28 results about "Local agent" patented technology

The invention provides a multi-type micro-grid cross-layer collaborative scheduling method based on multi-agent reinforcement learning, and relates to the technical field of power system and micro-grid scheduling, and the method comprises the steps: firstly constructing a multi-type micro-grid layered collaborative architecture, a multi-agent reinforcement learning scheduling model, and a centralized coordination layer where global agents are deployed in a platform layer; and the local agents are deployed at each distributed micro-grid node of the platform layer, a cross-layer cooperative training mode is adopted to train the model, and finally, cooperative decision results of the global agents and the local agents are integrated to generate a global-local cooperative scheduling scheme. According to the method, global optimization and local flexibility can be considered, and the hierarchical multi-agent architecture adopts a multi-type micro-grid hierarchical collaborative architecture, so that the method can be suitable for multi-type micro-grid operation scheduling of a novel power system, and multi-type micro-grid collaborative scheduling considering global optimization and local flexibility is realized.

The invention provides a service debugging method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of service debugging. Initiating a debugging request for a target service instance based on a local agent, and obtaining a short-term security credential by the local agent; wherein the short-term security credential comprises the target service instance identifier; establishing, by the local agent, an encrypted communication channel with a remote agent using the short-term security credential; the remote agent extracts the target service instance identifier from the short-term security credential and obtains a real-time network address of the target service instance; and sending, by the remote agent, the received debugging traffic to the real-time network address of the target service instance. Identity authentication and authorization are realized through a short-term security certificate, and security and credibility of a debugging request are ensured; the security risk in the debugging process is reduced, and the flexibility and efficiency of cross-network environment debugging are enhanced.

The invention discloses a one-way off-line security intelligent decision center system with local deployment as a main and external network service as an auxiliary and a control method, and belongs to the technical field of artificial intelligence security and edge intelligent decision. A three-level security domain physical and logic dual isolation architecture is adopted, extranet multi-model intelligent security enablement is performed on a local core decision domain through a physical one-way import channel, the local core decision domain completes reasoning calculation in combination with an enablement result in a non-extranet interaction state, a top-layer task scheme is generated, and a top-layer task scheme is generated. Through a multi-mode man-machine interaction mode, after human rigid authorization, issuing to an intranet equipment scheduling domain for execution, and feeding back a state in real time to form a full-link closed loop; the method realizes external network multi-model intelligent security enabling, local agent professional decision and rigid closed loop dominated by human beings in the whole process, aims to solve the problems of cloud dependence, data leakage, out-of-control AI, high cost, poor adaptability and the like, and is suitable for families, industries, confidential scenes, field non-support operation and other scenes.

The invention discloses a cloud and edge integrated industrial agent distributed task execution method and system. The method comprises the following steps: constructing a global agent at a cloud end, and deploying a local agent at an edge device; the global agent receives the production task and decomposes the production task into sub-tasks; sub-tasks are dynamically allocated based on the real-time state and capability of each edge device; the local agent executes the distributed subtasks and keeps state synchronization with the global agent; and establishing a cloud and edge collaborative learning mechanism. The system comprises a task analysis module, a resource sensing module, a dynamic allocation module, a collaborative execution module and an incremental learning module. According to the method, the problems of insufficient collaboration of cloud computing and edge computing and rigid task allocation in a traditional industrial system are solved, and intelligent collaboration of cloud edge resources and efficient execution of tasks are realized.

This invention relates to the fields of artificial intelligence, multi-agent collaborative scheduling, and federated security management. Specifically, it is a federated capability management and unified scheduling system for multi-agent collaboration, comprising: a central scheduler receiving collaborative task requests across multiple local agent nodes and obtaining capability declaration information containing only the capability tags of the local agent nodes; constructing prompt word templates and generating an initial fine-grained task decomposition scheme; extracting cross-domain call links, statistically analyzing the frequency of cross-domain calls and data interaction dimensions, predicting the communication interruption probability of local agent nodes and comparing it with a security risk threshold; if the probability is lower than the threshold, a fine-grained scheme is adopted; if the probability is higher than or equal to the threshold, a minimum-privilege interaction coarse-grained scheduling scheme is generated and distributed; and local agent nodes executing sub-tasks and reporting their status to maintain core business continuity without exposing underlying private data.

Embodiments of the present application relate to the technical field of artificial intelligence, and provide a communication method, apparatus, and system. The method comprises: when it is determined that a first subnet has been established, an agent instance corresponding to a first terminal determining a first user plane network element for the first terminal and a second terminal to establish the first subnet; then sending a first request message to an agent instance corresponding to the second terminal, the first request message being used for deciding to create a local agent instance corresponding to the second terminal on a first artificial intelligence proxy function network element; and finally sending a first rule to the first user plane network element to instruct the first user plane network element to send, to the first artificial intelligence proxy function network element, data requiring multimodal conversion in the first subnet. On the basis of the first rule, the local agent instance corresponding to the second terminal enables the first terminal and the second terminal communicating with each other without relying on a same protocol to implement conversion between different protocols and different modalities between terminals by means of the local agent instance corresponding to the second terminal, thereby implementing cross-ecological interconnection of the terminals.

The embodiment of the invention provides a communication method, device and system, and relates to the technical field of artificial intelligence. The method comprises the following steps: after an agent instance corresponding to a first terminal determines to establish a first sub-network, determining a first user plane network element used for establishing the first sub-network by the first terminal and a second terminal; and sending a first request message to the agent instance corresponding to the second terminal, wherein the first request message is used for deciding to create a local agent instance corresponding to the second terminal on the first artificial intelligence proxy function network element. And finally, a first rule is sent to the first user plane network element, and the first user plane network element is indicated to send data needing multi-mode conversion in the first subnet to the first artificial intelligence proxy function network element. According to the embodiment of the invention, the local agent instance corresponding to the second terminal can realize the conversion of different protocols and different modes between the terminals through the local agent instance corresponding to the second terminal on the basis of the first rule under the condition that the first terminal and the second terminal do not depend on the same protocol for communication, so that the cross-ecological interconnection of the terminals is realized.

The application discloses a kind of dynamic load balancing management methods, equipment and medium based on AI intelligent agent, method includes: calculating node correlation coefficient, constructs GCN-GRU hybrid load prediction model, predicts the space-time feature matrix of future node, obtains the predicted node load data;Calculate the normalization weight of each load data, load comprehensive value and load balancing degree;Local agent, coordination agent are deployed, and local agent state space vector is constructed;The mapping relationship of local policy network is constructed using local agent state space vector, and the action space vector and migration amount of node execution load migration action are generated, global value is calculated, and the parameters of local policy network are optimized based on global value, and the optimal action space vector and corresponding migration amount are generated to carry out load balancing migration.Computer equipment includes processor and memory;Computer readable storage medium stores at least one instruction, at least one program, code set or instruction set.

The application discloses a kind of cross-regional virtual power plant collaborative scheduling method, equipment, medium and product, it is related to data processing field.The method includes collecting the real-time characteristic data such as time-space position, output / demand prediction of each distributed energy resource and load, determines each characteristic dimension dynamic weight based on the global optimization target and data of current scheduling period;According to dynamic weight and real-time data, generate dynamic resource cluster division instruction containing member list and collaborative constraint condition, and send initial cross-regional collaborative scheduling instruction containing net exchange power target value and the like and compensation price signal to each dynamic resource cluster local agent;After receiving the aggregated response boundary information returned by local agent, update instruction and signal, obtain target collaborative scheduling instruction, finally sent to each dynamic resource cluster execution, realize the effective control of cross-regional virtual power plant resource.The application can alleviate the problem that the adaptability of cross-regional virtual power plant collaborative scheduling instruction and actual resource capacity is not enough.

A system for networking and computing is provided herein which can include one or more processors and a memory storing instructions that cause the system to receive a request for a SFC in a network having a plurality of data centers arranged in clusters, wherein the SFC includes one or more VNFs located in the data centers in which packets traverse for various services, provide details of the request to one or more local agents each covering a cluster, and is configured to determine provisioning aspects of the SFC in the corresponding cluster, and receive the provisioning aspects for the SFC from the one or more local agents.

The invention discloses a network slicing management method based on a large language model and a related device, and relates to the technical field of network slicing, and the method comprises the steps: calling a global agent to receive a network slicing demand in a natural language format, and obtaining a demand intention; calling a global agent to interact with the large language model based on the demand intention to obtain a slicing request; calling a global agent to obtain a resource state in a current network; calling a large language model to determine a global resource allocation strategy based on the slicing request and a resource state in a current network, obtaining a resource allocation strategy of each sub-region, and obtaining a region slicing request of each sub-region; calling a local agent to interact with the large language model based on the region slicing request of each sub-region to obtain an operation instruction of network slicing operation performed by each sub-region; and executing the operation instruction to perform network slicing operation on each sub-region. The purpose of improving the management efficiency of the network slices is achieved based on the large language model.

The invention provides an engine system based on integrated privacy federated computing capability, which comprises the following steps that: a federated computing coordinator receives a joint computing task, splits the joint computing task into a plurality of sub-queries, and issues the sub-queries to a deployed local agent process; the local agent process executes the sub-query to obtain a local intermediate result, performs secrecy processing on the local intermediate result based on the privacy computing protocol to obtain a secrecy intermediate result, and sends the secrecy intermediate result to the federal computing coordinator; the federal computing coordinator adopts a federal aggregation algorithm to aggregate the secret intermediate result to obtain a global computing result, and sends the global computing result to a version control system; the block chain module obtains aggregation operation information, generates an operation abstract based on the aggregation operation information and writes the operation abstract into a block chain account book; and the version control system generates new version data according to a global calculation result, and records a data version evolution path based on a content addressing storage structure and a copy-on-write mechanism. According to the method and the device, privacy protection cooperative calculation of distributed spatio-temporal data can be realized, and centralized processing risks are avoided.

A computer-implemented method comprising: receiving an electronic message at an endpoint device associated with an organization; analyzing the electronic message by a local agent executed on the endpoint device; classifying the electronic message into one of a plurality of trust levels by the local agent; generating, by the local agent, a real-time alert or action at the endpoint device based on the classification; transmitting, by the local agent, metadata associated with the electronic message and its classification to a centralized machine learning engine; and updating, by the centralized machine learning engine, one or more machine learning models based on aggregated metadata received from a plurality of endpoint devices, wherein the one or more updated machine learning models are used to improve subsequent classification of electronic messages at the endpoint devices.

The invention discloses a smart power grid adaptive control system and method, and relates to the technical field of smart power grid operation control, and the system comprises a physical sensing layer, an edge calculation layer, a cloud super-simulation digital twin platform and a communication network. According to the method, through a central coordination agent in a digital twinborn platform, a multi-agent reinforcement learning algorithm is adopted to carry out cooperative training on a local agent. According to the method, a super-simulation digital twinborn model integrating geometry, physics, behaviors and rules in multiple dimensions is constructed, and through a training normal form of deep fusion of a physical information neural network and a generative adversarial network, the model has both a physical rule following capability and an uncertainty deduction capability; and safe, rapid and adaptive optimization of power grid control in a large-scale and uncertain environment is realized.

The invention discloses a multi-agent-based power distribution network flexible load coordination control method, and relates to the technical field of power distribution network intelligent control and optimization, and the method comprises the following steps: S1, constructing a multi-agent system for power distribution network flexible load coordination control, the multi-agent system comprises a local agent, a cluster coordination agent and a system coordination agent; the local intelligent agents are deployed on the flexible load devices and used for collecting local state information and executing control instructions. According to the multi-agent-based power distribution network flexible load coordination control method, accurate coordination control of diversified flexible loads in a power distribution network is realized by constructing a layered multi-agent system and combining flexible load dynamic modeling and a multi-agent game coordination mechanism. The system can effectively improve the adaptability of the power distribution network to load fluctuation and new energy output change, improve the operation state of the power grid, and enhance the flexibility of the system to cope with different operation conditions.

The invention discloses a multi-agent dynamic defense game method and system based on federal reinforcement learning, and relates to the technical field of information security. The method comprises the following steps: modeling a distributed network defense scene into a multi-agent-based partially observable Markov environment, defining a seven-tuple environment model comprising an agent set, a global state space, a local observation space, an action space, a state transfer function, an observation function and a reward function, and designing a layered mixed reward function; and constructing a hierarchical collaborative defense architecture based on federal reinforcement learning, and executing a closed-loop online dynamic defense process based on the defense architecture. Distributed training and real-time decision are executed through the local agent layer, intra-group model encryption aggregation is performed through the edge layer, and global meta-strategy generation and dynamic role allocation are completed through the central layer. According to the architecture, communication overhead is reduced, single-point failure is avoided, original data privacy of each node is effectively protected, and efficient adaptive strategy learning and dynamic game collaboration can be realized.

The invention relates to the technical field of coal mine intellectualization and artificial intelligence, and discloses a cross-mine multi-agent collaboration and semantic alignment system based on a large language model, which comprises a local knowledge extraction agent, a semantic alignment agent and a collaborative decision agent. The local agent accesses underground unstructured data, converts the underground unstructured data into a structured causal triple by using a coal mine large model, and screens and uploads the structured causal triple based on confidence; performing vector coding and similarity calculation on the triple by the semantic alignment agent, and constructing a coal mine domain ontology map; the collaborative decision-making agent converts an abnormal event into a natural language problem, initiates semantic inquiry in a cross-mine manner to obtain a historical strategy, and generates a safety suggestion in combination with a graph. Through cooperation of a large model and multiple agents, efficient alignment and reuse of causal knowledge are realized under zero original data transmission, the problem of term isomerism is solved, and the accuracy of cross-mine safety decision is remarkably improved.

Methods and systems for version-control based provisioning and utilizing local agents in an information technology (IT) infrastructure system are disclosed. In embodiments a backend IT infrastructure system and a local system are provided and networked together. In embodiments the local provisioning system includes one or more of a local database, a local provisioning agent, and a version-control system. In embodiments, the backend IT infrastructure system is configured to receive a request to execute a run based on a configuration file including infrastructure as code instructions. In embodiments, the backend system is configured to submit one or more run tasks to the local provisioning agent. In embodiments the backend system is configured to output state files to the local database. In embodiments the version-control system is configured to manage the configuration file while the backend system is configured to monitor the file via the version-control system over a network tunnel.

This invention belongs to the field of intelligent operation and maintenance and fault tracing technology for power equipment, and relates to a method and system for fault tracing of converter transformers using a reinforcement learning dual-agent approach. The method includes: acquiring operation and maintenance information and fault information to construct a fault tracing knowledge graph containing entity sets and relation sets; acquiring fault tracing queries and encoding them into query representations to determine initial entities; determining a set of candidate relations by a global agent based on the query representations and current entities; performing graph attention weighting on the candidate entity sets associated with the candidate relations by a local agent and outputting the entity selection result; updating the inference trajectory and calculating soft rewards based on the matching degree between the inference trajectory, query representation, and target entity to update the dual-agent strategy; and outputting the inference trajectory as a tracing result containing root cause entities and evidence chain entity sequences when the termination condition is met. The technical solution of this application can improve the discrimination of inference path selection and ranking and enhance the stability of root cause localization.

Trust systems and methods are provided in a distributed architecture. In one implementation, a distributed system includes multiple Internet of Things (IoT) devices distributed throughout a network, wherein each IoT device is embedded with a local agent configured to perform processing and / or computing functionality. The distributed system further includes a backend entity, such as a device trust system, configured to manage the multiple IoT devices. In addition, the distributed system includes multiple Rendezvous Zone (RZ) proxy devices communicatively interposed at edge locations in the network between the backend entity and the multiple IoT devices. Each RZ proxy device is configured to perform trust and security functionality on behalf of one or more IoT devices of the multiple IoT devices.

The application is particularly a control method of an Internet of Things gateway fusing communication and AI computing power, and relates to the technical field of Internet of Things gateway control, comprising: constructing a machine-understandable structured function identity file and a dynamic digital gene model for a physical device; analyzing structured service requirements and fusing multi-source context information to generate a device and network joint execution strategy meeting QoS requirements through an AI decision model; and dynamically creating a virtual communication end point based on the execution strategy.In the application, the device structured capability declaration and the dynamic digital gene model realize the transformation of the device from passive response to active capability declaration; the device automatically pushes a standardized description file containing function fingerprints, capability vectors and the like after being powered on, so that the gateway can complete identification and adaptation without pre-configuration; in cooperation with the transparent adaptation characteristics of the virtual communication end point, a standard interface is simulated on the cloud and a local agent is simulated for the device, so that the device can be accessed without modifying the firmware.

Methods and systems for version-control based provisioning and utilizing local agents in an information technology (IT) infrastructure system are disclosed. In embodiments a backend IT infrastructure system and a local system are provided and networked together. In embodiments the local provisioning system includes one or more of a local database, a local provisioning agent, and a version-control system. In embodiments, the backend IT infrastructure system is configured to receive a request to execute a run based on a configuration file including infrastructure as code instructions. In embodiments, the backend system is configured to submit one or more run tasks the local provisioning agent. In embodiments the backend system is configured to output state files to the local database. In embodiments the version-control system is configured to manage the configuration file while the backend system is configured to monitor the file via the version-control system over a network tunnel.

The invention discloses a distributed mandatory access control method and system based on a chained proxy, and the control method comprises the steps: intercepting a security access event of a subject to an object through an LSM proxy module of a Linux kernel of a terminal host, collecting a security context to generate a first security request packet of a unique chained ID, and transmitting the first security request packet to a local proxy process; and the local agent process queries the cached security decision set, directly returns a decision result if the matching is hit, encapsulates a second security request packet and sends the second security request packet to the remote decision server if the matching is not hit, and the remote server generates a final decision and returns and executes the final decision. According to the embodiment provided by the invention, complete decoupling of the mandatory access control decision logic and the operating system kernel can be realized; through the remote decision server supporting cross hosts and cross security domains, the technical limitation that an existing mandatory access control scheme can only realize local decision is effectively broken through, the kernel mode processing overhead is reduced, and the security and availability are improved.

The invention discloses a dynamic load balancing management method and device based on an AI intelligent agent and a medium. The method comprises the steps of calculating a node correlation coefficient, constructing a GCN-GRU hybrid load prediction model, predicting a spatial-temporal characteristic matrix of future nodes, and obtaining predicted node load data; calculating a normalized weight, a load comprehensive value and a load balance degree of each type of load data; deploying a local agent and a coordination agent, and constructing a local agent state space vector; and constructing a mapping relation of the local strategy network by using the state space vector of the local agent, generating an action space vector and a migration quantity of a node executing a load migration action, calculating a global value, and optimizing parameters of the local strategy network based on the global value. And generating an optimal action space vector and a corresponding migration volume to carry out load balancing migration. The computer equipment comprises a processor and a memory; the computer readable storage medium stores at least one instruction, at least one program, a code set or an instruction set.

The invention relates to the technical field of network resource management, and relates to a cross-domain network resource arrangement method and system, a computer program product and electronic equipment. The method is applied to a resource arrangement system, the system comprises a global coordinator, domain managers registered by network domains in the global coordinator and local agents deployed under each domain manager, and state summary information of each network domain is calculated through the global coordinator based on an attention mechanism to obtain attention weights of each network domain; determining a cross-domain resource allocation strategy through a global coordinator according to the state summary information and the attention weight, and allocating the cross-domain resource allocation strategy to a domain manager; for each domain manager, carrying out intra-domain decision-making by taking the sub-policy as a constraint, obtaining an intra-domain resource allocation policy, and issuing the intra-domain resource allocation policy to a local agent for execution; based on the attention weight, the global coordinator is utilized to allocate the global performance reward as a domain level reward for the domain manager, and the domain level reward is allocated by the domain manager as a local reward for the local agent.

The application discloses an agent method, device and computer readable storage medium, wherein the agent method acquires transaction data of a user, the transaction data comprising a transaction request and private data, then sends the transaction request to a blockchain platform, the blockchain platform completes the transaction according to the transaction request and locally stores the private data, and sends the private data to a peer agent for transaction verification after the transaction is completed, thus, by using the tamper-proof characteristic of the blockchain technology, the credibility and traceability of the transaction are ensured, meanwhile, the private data is only transacted between the local agents of the two parties, and the private data is locally stored, that is, the private data will not be uploaded to the blockchain platform, thus effectively solving the problem that the private data of the blockchain platform is prone to leakage, the transaction security is high, the private data is effectively protected, the application is suitable for cross-blockchain platforms, and the private data does not need to be uploaded, thus the storage space occupied can be reduced.

A computer-implemented method for preventing loss of function in a local instance when there is a disturbance of a connection to a backend in a communication system that has a local instance having a local agent and a backend acting in part as a global agent, an agent designating a unit that is designed to achieve defined goals through autonomous behavior. The method includes, when there is a connection to the backend, the providing of the function by the global agent, which receives backend information from the backend during the providing of the function, and, if a connection to the backend is disturbed, the providing of the function by the local agent.

This application provides a server GPU computing power allocation method, system, and server, belonging to the field of server resource management, and solves the problems of low resource utilization and inflexible decision-making in traditional allocation methods. The method includes: a central agent constructing a directed weighted game graph of tasks and GPU clusters, and solving a global allocation strategy based on Nash equilibrium; a local agent allocating tasks to specific GPUs through reinforcement learning, achieving intelligent migration by combining dynamic thresholds and anomaly detection, and iteratively optimizing the strategy through federated learning. The system includes a central agent, local agents, GPU clusters, and an experience pool module, and the server runs the system to execute the method. The solution, through two-layer agent collaboration and multi-algorithm fusion, quantifies the task cluster adaptation relationship, dynamically adjusts the allocation strategy, significantly improves resource utilization and task processing efficiency, and enhances adaptability and reliability in complex scenarios.