69 results about "Centralized computing" patented technology

The invention relates to the technical field of virtual power plants, and discloses a virtual power plant optimal scheduling system and method, and the system comprises a data obtaining module, an edge calculation module, a prediction module, a scheduling controller, a topology reconstruction module, and an intelligent terminal device cluster. According to the invention, the edge computing module carries out localization processing and prediction on the sensing data, so that rapid generation and issuing of a scheduling scheme are realized, and the problem of response delay caused by network transmission and centralized computing of a traditional centralized architecture is avoided, thereby supporting millisecond scheduling feedback and improving the scheduling efficiency. The real-time response capability under the sudden load fluctuation or fault condition is remarkably improved, a multi-dimensional perception and prediction mechanism is constructed based on an LSTM neural network prediction model, the recognition and trend prediction capability of the system on meteorological disturbance, equipment aging and operation abnormity is enhanced, the intelligent level of the virtual power plant system is improved, and the real-time performance of the virtual power plant system is improved. The system can dynamically generate an optimal scheduling strategy to ensure stable operation of the virtual power plant under various working conditions.

A computing system including one or more processing devices configured to receive an image generation prompt and a reference image. Over a plurality of denoising timesteps, the one or more processing devices compute a guided image by applying denoising updates to a generated image at a denoising diffusion model. At a subset of the denoising timesteps, computing the guided image further includes applying guidance updates to the generated image based on the image generation prompt, the reference image, and a generated image set. The one or more processing devices compute each guidance update by performing a forward pass and a backward pass in first and second integration timesteps. A size of the generated image set and numbers of the first and second integration timesteps are each equal to a predefined integration timestep count. The one or more processing devices output a final generated image computed in a final denoising timestep.

The invention discloses an intelligent electric energy meter data acquisition device, method and equipment for electric carbon metering and a storage medium, and relates to the technical field of electric power systems and carbon emission management, and the device comprises a carbon emission hardware calculation circuit module which verifies frozen data and regional-level electric carbon factors and line loss power of a superior node, and generates carbon emission data; a first non-volatile processor storing carbon emission data; and the second nonvolatile processor and the first nonvolatile processor are backups for each other and store electric quantity freezing data. The device expands a carbon emission metering function at low cost through an idle RS485 port of an existing intelligent electric energy meter, generates a dynamic electric carbon factor through local calculation, accurately calculates carbon emission data, and improves the accuracy of the carbon emission data. The data accuracy, real-time performance and reliability are improved, the centralized calculation burden is relieved, the system modification cost is reduced, large-scale popularization and application are supported, and low-carbon transformation and green power management are effectively promoted.

The invention discloses a global cooperative control method and system for unbalanced fault ride-through of a multi-terminal DC power grid, and the method comprises the steps: constructing a fault detection model based on the positive sequence voltage drop depth and the negative sequence voltage size of the AC side of each converter station, automatically judging an unbalanced fault, and carrying out the fault detection when the fault is detected. Calculating the maximum allowable active current based on a fault station unbalanced current control model to carry out current limiting control on the converter station; and carrying out output power adjustment on the DC voltage deviation of the non-fault station based on the non-fault station power cooperation model. Asymmetric fault ride-through is carried out by using a constrained and optimized current instruction generation algorithm, and the capacity of a converter is dynamically and accurately distributed and utilized in a limited manner, so that the voltage support capability of a fault alternating-current power grid is maximized, and globally optimized adaptive power distribution is realized; the power of the non-fault station is calculated and distributed in a centralized manner and is dynamically adjusted according to the real-time available capacity, so that the self-adaption and global optimization of the running state change of the power grid are realized, and the fault handling toughness and reliability of the power grid are improved.

The invention provides a centralized computing power scheduling method applied to a computing power network, and the method comprises the steps: for a service deployment scheduling task, obtaining computing power resource information in the computing power network, and determining computing power resources needed by the service deployment scheduling task; filtering all computing nodes to obtain a plurality of candidate computing nodes meeting computing power resources; all the candidate computing nodes are scored, and the candidate computing node with the highest score is selected from scores to deploy the service instance; for a service request scheduling task, acquiring computing power resource information and network resource information of a computing power network, determining computing power resources and network resources required by the service request scheduling task, filtering all service instances which do not meet the computing power resources and the network resources, and taking the remaining service instances as candidate service instances; and scoring all the candidate service instances, and selecting the candidate service instance with the highest score from scores to execute the service request.

The invention relates to an intelligent network connection vehicle formation centralized control method based on deep reinforcement learning, and belongs to the technical field of intelligent traffic. A communication network architecture in a vehicle-road-cloud set is constructed, a vehicle-mounted computing task is unloaded to an edge cloud node, and global optimization control is realized in combination with a DDPG algorithm. The vehicle node obtains state information through a vehicle-mounted sensor and transmits the state information to the edge cloud node through the roadside node; and the edge cloud integrates multi-vehicle data collected by the RSU into formation global state information, generates a control instruction in real time by using a pre-trained DDPG intelligent agent, and broadcasts the control instruction to formation vehicles for execution through C-V2X communication. And the DDPG intelligent agent is guided to train through a reward function and is combined with a fixed time interval strategy and a second-order kinematics model to realize rapid convergence of a vehicle distance and a speed error. According to the invention, the load of vehicle-mounted hardware is reduced through a centralized computing architecture, and the real-time performance and safety of formation control are improved; complex algorithm global optimization is supported by using cloud computing power.

The invention discloses a free airship cluster autonomous coverage method based on distributed learning and model predictive control, and belongs to the technical field of automatic control. The method comprises the following steps: firstly, establishing a six-degree-of-freedom dynamic model considering quality, buoyancy characteristics and wind field disturbance, and ensuring the physical accuracy of airship motion modeling; secondly, acquiring local wind field data by each airship through a sensor, performing local estimation by using a distributed Gaussian process regression algorithm, and realizing global wind field prediction through induction point approximation and hyper-parameter iteration; then, carrying out Voronoi division based on a global wind field sensitivity function, and determining a coverage area and a weighted centroid target of each airship; and finally, constructing an optimization problem based on model predictive control, constraining thrust and torque input, and carrying out rolling optimization by adopting a sequential quadratic programming method to realize stable track tracking and autonomous region coverage. The sensing, planning and cooperative control capabilities of an airship cluster in a complex non-uniform wind field environment are effectively improved, the centralized calculation overhead is reduced, and the robustness and adaptability of the system are enhanced.

A system receives ETL specification for processing stream data, including a transform operation represented using a database query specification. The system generates a dataflow graph of a sequence of database queries by decomposing the database query into a first database query that generates an intermediate results table, and a second database query that receives as input the intermediate results table and outputs data used for performing the transform operation. The system executes the sequence of database queries for performing the transform operation on stream data received from the source. When receiving an incremental data set, the system determines an output change set based on the received incremental data set by traversing an execution plan and processing each operator in the execution plan, and computing a change set of a particular operator from the change sets output by the one or more other operators based on the incremental data set.

The invention discloses a retrieval question and answer method and system combined with a time sequence context, a medium and equipment, and belongs to the crossing field of artificial intelligence and intelligent operation and maintenance of an electric power system. Extracting time sequence semantic elements in the event sequence, and retrieving a corresponding event sub-graph in a time sequence directed graph constructed based on an event sequence, a time sequence and a logic link of the power equipment; performing feature aggregation on the sub-graph nodes to obtain context vectors; performing weighted fusion on the query semantic vector and the context vector to obtain a joint vector; calculating the similarity between the knowledge fragment vector set and each knowledge fragment in the knowledge fragment vector set, and screening out a plurality of most relevant knowledge fragments; and inputting the original query and the knowledge fragments into a text generation model, outputting answers, and completing questions and answers. By implementing the method and the device, the technical problem that an intelligent question and answer retrieval method in the prior art lacks pertinence and professional traceability of retrieval results in a complex event sequence scene in the power field can be solved.

Virtual Machine (VM) creation based on a dynamically-calculated feature set. A plurality of feature sets are identified. Each feature set in the plurality of feature sets corresponds to a different VM host node of a plurality of VM host nodes that are part of a common VM migration group, and indicates a set of features available at a corresponding VM host node. From the plurality of feature sets, a group feature set is calculated. The group feature set includes a subset of features that are common among the plurality of feature sets. A VM created within the plurality of VM host nodes is configured to use the group feature set.

The present invention discloses a method for optimizing the weight of a conical spring based on the Harris hawk algorithm with random unit permutation. First, the objective function for optimizing the weight of the conical spring and the key parameters to be solved are determined. Then, the structure of the original Harris hawk algorithm is optimized. The specific optimization measures are as follows: First, cancel the steps of centrally calculating the objective function value in the original Harris hawk algorithm except for the step of initializing the population, but record the results while the algorithm obtains a better solution. Second, introduce a random unit permutation mechanism at the position before the end of each iteration optimization in the original Harris hawk algorithm. Third, delete the search steps in the original Harris hawk algorithm where the absolute value of the energy factor E is greater than or equal to 1. The optimized Harris hawk algorithm is called the Harris hawk algorithm based on random unit permutation, and the key parameters are solved using this Harris hawk algorithm based on random unit permutation to obtain the key parameters for optimizing the weight of the conical spring. The advantages are fast convergence speed and high solution accuracy.

The invention provides a handwriting elimination method, a model training method and a related device. An electronic device obtains a to-be-processed image and a color mask of the to-be-processed image; wherein a handwriting area in the image to be processed is marked in the color mask, the handwriting elimination model of the image to be processed comprises an encoder and a decoder, and the encoder comprises a feature extraction layer; the feature extraction layer processes various image features in the to-be-processed image by using the attention information in the color mask to obtain output features of the feature extraction layer; and processing the output characteristics of the encoder through a decoder to obtain a clean image after handwriting elimination. Therefore, the attention information in the color mask is directly utilized to guide the feature extraction process, so that the model can concentrate computing resources in a local range, the unnecessary global computing burden is reduced, and the overall computing efficiency in the handwriting elimination process is improved.

Systems and devices for network data collection and processing are provided. An example system includes a first networked device and a centralized computing device communicably coupled with the at least one networked device. The first networked device operates to generate event-driven data entries associated with the first networked device and generate first data packets including the event-driven data entries and / or manipulated outputs generated based on manipulations to the event-driven data entries. The centralized computing device receives the first data packets from the first networked device and determines configuration updates based on the first data packets. The configuration updates are generated locally by the centralized computing device, and the centralized computing device transmits the one or more configuration updates to the first networked device.

This invention discloses a smart energy meter data acquisition device, method, equipment, and storage medium for carbon emission metering, relating to the fields of power system and carbon emission management technology. It includes a carbon emission hardware calculation circuit module that verifies frozen data and regional-level carbon emission factors and line loss power from the upper-level node to generate carbon emission data; a first non-volatile processor that stores the carbon emission data; and a second non-volatile processor that serves as a backup for the first non-volatile processor and stores frozen electricity data. The device of this invention expands the carbon emission metering function at low cost through the idle RS485 ports of existing smart energy meters, generates dynamic carbon emission factors through local calculation, accurately calculates carbon emission data, improves data accuracy, real-time performance, and reliability, reduces the burden of centralized calculation, lowers system transformation costs, supports large-scale application, and effectively promotes low-carbon transformation and green power management.

A method for deterministic execution of an instruction set (I) containing instructions (M) to be carried out by a processing unit of the computing device includes the steps of executing a predefined number of pre-branch instructions (A) up to a conditional branch (C) of the instruction set (I) containing at least one branch condition determining a subsequent instruction which is to be executed after the conditional branch (C); predicting at least one of the pre-branch instructions (A) and / or a target instruction (T) to be executed as the subsequent instruction if the at least one condition is met; and inserting the at least one predicted pre-branch instruction (A) and / or the target instruction (T) into the instruction set (I) after the conditional branch (C). A computer program, computer-readable data-carrier, computing device, and apparatus, such as a vehicle, in particular an aircraft may be configured to perform the method.

A system receives ETL specification for processing stream data, including a transform operation represented using a database query specification. The system generates a dataflow graph of a sequence of database queries by decomposing the database query into a first database query that generates an intermediate results table, and a second database query that receives as input the intermediate results table and outputs data used for performing the transform operation. The system executes the sequence of database queries for performing the transform operation on stream data received from the source. When receiving an incremental data set, the system determines an output change set based on the received incremental data set by traversing an execution plan and processing each operator in the execution plan, and computing a change set of a particular operator from the change sets output by the one or more other operators based on the incremental data set.

Systems and methods for a patient monitoring system enable a hub (e.g., centralized computing device) to automatically receive data from varying different types of patient monitoring devices. In particular, the patient monitoring system may utilize polarity selection circuitry to automatically select (e.g., swap, switch) polarity of one or more input ports of the hub of the patient monitoring system. As such, the hub may include one or more universal ports to successfully receive or accept multiple different connectors of the varying patient monitoring devices that may include varying pin definitions.

A comprehensive and centralized system for managing and tracking communications being delivered through both digital communications channels and physical communication channels. A centralized computing hub receives communication requests from originating systems of record or via user input. In response to receiving a communication request, content is generated or otherwise obtained and quality assurance checks are performed to assure that the content meets the requirements of the originating system of record and is applicable to the intended communication recipient. Once the quality checks have been performed and the content is deemed verified, delivery of the communications is orchestrated in accordance with the chosen communication channel. The timeliness of the communications is tracked throughout the system and alerts are generated in response to the timeliness exceeding thresholds. Thresholds are intelligently determined on a continuous basis using rolling averages of previous communications that result in predictions of future communication averages.

The invention provides a centralized computing power network system aiming at transmission scheduling, which comprises a transmission scheduling subsystem, a computing power scheduling subsystem and a plurality of computing power clusters, and is characterized in that the transmission scheduling subsystem is configured to divide the computing power network system into different levels of regional networks, sense and manage the network resource state of each regional network in a level-to-level manner, and transmit the network resource state to the computing power clusters; a user terminal is accessed to each regional network; receiving a resource request message sent for the application by the user terminal or the computing power cluster, and determining a resource demand of the application according to the request message; the computing power scheduling subsystem and the computing power scheduling subsystem cooperatively execute resource scheduling, and the method comprises the steps that the computing power scheduling subsystem schedules computing power resource instances meeting resource requirements from other computing power clusters according to the network resource state of the regional network and the sensed computing power resource state of the computing power clusters, and a scheduling result is obtained; the transmission scheduling subsystem generates a routing strategy for accessing the computing power resource instance when the scheduling is successful; and the user terminal or the computing power cluster communicates with the computing power resource instance based on the routing policy.

This invention discloses a distributed safety learning control method for mobile robot swarms, relating to the field of robot cooperative control technology. The method includes: calculating the future short-term motion position based on a nominal control quantity, and calculating the predicted neighborhood distance by combining it with a neighborhood state set to form multiple sets of safety constraints; combining these multiple sets of safety constraints with environmental boundary information to form obstacle avoidance constraints, and establishing constraint priority rules with actuator limits; obtaining a hierarchical safety constraint set by adding constraint level identifiers; using the nominal control quantity as the optimization objective and the hierarchical safety constraint set as the constraint conditions to construct a local optimization subproblem, and performing a distributed solution using the alternating direction multiplier method to generate a safe and feasible control quantity. This invention achieves the output of safe and feasible control quantities without centralized computation, thereby improving the real-time performance and reliability of swarm cooperative control.

The automated closed-loop gravity infusion system includes a fluid supply source, a drop counter operatively engaging a drip chamber of the fluid supply source, a roller clamp operatively associated with the drop counter, a processor operatively associated with the roller clamp, and a human-machine interface (HMI) operatively connected to the processor. Patient data is accessed through the HMI. The gravity infusion system utilizes the patient data and data from the drop counter to determine the appropriate position of the roller clamp. The drop counter continuously monitors the drip rate of the drip chamber and transmits the data to the processor. The processor analyzes the data and automatically adjusts the position of the roller clamp to administer the appropriate volume of fluid to the patient. The HMI is operatively linked to a remote centralized computing system capable of simultaneously monitoring multiple systems used to infuse different patients.

The invention discloses an Internet of Vehicles data warehouse method and device, electronic equipment and a storage medium, and relates to the field of data processing, and the method comprises a data carding step, a data scene analysis step, a data application demand analysis step, a data set warehouse architecture design step and a data warehouse development step. Wherein the data sorting step comprises the following steps of: determining data stored in a data warehouse; the data scene analysis comprises the following steps: determining a subject domain to which data belongs; the data application demand analysis comprises the following steps: determining various indexes; the data set warehouse architecture design comprises the following steps: defining hierarchical division of a data warehouse; the data warehouse development step comprises designing a full-link data calculation process. According to the scheme, the problem of data islands formed by scattered storage of Internet of Vehicles data and the problem of resource waste caused by repeated calculation of indexes in different business scenes are solved. Unified data acquisition, cleaning and management are realized, and centralized calculation and sharing multiplexing of general indexes are realized.

The invention discloses a database generation method, system and device, a program product and a storage medium, and the method comprises the steps: obtaining target risk data based on a target business line and a target business system; performing standardization processing on the target risk data to obtain standardized risk data; and storing the standardized risk data to obtain a risk database. Standardized risk data are stored in a centralized mode to form a risk database, the database can directly provide data services for an external application system, a traditional repeated storage mode is replaced to save IT resources, and meanwhile a foundation is laid for improving centralized computing power; and the requirements of digital risk management of the insurance group on data integrity, timeliness, accuracy and sharing are comprehensively met.

The present invention relates to a central computing unit (CCU) (105) configured for a vehicle (800), such as an on-board computing unit of an automobile, to centrally control different functions of the vehicle (800), the CCU (105) comprising: a housing structure (605); a plurality of electronic modules (115, 120, 125, 135) housed (600) by the housing structure (605), including a first module (115, 120, 125, 135) and at least one second module (115, 120, 125, 135), the modules individually or collectively defining at least one computing function of the CCU (105); and passive connection means (130) for indirectly connecting the first module (115, 120, 125, 135) with one or more of the second modules (115, 120, 125, 135) to provide one or more gain-free signal paths and / or gain-free power (PWR) connections between the first module (115, 120, 125, 135) and one or more of the second modules (115, 120, 125, 135); wherein each of the modules (115, 120, 125, 135) comprises at least one respective first connector (1110), and the connecting device (130) comprises, for each first connector (1110), a corresponding second connector (920, 925) mating therewith to connect the respective module (115, 120, 125, 135) via (1400) the mating first and second connectors (920, 925). 135) and the connecting device (130) are connected in a detachable power connection and / or a detachable signal connection.

The application discloses a regional power grid transformer group anti-short-circuit capability distributed checking system and method, and relates to the technical field of electric power engineering. The system comprises a distributed sensing module, a transformer group management and control module, a short-circuit checking module, a terminal centralized control module, and is distributed in a local power grid information measurement center, a transformer group management and control center and a regional power grid anti-short-circuit checking and analysis center. The application collects data and preliminarily judges faults through the distributed sensing module, filters faults and transformers to be checked through the transformer group management and control module, performs multi-physical field checking through the short-circuit checking module, and divides risk levels and gives early warnings through the terminal centralized control module. The application solves the problems of low static checking efficiency, difficulty in adapting to power grid changes and insufficient centralized computing power in the prior art, realizes dynamic and distributed checking, and improves the safety and stability of power grid operation and maintenance.

Embodiments described herein are directed to efficient content management within a data-processing environment. In some embodiments, data is obtained at an edge processor. The edge processor may perform data processing in proximity to generation of the data and provide the processed data to a centralized computing service for subsequent processing or storage. At the edge processor, basic data processing may be performed to process the data based on a basic data processing pipeline executed on the edge processor. The basic data processing pipeline may use basic data processing content to perform the basic data processing. The processed data may be provided to a component of the centralized computing service.

Methods, systems, devices, and computer program products for machine learning in datacenter applications are provided. An example method includes receiving, by a centralized computing device, data packets from a networked device communicably coupled with the centralized computing device. The networked device is associated with performance of at least a first machine learning based task, and each of the data packets include data entries generated by the networked device based on data traffic associated with the at least one networked device and / or one or more modifications thereto. The method further includes generating updated operational parameters associated with the first machine learning based task based on the data entries forming the plurality of data packets where the updated operational parameters are generated locally by the centralized computing device. The method also includes transmitting, by the centralized computing device, the updated operational parameters to the networked device.

The invention discloses an expressway traffic monitoring method and system and a medium, mainly relates to the technical field of traffic monitoring, and is used for solving the problems of insufficient multi-source heterogeneous data fusion capability and serious response delay caused by a centralized computing architecture in the existing scheme. Comprising the following steps: segmenting a highway into a plurality of to-be-detected road sections according to traffic flow detection data; determining the monitoring type of each road section to be detected according to the video monitoring data; determining a monitoring data type and a monitoring program of each monitoring type, issuing the monitoring data type to the corresponding data extraction node, and issuing the monitoring program to the corresponding edge computing node; extracting specific data corresponding to the monitoring data type from the multi-source heterogeneous real-time data and the external influence data through a data extraction node; and transmitting the specific data to the corresponding edge computing node, and inputting a monitoring program to obtain an output monitoring result.

The application provides a train braking force distribution system, method and train, the system fuses a train ATO module and a CCU module to form an OCU module, and fuses a single-car corresponding TCU module and a BCU module to form a VDU module corresponding to a car, reconstructs a train braking force distribution system architecture, and can effectively solve the problems existing in the current distributed braking force distribution mode. The OCU module serves as a centralized computing core, analyzes and calculates the braking force demand values of each axle of the whole train according to the data uploaded by each VDU module, and issues the demand values to the corresponding VDU module, without the cumbersome connection process of the original TCU feeding back the actual electric braking value to the BCU and the BCU issuing an execution instruction again, so that the connection of the braking force distribution processes is smoother, the problem of poor process connection is effectively solved, and the distribution efficiency of the train braking force is improved.

The embodiment of the invention provides a railway project rapid modeling method and system based on multi-source data fusion, and the method comprises the steps: obtaining multi-source modeling data, and carrying out the data fusion based on the multi-source modeling data, and determining the fusion data; preprocessing the fusion data, and determining preprocessed fusion data; performing parametric modeling based on the preprocessed fusion data, and determining a target model; and performing model verification through digital twinning based on the target model, and determining a verification result. Compared with a traditional BIM platform, according to the scheme, through dynamic parameter driving, hybrid simulation acceleration and knowledge enhancement decision making, the limitation of linear modeling, centralized calculation and experience dependence is broken through, a'data-model-decision 'closed loop is achieved in complex railway engineering, and a technical normal form capable of being quantitatively verified is provided for industry digital transformation.