25 results about "Streaming computation" patented technology

This invention discloses a fully embedded linear power flow calculation method and apparatus based on adaptive piecewise segmentation. The method includes: processing the system state... S max , S min The power flow solution at point 0 is used as the Germ solution to construct the FFHE-LPF model, update the correction coefficients, and calculate the maximum power mismatch value. When the error first exceeds 0.5%, the model is then used to calculate the maximum power mismatch value. max Record the current system status S 1 and S 2; Change the current system status S 1. S The power flow solutions at points 2 are used as Germ solutions to construct new FFHE-LPF models. The upper and lower bounds of the convergence range of these models are calculated, and the corresponding segment lengths are also calculated. l b1 and l b2 Take the segment length l c =min{ l b1 , l b2} as the basis for partitioning; based on segment length l c Divide the fluctuation range of new energy output into equal parts G The system constructs FFHE-LPF models for different sub-intervals, determines the sub-interval to which a power source belongs based on its current output, and updates the correction coefficients to obtain the power flow solution. This solution is then used to identify line overload or voltage exceedance risks in the power system. The device includes a processor and a memory.

This invention discloses a cluster resource scheduling method, system, terminal, and storage medium. Based on mainstream AI computing platforms, it innovatively designs a resource manager, task lifecycle manager, and scheduler. When the job to be processed is an elastic training job, the allocated computing resources for the elastic training job are determined by the predicted idle resources of the cluster within a first preset time period and the required computing resources of the elastic training job. Based on the real-time idle resources of the cluster and the allocated computing resources, a job node is allocated to the job, and the job node is invoked to execute the elastic training job within a second preset time period. Based on the predicted idle resources of the cluster in the next first preset time period, the allocated computing resources for the elastic training job are adjusted. Based on the real-time idle resources of the cluster and the adjusted allocated computing resources, a job node is reallocated to the job, and the job node is invoked to continue executing the elastic training job in the next second preset time period, effectively improving cluster utilization.

The application discloses a heat release tracing method for new energy station debugging quality problems and a related device, comprising: preprocessing environmental data and equipment data of the new energy station to obtain preprocessed data; constructing a standardized event stream according to the preprocessed data; calculating the dynamic failure rate of each basic event in the standardized event stream; based on the dynamic failure rate of each basic event in the standardized event stream, hierarchically combining each basic event according to system function relationship and logical dependence to construct a dynamic fault tree model of the new energy station; based on the dynamic fault tree model of the new energy station, through a time progression logic reasoning algorithm, combining the dynamic failure rate of each basic event, calculating the delayed probability contribution degree of each basic event, identifying the key root cause event and the propagation path, and the method and the related device can improve the tracing accuracy and response efficiency of the new energy station debugging quality problems.

A computing device may receive a stream of content (e.g., a live stream of content). The computing device may determine a plurality of representations for the stream of content. The computing device may determine a primary encoder associated with each of the plurality of representations. Determining the primary encoder may be based on the plurality of representations. The computing device may determine viewer data associated with the stream of content. The computing device may determine whether to encode a backup representation for at least one of the plurality of representations. Determining whether to encode the backup representation may be based on the viewer data associated with the stream of content. The computing device may cause at least one backup encoder to encode the backup representation for the at least one of the plurality of representations.

The application discloses a kind of real-time financial transaction anti-fraud data processing method, equipment and medium, it is related to data processing technical field.The method includes: providing visual rule configuration interface, visual rule configuration interface includes transaction source node and alarm node, and the processing node of corresponding fraud detection logic;In response to user operation, in visual rule configuration interface, by dragging, connecting transaction source node, processing node and alarm node, obtain real-time anti-fraud rule data stream;Real-time anti-fraud rule data stream is converted into structured query language SQL statement, and the detection logic of processing node is integrated to SELECT query projection list;SQL statement is submitted to stream computing engine cluster, to carry out real-time fraud detection to the transaction data of inflow.The application realizes the detection logic of each processing node by the above method Integrated into SELECT query projection list, while maintaining the ease of use of low-code development, ensure the optimization of processing efficiency.

The application discloses a section transmission power probability prediction method and device, equipment and storage medium, relates to power prediction technical field, and includes;Input power grid topology, take the bus in the power grid topology as a node, take a line as an edge, convert the power grid topology into a power system graph;The power system graph is input into a graph neural network, and a node embedding vector is obtained;Generate an initial value based on the power system graph, perform power system load flow calculation on the initial value to construct a training data set;Wherein, the initial value includes load demand and unit output data, and the training data set includes the initial value and corresponding section transmission power;The node embedding vector is input into a space-time feature extraction model to obtain space-time feature information;The space-time feature information and the training data set are input into a hybrid density network to obtain a section transmission power probability distribution. The application can improve the accuracy and efficiency of power probability prediction.

The present application relates to the technical field of power system analysis and calculation, and particularly relates to a kind of parallel computing method for probability power flow suitable for dynamic distribution network, including generating multiple scene sample matrix with correlation, task division is carried out to multiple scenes, using CPU-GPU heterogeneous parallel architecture to carry out segmented linearization power flow calculation and statistical analysis to mass scene.The present application effectively solves the dimension disaster problem of distribution network probability power flow calculation containing strong correlation random source and load, improves the calculation efficiency and scale processing capacity of related probability power flow calculation.

This invention discloses a meshless numerical simulation method for wellbore-reservoir coupling based on EFVM, belonging to the field of reservoir numerical simulation technology, including: S1, constructing the reservoir computational domain and discretizing the point cloud; S2, initializing simulation parameters; S3, EFVM reservoir seepage simulation; S4, Beggs-Brill wellbore multiphase flow calculation; S5, coupling iterative convergence and relaxation update; S6, stage advancement and simulation termination. The meshless numerical simulation method for wellbore-reservoir coupling based on EFVM provided by this invention solves the problems of grid dependence, low coupling accuracy, and poor convergence in existing technologies.

The embodiment of the present application provides a power flow calculation method and system based on an unsupervised physical information graph neural network, and relates to the technical field of power system power flow calculation. The method comprises the following steps: obtaining power system power flow operation data; converting the power flow operation data into an undirected graph structure to obtain graph data; preprocessing the graph data to obtain standardized graph data; constructing an operation model for calculating power system power flow values, and pre-training the operation model; inputting the standardized graph data into the pre-trained operation model to obtain the power flow values of the power grid system. The present application solves the problems of low calculation efficiency of traditional numerical calculation methods and the inability of traditional machine learning methods to fully utilize the power grid topology structure and rely on a large amount of labeled data, and realizes fast and accurate power system power flow calculation by fully utilizing the topology structure information, branch attributes and physical mechanism in the power grid.

A system is described that includes a computing device having one or more processors and memory. The computing device is configured to receive a first stream of first data. The computing device is further configured to perform an analysis of the first stream of first data at a first time and a second time different from the first time, and determine whether the first stream of first data changes between the first time and the second time. Additionally, the computing device is configured to adjust at least one adjustment parameter of an audio adjustment algorithm of an audio processing system if the first stream of first data changes between the first time and the second time. The audio processing system is configured to process an audio input through the audio adjustment algorithm to generate an audio output.

The application discloses a short-circuit current fast calculation and inhibition method and system based on day-ahead production simulation, and the method comprises the following steps: based on load prediction data, power generation prediction data and various production plan curves, a scenario set covering multiple time sections is generated by using a probabilistic method, and power flow distribution is calculated as a multi-scenario production simulation result; through a preset data mapping and conversion rule library, the production simulation result is converted into simulation data files suitable for electromechanical transient simulation software; by using a parallel computing framework, a short-circuit current calculation module of the electromechanical transient simulation software is called in batches, the maximum short-circuit current of each bus and the corresponding time section and scenario are identified; a suppression measure knowledge base is constructed, a candidate suppression measure combination is generated, a multi-objective optimization algorithm is used to optimize, screen and simulate and check the candidate suppression measures under the condition of safety and stability constraints, and an optimal suppression measure scheme is output. The application couples production simulation and transient simulation, and realizes intelligent decision-making.

This invention discloses a method and system for calculating harmonic power flow in distribution networks based on generative adversarial networks (GANs) and deep reinforcement learning, belonging to the field of power system analysis and control technology. The method includes: collecting real-time operating data of the distribution network and preprocessing it; inputting the preprocessed data into a GAN, generating high-precision harmonic power flow data by minimizing the total loss function of fusion adversarial loss and physical constraint loss; constructing a virtual training environment using the generated data, and using deep reinforcement learning to enable the agent to learn the optimal iterative update strategy; finally, performing harmonic power flow calculation based on the learned strategy, and performing physical consistency correction based on Kirchhoff's laws, power balance, voltage limits, and impedance positive definiteness, outputting power flow results that meet engineering requirements. This invention significantly improves the computational efficiency, accuracy, and adaptability of harmonic power flow calculation while ensuring physical reliability, and is suitable for real-time analysis and optimization control of dynamic distribution networks.

An acceleration system for stream computing engines in single-machine multi-core architectures includes a monitoring thread unit and a worker thread unit. The monitoring thread unit performs intra-Socket dispatch processing based on cross-Socket task dispatch information. The worker thread unit performs task execution processing based on specified dispatch tasks and intra-Socket dispatch task information, obtaining worker thread load results, and performs specified dispatch and cross-Socket dispatch processing based on Socket load information, obtaining task dispatch results. This invention, while retaining the scalability of a general scheduling strategy, deeply considers NUMA architecture, data dependencies between operators, and the characteristics of multi-core processors. By utilizing operator execution paths and data distribution information to dynamically adjust tasks, it reduces unnecessary thread switching, message passing, and state transition overhead. This effectively improves system throughput and reduces latency when facing real-time data streams with dynamically changing loads, ultimately achieving cost reduction and efficiency improvement in high-performance computing scenarios.

The application relates to a power system flow calculation method and device based on trust domain model switching, wherein the method comprises the following steps: a unified mathematical model of power system flow is established, a complex alternating current flow problem is converted into a standard nonlinear optimization problem; a numerical solver based on a trust region framework is constructed to ensure that an iteration process is carried out in a reliable region; dynamic model selection and parameter adaptive adjustment are realized, a system real-time state optimization calculation strategy is adopted, and therefore the convergence and robustness of the flow calculation are significantly improved. Thus, the problems in the prior art, such as single model, insufficient adaptability, limited convergence reliability, lack of numerical stability, lack of perfect adaptive mechanism and the like when a sick state power system is processed, are solved, and the reliable application of the flow calculation method in the sick state power system is seriously restricted.

The application discloses a compression and decompression method and system of DEMURA compensation values. The compression method comprises the following steps: dividing the DEMURA compensation values; calculating the predicted values and the residual errors of each sub-data; calculating the maximum continuous coded residual error number and the maximum continuous uncoded residual error number of a data block; calculating the residual error coding table of the data block; and writing the sub-data coding, the residual error coding table, the maximum continuous coded residual error number and the maximum continuous uncoded residual error number of the first M sub-data of the data block into the compressed binary bit stream. The decompression method comprises the following steps: analyzing the binary bit stream; calculating the predicted values of each sub-data; calculating the residual error decoding table of the data block; and reconstructing the data block. The technical scheme of the application fully utilizes the correlation of the data in the spatial domain by means of the sliding window prediction, residual error coding and residual error decoding of part of the data, so that the compression rate is improved and the accuracy of the reconstructed values is further improved.

The application discloses a kind of setting methods, devices, media and equipment of relay protection setting value, belong to setting value setting field, and the application is obtained by acquiring the multi-source data of main grid, distribution network and user side, and is preprocessed and topologically dynamic mapping using digital twin modeling model, and constructs the digital twin that reflects the operation state of three-level distribution network.Based on this, combined with the short-circuit current calculation model of multiple types of distributed energy, the total short-circuit current of the region is calculated, and the relay protection setting rule parameters are dynamically adjusted through the reinforcement learning rule optimization model. With the main grid setting value as the constraint, the distribution network and user side setting value are matched, and the preliminary setting value scheme is generated. The scheme is checked using the sensitivity dynamic checking model, and finally a reliable distribution network relay protection setting value is generated through the blockchain collaboration model, ensuring the reliability and credibility of the setting value scheme. The application effectively solves the problem that the existing technology cannot accurately and efficiently set the relay protection setting value of the distribution network.

The application provides a stream management method and a computing device for writing data to a multi-stream solid state disk. The stream management method is based on a user state and includes: receiving a write file request for writing a file; distributing the write file to at least one virtual stream according to an adaptive distribution strategy, each virtual stream including at least one write data block request of a data block, and the at least one write data block request of the data block being assigned with a same virtual stream ID; calculating the heat of each virtual stream; and mapping each virtual stream to a physical stream based on the calculated heat. According to the technical solution of the embodiments, the performance of the multi-stream solid state disk can be improved through the adaptive distribution strategy and the virtual stream mapping.

This invention relates to the field of big data-driven software development management technology, and discloses a big data-based software development management method, including the following steps: S1, big data collection and preprocessing based on a distributed architecture; S2, requirements analysis and decomposition based on the GBDT machine learning model; S3, project planning and resource allocation based on intelligent scheduling algorithms; S4, development process control and quality inspection based on Flink real-time stream computing; S5, risk warning and handling based on the DNN deep learning model; S6, project acceptance and data review based on data mining algorithms. By adopting a distributed architecture-driven full-process data collection, standardized preprocessing, and dedicated storage management solution, this invention solves the problems of data disorder, poor reusability, and insufficient support, which lead to a lack of scientific data basis in subsequent requirements analysis, project planning, and other stages.

The application relates to a distributed computing engine based on unified search and big data processing, which comprises a unified interface module, a distributed execution module and a storage management module; the unified interface module is used for receiving a computing task and performing task analysis, so that a corresponding computing engine is started according to a task type obtained through task analysis; the distributed execution module comprises a search computing engine, a graph computing engine and a stream computing engine, which are respectively used for reading and executing corresponding computing tasks; and the storage management module is used for uniformly performing data table and storage management on data corresponding to the search computing engine, the stream computing engine and the graph computing engine. The application solves the problem that the data computing engine in the prior art cannot provide parallel and efficient real-time computing for multiple types of data processing tasks to perform real-time and rapid response when facing multiple complex computing scenes.

The application discloses a kind of setting methods, devices, media and equipment of relay protection setting value, belong to setting value setting field, and the application is obtained by acquiring the multi-source data of main grid, distribution network and user side, and is preprocessed and topologically dynamic mapping using digital twin modeling model, and constructs the digital twin that reflects the operation state of three-level distribution network.Based on this, combined with the short-circuit current calculation model of multiple types of distributed energy, the total short-circuit current of the region is calculated, and the relay protection setting rule parameters are dynamically adjusted through the reinforcement learning rule optimization model. With the main grid setting value as the constraint, the distribution network and user side setting value are matched, and the preliminary setting value scheme is generated. The scheme is checked using the sensitivity dynamic checking model, and finally a reliable distribution network relay protection setting value is generated through the blockchain collaboration model, ensuring the reliability and credibility of the setting value scheme. The application effectively solves the problem that the existing technology cannot accurately and efficiently set the relay protection setting value of the distribution network.

This application relates to a topology adaptive power flow calculation method, apparatus, computer device, computer-readable storage medium, and computer program product based on a physical information graph neural network. The method includes: constructing a graph structure corresponding to the target power grid, using different types of power grid bus nodes as graph nodes and the topological relationships between power grid bus nodes as edges of the graph nodes; inputting the graph structure corresponding to the target power grid into a pre-trained power flow calculation model, and outputting the power flow calculation results of the target power grid; wherein, the power flow calculation results include the voltage magnitude and voltage phase angle of all graph nodes in the graph structure; the power flow calculation results are calculated by the pre-trained power flow calculation model based on node characteristics and edge characteristics; the pre-trained power flow calculation model is trained using power grid graph structure samples under different operating conditions and / or different data acquisition times, as well as physical constraints. This method can improve the intelligence of power flow calculation.

This application provides a carbon flow method based on a hierarchical time-colored Petri net model, belonging to the field of carbon management technology. It constructs a formally defined hierarchical time-colored Petri net basic model, and based on this basic model, obtains an application model containing enterprise, workshop, and equipment layers. In the application model, transitions are distinguished into decision-making transitions and execution transitions. A potential carbon flow calculation function is configured for decision-making transitions to calculate the potential carbon emission increment generated by the current decision-making activity and update it in the carbon account data of the flowing token. An actual carbon flow calculation function is configured for execution transitions to calculate the actual carbon emission increment generated by the current execution activity within its time delay and update it in the carbon account data of the flowing token. Based on the calculated potential and actual carbon emission increments, carbon footprint data is automatically collected from the equipment layer through the workshop layer to the enterprise layer. This enables real-time, dynamic, and cross-level automatic collection of the carbon footprint throughout the product's entire lifecycle.

This invention relates to the field of internet advertising technology and provides an intelligent decision-making system for advertising placement based on real-time stream computing. The system includes: a heterogeneous stream computing engine module; a cold-start state space prediction module; a stream complex event detection module; and a real-time online analysis and storage cluster module. This module employs a large-scale parallel processing architecture and a primary key update model to achieve second-level writing of business metrics and prediction results and accelerated materialized view queries. By constructing a full-link real-time stream closed-loop architecture, it achieves millisecond-level response from user behavior perception to automatic execution of placement decisions. By introducing a dynamic prediction model and a complex event detection mechanism, the system effectively solves the problems of decision blind spots and nonlinear risk identification in the early stages of placement, significantly enhancing the predictability and accuracy of placement strategies.