126 results about "Smart computing" patented technology

The invention relates to the technical field of spiking neural networks, in particular to a spiking neural network conversion method and system based on differentiable adaptive optimization, and the method comprises the steps: constructing and pre-training a target ANN model; in a pre-trained ANN model, performing simulation processing on the activation value of each layer by using a global differential value representation operator; on the basis of a loss function, an original weight parameter in the target ANN model and a trainable excitation threshold in a global differentiable value representation operator are placed in the same optimization framework for joint fine tuning, and an optimal weight and an optimal threshold learned by each layer are obtained; and an equivalent SNN model is constructed. According to the method, precise correspondence between continuous activation and pulse sequences can be optimized, the method is suitable for multiple pulse coding modes, quantization errors are reduced to the maximum extent, the deployment requirements of neuromorphic hardware and an edge intelligent computing platform can be deeply adapted, and an efficient and universal model conversion solution is provided for the neuromorphic computing industry.

The invention provides a method and device for an intelligent computing cloud platform to realize agent self-evolution through computing power, and relates to the technical field of intelligent computing cloud platforms and computing power infrastructures, and the method comprises the steps: S1, an agent obtains target behavior data of a user using the agent; s2, the intelligent agent analyzes the target behavior data by adopting a large language model to obtain behavior statistical characteristics; s3, the intelligent agent adopts a large language model to identify user preference features according to behavior statistical features; s4, determining a quality evaluation result of the intelligent agent by the intelligent agent according to the behavior statistical characteristics by adopting a large language model; and S5, the intelligent agent determines an evolution scheme of the intelligent agent according to the quality evaluation result and the user preference characteristics by adopting a large language model, and the intelligent agent is optimized based on the evolution scheme. According to the invention, the development cost of the intelligent agent can be effectively saved, and the optimization of the intelligent agent better meets the actual use demand of a user.

The invention belongs to the field of edge computing, and discloses an edge intelligent computing device and method supporting multi-level storage and computing integrated acceleration, and the device comprises a main processor system, a near memory computing module and a near flash memory computing module. The near memory calculation module integrates an NPU (Network Processing Unit) with a distributed SRAM (Static Random Access Memory) and an HBM DRAM (Dynamic Random Access Memory) with an MAC (Media Access Control) array, and realizes high-throughput reasoning acceleration And the near flash memory calculation module is used for tightly coupling the SSD controller and the retrieval acceleration engine to realize the nearest retrieval acceleration of mass data at the SSD level. All the modules are interconnected through a PCIe bus, a three-level storage and calculation integrated framework of SRAM-HBM DRAM-SSD is constructed, cross-level data migration is effectively reduced, and the bottleneck of the PCIe bus is avoided. According to the method, high-throughput video reasoning and TB-level data retrieval tasks on the edge side can be efficiently supported at the same time, and the calculation energy efficiency ratio and the system throughput rate are remarkably increased.

The invention provides a modularized positioning navigation time service system and a positioning navigation time service method, the modularized positioning navigation time service system comprises a plurality of PNT assembly modules and an information fusion module, and the modules realize efficient data interaction through a standardized interaction protocol. The PNT component module is composed of a sensing element, an edge calculation chip and an information transmission interface, the standardized information extraction sub-module is deployed in the edge calculation chip, comprises a parameter setting unit, a parameter characteristic extraction unit, a constraint construction unit and a statistical characteristic extraction unit, and converts output information of a PNT component into a standard form; and the information is transmitted to the information fusion module. The information fusion module comprises an information transmission interface and an intelligent computing platform, intelligent optimization and fusion processing of multi-source PNT information is achieved through an information monitoring sub-module, a parameter adjustment sub-module, a space-time unification sub-module and an intelligent fusion sub-module which are deployed in the intelligent computing platform, and state estimation information is output.

AI-DRIVEN SMART COMPUTER DEVICE FOR LOAN APPROVAL AND CREDIT EVALUATION.

The application relates to a digital twin architecture and an intelligent control method of an electric vehicle thermal management system, and belongs to the electric vehicle thermal management field. The application utilizes a digital twin technology to establish digital twins corresponding to each module and each component of physical real vehicle thermal management in a cloud platform thermal management digital space; real-time operation data of the physical real vehicle is collected through a sensing device and uploaded to a cloud platform digital space digital twin engine; the digital twin engine utilizes the real-time operation data to realize accurate mapping of the digital twins and the physical real vehicle; an intelligent computing module in the digital twin engine realizes real-time calculation of an optimal electric vehicle thermal management strategy through coupling of the digital twins and operation results; the physical real vehicle receives the optimal thermal management strategy from the digital space, and completes one-time vehicle thermal management control. The application can improve the control efficiency and precision of integrated thermal management of the electric vehicle, improve the cruising range of the electric vehicle, and reduce the risk of battery thermal runaway.

This invention discloses a ship cloud computing load balancing method based on a hybrid intelligent algorithm, relating to the field of ship intelligent computing and cloud computing integration technology. The method includes the following steps: task classification modeling, edge-cloud resource status monitoring, multi-objective optimization decision-making, and dynamic load adjustment. This invention reduces the load standard deviation of ship edge-cloud nodes through a multi-objective optimization algorithm and a dynamic adjustment mechanism. It optimizes task processing efficiency, reducing the average latency of real-time tasks and shortening the total processing time of non-real-time tasks. It reduces system energy consumption by optimizing the energy consumption coefficient, thereby reducing the total energy consumption of edge nodes and cloud servers. It adapts to ship scenarios, improving the robustness and adaptability of scheduling strategies to address the heterogeneity of ship tasks and network dynamics.

The invention discloses an encrypted traffic high-speed shunting and content-level security collaborative identification method and system based on a heterogeneous core particle architecture, and belongs to the technical field of network security and intelligent computing. The method comprises the following steps of: 1, initializing a system and configuring a core particle framework; step 2, executing flow aware scheduling to realize load balancing and task migration; step 3, executing multi-level cache and main / standby switching; 4, performing intelligent fingerprint restoration and encrypted content identification; 5, generating a bimodal watermark and establishing a dynamic fusing contract mechanism; and step 6, executing federal collaborative learning and security recommendation. According to the method, the user privacy and the data encryption integrity are guaranteed, high-throughput and low-delay intelligent security identification is realized, and the security collaboration and credible defense capability in a distributed network environment can be remarkably improved.

The application provides a method and device for quantifying elastic KV cache through computing power of an intelligent computing cloud platform, and relates to the technical fields of intelligent computing centers, intelligent computing centers, computing power infrastructure and intelligent computing cloud technology.The method comprises the following steps: S1, dividing historical tokens into multiple cache blocks, quantifying KV data and writing the data into corresponding cache blocks, and selecting multiple candidate anchor points; S2, calculating block-level summary data; S3, in response to a new target token, scoring the cache blocks to generate cache block scores and screening out candidate cache blocks; S4, calculating uncertainty index data and determining a target cache block with to-be-restored precision according to the uncertainty index data; and S5, locating an upstream target anchor point and locally playing back the historical tokens based on the upstream target anchor point to generate target high-precision KV data of the target cache block.The application can greatly improve the quantification effect of KV cache of the intelligent computing cloud platform.

A satellite-borne safety intelligent computing system is used for guaranteeing safety of a satellite edge computing system and comprises a host processor, an AI accelerator and a safety bus connected with the host processor and the accelerator. The satellite-borne safe intelligent computing system further comprises a bus stream encoding and decoding engine which is embedded in the interface controller of the data bus and used for encrypting and decrypting transmission data of the data bus. By encrypting the cache stream on the data bus, key data such as Bias bias cache and the like can be protected. In addition, an attacker can be prevented from stealing structural information of the neural network model through interrupt signals by interrupting obfuscated codes, dual protection of'signal + data 'is formed, and time sequence detection Trojan horse and side channel attacks can be effectively resisted.

The invention provides a computing power anti-fragmentation method and device for an intelligent computing center, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructure, and the method comprises the steps: under the condition that a to-be-processed task is obtained, based on the computing power of the intelligent computing center, according to the task type and / or demand resource size of the to-be-processed task, determining the computing power of the to-be-processed task; allocating computing power resources corresponding to the first device in the at least one device to execute the to-be-processed task; the computing power resource corresponding to each device comprises a first computing power resource and / or a second computing power resource, the first computing power resource is divided into at least two parts of computing power resource, the first part of computing power resource is used for processing a first type of task, and the second part of computing power resource is used for processing a second type of task. According to the method, the computing power of the intelligent computing center is utilized to allocate the computing power resources of the corresponding equipment to execute the tasks according to the task types of the tasks to be processed and the size of the required resources, the computing power resources of the equipment can be fully utilized, and waste of the computing power resources is reduced.

The invention provides a computing power resource operation fault defense method and device of an intelligent computing center, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructure, the method comprises the following steps: S1, obtaining historical data corresponding to computing power resources of target equipment, the target equipment being equipment providing the computing power resources in the intelligent computing center; s2, performing feature extraction on the historical data to obtain a correlation feature vector and a time sequence feature vector corresponding to each physical index in a plurality of physical indexes included in the computing power resource; s3, splicing the associated feature vector and the time sequence feature vector corresponding to each physical index to obtain a prediction feature vector corresponding to each physical index; and S4, predicting a target parameter value corresponding to each physical index based on the prediction feature vector, wherein the target parameter value is used for determining whether the computing power resource of the target equipment is abnormal or not. The fault rate of equipment can be greatly reduced.

The present invention belongs to the field of intelligent computing, and specifically relates to a multi-task collaborative attention TSK fuzzy system modeling method, which is used to provide an interpretable multi-task intelligent evaluation model for fermentation food safety assessment. The method includes two main parts: a new multi-task TSK fuzzy system model structure and a multi-task collaborative optimization process. In the intelligent evaluation model, the proposed multi-task collaborative processing unit is used to perform collaboration among multiple evaluation tasks. By using the multi-task feature selection layer and task attention structure respectively to extract the unique information of each task and the relevant information between tasks, the evaluation performance of each evaluation task can be better improved. In the collaborative optimization process, the present invention uses multi-task collaborative regularization to achieve more efficient mining and utilization of the unique information of each task.

This invention provides a method and apparatus for generating tokens using computing power acceleration on an intelligent computing cloud platform. It relates to the fields of intelligent computing centers, smart computing centers, computing infrastructure, and smart computing cloud technologies. The method for generating tokens using computing power acceleration on an intelligent computing cloud platform includes: Step S1, establishing a full-text index of a target data set; Step S2, responding to a received data query request, determining whether to enable document index acceleration; Step S3, if document index acceleration is enabled, extracting scalar filtering conditions and vector filtering conditions; Step S4, performing a full-text search based on the scalar filtering conditions to obtain a first data set; Step S5, performing a vector search on the vector index of the first data set based on the vector filtering conditions to obtain a matching second data set, and returning it to the user. This invention can improve the speed of vector database processing with scalar constraints on intelligent computing cloud platforms, reduce resource consumption, and increase token generation speed.

The present disclosure relates to the technical field of computational power. Provided are a method for measuring the computing capability of an intelligent computing center by means of 1-degree computational power, and a device. The method comprises: step S1, acquiring an initial half-precision floating-point computational power number, an initial video memory bandwidth and an initial video memory capacity of a processor of a first intelligent computing center; step S2, calculating a first ratio of the initial half-precision floating-point computational power number to a preset half-precision floating-point computational power number, a second ratio of the initial video memory bandwidth to a preset video memory bandwidth, and a third ratio of the initial video memory capacity to a preset video memory capacity; and step S3, setting a product of a weighted sum of the first ratio, the second ratio and the third ratio, a preset unit of floating point operations, and a preset time unit as an initial 1-degree computational power value of the first intelligent computing center.

The application discloses a kind of modularization on-orbit edge intelligent computing platform, belong to spacecraft electronic information technical field.The platform uses distributed architecture, including at least one network and management module NMM and at least one general edge AI module UAM, NMM as the data exchange and control center of platform, including system management unit SMU, redundant network exchange unit and controlled wireless communication unit;UAM as the core processing unit of platform, responsible for executing data and computationally intensive tasks, using the separation type design of core board plus bottom plate, each UAM is connected the redundant network exchange unit by independent Ethernet link.The platform is through the distributed architecture with network exchange as center, in combination with hierarchical module level fault self-healing mechanism, multi-level redundant communication architecture facing application optimization and the flexibility of elastic computing power scheduling mechanism based on microservice, while guaranteeing high performance and flexible scalability, greatly improve the on-orbit survivability and task execution efficiency of platform.

The invention provides a multi-valued spiking neuron processing method based on dynamic threshold feedback reset, and relates to the technical field of brain-like intelligent calculation, and the method comprises the steps: introducing a dynamic threshold in explicit association with an input current, a multi-valued issuing mechanism, and combining the feedback reset adjustment of a historical pulse behavior; the neurons have higher information expression ability and adaptive state regulation and control ability in a single time step; meanwhile, a substitution gradient coupled with a dynamic threshold value is introduced for multi-valued distribution to guarantee gradient stable propagation in the training stage, and multi-valued pulses are equivalently converted into sparse binary events in the reasoning stage, so that the low energy consumption characteristic of event-driven calculation is kept while the expression ability is improved. According to the invention, neuron processing of multi-valued pulse expression, threshold and reset adaptive adjustment, training stage gradient stability and inference stage event-driven energy efficiency maintenance problems can be simultaneously solved on the whole.

The invention provides a computing power aggregation method and device of a common computing power intelligent computing cloud platform, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructure, the intelligent computing cloud platform comprises a logic computing power cluster and a plurality of physical computing power clusters, each physical computing power cluster comprises a plurality of computing power nodes, and each physical computing power cluster comprises a plurality of computing power nodes. The method comprises the steps that a logic computing power cluster synchronizes information of each computing power node in a physical computing power cluster in real time, the computing power nodes are mapped into virtual computing power resources, and the virtual computing power resources are put into a computing power resource pool; receiving a container creation request or a calculation task creation request of a user, and creating a container based on the request; based on the information of each virtual computing power resource in the computing power resource pool, screening out a target virtual computing power resource matched with the container from the computing power resource pool; and determining a target computing power node corresponding to the target virtual computing power resource, and scheduling the target computing power node operation container through a physical computing power cluster where the target computing power node is located.

This application relates to the field of food processor control technology, and discloses a control method and system for a food processor based on intelligent adjustment. The method or system includes: a data sensing unit that collects real-time data on the physical properties of ingredients and motor operating parameters; an ingredient characteristic identification unit that identifies the hardness, toughness, and processing stage of the ingredients; an adaptive control unit that intelligently calculates the optimal motor control command based on the identification results, motor status, and preset curves; a motor drive unit that executes the command; and a safety protection unit that monitors and triggers protection. By adopting the above technical solution, this application can avoid motor overload, food splashing, and uneven grinding, thereby improving processing efficiency, uniformity, and safety.

The invention provides a method and device for an intelligent computing cloud platform to call a self-constructed knowledge base through computing power, and relates to the technical field of intelligent computing cloud platforms and computing power infrastructure, the method comprises the following steps: carrying out content analysis on an obtained file to obtain first content information; generating first summary information of the file according to the first content information, and extracting to obtain structured metadata information of the file; respectively storing the first summary information and the structured metadata information into a first database and a second database of a knowledge base, and establishing a retrieval list of the first summary information and the structured metadata information; under the condition that query information is received, to-be-called information stored in the knowledge base is called through the retrieval list, and the to-be-called information comprises at least one of the first summary information and the structured metadata information. In this way, the calling effect of the knowledge base is enhanced.

The application provides a kind of intelligent computing resource scheduling terminal of vehicle cloud cooperation, is deployed in edge cloud node, including full source data sensing and interactive interface, cognitive front-end processing, strategy reasoning decision, computing resource arrangement instruction distribution and performance monitoring and reward calculation module, can realize with scheduling task description generation and receive candidate computing node's node state report and maintain the resource index table of node, node matching feature extraction, target node decision, instruction issuing and execution feedback whole process operation.The application faces intelligent transportation, autonomous driving and other cross-domain collaborative tasks, and for the decision-making uncertainty area formed by the lack of priori discriminability of the execution side selection executed by the vehicle end / cloud end, through the priori matching evaluation of cognitive front-end and strategy reasoning cooperation, the decision-making computing burden of high-dimensional multi-source state can be reduced, and the risk of improper execution side selection can be reduced, so as to improve the computing real-time performance, success rate and system operation safety of vehicle cloud cooperation task.

The invention relates to an intelligent calculation platform for a photovoltaic support, and the platform comprises an interaction module which is used for enabling a user to input basic information needed by the design calculation of the photovoltaic support and a design standard needed to be adopted, and displaying a design calculation result of the photovoltaic support to the user; and the data calculation processing module is used for carrying out mechanical and cost single modeling analysis calculation on the photovoltaic support according to the basic information and the design standard, or carrying out iterative optimization design calculation so as to obtain a design calculation result of the photovoltaic support. According to the method, the design calculation of the photovoltaic bracket can be quickly performed, so that the design scheme of the photovoltaic bracket can be quickly obtained, the design time can be shortened, and the design efficiency can be improved.

The invention provides a method and device for expanding a training data set through computing power of an intelligent computing center, and relates to the technical field of intelligent computing centers, intelligent computing centers, computing power infrastructures and automobiles, and the method comprises the steps: obtaining a first training data set, an abnormal identifier inputted by a user, and size information corresponding to the abnormal identifier; identifying the first picture according to the size information to obtain a blank area of the first picture; generating a second picture based on the blank area and the abnormal identifier; and in the first training data set, replacing the first picture with the second picture to obtain a second training data set. Through the above steps, the second picture which contains the abnormal identifier and is different from the first picture can be continuously generated, and different types of abnormal data can be automatically amplified in different detection scenes until the data volume of the second training data set can support the training of the image recognition model so as to improve the training effect of the model.

The invention discloses a generalized quantitative perception training method and system based on knowledge feature distillation, and belongs to the technical field of intelligent computing. The method comprises the following steps: constructing a full-precision floating-point teacher model and a mixed bit width quantification student model based on a neural network architecture and a target task scene; carrying out pre-training on the full-precision floating point teacher model; taking the pre-trained full-precision floating-point teacher model as a knowledge transmission subject and the mixed bit width quantification student model as a knowledge receiving subject to carry out quantification perception training; the quantitative perception training is integrated with an attention feature fusion module and a multi-level context knowledge module; integrating a knowledge feature distillation strategy in the quantitative perception training process to realize knowledge extraction, migration and fusion of multilayer features; and a knowledge distillation loss function is calculated, the knowledge distillation loss function is taken as an optimization target, training optimization of the mixed bit width quantification student model is completed, and universal quantification perception training is realized.

The invention discloses a fusion calculation method of a quantum classical hybrid calculation system and a related device, and the method comprises the steps: carrying out the splitting processing of a task request applied by the quantum classical hybrid calculation system, so as to obtain calculation subtasks independently completed by quantum calculation, classical calculation and intelligent calculation; in response to the split calculation sub-task type, configuring the heterogeneous calculation resources to the corresponding calculation sub-tasks; the heterogeneous computing resources comprise quantum computing resources, classical computing resources and intelligent computing resources; the computing resources are evaluated and analyzed according to the resource evaluation and intelligent analysis module to obtain a corresponding scheduling decision basis; inserting the calculation subtasks into the corresponding task queues, and scheduling the task queues based on the scheduling decision basis, the user-defined scheduling strategy and the first linkage relationship of the calculation subtasks; according to the method, a classical and efficient fusion calculation technical scheme of quantum is provided, and an efficient solution is provided for a complex calculation problem through computing power complementation.

This invention provides an intelligent computational method for predicting effector proteins in the secretory system of Gram-negative bacteria, comprising: S1: collection and preprocessing of protein sequence datasets; S2: model construction; S3: model training; and S4: model prediction. This invention has advantages such as high prediction accuracy and good anti-interference performance.

This invention provides a method and apparatus for measuring computing power on an intelligent computing cloud platform based on GPU computing power consumption and inference token values. It relates to the fields of intelligent computing centers, smart computing centers, computing infrastructure, and smart computing cloud technologies, and includes the following steps: Step S1, collecting task data, GPU runtime data, and model and engine parameters during inference on the intelligent computing cloud platform; Step S2, extracting first feature information, second feature information, and third feature information from the task data, GPU runtime data, and model and engine parameters; Step S3, performing token fitting computing power calculation based on the first feature information, GPU model, and third feature information to obtain a first computing power consumption value; Step S4, performing GPU computing power consumption benchmark calculation based on the first, second, and third feature information to obtain a second computing power consumption value; Step S5, determining a reliable computing power consumption value based on the first and second computing power consumption values. This invention can significantly improve the accuracy of computing power measurement on intelligent computing center cloud platforms.

The invention provides a computing power resource model test method and device of an intelligent computing center, and relates to the technical field of computing power infrastructure, the method comprises the following steps: determining a target computing power resource associated with a test model, the target computing power resource being a computing power resource for the test model to execute a training task and / or a reasoning task; the target computing power resource is tested based on gradient increasing preset data, the average test bandwidth of the target computing power resource is calculated, and the average test bandwidth is used for representing the communication performance of the test model. The communication performance of the target computing power resource under different data scales is quantified, and the communication performance of the test model is represented by the average test bandwidth. Effective performance testing of the model is achieved, the situation that the computing power resource operation efficiency of an intelligent computing center is reduced due to the fact that communication bottlenecks are found only in actual application is avoided, the gradient increasing preset data cover multiple transmission scenes, and the reliability of the model testing result is improved.

The invention provides a large-scale parallel sampling environment construction method and device for reinforcement learning of an intelligent computing cloud platform through computing power, and the method comprises the steps: obtaining a plurality of physical disks of a plurality of computing power nodes of the intelligent computing cloud platform, and constructing the plurality of physical disks into a bottom storage pool through an RAID technology, storing the basic mirror image of the operating system in a qcow2 format, and further generating a plurality of virtual machine instances and running the virtual machine instances; packaging the QEMU process and the running environment of the QEMU process into a container, and performing distributed scheduling on the container according to the disk topological relation among the plurality of bottom storage pools on the plurality of computing power nodes and the real-time I / O load condition. Therefore, the construction of a large-scale parallel sampling environment of reinforcement learning can be realized by utilizing the computing power of an intelligent computing cloud platform and integrating storage optimization, hardware virtualization and container arrangement technologies.