and uploading module is used for outputting a local alarm by taking the value of the 95% position as a statistical result. According to the invention, the overall end-to-end time delay of the acquisition terminal is reduced.

The invention discloses a high-speed rendering method and system for a three-dimensional large-scale dynamic net rack of a one-graph of a power grid. The method comprises the following steps: generating three-dimensional model parameters of a power transmission line according to tower account data; dynamically scheduling rendering of the three-dimensional model of the power transmission line according to the visible area range; slicing the power transformation multi-source heterogeneous three-dimensional data; optimizing a three-dimensional rendering pipeline of the power grid graph; and three-dimensional net rack high-speed rendering is carried out through visual cone removal, LOD and instantiation rendering technologies. According to the invention, based on the machine account data of the power transmission line, accurate rendering of the three-dimensional line on one graph of the power grid is realized through parametric modeling; for a multi-source heterogeneous transformer substation model, three-dimensional slices in a unified format are formed through a scene division technology, and support of a power grid graph on a three-dimensional slice rendering pipeline is expanded; in order to improve the rendering performance, dynamic rendering scheduling is carried out on the three-dimensional net rack in the viewport, repeated drawing of a large number of same devices is avoided, the performance overhead is reduced, and the performance bottleneck of loading the large-scale three-dimensional net rack is effectively solved.

The invention relates to an electric power multi-source data slicing processing method and system based on reinforcement learning, and the method comprises the following steps: S1, obtaining electric power multi-source data, and carrying out the preprocessing of the electric power multi-source data, and obtaining a structured original data stream; s2, obtaining a multi-dimensional feature vector through feature extraction; s3, adopting a clustering algorithm to identify a behavior pattern to obtain a clustering label, calculating the distance from a sample to each clustering center, and when the distance is obviously higher than the average value plus threshold value beta times standard deviation of the cluster, additionally marking as abnormal; s4, intelligent slicing decision making is carried out based on reinforcement learning, and a slicing strategy, data and labels are obtained; s5, according to the obtained slicing strategy, data and label, through knowledge graph service rule verification, it is ensured that domain constraints are met; and S6, sending the verified strategy to distributed slice storage and management, and carrying out data slice storage and index construction according to the strategy. According to the method, the structured governance level and the high-value utilization capability of the power data assets are effectively improved.

The invention provides a key frame extraction method for a multi-modal large model. The multi-modal large model video data processing efficiency and generation quality are improved through cooperative computing of a user side and a cloud side. According to the technical scheme, the method comprises the following steps: 1) after a client application submits data, an interface service performs primary processing on the data, and video frames and audio frames are divided and merged into data pieces according to a time sequence; 2) dynamically extracting n key video frames in each data piece of the user side, and compressing, packaging and transmitting the n key video frames and corresponding audio frames to the cloud side; 3) the cloud performs coding fusion on the multi-modal data and then inputs a large model generation result; and 4) the cloud evaluates the generation quality, and the user side adaptively adjusts data preprocessing and key frame extraction parameter configuration according to the generation quality and the response time delay. By reducing redundant data transmission and dynamically optimizing calculation and transmission resources, the response speed and the output quality of the multi-mode large-model cloud service are remarkably improved.

The invention provides a UCIe standard protocol layer data transmission circuit, a related method and a chip, and the circuit comprises a slicing circuit which is configured to segment a data packet from a service port; sending the slice frame and slice description information associated with the slice frame to a framing device; the framing device is configured to fill a flow control unit Flit with the slice frames from the slice devices and slice description information associated with the slice frames, and send the Flit to the deframing device; the frame deframer is configured to determine slice frames belonging to the same service port from the Flit, and send the slice frames and slice description information associated with the slice frames to a packing circuit corresponding to the service port; and the packing circuit is configured to splice the data slices belonging to the same data packet in the slice frame based on the slice description information and output the data packet. The technical problem that extra delay is introduced due to beat insertion due to the fact that large multiplexers need to be introduced in existing protocol layer data transmission of the UCIe standard is solved.

The invention discloses an RAG knowledge base construction method and device and a medium, and the method comprises the steps: carrying out the analysis of the format, structure and semantic features of input data through a multi-level priority judgment logic, and automatically recognizing the specific data type. And calling a corresponding adaptive segmentation strategy for processing according to the identified data type, and generating data slices with semantic coherence. And asynchronously calling the large language model to generate a keyword set for each data slice, and uniformly storing the data slices, the keyword sets and associated multi-modal resource links by adopting a mixed storage mode to construct an enhanced RAG knowledge base. In the retrieval stage, query keywords are generated for user query according to the same logic, and related results are accurately retrieved from the knowledge base through a keyword matching mechanism. According to the method, the semantic segmentation problem is effectively avoided through intelligent segmentation with adaptive types, multi-modal data access is supported, and unified management and efficient retrieval of heterogeneous resources are realized.

The invention discloses a method and device for automatic roasting and process intelligent control of a belt type roasting machine, and relates to the technical field of pellet roasting of the belt type roasting machine in the metallurgical iron and steel industry. The method comprises the four steps of historical data slice acquisition and modeling, target process parameter determination, feedback verification and online correction and rolling prediction control, a model historical library is established through service slices and time slices, modeling is performed based on a three-layer feed-forward neural network, flexible weighted deviation calculation combined with dynamic weight and rolling optimization of a softening coefficient are combined, and the rolling prediction control is performed. And accurate regulation and control of equipment parameters are realized. The automatic control system is mainly used for automatically controlling the pellet production process of the belt type roasting machine, the product quality stability is improved, the energy consumption is reduced, the labor intensity of workers is relieved, and support is provided for unmanned and intelligent production.

The invention belongs to the field of computer vision, relates to a rockfall monitoring method and system based on visual identification, and aims to solve the technical problems that an existing scheme is difficult to achieve good balance among reliability, accuracy and economy and is poor in application effect. The method comprises the following steps: acquiring video stream data and vibration waveform data of a target area; monitoring according to the vibration waveform data, and when the vibration energy exceeds a preset first threshold value, generating a primary vibration event trigger and extracting video data slices from the video stream data; inputting the video data slices into a target detection network to identify candidate moving targets in the video frames and extract kinematic features; performing spectral analysis on the vibration waveform data to determine a comprehensive confidence coefficient; and counting the number of candidate targets of which the comprehensive confidence exceeds a threshold value in a specific time window, and determining a current rockfall risk level. Based on the method, video detection and vibration detection are combined, and the actual application effect is better.

The invention discloses a quantum random encryption and distributed synchronization information transmission method and device, and is applied to the technical field of power systems. Acquiring initial data of the state of the power equipment; generating an unpredictable quantum true random number; encrypting the initial data by using the quantum true random number to obtain target data; synchronizing the clock synchronization precision of the plurality of nodes; generating data slices according to the target data and distributing the data slices to the plurality of nodes; and synchronizing the data slices of the plurality of nodes. According to the invention, the technical problem of safe and reliable information transmission of the power system is solved at least to a certain extent.

The invention discloses a six-dimensional sensor noise covariance matrix adjusting system based on impact response, and relates to the technical field of data processing, and the system comprises a base covariance construction module which is used for obtaining multi-sensor channel impact data, calculating a channel-level indication quantity, and fusing the channel-level indication quantity to obtain an impact score; determining an allowed adjustment channel set and an adjustable sub-block index, generating a diagonal priority mark, and obtaining a base covariance through weighted smoothing; the increment calculation verification module intercepts data, constructs a spatio-temporal data slice matrix, carries out shrinkage robustness estimation, generates candidate trial increment by combining channel weight and characteristic decomposition, and outputs unique trial increment after verification of an improved gate and a certificate margin gate; the adjustment fusion optimization module superposes the trial writing increment to the base covariance, after stability and consistency certificate inspection, self-adaptive adjustment is performed according to a three-section strategy, and a current noise covariance matrix is output through forgetting fusion; and the problems of numerical instability and distortion caused by redundancy adjustment in an impact environment are avoided.

Methods for a machine-learning network that provide efficient, scalable, and granular analyses during validation of a machine learning model are disclosed. Validation of models depends upon many factors, including the real-world application of the model, the type of model being trained, and the types of data samples it is being trained on. In order to provide relevant edge case information to users that pertains to their specific model, data slice finding techniques may be used to identify subsets of the dataset that are particularly problematic. By limiting a length of the slice description that the algorithm searches and by configuring the algorithm to target specific types of errors, users are provided with a more granular analysis that then allows them to determine how or if they need to retrain the model.

The invention relates to the field of capacitor monitoring, and provides a capacitor quasi-online state monitoring method and system for a wind power converter, and the method comprises the steps: extracting a data slice comprising DC bus voltage, capacitor branch current, housing temperature and output power from a data buffer area; furthermore, output power data is utilized, intelligent screening is carried out through a quasi-steady-state slice trigger, and only effective data with the power average value large enough and the fluctuation small enough are reserved. And for the screened effective data slices, extracting the DC bus voltage and the capacitance branch current, and accurately identifying the voltage, current amplitude and phase information of the main harmonic through fast Fourier transform. Based on the key harmonic data, the complex impedance of the capacitor is calculated, and then the original ESR and capacitance estimation value are calculated. And finally, in combination with the capacitor shell temperature extracted from the same data slice, performing temperature compensation on the original parameters, and outputting a final result capable of truly reflecting the aging state of the capacitor.

The invention provides a physical frame structure design method suitable for satellite-ground high-order high-bit-rate wireless communication, which comprises the following steps of: slicing service data according to a preset length, and generating a baseband frame based on the obtained data slices; performing scrambling processing on the baseband frame to obtain a scrambled baseband frame; sequentially executing preset error correction coding processing on the scrambled baseband frames to obtain coded frames; performing internal interleaving processing on the coding frame to obtain internal interleaving data, and forming a first physical frame; performing external interleaving processing on the first physical frame to obtain external interleaving data; mapping the outer interleaved data into a symbol stream; performing physical frame framing based on the symbol stream to obtain a second physical frame; and performing scrambling processing on the to-be-scrambled symbols except the physical frame header in the second physical frame to obtain a scrambled physical frame, and performing modulation based on the scrambled physical frame to perform data transmission, thereby improving phase noise resistance and channel distortion resistance, enhancing Doppler frequency shift resistance, and optimizing transmission efficiency and reliability.

The invention discloses a method and a system for detecting vertical floating of duct pieces in a shield tunnel splicing stage, and belongs to the field of tunnel engineering monitoring. The method comprises the following steps: acquiring a laser point cloud and an apparent image of the inner surface of a segment, and constructing a ring position data slice set indexed according to ring numbers; performing refined section fitting considering segment local deformation on the slice set, and extracting an actual measurement center of each ring through physical rigid body separation or statistical joint optimization; self-adaptive parameterization interpolation is executed based on the fitting sensitivity features, and an actual measurement tunnel center line is generated; constructing a self-calibration optimization objective function containing multi-source constraints, and performing incremental updating on external parameters and ring-by-ring attitudes of the sensor by using a sliding window; and establishing a multidirectional component unified coordinate system, decomposing spatial deviation and judging a segment floating mode. The problem of coupling of rigid body displacement and elastic deformation is solved, and high-precision and anti-drifting duct piece floating closed-loop detection is achieved.

The invention discloses a supercomputing optical module chip EMC multi-dimensional signal intelligent analysis and diagnosis system, and relates to the technical field of chip testing. A data acquisition module applies a test excitation signal to a chip and synchronously acquires a real-time multi-channel EMC data stream; the data slicing module splits the data stream into real-time data slices according to a preset time window, and extracts real-time feature vectors to generate an aggregation ball; the fault template library module stores a fault template containing a template slice sequence, and each template slice defines a template feature data interval corresponding to a real-time data slice space-time; the intelligent diagnosis module realizes parallel matching of the aggregation ball and the template slice through a routing scheduling unit, identifies a fault mode based on a matching result in a continuous time window and locates a specific physical channel; on the basis of real-time data slicing and a dynamic matching mechanism, real-time intelligent diagnosis and accurate fault positioning of the chip EMC test are realized, and the test efficiency and the diagnosis accuracy are remarkably improved.

The embodiment of the invention discloses a method for generating a video abstract, which comprises the following steps of: after slicing video data, respectively extracting a visual description text, a text abstract and an emotion label from each video clip, and fusing all modal extraction results to determine a comprehensive description text of each video clip; and then determining the logic relationship of each video clip by analyzing the semantic relevance of the comprehensive description text, integrating each video clip based on the time sequence and the logic relationship to obtain a text sequence, and finally generating a video abstract based on the text sequence. Through division processing of multi-modal data and later fusion, the processing efficiency and universality are improved while the accuracy and coherence of the abstract are ensured, and through cross-fragment semantic correlation analysis, a text sequence with smooth logic coherence and narration is constructed, and the accuracy of the generated video abstract is improved.

Methods for a machine learning network that provide efficient, scalable, and granular analyses during validation of an object detection model are disclosed. The system described herein is configured to use extraction of visual concepts to provide interpretable metadata to a data slice finding technique. The identified, poor-performing slices are then provided to a user for selection as to which slice or slices to focus on when preparing a subsequent training dataset that is to be used to further refine the object detection model. The system then coordinates with a large language model and with a vision and language foundational model to augment the original validation dataset with supplementary image samples that are determined to be associated with the problems currently causing poor performance of the model.

The invention relates to the technical field of data processing security, in particular to a retrieval enhancement generation method and device based on data grading and classification management and control, and the method comprises the following steps: determining the grade and category of each piece of local knowledge base data, and setting data permissions of all users according to the data grade and category information; carrying out blocking and vectorization processing on each file; according to the user identity, confirming the data grade and the data type that the user has access to; carrying out vectorization processing on the retrieval problem of the user; screening the data slices according to the data authority of the user, and performing vector similarity calculation on the user problem to obtain the data slices within a similarity threshold range; fusing the user question and the retrieved data slices into a cue word template to obtain cue words; answers are generated based on the large model and cue words, answers to questions are given, and grading and classification management and control can be carried out on data of a local knowledge base and question and answer applications based on the large model. The device is used for implementing the method.

The invention relates to a sleeping posture recognition and self-adaptive pillow height adjustment system and method and a storage medium, and the sleeping posture recognition and self-adaptive pillow height adjustment system comprises a data collection module, a sleeping posture recognition module, a pillow height adjustment module and a user interaction module; the user interaction module is used for acquiring personalized prior data and historical sleep comfort score of a user; the data acquisition module is used for acquiring a pressure data set of the pressure sensor array in real time; the sleeping posture recognition module is used for constructing a plurality of groups of time domain data slices for the pressure data set according to a preset time window, inputting the time domain data slices into a time domain feature model to obtain a plurality of groups of low-dimensional pressure data feature parameters, and further recognizing an effective sleeping posture state; and the pillow height adjusting module is used for calculating and calibrating the pillow height based on the personalized prior data of the user, the effective sleeping posture state and the historical sleeping comfort score, and adjusting the pillow surface. According to the method and the device, the problems of high computing power consumption and difficulty in meeting precise and personalized sleep requirements of the user in related technologies are solved.

Methods for a machine-learning network that provide efficient, scalable, and granular analyses during validation of a machine learning model are disclosed. Providing quantitative analysis information to machine learning experts when they are deciding how to proceed with further optimizing their given machine learning model allows for more directed procedures during edge case detection. Following the execution of a principal machine learning model using a validation dataset, a shallow learning model may be trained to provide simulations about how the principal model may be improved, or not, given different re-training scenarios. By using slice-based schemes, the validation dataset is divided into certain problematic data slices, and then, during inference of the shallow learning model, additional quantitative information about the effect on other data slices given a subsequent re-training of the principal model using a certain problematic data slice allows the ML expert to make more informed decisions.

The invention discloses a social media harmful information identification method and device based on large-small model collaborative optimization, and the method comprises the steps: firstly, carrying out the feature analysis of social media information through an LLM, and automatically matching an optimal professional small model; secondly, an automatic deployment engine is utilized, model deployment codes are generated through LLM, and localized deployment is completed; then, constructing a dynamic code generation unit, automatically generating a code and a fine tuning code for calling a professional small model to identify social media harmful information according to the model ID and deployment parameters, and performing grammar verification, performance evaluation and safety detection on the generated code by adopting LLM; further, a social media information pipeline is designed, and functions of data slice input, recognition result classification, challenging sample screening and fine tuning data set construction are included; and finally, implementing a model persistent evolution strategy, and completing model version upgrading by iteratively executing a recognition-screening-fine tuning process. Through a collaborative decision-making mechanism of the LLM and the professional small model, end-to-end automation of a harmful information identification process is realized, the manual intervention cost is remarkably reduced and the identification efficiency is improved on the premise of ensuring the identification quality, and the method is particularly suitable for a large-scale harmful information identification scene of multi-field heterogeneous data.

A data distribution server includes a first interface for communicating with a plurality of content providers and a second interface for comminating with a plurality of wireless transceivers. The data distribution server includes memory storing operational instructions and processing circuitry operably coupled to the first and second interfaces and to the memory, where the processing circuitry is configured to execute the operational instructions to receive a portion of a data object from a content provider of the plurality of content providers, generate a plurality of data slices from the portion of the data object, distribute the plurality of data slices between the plurality of wireless transceivers for temporary storage and facilitate transmittal of the plurality of data slices to a user device that is adapted to wirelessly communicate with each wireless transceiver of the plurality of wireless transceivers.

A computer is operable to store a set of encoded data slices of a data segment, where each encoded data slice includes a corresponding checksum of a plurality of checksums. Based on ones of the plurality of checksums corresponding to the set of encoded data slices, it is determined when the set of encoded data slices includes at least one corrupted encoded data slice. At least a threshold number of encoded data slices of the set of encoded data slices are retrieved, and a reconstructed data segment is generated using the at least the number of encoded data slices of the set of encoded data slices. At least one at least one reconstructed encoded data slice is formed utilizing the error correcting information dispersal algorithm on the reconstructed data segment, and the at least one corrupted encoded data slice is replaced with the at least one reconstructed encoded data slice.

The invention belongs to the technical field of computer storage, and particularly discloses an encryption difference compressed data restoration method and device based on similarity perception. According to the method and the device, the differential data block and the reference block which are subjected to aggregation storage are read from the continuous disk space at one time; in the data recovery process, pipelining processing is carried out on a difference decoding task and an erasure code decoding task; and performing data restoration according to the size of the data slice, the difference data block and the reference block by executing the pipelined difference decoding task and the pipelined erasure code decoding task. According to the mode, the whole aggregation storage data block containing the differential data block and the reference block is read from the continuous disk space, compared with respective reading, one-time extra disk overhead can be reduced, then the differential decoding task and the erasure code decoding task are subjected to pipelining processing, and the decoding efficiency is improved. And data recovery is carried out by executing the decoding task, so that the data recovery performance of the encryption difference compression system can be effectively improved.

The application discloses a kind of indexless quick slice transaction coding storage methods for alliance chain transaction data.The application realizes the indexless quick reading of transaction and the storage slice of transaction data by decoupling the verification function and data index function of transaction tree, improves the reading efficiency of transaction.For transaction verification function, the application proposes a verifiable data structure suitable for transaction data storage slice-indexless transaction tree, reduces the size of verification information, reduces the number of node interactions required for transaction data reading in data slice scenario.For transaction index function, the application proposes a new transaction data content coding method, realizes the indexless reading of transaction, improves the access efficiency of transaction, reduces the storage overhead of index.

The invention discloses a carbon emission factor calculation verification method based on a block chain, particularly relates to the technical field of carbon emission accounting and block chain credible measurement, and is used for solving the problems that an existing carbon emission factor calculation process depends on a dispersed system and manual summarization, the publication caliber is not uniform, a traceable evidence chain is lacked, and the calculation efficiency is low. And a supervision mechanism and a third party are difficult to carry out independent re-calculation and credible verification on the factor result submitted by the enterprise. A carbon emission factor calculation template is taken as a core at an enterprise node side, a device range, a statistical period and a data preprocessing rule are configured in a unified manner, abstract identifiers are generated step by step for dimensionalized data abstracts and calculation evidence chain fingerprints, calculation certificates are written into an alliance chain, and a calculation result is obtained. And the verification node obtains a verification fingerprint and registers a verification result based on the same template fingerprint and data slice recalculation, so that an integrated evidence closed loop of factor calculation, recalculation verification and factor selection is formed, and the problems that a factor source is non-transparent, cross-subject recalculation is difficult and a credible basis is lacked are solved.