20 results about "Task support" patented technology

To reduce a burden on a service providing side for dealing with a user.SOLUTION: A response task support device 10 receives input of desired information D2 indicating action contents desired by users. The dealing work support device 10 inputs the received desired information D2 to a dealing generation model 31 which is a learning model, and acquires dealing information D6 which is the dealing information D2 output by the dealing generation model 31 and indicates dealing contents on the service providing side for achieving the action contents indicated by the desired information D6. The dealing service support device 10 generates code information D6 for displaying the acquired dealing information D7, and outputs it to the user terminals 36. The clerk terminals 37 read the code information D7 displayed on the user terminals 36 and display the corresponding information D6.SELECTED DRAWING: Figure 1

The embodiment of the invention provides a resource operation task processing method, device and system, a storage medium and a program product, and relates to the technical field of cloud computing. In the embodiment of the invention, the resource operation task submitted by the user is arranged based on the arrangement relationship among various resource operation tasks supported by the target resource object and in combination with the task queue of the target resource object, and the resource operation task is submitted by the user in the arrangement process. Therefore, the resource operation tasks having the mutual exclusion type arrangement relationship with the existing resource operation tasks in the task queue are not enqueued, and the resource operation tasks having the non-mutual exclusion type arrangement relationship are enqueued. After enqueue, scheduling is carried out according to the non-exclusive arrangement relation between the tasks, the conflict problem between different resource operation tasks can be solved, and under the condition that reasonable arrangement scheduling is achieved, a user does not need to pay attention to the completion condition of the resource operation tasks and can submit the resource operation tasks at any time according to needs; and the efficiency and experience of resource operation on the cloud product by the user are improved.

The invention discloses a relay satellite multi-access resource multi-satellite multi-target scheduling method and system, and relates to the field of spaceflight testing, operation and control resource scheduling, and the method comprises the steps: obtaining a to-be-scheduled task set and constraint conditions, and the constraint conditions comprise intra-task constraints generated by limiting schemes and inter-task constraints coexisting by the limiting schemes; constructing a feasible scheme set meeting the constraint in the task according to the visible window and the beam transmission rate; executing a non-subtraction domain driven algorithm, and calculating a task exchange ratio to represent the capability of a task to unblock a frozen state; and constructing a non-reduced neighborhood according to a task exchange ratio, and realizing iterative optimization of a solution by releasing a high-exchange-ratio task and reallocating a free task in the neighborhood until a global optimal scheduling solution is obtained. According to the method, a task exchange ratio and non-subtractive neighborhood construction strategy is introduced, the resource conflict problem under multi-satellite multi-target complex constraints is effectively solved, constraint dynamic configuration based on a rule engine is supported, and the task guarantee capability and the system flexibility are improved.

The invention provides an intelligent dispatching method and system for airport cargo station forklifts based on a dynamic camping mechanism, and relates to the technical field of air freight, and the method comprises the steps: S1, inputting related information, and building a mapping relation rule; s2, responding to the transportation application, analyzing application information, determining a wagon balance according to a mapping relation rule, generating a transportation task, initializing the transportation task into an idle state, and storing the transportation task into a task pool; s3, for the online forklift terminals, idle tasks are screened from the task pool, and the tasks are pushed to the forklift terminals to be pre-occupied; s4, the forklift terminal starts task execution, the task state is updated to be in execution, and the task state is updated to be completed after the task is completed; s5, task state changes are tracked in real time through a distributed state machine, and new task allocation is automatically triggered when the tasks are completed; s6, automatically releasing the task with overtime pre-occupation back to the task pool; and for the task in execution, manual release is supported. According to the invention, the intelligent and self-adaptive scheduling process can be realized.

The invention relates to a task allocation method and device, a readable storage medium and a program product. The method comprises the steps that a to-be-executed file is obtained, a to-be-executed target task can be determined according to the to-be-executed file, under the condition that it is determined that the target task supports asynchronous execution, a specified task flow corresponding to the target task is determined, and flow configuration information of the specified task flow and running data of the target task are determined according to the to-be-executed file. And sending the flow configuration information and the operation data to the second processor, the flow configuration information and the operation data being used by the second processor to execute the target task on the specified task flow. By adopting the method, the task allocation flexibility can be improved.

The invention belongs to the technical field of mobile computing and intelligent software, and particularly relates to an interactive GUI (Graphical User Interface) intelligent agent system oriented to mobile equipment cross-application control. The system comprises a global task management module for receiving a natural language task instruction input by a user, disassembling a task through a global task planner in combination with installed application information in user equipment, and generating standardized subtasks; the application-level task execution module is used for refining the subtasks into atomic operations and cooperatively completing operation execution and user interaction through multiple agents; and the self-learning module is used for extracting key information and related knowledge in the task execution process after the task is completed, and updating the key information and the related knowledge to long-term memory. The method has good cross-application cooperation capability, high-robustness long-range task execution capability and interaction-based dynamic task correction capability, can be widely applied to intelligent assistants, automatic operation platforms and user task support systems in mobile terminals, and remarkably improves user operation efficiency and intelligent experience.

The invention discloses a GPU resource optimization method and system based on dynamic prediction and a topological thermodynamic diagram. The method comprises the steps that a server thermodynamic diagram is constructed, and a thermodynamic value of the server thermodynamic diagram is subjected to weighted calculation based on historical task characteristics and the number of node topology connection edges; on the basis of the to-be-scheduled task queue and the historical task data, predicting a future integrated GPU resource demand to generate a resource demand popularity map; based on the resource demand heat map and a dynamic reference value calculated based on a server thermodynamic diagram, GPU resource fragment risks are quantitatively identified; in response to the fragment risk, selecting an optimal optimization strategy; and executing the selected strategy, including dynamically migrating the tasks supporting the check points and / or performing resource scaling on the tasks supporting the elastic scaling, so as to realize lossless reconstruction of GPU resources. Through intelligent closed loop of prediction-decision-execution, resource fragments are actively resolved, and the resource utilization rate and task scheduling efficiency of a large-scale GPU cluster are remarkably improved.

This invention discloses a multi-dimensional performance evaluation method and simulation system for large-scale distributed space systems, implemented using low Earth orbit satellites. The system includes end nodes and computing satellite nodes. The method specifically includes: S1. Constructing a distributed space system performance evaluation index system, which includes heterogeneous task support, massive task support, task real-time performance, resource utilization, load balancing, and application business indicators; S2. Constructing a relative matrix based on the analytic hierarchy process (AHP); S3. Obtaining a comprehensive score using a multi-dimensional fuzzy evaluation method; and S4. Designing the architecture and functional modules of the distributed space system simulation system. This invention, through kinematic simulation of satellite nodes, configuration and detection of node resources, design of various simulation scenarios and schemes, and multi-index evaluation of scenarios, can meet the performance evaluation and optimization needs of distributed space systems in complex application scenarios.

The application provides a scientific research auxiliary system and method based on port and shipping big data AI intelligent agents, which comprises collecting port and shipping multi-modal heterogeneous historical data, constructing meta-training tasks containing support sets and query sets according to scientific research tasks; extracting data feature mutual information inside each meta-training task to generate a task graph, encoding into a task feature vector and clustering into K task clusters; iteratively training the meta-tasks in each cluster to obtain the initial model parameters of the cluster and the plasticity mask generated by the gradient variance; inputting a new task, extracting the task features and calculating the affinity weight of each cluster, weighting and fusing the cluster parameters and the mask to complete multi-task knowledge transfer; modulating the basic mask into a dynamic mask that adapts to the distribution of the new task through a super network; fine-tuning the fusion parameters with the support set of the new task, and multiplying the gradient and the dynamic mask element by element to protect the general core knowledge; the fine-tuned model processes the query set to output the results, supporting intelligent optimization and deployment of port and shipping resources.

The invention discloses an intelligent agent task planning reliability verification method, and relates to the technical field of data fusion processing, and the method comprises the following steps: collecting windproof and flood prevention multi-source scene data, recognizing abnormity, and carrying out the screening to form a qualified scene data set; standardizing the interpolated data and adding an affiliation label; generating task support data; screening and generating a task support data set; matching corresponding task nodes, recording a mapping relation log and generating a list of to-be-executed task nodes; assigning task nodes to the agents; labeling task execution results and forming an agent task execution state table; integrating the quantitative index and the qualitative problem feedback to form an execution feedback data set; associating the problem phenomenon with the process link data, carrying out optimization and recording to form an optimization measure log; three types of evaluation indexes are calculated, and a comprehensive reliability score is calculated, fed back and output. The method can effectively improve the condition that the intelligent agent task planning reliability is difficult to effectively quantify and verify in the prior art.

The invention provides a task management method which is applied to task management. The method includes: acquiring first description information about a first task, the first description information including at least one of the following items: a time validity condition indicating when the first task is valid; a spatial validity condition indicating where the first task is valid; the reusability indication is used for indicating whether the first task can be used as a computing block (CB) in another task or not; the application indication indicates at least one application supported by the first task; describing interface information of an interface through which the first task can be accessed; the task intention information describes a task intention of the first task; the task specification describes intercommunication logic among one or more CBs of the first task, and the one or more CBs are used for realizing the task intention described by the task intention information; or at least one task parameter to be specified; a first request is sent to a task open function (MEF) network entity, where the first request includes the first description information about the first task, and the first request indicates a management related to the first task.

The invention relates to the technical field of real-time operating systems, and discloses a method for creating real-time operating system tasks, which comprises the following steps: determining a real-time task scheduling mode; the real-time task scheduling mode is any one of task scheduling modes supported by the real-time operating system; and creating a task according to the real-time task scheduling mode, and obtaining a real-time task supporting the task scheduling mode. When the task is created, the real-time task scheduling mode used by the created real-time task is determined in advance, so that two real-time task scheduling modes, namely a preemptive mode or a round-robin mode according to time slices, can exist in a real-time operating system at the same time, and a more flexible task creating mode is provided to realize better allocation of CPU (Central Processing Unit) resources. The invention further discloses a real-time operating system task creating device, electronic equipment and a storage medium.

We provide the system. [Solution] means of collecting information, A means of constructing a machine learning model using collected information, Support means for providing task assistance based on constructed machine learning models, An analytical method for analyzing and improving the execution data of task support, A condition diagnostic means for acquiring information from mechanical devices and identifying abnormalities, A means for proposing maintenance and management methods based on the condition diagnosis results, A system that includes this.

The application discloses a three-stage task automatic generation method and system of a general intelligent agent, in a task variable setting stage: a diversified environment is formed by changing environment parameters; in a task interaction stage: a task execution-task generation cycle is entered, infinite task generation is realized, and a task table is updated; in a task evolution stage: existing tasks are reused and recombined to generate new tasks, the generated new tasks are transported to the task interaction stage for task execution, and the same type of nodes with different attributes in a task graph are replaced to recombine and generate new tasks. The application can fully utilize existing task data, generate infinite tasks for intelligent agents, and support task generation in new scenes and dynamic scenes; the application solves the problems that the prior art cannot generate tasks for new scenes and cannot deploy a task generation and execution cycle link, and reduces the dependence of intelligent agents on a large number of environment parameters.

The invention discloses a universal image coding method based on a multi-expert model, system equipment and a medium, which are corresponding schemes, which can give consideration to diversified application requirements and optimal performance of various tasks, remarkably reduce the development and deployment cost supported by new tasks, and improve the development and deployment efficiency. Specifically, various machine vision and human vision related tasks are uniformly supported by a single framework, and diversified application requirements are considered; moreover, efficient knowledge sharing and decoupling are realized through an expert model mechanism, and the optimal performance of various tasks is considered; meanwhile, flexible task adaptive capacity and an efficient incremental expansion mechanism are provided, and development and deployment cost supported by new tasks is remarkably reduced; in addition, the requirements of high universality, high performance and low maintenance of current and future intelligent vision applications on the image coding system are met.

This invention discloses a method, apparatus, device, and computer-readable storage medium for industrial defect visual detection based on hybrid meta-learning, belonging to the field of computer vision technology. The method includes: acquiring a target query image as input to a pre-set model; extracting target feature vectors using dedicated model parameters, and calculating their similarity score with a baseline feature vector in a feature cache to determine the defect category. Model construction includes: performing meta-training on a deep network to obtain meta-model parameters; self-supervised fine-tuning of the meta-model parameters based on a new task support set to obtain dedicated model parameters; and re-extracting support set features using the dedicated model parameters to generate a baseline feature vector and storing it in a cache. This invention effectively solves the problems of low detection accuracy and low deployment efficiency in scenarios with few samples and across different domains by combining meta-training with self-supervised fine-tuning and feature refresh mechanisms.

The application discloses a GPU resource optimization method and system based on dynamic prediction and topology heat map. The method comprises the following steps: constructing a server heat map, the heat value of which is calculated based on historical task characteristics and node topology connection edge quantity weighting; predicting future integrated GPU resource demand based on a to-be-scheduled task queue and historical task data to generate a resource demand heat map; quantitatively identifying GPU resource fragmentation risks based on the resource demand heat map and a dynamic reference value calculated based on the server heat map; selecting an optimal optimization strategy in response to the fragmentation risks; and executing the selected strategy, including dynamically migrating tasks supporting checkpoints and / or scaling resources of tasks supporting elastic scaling, to realize lossless reconstruction of GPU resources. Through the intelligent closed loop of prediction-decision-execution, the application actively resolves resource fragmentation, and significantly improves the resource utilization and task scheduling efficiency of a large-scale GPU cluster.

We provide the system. [Solution] A sensor means that detects user behavior and acquires data, A processing means for inferring the user's intent based on the analysis of the aforementioned data, A means of providing advice to users, A household task support system that includes this.

The invention provides an intelligent right steward implementation method and device based on AI capability. Through a keyword tag knowledge base + AI large model generation dynamic Prompt, cross-scene demand accurate matching is realized; tasks are dispatched in combination with an agent arrangement model, multi-scene right and interest combination recommendation is supported, and meanwhile, requirements are determined by actively asking a user; based on an operator flow monitoring technology, after user authorization, a currently used APP is judged in real time, data are synchronized to a background to match rights and interests, the data are actively pushed through a notification bar, the current scene requirement of the user is directly met, and the problem that recommendation cannot respond to the real-time requirement is solved; the whole process of'interface capture, analysis and automatic operation 'is driven by AI, manual intervention of a user is not needed, a complete link closed loop of'rights and interests recommendation, ordering and automatic use' is achieved, the user operation path is greatly shortened, and the pain point of'low rights and interests use efficiency 'is solved.

To improve productivity by reducing a work burden of a staff working in a restaurant.SOLUTION: A business support device capable of communicating with an earphone microphone terminal worn by a staff includes: a determination unit configured to determine whether or not to output voice task instruction data in a predetermined task based on output time data indicating a time when the voice task instruction data is output to the earphone microphone terminal; an extraction unit configured to extract a corresponding staff by referring to task-in-charge data in which a task of the staff, a use language, and an ID of the earphone microphone terminal are associated with each other; an acquisition unit configured to acquire the voice task instruction data based on the task of the extracted staff and the use language of the extracted staff; and a distribution unit configured to distribute the acquired voice task instruction data to the earphone microphone terminal of the extracted staff.SELECTED DRAWING: Figure 1