91 results about "Hybrid Scheduling" patented technology

A system for multi-stage planning of construction processes and resource allocation, consisting of: a central planning engine configured to receive input data, including architectural design models, structural constraints, procurement schedules, and historical performance indicators; a task decomposition processor that is operationally connected to the central planning engine and configured to generate a hierarchical construction task graph by decomposing macro-level construction milestones into mid-level and micro-level subtasks, with each subtask having time estimates, location identifiers, resource requirements, and mutual dependencies; a hybrid planning processing unit configured to resolve time and resource constraints across the entire task diagram; a resource coordination controller that is operationally connected to the central planning engine, wherein the resource coordination controller includes a real-time database of work units, machines and material stocks, each resource being tagged with attributes such as availability, usage history, operating status and spatial location; a multitude of distributed execution units distributed across the construction zones, each distributed execution unit comprising an embedded controller, sensor interfaces, task status processing logic, and communication circuitry, each distributed execution unit being configured to receive planning instructions from the central planning machine, execute localized control logic for task confirmation and resource activation, and transmit task execution data back to the central planning machine; an adaptive conflict resolution processing unit that is operationally connected to the central planning engine and configured to detect conflicts in task execution or resource conflicts, simulate alternative task-resource allocation scenarios using a real-time multi-agent model, and autonomously update the task graph with revised task sequences and resource allocations; and A dashboard for the construction process, configured to visualize task progress, deviations from the planned schedule, and resource efficiency metrics, with the dashboard also being able to receive manual override inputs or approve automated conflict resolution proposals generated by the adaptive conflict resolution module.

The invention relates to the technical field of smart community data management, in particular to a smart community metadata interaction method and system based on an edge computing framework, and the method comprises the steps: carrying out the dynamic storage of multi-dimensional data through a hierarchical rolling cache; generating a uniform resource vector through sequential progressive mapping; community semantic tags are added for the uniform resource vectors through extensible tag slots, connection is established according to association rules, and a resource semantic graph oriented to community services is constructed; receiving a resident task request, constructing a multi-factor weighted scoring engine based on the resource semantic graph, and generating a task data packet with a priority label; a two-channel hybrid scheduling network is constructed, a first channel outputs resource availability and node association degree, and a second channel outputs time sequence characteristics and load prediction parameters; and fusing dual-channel output to obtain task-node correlation and generate an optimal matching strategy. According to the method, task collaboration and resource adaptive management are realized through semantic resource modeling and hybrid scheduling.

The invention relates to the technical field of artificial intelligence and edge computing, in particular to an end-side cloud collaboration system based on a hybrid scheduling strategy. The end-side cloud cooperation system based on the hybrid scheduling strategy comprises a model segmentation module, a state sensing module, a reasoning scheduling module, a model compression and deployment module, an edge optimization engine and a communication synchronization module. According to the end-side cloud cooperation system based on the hybrid scheduling strategy, elastic deployment and dynamic reasoning of a complex model in a multilayer heterogeneous environment are supported, the system state can be sensed in real time, a scheduling path can be optimized, and the robustness of the system is improved; and in combination with model compression and edge operator optimization, the reasoning efficiency is remarkably improved, the communication load is reduced, and then low-delay cooperation between the terminal and the cloud is ensured.

The invention discloses a medical logistics robot task allocation system based on time division multiplexing and dynamic scheduling, which comprises a cloud platform, a plurality of robots and a cabin body replacement station, and is characterized in that the cloud platform adopts a conflict resolution strategy based on priority and time windows on the basis of task time division and class division allocation logic of a mixed AI scheduling algorithm; tasks are dynamically and intelligently distributed according to hospital business laws and real-time states of the robots; the robot communicates with the cloud platform through the communication equipment, various tasks issued by the cloud platform are executed by carrying various cabins, and the tasks comprise material distribution, cleaning, disinfection and polling; and the cabin body replacement station is used for providing cabin body replacement equipment and assisting the robot in replacing the cabin body. According to the invention, tasks can be dynamically and intelligently allocated according to hospital business rules and real-time states of the robots, so that the task processing efficiency and the resource utilization rate are remarkably improved, and the hospital logistics operation cost is reduced.

The invention discloses a hospital multi-scene transportation robot hybrid scheduling method based on dynamic priority, and belongs to the technical field of robot control and scheduling. The method comprises the following steps: acquiring hospital dynamic environment data; receiving and analyzing the multi-mode instruction to form task parameters; preliminarily grading according to task parameters and preset rules, and dynamically adjusting task priorities in combination with a dynamic map and the current state of the robot; based on the adjusted task priority and robot type, distributing robots and generating an initial path considering distance and people flow density; the robot is driven to track along the path trajectory, and obstacle behaviors are predicted in real time and dynamically avoided; decomposing a transportation task into task chains, and when the electric quantity is lower than a threshold value, deciding whether to insert a charging task in combination with the priority; and after the task is completed, prompting and transmitting feedback data. According to the invention, dynamic priority, behavior prediction and hybrid scheduling are adopted, intelligent distribution and path optimization of transportation tasks are realized, and the efficiency and safety of internal logistics of hospitals are improved.

The invention discloses a distributed real-time target detection tracking method and system based on dynamic perception and adaptive regulation and control, and belongs to the technical field of intelligent monitoring and computer vision. The method comprises the following steps: firstly, calculating the weight of each camera in real time by using the time, position and map information of an event and camera network topology, generating an optimal search path by predicting a target motion track, and decomposing a global tracking task into multiple levels of subtasks; then, according to the task complexity and the node resource state, adopting a hybrid scheduling strategy to realize short-term container elastic expansion and contraction and long-term resource prediction, and ensuring efficient and stable resource scheduling; then, through multi-modal feature fusion and spatio-temporal joint modeling, cross-camera video data alignment, target association and continuous tracking are realized; and finally, performing multi-source data fusion and risk reasoning by using the large model, and outputting graded early warning and three-dimensional visual decision information. According to the method, the tracking accuracy can be improved, and resource utilization optimization and real-time response are realized.

The invention provides a graph card-based hybrid scheduling rendering method and system, and the method comprises the steps: dividing a frame buffer area, binding the divided frame buffer area with a task, completing the rendering of each card through a hybrid task scheduling strategy, and finally synthesizing and outputting a complete graph; a frame buffer area of a rendering graph is divided into a plurality of graph cards according to display criticality, each graph card is bound with an independent rendering task and a video memory area, a unique task division and resource allocation mode is formed, and the mode is different from a traditional overall rendering mode; and in combination with a hybrid scheduling strategy of high-priority task preemptive scheduling and dynamic card task parallel rendering, the problems of rendering task priority conflict and resource utilization are effectively solved.

The invention discloses a remote sensing rapid diffusion super-resolution reconstruction method based on adaptive scene perception, belongs to the technical field of mode recognition, and is used for super-resolution reconstruction of remote sensing images. The method comprises the following steps: preprocessing a high-resolution image, and calculating a residual error between the high-resolution image and a bicubic up-sampling result as a training target of a diffusion model; performing wavelet transformation on the bicubic up-sampling image through a frequency domain enhancement module, extracting frequency domain features, and fusing the frequency domain features with original spatial domain features to serve as condition input of a model; inputting the fused features into a conditional diffusion model, carrying out adaptive rapid sampling by adopting a linear-cosine hybrid scheduling strategy, and completing high-quality reconstruction within 20 steps; the feature expression capability is enhanced through an overlapping and crossing attention module integrated in the U-Net denoising network; and finally, fusing the residual error reconstruction module and the bicubic up-sampling image to obtain a final super-resolution reconstruction result. Experiments carried out on two common data sets (Potsdam and Toronto) show that the method of the invention significantly improves the sampling efficiency while maintaining the reconstruction quality, and is superior to an existing remote sensing image super-resolution method.

The invention relates to the field of container and virtual machine mixed scheduling, in particular to a container and virtual machine mixed scheduling method and device based on a domestic AI server, and the method comprises the steps: judging whether the server is used for operating a virtual machine or a container, if the server is used for operating the virtual machine, operating the virtual machine and executing corresponding operation, and if the server is used for operating the container, executing corresponding operation; if yes, the container is operated, corresponding operation is executed, a Volcano scheduling framework is integrated, and a Gang scheduling strategy and a BinPack scheduling strategy are defined; and a resource pooling management and dynamic adjustment mechanism is adopted, so that the resource utilization rate and the task execution efficiency are improved. According to the method, the container and the virtual machine can share and efficiently utilize the Ascend-910B resources, collaborative operation of multiple types of tasks is achieved, deep fusion and efficient collaboration of the container, the virtual machine and the domestic AI resources are achieved, and the scheduling intelligence level and the resource utilization efficiency of the domestic AI infrastructure are improved.

The invention belongs to the technical field of big data processing and resource management, and discloses a resource optimization-oriented big data scheduling method and device, equipment and a storage medium. The method comprises the following steps: synchronizing metadata of an off-line production task from an existing data processing system through a preset interface, and carrying out sorting and resource allocation on a task queue by adopting a hybrid scheduling algorithm according to a task priority, a predicted execution time length and a task dependency relationship, so as to generate a task scheduling sequence and a resource allocation scheme, then starting and executing the task, monitoring the task execution state and the resource use condition in real time in the execution process, and dynamically adjusting the task execution sequence to realize bidirectional synchronization of production data and scheduling information without additionally designing an external data acquisition component; the method completely depends on off-line production data acquired by an existing data processing system of a semiconductor enterprise, realizes efficient and reasonable allocation of resources, and remarkably improves the scheduling efficiency and the overall performance of the system.

The invention discloses a large machine tool production process-oriented real-time hybrid scheduling method, which comprises the following steps of: reading a current state st of a workshop after a machine tool completes a processing procedure of a current processing part in the workshop, and selecting an action at under the current state st based on a decision network; based on a path rule corresponding to the action at, performing path decision for the current process of each part in the buffer area, selecting a suitable machine tool, forming a to-be-processed workpiece set of all the machine tools in a workshop, and extracting the to-be-processed workpiece set of the current decision-making machine tool; determining a next machined part Js for the decision-making machine tool based on a sorting rule corresponding to the action at, selecting the part Js from the buffer area and sending the part Js into a corresponding procedure of the decision-making machine tool for machining, and deleting a to-be-machined procedure set of all machine tools when the part Js is selected and sent into the decision-making machine tool for machining; the path decision and the sorting decision are mutually combined, the simultaneous decision also provides regret and re-decision opportunities for the path decision, the mechanism is introduced into the decision process of the DRL, and the flexible workshop dynamic scheduling problem can be solved through a decision network.

The invention relates to the technical field of artificial intelligence and distributed computing, in particular to an AI service distributed deployment method and device, equipment and a storage medium. According to the AI service distributed deployment method disclosed by the invention, access to diversified AI services is unified and simplified through a single interface, and optimal allocation of heterogeneous resources, high-concurrency request processing, low-delay response and fine-grained authority control are realized through a hybrid scheduling mechanism, multi-stage security authentication and remote model localization calling; the AI service deployment efficiency, the resource utilization rate and the safety are remarkably improved, and the method is suitable for deployment, integration, management and unified service providing scenes of various artificial intelligence applications.

The invention provides a domestic large model edge cloud collaborative management platform based on a quantum-classical hybrid architecture and an adaptive optimization method, and the platform comprises a quantum heuristic computing power optimization engine which carries out the heterogeneous computing power scheduling based on a quantum annealing algorithm, and enables a large model task to be divided into a quantum bit mapping problem capable of parallel computing, hybrid scheduling of classical calculation and quantum calculation is completed; the edge end dynamic knowledge distillation architecture is used for collecting real-time data through edge equipment, dynamically extracting cloud large model knowledge and compressing the cloud large model knowledge into a deployable edge small model; reverse synchronization of an incremental learning mechanism is set; the multi-modal data sovereignty management mechanism is used for realizing verifiable auditing of data use compliance in a model training process; the management platform realizes secure data input, completes cloud large model training and quantum acceleration, and establishes a bidirectional knowledge flow closed loop. According to the method, large model computing power efficiency and training bottlenecks are overcome, the problems of real-time performance, privacy and knowledge islands are solved, and it is guaranteed that data are safe and compliant.

The embodiment of the invention provides a GPU computing power scheduling method and device, equipment, a medium and a program product, and relates to the technical field of computing power scheduling, and the method comprises the steps: dividing a computing unit and video memory resources of a GPU into a plurality of dynamically adjustable partitions according to a real-time task demand; calculating the priority weight of each task; preemptive scheduling and cooperative scheduling are combined, resource preemption is carried out on the first priority task, and time division multiplexing is carried out on the non-preemptive task; executing a dynamic voltage frequency adjustment (DVFS) and task merging strategy; through dynamic resource partitioning, dynamic priority evaluation, a hybrid scheduling strategy and an energy efficiency optimization algorithm, the GPU resource utilization rate and the energy efficiency ratio in a multi-task scene are remarkably improved, and meanwhile, energy consumption is reduced by combining a DVFS and a task merging strategy.

The application is suitable for the technical field of electric power metering and prepayment system, and relates to a Token server heterogeneous security module mixed scheduling method and system, which comprises the following steps: setting an active module set of the system, m is the number of active modules, acquiring the current remaining Token quantity and concurrent processing capacity of each active module i, and acquiring the total number N of current Token generation requests to be processed; calculating the remaining Token proportion after the distribution of the active module i and the proportion of the concurrent capacity of the active module i in the comprehensive concurrent capacity of all modules; establishing an optimization model with and as a target function; solving the optimization model to obtain the Token request quantity of each module, completing the distribution of the Token generation request and updating the remaining Token quantity of each module. The application can greatly reduce the Token purchase quantity and the number of security module deployments, and greatly improve the economy of system configuration.

The application relates to a mixed scheduling method based on a containerized mixed key system, and specifically comprises the following steps: configuring a first-level scheduling strategy in secondary real-time scheduling in a scheduling blueprint, establishing a main time framework and a container time framework, setting the time of the main time framework, setting the execution sequence and execution time of each real-time container in each main time framework, configuring a second-level scheduling strategy in the secondary real-time scheduling, configuring the number of time slices that can be occupied by each non-real-time task after being scheduled, creating a container and starting the container, and the system runs according to the scheduling blueprint; and when the real-time container, the real-time task, the non-real-time container or the non-real-time task is paused or exited, the system automatically adjusts the corresponding execution sequence and execution time. The application solves the problem that the existing hierarchical scheduling strategy for the real-time container cannot guarantee the minimum response time of the real-time task, ensures that the response of the real-time task can be more timely, and meets the actual needs of different types of key tasks.

The present application relates to the technical field of data transmission, and specifically provides a data transmission method and system for solid waste sorting equipment, a terminal and a storage medium, comprising: adding basic classification marks and dynamic priority marks to data packets according to the function types of sorting data, and dividing the data into synchronous control flow, real-time event flow, process awareness flow and background management flow; a data source end acquires link state information before sending, and adaptively adjusts the dynamic priority of the real-time event flow; a network switch performs mixed scheduling transmission on the data packets based on time-aware shaping, credit shaping and preemption mechanism, wherein the preemption of low-priority queues is only allowed when the data packets of the highest-priority sub-queue carry the highest dynamic priority mark. Through the cooperative mechanism of basic classification and dynamic priority, the present application realizes fine classification transmission of heterogeneous data flow, solves the problem that key instructions are easily affected by network congestion, and effectively reduces transmission delay and jitter.

The present application relates to the technical field of database, in particular to an index updating method, product, medium and equipment. The index updating method comprises: triggering index page inspection by using a preset hybrid scheduling strategy, and identifying a to-be-merged page in an index file; judging whether a target page matching the to-be-merged page exists in the index file; if yes, acquiring a page lock of the to-be-merged page and the target page; merging the to-be-merged page and the target page, and releasing the page lock. The present application locks and releases by page granularity before and after page merging, ensures the atomicity and concurrent safety of the merging operation, avoids locking the whole table or the whole index, effectively suppresses the index file expansion without blocking normal read-write business, and improves the system availability.

The application discloses a vertical and plane transportation mixed scheduling system and method based on space-time coordination, relates to the technical field of intelligent scheduling, and comprises the following steps: obtaining new transportation equipment information, determining initial task allocation weights of the transportation equipment based on rated operation capacity of the transportation equipment; constructing a historical coordination efficiency set and a real-time coordination efficiency set; constructing a historical load characteristic curve; obtaining a current task allocation request quantity, matching a composite load value of the same request quantity in the historical load characteristic curve, calculating an average value as a standard operation load and determining a fluctuation range; combining the total number of the same type of equipment before and after expansion to calculate a theoretical load value; calculating a current actual load value; comparing the actual load value with the theoretical load value, and adjusting the task allocation weights when the actual load value exceeds the fluctuation range; and counting the total number of paired tasks of the vertical and plane equipment and the space-time coordination hit number, calculating a hit rate, and allocating the task weights based on the hit rate, so that the vertical and plane transportation mixed scheduling is realized.

The invention discloses an end-cloud collaborative flexible production line dynamic production scheduling method and system and a storage medium, and relates to the technical field of intelligent manufacturing, and the method comprises the steps: carrying out the modular decomposition of a flexible production line, and dividing the production line into a rigid execution unit and an elastic adjustment unit; constructing a hierarchical control network comprising a plurality of edge computing ends, and establishing a two-way communication channel with the cloud decision center; establishing a real-time resource portrait window, forming a dynamically updated resource situation map, pushing real-time resource data to a cloud decision center, performing global production scheduling planning, generating a mixed production scheduling scheme, and distributing the mixed production scheduling scheme to corresponding computing nodes for execution. According to the invention, the technical problems of low production scheduling efficiency and difficulty in flexibly responding to emergencies due to the lack of real-time adaptive capacity to dynamic changes in the existing flexible production line are solved, and hierarchical control and real-time resource monitoring of end-cloud collaboration are achieved. And the production flexibility, the resource utilization rate and the overall efficiency of the flexible production line are improved.

The application provides a large model inference optimization method, system, medium, terminal and program product for Ascend processors. The method comprises: comparing the automatically detected version with a built-in compatibility table; if the versions do not match, automatically loading a runtime patch or calling a backup library; if the versions match, processing a user request based on a preset hybrid scheduling mechanism; dynamically selecting a MindIE inference framework or a vLLM inference framework based on a runtime adaptation layer according to the user request parameters; converting large model weights into low-bit weights and compensating based on an error compensation matrix; identifying subgraphs and fusing operators for a model graph according to the computing characteristics of the Ascend processor, and rearranging intermediate tensors according to the 512B memory access alignment rule of the Ascend processor; performing pipelined overlap processing on the matrix multiplication calculation of the current token and the full aggregation operation based on an asynchronous execution flow; and performing inference and generating an inference result according to the user request. The application can improve the efficiency and reliability of large model inference.

The invention provides an intelligent calculation center cloud scheduling system, and relates to the technical field of artificial intelligence and big data, and the system comprises a GPU cluster which is formed by the mixed deployment of at least a first type of GPU cards and a second type of GPU cards; and the scheduler is connected with the GPU cluster, a plurality of scheduling strategies are configured in the scheduler, and the scheduler is used for carrying out hybrid scheduling on the first type of GPU cards and the second type of GPU cards based on the scheduling strategies. The method has the beneficial effects that multi-strategy hybrid scheduling of different types of GPU cards is effectively realized, the use efficiency of computing power resources is improved, the operation cost is reduced, the operation efficiency of a system is improved, and the system risk is reduced.

The invention provides a long text-oriented analysis checking method, device and equipment and a storage medium. The checking efficiency and accuracy are remarkably improved through intelligent fragmentation and mixed scheduling. The core of the method is that according to semantic relevance among text fragments, fragments with reference or dependency are distributed to serial auditing nodes for deep and coherent context verification, independent fragments are distributed to parallel auditing nodes for concurrent processing, and therefore the problem of long text semantic segmentation is solved, and meanwhile the content of the text fragments is improved. And the auditing time is greatly shortened from the hour level to the minute level. Besides, the output behavior of the large language model is restrained by loading a predefined structured instruction set in the auditing node, so that'illusion 'and misjudgment generated when the model processes complex tables and nested terms are effectively inhibited, and the reliability of a checking result is remarkably improved. The method has flexibility and expandability, and is suitable for automatic and intelligent checking scenes of various professional long texts such as contracts and bid invitation files.

The application discloses a hybrid scheduling system and method based on clothing flexible production line rules and machine learning. The system contains rule algorithm module, machine learning algorithm module and coordination and decision module, through rule processing deterministic task, machine learning optimization dynamic scene, and using dynamic weight adjustment to realize collaborative scheduling. The method includes initialization, rule preliminary scheduling, machine learning optimization, coordination decision and production monitoring steps, introduces priority quantization model, equipment matching degree formula and dynamic weight algorithm, solves the problem of insufficient adaptability of traditional scheduling. The application combines rule algorithm and machine learning algorithm, realizes reasonable allocation of production task and optimization scheduling of equipment, improves production efficiency, resource utilization rate and product quality of clothing flexible production line, and reduces production cost.

The invention discloses a computing resource allocation method and system for heterogeneous GPU hybrid scheduling, and belongs to the technical field of computer equipment.The method is characterized by comprising the steps that allocation types corresponding to tasks to be allocated are identified, and the allocation types comprise single task allocation, group task allocation and emergency task allocation; if the allocation type is grouped task allocation, acquiring the priority of each task in the to-be-allocated task group; the first task is allocated to a GPU node according to the priority, so that the GPU node generates an affinity label of the first task and the unallocated task, and the affinity label responds to the identity label of the unallocated task; according to the method, the affinity labels are obtained through the response mechanism, the tasks are distributed according to the affinity labels, and compared with the method that the affinity relations of the undistributed tasks are obtained by traversing a query table and are distributed, the method has the advantages that the distribution time is shortened, and the scheduling efficiency is improved.

The invention discloses an AGV scheduling method, scheduling equipment, AGV equipment and a storage medium, and the method comprises the steps: a scheduling equipment side creates a map information model based on the topological information of a to-be-scheduled region; under the condition that vehicle body registration messages sent by the multiple AGV devices are received and registration succeeds, a map information model is sent to the multiple AGV devices, and the map information model is used for indicating that the multiple AGV devices have unified attitude information in the to-be-dispatched area. The AGV equipment side receives the map information model sent by the scheduling equipment under the condition that the AGV equipment successfully registers to the scheduling equipment; and carrying out AGV scheduling in the to-be-scheduled area based on the map information model. Therefore, for the multiple brands of AGVs operating in the same area, the unified map information model is created, so that the hybrid scheduling of the multiple brands of AGVs can be compatible, and the operation efficiency of the AGVs is improved.

The application relates to the technical field of cloud computing, in particular to a computing resource scheduling method and device, equipment and a storage medium, the method comprises the following steps: constructing a self-defined type of resource, and constructing a corresponding controller according to the self-defined type of resource; when the category of the controller is a virtual machine, a container containing the virtual machine is constructed; according to the type corresponding to the self-defined type of resource, corresponding resource configuration is extracted, and the creation instruction of the container containing the virtual machine is rewritten, the virtual machine is scheduled, and thus a single physical computer can support mixed scheduling of virtual machines and containers simultaneously.

The invention discloses a CAN bus hybrid scheduling and fault-tolerant communication method and system for multi-module equipment, and the method comprises the steps: receiving and confirming an instruction of an upper computer in real time through a hardware interruption mechanism, and endowing the instruction with the highest priority through an event-driven task; serial polling communication is executed on a plurality of lower computer modules based on the state machine model, and different fault-tolerant strategy parameters including the maximum number of retry times and timeout are called according to instruction types; in a response receiving stage of the lower computer module, establishing an application layer session verification mechanism based on a two-factor session verification mechanism of an identifier and a data field; and under the condition that the lower computer module has no response, fault isolation is realized by adopting a limited number of retry and fault node skipping strategy, and system-level communication blockage is prevented. Compared with the prior art, the real-time performance, the certainty and the robustness of multi-device CAN bus communication are remarkably improved, and the method is suitable for the fields of power supply control, embedded systems and the like.

The application discloses a hybrid scheduling system of a FaaS platform based on KVM virtualization, which is based on a lightweight user mode scheduling mechanism, and realizes task scheduling and dynamic load balancing through the cooperative work of four core modules, i.e., a pre-execution processor, a task detector, a task distributor and a load balancer. The pre-execution processor analyzes the hybrid execution time of different task types under different scheduling methods, and provides data support for scheduling decisions; the task detector monitors the submitted tasks in the system in real time and judges the task types; the task distributor makes a distribution decision in combination with pre-processing information, task characteristics and real-time load proportion; and the load balancer ensures the load balancing between scheduling sets through real-time load monitoring, and avoids the overload of single-core sets. The application can reduce the average execution time of tasks of the FaaS platform in a high-concurrency scenario, improve the resource utilization rate and task execution efficiency, and provides an efficient and reliable solution for serverless computing task scheduling.

The invention provides a self-adaptive hybrid scheduling accelerator based on an FPGA (Field Programmable Gate Array) and an implementation method thereof. The accelerator comprises a control interface unit (CIU), a multi-level priority queue array (MPQA) and an adaptive scheduling core (ASH). The method aims at solving the problems of large task scheduling delay, slow interrupt response, system throughput bottleneck and the like caused by a traditional software scheduler in a heterogeneous multi-core SoC, a high-requirement embedded system and an edge computing node.