95 results about "Data quantization" patented technology

The invention relates to the technical field of data processing compression and machine learning, and discloses a non-stationary data quantitative processing method based on dynamic volatility perception, which comprises the following steps of: continuously tracking real-time volatility statistical measurement of a photovoltaic data stream by adopting a Welford algorithm, and parallelly calculating a first-order differential sequence of the data stream to capture time dynamic characteristics; constructing a non-uniform quantizer with learnable parameters, setting a non-uniform input threshold by using a quantile initialization strategy, and mapping a non-uniform interval into a fixed and uniform integer output in combination with a non-uniform-to-uniform quantization mechanism; in order to guide learning of quantization parameters, a dynamic composite loss function based on volatility adaptive Beta divergence and time gradient consistency is constructed, the function comprises a Beta divergence distribution loss item dynamically controlled by real-time volatility, and the method can adapt to non-stationary characteristics of energy data such as photovoltaic power and the like. The robustness is remarkably improved while the high signal-to-noise ratio is guaranteed, and the method is suitable for edge computing scenes with limited resources.

The invention provides an information visualization method and system, a medium, equipment and a program product, and relates to the technical field of information processing, and the method comprises the steps: obtaining heterogeneous input information; performing category detection on the heterogeneous input information by utilizing the classification model, and determining a category corresponding to the heterogeneous input information; calculating entity relevancy, information relevancy and information weight corresponding to the heterogeneous input information according to the category; inputting structured label information corresponding to the heterogeneous input information according to the entity relevancy, the information relevancy and the information weight; constructing an atlas data structure corresponding to the structured label information; and mapping the map data structure into visual elements to obtain visual information. According to the method, heterogeneous input information is converted into visual information, so that leap-type improvement from bottom-layer data quantification to top-layer cognitive perception is realized.

The invention discloses a new energy station primary frequency modulation capability evaluation method considering a power grid structure and operation characteristics. The method comprises the following steps: collecting data; the data comprises complex network data of a power system, historical operation data of the system, virtual inertia control parameters of a wind generating set, droop control parameters of a photovoltaic generating set and primary frequency modulation control performance data; quantifying the primary frequency modulation comprehensive capability of the new energy station according to the collected data; and training the new energy station primary frequency modulation capability evaluation model and obtaining the new energy station primary frequency modulation capability evaluation model with the highest prediction accuracy. According to the method, the new energy station primary frequency modulation capability evaluation model is established based on the XGBoost algorithm, compared with a traditional time domain simulation method, when the method is applied to different new energy stations, modeling does not need to be conducted again, and the method has obvious advantages in the aspects of universality and calculation efficiency.

In a data storage device with a multi-level-cell memory, bits of data are stored in different levels of each of the memory cells. In some situations, a quantized version of the stored data (e.g., the most-significant bits) may be requested. Responding to such a request can involve reading all levels of the memory cells to retrieve a full version of the data and then selectively providing only the quantized version. To improve performance, the data is stored in the memory in an interleaved manner in which the quantized version of the data is stored in the same level(s) instead of being spread across all levels of the memory cells. That way, when the quantized version of the data is later requested, only the relevant level(s) are sensed, thereby avoiding the time and resources needed to read memory level(s) that do not store the quantized data.

The invention relates to a processing device and a quantization method for quantization and inverse quantization of numeric data. In one or more aspects, a processing apparatus for numeric data quantization includes processing circuitry to determine a maximum exponent from a set of exponents of a set of digit representations of a set of digits, obtain a set of scaled exponents based on the maximum exponent, and quantize the scaled exponents based on the set of scaled exponents. And one of (i) obtaining a set of quantized significant numbers based on the set of digit representations and a set of mantissas of the set of scaled exponents, or (ii) obtaining a set of quantized mantissas based on the set of mantissas. The processing circuitry is configured to output a set of quantized digit representations of the set of digits based on the set of quantized significant numbers or based on the set of quantized mantissas and the set of scaled exponents; and outputting an offset exponential scaling factor based on the maximum exponent.

A computing system for performing runtime data handling optimization for generative models is provided. The computing system comprises at least one processor and memory comprising a first memory and a second memory, wherein the memory stores instructions that, when executed by the at least one processor, cause the at least one processor to execute a generative model. The computing system computes a first value matrix entry based upon the processing of an input to the generative model. The first value matrix entry is stored in a first memory wherein a first group of value matrix entries is identified. The computing system executes data quantization on the first group of value matrix entries which results in a first quantized value matrix. The first quantized value matrix is added to a second memory where it can be used during generation of the generative model. When the first group of value matrix entries is less than a group size parameter, data padding matrix values are generated and used during execution of the data quantization.

This invention discloses a big data-based method for maximizing efficiency management in a hardware processing workshop, relating to the field of big data management technology. The method includes: acquiring historical video data of hardware workpiece processing tasks; quantifying the skill levels of processing personnel in the hardware processing workshop; clustering and classifying the processing personnel according to their skill levels; initializing the time distribution of hardware workpiece processing actions under each skill level; constructing a hardware workpiece processing action-effective / ineffective time identification model; marking the types of ineffective time-consuming tasks for each worker's skill level; performing causal tracing for these ineffective time-consuming task types; constructing a causal list of ineffective time-consuming tasks; and generating a sequence of causal improvement tasks for ineffective time-consuming tasks. The beneficial effects of this invention are: improving the accuracy of ineffective time analysis and improvement, and maximizing the efficiency of the hardware processing workshop.

A data quantization method and device, a computer readable storage medium and a chip, the data quantization method comprising: obtaining a data block to be quantized, dividing the data block to be quantized into one or more data groups to be quantized according to a preset grouping method; performing quantization calculation on each data group to be quantized according to a target quantization format, to obtain quantization data corresponding to data in each data group to be quantized and quantization metadata corresponding to each data group to be quantized, wherein the quantization metadata is used for dequantization of the quantization data; and storing the quantization metadata corresponding to the data block to be quantized and the quantization data as a continuous data structure.

The application discloses a kind of identification methods of the influence factor of aero-engine power control device parts working process, it is related to aero-engine maintenance technical field, by the risk level and repair means effectiveness level of data quantization baseline unit are matched diagnosis, to judge whether the current maintenance means can be matched with the risk level of parts, effectively improve the control precision of influence factor in repair process, eliminate internal and external field product quality hidden danger, comprehensively improve the quality prevention ability, process control ability and quality control ability in product repair process, it plays a guiding role to the systematic method of process development and optimization.

This application discloses a wearable EEG signal processing method and device based on feature fusion. The method includes: synchronously latching the main channel EEG signal from bioelectric electrodes and multimodal auxiliary signals from various auxiliary sensor interfaces within a preset period; inputting the multimodal auxiliary signals into a pre-trained interference identification model, outputting a structured interference quantization vector, which includes the confidence and intensity of different interference sources; processing the main channel EEG signal according to the confidence and intensity of different interference sources and a preset parameterized anti-interference algorithm library to obtain the final EEG data; quantifying the quality confidence of the final EEG data, storing the ternary mapping relationship between the current period, quality confidence, and the final EEG data, and sending it to a downstream application. Using this application, adaptive matching of the processing strategy with the real-time interference scenario is achieved, effectively avoiding overprocessing or underprocessing, and greatly improving the data robustness and accuracy of downstream applications.

This invention provides a geological disaster monitoring and early warning method and system based on BeiDou signals, belonging to the field of data monitoring technology. It includes: extracting spectral, texture, and shape features from monitored real-time images for analysis to identify abnormal areas and corresponding geological disaster types in the monitored region; analyzing surface deformation data of the identified abnormal areas based on BeiDou satellite data to obtain deformation data features of the abnormal areas, assessing the degree of deformation, and obtaining a deformation degree assessment value; determining the geological disaster index of the monitored region based on the deformation degree assessment value and the geological disaster type, and generating a corresponding early warning strategy. This invention achieves early automatic delineation of potential hazards by analyzing image features from BeiDou satellite imagery, and combines BeiDou deformation data to quantitatively analyze the activity status of disaster bodies, generating a tiered early warning strategy, thus improving the spatiotemporal coverage, accuracy, and timeliness of geological disaster monitoring and early warning.

The invention discloses a real-time myocardial infarction risk prediction data processing system based on edge calculation, and relates to the technical field of medical data processing. The system comprises a prediction data acquisition-quantification module, a risk prediction data intelligent optimization module, a drift rate data acquisition-quantification module and a drift correction intelligent optimization module. According to the method, the accuracy of myocardial infarction risk prediction data is obtained through quantification of obtained risk prediction data, whether intelligent optimization of the myocardial infarction risk prediction data is executed or not is judged, dynamic drift rate data in an edge-cloud collaborative management link is obtained through quantification of dynamic drift rate data to obtain the dynamic drift rate of an edge model, and whether intelligent optimization of dynamic drift correction is executed or not is judged. The accuracy of real-time myocardial infarction risk prediction data processing is improved, and the problem that in the prior art, the accuracy of real-time myocardial infarction risk prediction data processing is low is solved.

The invention discloses a defect identification method and system for a substation door type structural member, and belongs to the technical field of substation operation and maintenance. According to the method, feature recognition is performed on the image sequence and the switching angle sequence through the large electric power model to obtain the structural feature matrix and the behavior feature matrix, so that the problem that a single data model cannot effectively detect dynamic defects is avoided; determining structural behavior consistency data through the structural feature matrix and the behavior feature matrix, and quantifying the dynamic consistency degree of the substation door type structural member between the structural features and the behavior features by the structural behavior consistency data, so as to fuse the structural feature matrix and the behavior feature matrix; obtaining a transformer box door type structural member defect identification model suitable for the transformer box door type structural member, and obtaining a model defect identification result; by aggregating the model defect identification results, the defect identification result of the substation door type structural member is obtained through multi-source scores, and the defect identification accuracy of the substation door type structural member is improved.

The application provides a parallel world simulation resource dynamic scheduling optimization method and system. In the method, the dependence relationship between tasks is detected in the parallel world simulation resource to obtain topology data, and the resource competition intensity and the sequence constraint are quantified to form an association feature matrix; a cross-dimension optical signal transmission path is constructed by analyzing the dynamic change of the matrix through a programmable photon array, and a low-conflict resource transmission channel atlas is formed; a global task execution sequence is generated by synchronously processing the matrix and the channel atlas; a resource occupation relationship atlas is generated by performing a multi-world resource occupation operation; the cross-dimension resource conflict probability in the atlas is calculated, and the task execution sequence is dynamically adjusted to avoid the simulation deadlock risk. The application realizes the nanosecond-level dynamic resolution of the cross-dimension resource conflict, the simulation deadlock rate is zeroed, and the task throughput is improved.

A graphics processor includes circuitry configured to generate a bounding volume hierarchy (BVH) data structure for a graphics scene to be rendered. The BVH includes a leaf node that includes: node data that specifies quantization frame data for quantized primitive children of the leaf node, a quantized primitive block that includes quantized vertex components of the primitive children, and a vertex residual block that includes additional vertex data for the primitive children. The vertex residual block may not include redundant data with the quantized primitive block and the quantized primitive block and vertex residual block may include, for the primitive children, sufficient data to reconstruct an original-precision representation of their vertices. The circuitry may traverse the BVH to perform ray tracing operations during render of the graphics scene, including to perform one or more ray primitive intersect tests based on the quantized primitive block and the vertex residual block.

The invention discloses a construction progress and equipment cluster collaborative virtual display system based on digital twinning, relates to the technical field of digital twinning, and solves the problems of data packet loss, error sequence and the like easily caused by blind calculation power distribution during protocol conversion. An efficient processing rhythm of complete coverage and less repetition is realized, display lag caused by data backlog is avoided, resource waste caused by invalid processing is reduced, the real-time synchronization capability of the digital twin model is ensured, the acquisition frequency and time of each monitoring node are determined, the normalization and timeliness of data acquisition are ensured, and the real-time performance of the digital twin model is improved. Data source errors caused by fuzzy acquisition logic are avoided, high-quality and traceable original data support is provided for subsequent processing and display, protocol difference node segments are accurately positioned, and dynamic pre-distribution of computing power is realized by combining historical data quantization conversion characteristics and required computing power.

In a data storage device with a multi-level-cell memory, bits of data are stored in different levels of each of the memory cells. In some situations, a quantized version of the stored data (e.g., the most-significant bits) may be requested. Responding to such a request can involve reading all levels of the memory cells to retrieve a full version of the data and then selectively providing only the quantized version. To improve performance, the data is stored in the memory in an interleaved manner in which the quantized version of the data is stored in the same level(s) instead of being spread across all levels of the memory cells. That way, when the quantized version of the data is later requested, only the relevant level(s) are sensed, thereby avoiding the time and resources needed to read memory level(s) that do not store the quantized data.

The present application relates to the technical field of wisdom teaching, disclose a kind of university VR teacher training simulation teaching data quantification evaluation method and system, the method includes: the teaching behavior data of teacher is carried out multimodal feature fusion, obtains teaching behavior feature;Teaching behavior feature is carried out semantic network reconfiguration, obtains teaching behavior feature atlas;According to the node attribute of teaching behavior feature atlas, teaching behavior feature atlas is carried out feature dimension mapping, obtains teaching ability quantitative index;The dynamic evolution law of teaching ability quantitative index is analyzed, obtains the change trend of teaching ability quantitative index;Teaching ability quantitative index is carried out stage trajectory fitting, obtains teaching ability development trajectory;Integrate teaching ability quantitative index and teaching ability development trajectory, generate comprehensive quantitative evaluation index;The present application can improve the efficiency of a kind of university VR teacher training simulation teaching data quantification evaluation.

The application discloses a kind of based on AI big model generation endowment evaluation scheme method and system, it is related to artificial intelligence and endowment service technical cross field, the specific steps of this method are as follows: by preset sensing device every 5 minutes collection old unstructured microfeature data and encryption upload;Call algorithm and carry out dimension transformation;Read cross-modal knowledge graph data quantization evaluation;Matching resource and calculating accessibility index screening target resource;AI big model integrates information and generates scheme containing multiple data and pushes to preset terminal.The application can capture the core physiological and behavioral data of the old people in real time without interference, and the full-link closed-loop protection ensures that privacy is not disclosed;Relying on cross-modal knowledge graph, multi-source data is efficiently linked, and the state of the old people is dynamically quantized and evaluated to adapt to changes;By calculating the accessibility index to screen practical resources, terminal hierarchical authorization and double authentication, it not only meets the information needs of each subject, but also ensures the safe and standardized circulation of the evaluation scheme.

A chip system, a method of processing data, and a computer readable storage medium are provided. The chip system comprises: a memory for storing first data to be input to a current layer of a neural network and weight data of the current layer; the quantization circuit is used for quantizing the first data by adopting a plurality of quantization modes, and the plurality of quantization modes are used for quantizing the first data into second data with a plurality of data types; the operation circuit is used for determining third data according to the second data and the weight data of the current layer; and the inverse quantization circuit is used for performing inverse quantization on the third data and determining fourth data output by the current layer. According to the scheme, the data of the current layer is subjected to hybrid quantization by adopting multiple quantization modes, so that the quantization error can be reduced, and the operation performance of a chip system is improved.

The invention discloses a data quantification method, which comprises the steps of operating an artificial intelligence AI model, acquiring data generated in the operation process of the AI model, the data comprising a plurality of groups, determining the abnormal degree of the data in the first group in the plurality of groups, the abnormal degree indicating the difference degree of the size of the data in the first group, and determining the difference degree of the size of the data in the second group in the plurality of groups. And determining parameters for performing exception suppression on the data in the first group according to the exception degree, suppressing the data in the first group according to the parameters for exception suppression, and quantifying the suppressed data in the first group. According to the method, the abnormal degrees of different groups are determined respectively, and the appropriate abnormal suppression parameters are selected for the different groups based on the abnormal degrees to perform abnormal value suppression, so that adaptive suppression of data in the different groups is realized, the situation of insufficient suppression or excessive suppression can be avoided, and the quantization precision of the AI model is improved.

The first processing unit performs a first quantization operation on X first data to obtain X second data, each second data comprising M bits, and the second processing unit supports processing of data with M bits; the first processing unit performs a second quantization operation on the X second data to obtain X third data, each third data comprising N bits, X being a positive integer greater than or equal to 2, M and N being any positive integers and N being less than M; the first processing unit converts the X third data into N fourth data, each fourth data comprising X bits, and the i-th fourth data in the N fourth data being composed of X i-th bits of data in the X third data; the second processing unit acquires the N fourth data, restores the N fourth data into the X second data, and processes the X second data. The flexible data quantization can be realized, and the storage space and transmission bandwidth of data can be saved.

The invention relates to the technical field of engineering control, in particular to a bucket load weight calculation method and system applied to an excavator, and the method comprises the steps: obtaining the hydraulic thrust data of the excavator at the current moment, and carrying out the geometric calculation of the hydraulic thrust data according to a preset device dynamics model, and obtaining the target torque data; the hydraulic thrust data is obtained by collecting data of a movable arm hydraulic cylinder, a bucket rod hydraulic cylinder and a bucket hydraulic cylinder of the excavator; determining theoretical torque data of the excavator at the current moment according to the device kinetic model; and calculating the load weight of the bucket according to the target torque data and the theoretical torque data to obtain the target load weight. The influence of the bucket load on the torque is quantified according to the actually measured target torque data and the theoretical torque data, so that the bucket load weight of the excavator is calculated, manual estimation does not need to be performed depending on personal experience, and the accuracy of calculation of the bucket load weight of the excavator is remarkably improved.

The present application relates to the technical field of vehicle network communication and safety management, in particular to a whole vehicle multi-channel CAN bus data collaborative processing method and system, comprising: collecting the multi-channel controller area network bus data in the whole vehicle in real time, and parsing the bus data into bus atlas data; quantitatively calculating the system determinism entropy; combining the current system determinism entropy and the bus atlas data, predicting the failure probability in the preset future time window through the pre-trained graph neural network model; in response to the failure probability being greater than the preset failure probability action threshold, locating the abnormal node with the largest contribution to the failure probability through the graph attention mechanism of the graph neural network model; based on the abnormal node, performing forced silence processing on the message frame from the abnormal node through the central gateway controller to reduce the system determinism entropy; the present application can accurately perceive the instantaneous threat to the safety instruction determinism, and solves the problem of insufficient sensitivity of traditional indicators.

The invention relates to the technical field of vehicle network communication and safety management, in particular to a vehicle multi-path CAN bus data cooperative processing method and system, and the method comprises the steps: carrying out the real-time collection of multi-path CAN bus data in a vehicle, and analyzing the bus data into bus atlas data; performing quantitative calculation to obtain system certainty entropy; predicting a failure probability in a preset future time window through a pre-trained graph neural network model in combination with the deterministic entropy of the current system and the bus graph data; in response to the failure probability greater than a preset failure probability action threshold value, positioning an abnormal node which contributes to the failure probability maximally through a graph attention mechanism of a graph neural network model; based on the abnormal node, performing forced silence processing on a message frame from the abnormal node through the central gateway controller so as to reduce the deterministic entropy of the system; instantaneous threats to security instruction certainty can be accurately sensed, and the problem that traditional indexes are insufficient in sensitivity is solved.

Some examples described herein provide for an encrypted data quantization apparatus and method, for example a memory controller to quantize encrypted data using a cache. One or more embodiments includes obtaining a first set of plaintext data bits to be stored in a memory device using an encryption scheme. A memory address for encrypted data bits to be stored in the memory device is identified for a first subset of plaintext data bits. A second set of plaintext data bits associated with the memory address is obtained from a cache, if present. The second set of plaintext data bits are modified according to the first set of plaintext data bits to be stored in the memory device to generate a third set of plaintext data bits that are then encoded according to the encryption scheme for storage in the memory device.

Provided in the present disclosure are a model data quantization method and apparatus, and an electronic device and a readable storage medium. The model data quantization method comprises: acquiring a first data matrix corresponding to model data; on the basis of a preset constraint condition, generating a scaling matrix corresponding to the first data matrix, and on the basis of the scaling matrix, performing scaling processing on the first data matrix, so as to obtain a second data matrix, wherein the preset constraint condition is used for constraining a distribution feature of matrix element values in the second data matrix; and on the basis of the second data matrix, obtaining a quantization result of the model data. The method helps reduce the data volume of model data, and reduce the occupancy of the transmission bandwidth and the consumption of computing resources, thereby improving the processing efficiency of a model.

The invention discloses a partition-oriented climbing auxiliary service demand evaluation method and system. The method comprises the following steps: calculating a prediction deviation; an uncertainty interval is constructed; determining a contribution coefficient; evaluating the theoretical contribution amount; constructing a power flow constraint; optimizing local requirements; the power increment uncertainty is quantified based on historical data, the power flow safety constraint is built in combination with the partition sensitivity, a climbing auxiliary service clearing optimization model considering the power transmission section safety limit is formed, and accurate evaluation and distribution of partition climbing requirements are achieved on the premise that the power grid safety is ensured. According to the method, the operability and economy of resource allocation are improved, the problems of inaccurate partition climbing demand assessment and unreasonable pricing caused by difficulty in estimating the support capability of the interval tie line in the prior art are solved, and the reliability and economy of high-proportion new energy power grid operation and the pricing rationality of the climbing auxiliary service market are remarkably improved.

At least one embodiment of the present disclosure provides a lookup table based data quantization method, device, processor and non-transitory computer readable storage medium. The method comprises: receiving data to be quantized, wherein the data to be quantized is of a first precision type; obtaining an index of the data to be quantized for a lookup table based on a first exponent part and a first mantissa part of the data to be quantized, wherein the index comprises a high bit part of the first mantissa part of the data to be quantized; performing a lookup in the lookup table based on the index, and obtaining a quantization result of a second precision type for the data to be quantized according to the lookup result, wherein the second precision type has lower precision than the first precision type. At least one embodiment of the present disclosure saves hardware storage space by using a smaller number of index structures, improves lookup table efficiency, and efficiently performs quantization. Furthermore, the bit width of the lookup table index can be adjusted according to the precision requirement, further reducing the error of the quantization result and maintaining the accuracy of the quantization result.

The present invention proposes a system that quantifies material levels from EEG data using non-linear equations, based on the same essential features previously disclosed, using alternative methods. These essential features consist of a predetermined frequency band, a merge ratio and a regression formula: (formula (I)) wherein Y is a material level quantification based on EEG.