34results about How to "Improve scheduling ability" patented technology

One or more embodiments of the invention provide a load-aware scheduling method of an inference system and the inference system. The method is applied to the inference system comprising a global scheduler and an inference engine. The inference engine comprises a Prefill engine and a Decode engine; the Prefill engine and the Decode engine respectively comprise at least one calculation instance; computing resources of the computing instances comprise GPUs carried on the computing nodes where the computing instances are located; the method comprises the following steps: a global scheduler obtains a target reasoning request to be executed, determines a first calculation instance in a Prefill engine of which the GPU load meets a first preset condition based on maintained GPU load information, and sends the target reasoning request to the first calculation instance to execute reasoning calculation in a Prefill stage; and a first local scheduler corresponding to the first calculation instance in the Prefill engine determines a second calculation instance in the Decode engine with the GPU load meeting a second preset condition based on the maintained GPU load information, and sends an inference calculation result of the Prefill stage to the second calculation instance to execute inference calculation of the Decode stage.

The invention discloses an RISC-V-based thread bundle scheduling method and device and a medium, belongs to the technical field of general graphics processors, and aims to solve the technical problem that the execution efficiency of instructions is affected by the fact that an existing thread bundle scheduling module is affected by a synchronous fence mechanism and a thread bundle scheduling strategy, and the thread bundle scheduling efficiency is easily reduced. The method comprises the following steps: performing instruction configuration on a current thread bundle state to obtain a real-time thread bundle state instruction; performing state control on the thread bundle queue to obtain a state change control strategy; performing instruction fetching sequence control on the thread beam queue, and determining a thread beam instruction fetching control strategy; performing address fetch control on the thread bundle queue, and determining an address type control strategy; and performing thread beam scheduling processing based on a thread beam scheduling system architecture on the thread beam queue through the real-time thread beam state instruction and the thread beam scheduler after the strategy configuration is completed to obtain a thread beam scheduling result.

The invention discloses a configurable flexible job shop scheduling method based on a multi-agent game model, and the method comprises the steps: mapping a scheduling problem into a dynamic disjunction graph model to describe the relation between a process and equipment, representing process nodes, equipment nodes and the mutual relation through a node set and an edge set, and combining the flexibility of a dynamic disjunction graph to obtain a flexible job shop scheduling model. And matching of equipment and processes is dynamically adjusted according to changes of production requirements, so that the equipment utilization rate is optimized. And a clustering method based on structure entropy is adopted, and process nodes are clustered into sub-graphs, so that the search space of a scheduling problem is reduced, and the calculation complexity is reduced. Construction of an order agent and an equipment agent and a game and cooperation process between the order agent and the equipment agent are covered. And each agent optimizes a decision through a deep Q network, and cooperates or competes according to a respective target. An automatic optimization configurator is used for optimizing hyper-parameters and a network structure of a deep Q network in a multi-agent system, and scheduling performance and adaptability are further improved.

The invention relates to the technical field of intelligent manufacturing, and discloses a product production process scheduling method and system based on intelligent manufacturing, and the method comprises the steps: 1, data collection and analysis; 2, establishing a production process model; 3, applying an intelligent scheduling algorithm; 4, a real-time monitoring and feedback mechanism; and 5, continuously optimizing and learning. According to the invention, in a real-time monitoring and feedback mechanism, the scheme combines historical data to dynamically correct real-time production data, and performs anomaly detection. By comparing the trend of the current production data with the trend of the historical data in real time, the system can identify and correct data abnormity in time, and it is ensured that each link in the production process is smoothly carried out according to a predetermined plan. The mechanism has the advantages that the accuracy of production data can be ensured by quickly detecting and correcting the deviation, the production delay caused by data errors or lagging is avoided, and the production efficiency and the scheduling precision are improved.

The invention discloses an electric quantity adjustment method and device for coping with extreme weather, and the method comprises the steps: obtaining the area of a disaster region when the extreme weather occurs, the duration of the extreme weather, and the electric quantity adjustment demand parameter data of the duration, and obtaining a time sequence of electric quantity adjustment demand parameters; performing normalization processing to obtain an electric quantity adjustment demand parameter normalization value; calculating an influence coefficient of the electric quantity adjustment demand index in the next time period based on the area of the affected area, the normalized value and the electric quantity adjustment demand parameter data; and calculating a predicted value of the extreme weather electric quantity adjustment demand index in the next time period to obtain the adjusted output electric quantity. According to the method, the power demand under the extreme weather condition can be predicted, a basis is provided for reasonable configuration and resource scheduling of a power system, and the method is very necessary. The method can effectively improve the adjustment capability of the power system and the multi-energy scheduling capability, promotes the power system to quickly recover stability after extreme weather, and guarantees the stable operation of the power system.

The invention discloses a power grid dispatching method and device and a storage medium. The method comprises the following steps: constructing an optimal power flow optimization model of a power grid system and a mobile energy storage vehicle optimization dispatching model; setting a first objective function and a first constraint condition of the optimal power flow optimization model, and setting a second objective function and a second constraint condition of the mobile energy storage vehicle optimization scheduling model; modeling the mobile energy storage optimization scheduling problem into a mixed integer linear programming problem, and solving the mixed integer linear programming problem based on the first objective function, the first constraint condition, the second objective function and the second constraint condition according to the node load and the carbon emission factor of the power grid system to obtain an updated path planning and charging and discharging scheme. According to the method, the carbon emission factor is introduced to fully consider the carbon emission characteristics of energy storage, the carbon emission is taken as a main optimization target, and the charging and discharging behaviors of the energy storage system can be dynamically optimized, so that the carbon emission can be effectively minimized, and the carbon emission effect can be effectively improved.

The invention relates to the technical field of unmanned aerial vehicle communication, in particular to an avoidance method and device based on optical communication unmanned aerial vehicle identification information automatic broadcasting, and the method comprises the steps: obtaining the current geographic position information of a target unmanned aerial vehicle, and calculating the kinematics information of the target unmanned aerial vehicle in combination with the historical geographic position information; calculating configuration information of a wireless light source of the target unmanned aerial vehicle according to the kinematics information so as to set a communication boundary of the wireless light source through the configuration information; constructing to-be-transmitted state information, and constructing the to-be-transmitted state information into a data frame conforming to a preset frame structure; controlling a wireless light source to broadcast a data frame into the space; and receiving state information of other unmanned aerial vehicles, and controlling the target unmanned aerial vehicle to avoid according to the state information. Therefore, the interference to the existing radio frequency network is reduced, the security of data transmission is improved, the cooperative control performance of the unmanned aerial vehicle cluster is improved, the real-time monitoring and scheduling capability is enhanced, and the method has important significance for constructing a safe and efficient unmanned aerial vehicle system.

The invention relates to a cloud resource scheduling method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring scheduling parameters of a virtual machine; matching the user identifier in the scheduling parameter with the user identifier configured by each preset computing node group to obtain a matched computing node group, each preset computing node group being adjusted in real time in the running process of the cloud resource service; performing static resource scheduling screening on each computing node in the matched computing node group based on the scheduling parameter to obtain a target computing node; and creating a virtual machine at the target computing node. By adopting the method, refined resource scheduling can be realized.

The present application relates to the technical field of flexible job shop scheduling in intelligent manufacturing and production scheduling, and particularly relates to an optimization method for solving flexible job shop scheduling with adjusted resource constraints. The method comprises the following steps: initializing parameters and randomly generating an initial population; using a crossover operator and a mutation operator in sequence to evolve the current population; executing a hybrid decoding strategy on the current population; sorting the individuals in the current population in ascending order of maximum completion time to form an elite population, and updating the elite population through problem-specific local search; judging whether the evolution condition is met, and if yes, executing mathematical evolution based on CP, and outputting a final solution when the running time reaches the total running time. The present application has the positive effects of reducing resource waiting time, improving machine utilization, and improving the resource utilization efficiency and scheduling performance of the entire workshop production.

This application discloses a multi-tenant task scheduling method and related apparatus, relating to the field of cloud computing technology. It receives tasks to be scheduled and their metadata, determines dynamic priority information, and sends the task to a target priority queue if the tenant corresponding to the task has an effective quota. When a tenant has a concurrent consumption token in the target priority queue, it acquires the token, executes the task, and then releases the token. This application sets different dynamic priorities for tasks to be scheduled, ensuring that tasks are distributed to resource-isolated independent queues, guaranteeing priority processing of each task, and improving resource utilization. Before task publication, sliding window rate limiting is implemented, and tenant-level concurrent consumption tokens are distributed during task consumption, isolating tasks between tenants. In complex multi-tenant, high-concurrency production scenarios, this enables intelligent task adjustment, significantly improving resource utilization and scheduling efficiency.

The present application relates to the technical field of intelligent manufacturing and intelligent scheduling, and particularly relates to a flexible processing unit dynamic regulation method, system and model based on multiple agents. The method comprises the following steps: S1: establishing an event-driven Markov decision process model; S2: constructing corresponding task selection agents and workpiece selection agents to form a hierarchical multi-agent deep reinforcement learning architecture; S3: designing the system state features, action space and reward function of the task selection agents and the workpiece selection agents; S4: jointly training the task selection agents and the workpiece selection agents based on a multi-agent proximal policy optimization training algorithm to obtain optimal policy parameters of the task selection agents and the workpiece selection agents; and S5: the task selection agents and the workpiece selection agents output task actions and workpiece actions respectively based on the optimal policy parameters obtained through training. The present application can improve the adaptability of the system to dynamic disturbances and the real-time decision-making capability.

The application provides a kind of desertification irrigation water resource intelligent scheduling system and method based on deep learning.The existing desertification irrigation water resource scheduling mode is mostly traditional experience type scheduling or simple quantitative scheduling.The application collects vegetation growth state data, soil moisture data, meteorological data, water resource supply and demand data, water quality data and water quantity data in desertification area, and transmits to data preprocessing module;standardized data is input into water resource demand prediction model, and vegetation water resource demand prediction information is output;based on water resource demand prediction information, the optimal water resource allocation scheme is selected;the execution module completes the scheduling supply of water resources.The application realizes the accurate scheduling of desertification irrigation water resources, improves water resource utilization rate, reduces waste and guarantees vegetation growth.

The application discloses a coal furnace mixed coal parameter determination method, device and equipment, wherein the method comprises the following steps: obtaining historical power plant operation data and historical coal type data; based on the historical power plant operation data and the historical coal type data, a coal power coupling model for representing the mapping relationship between the coal quality parameters and the coal furnace performance indexes is trained; current power grid operation data and current coal furnace operation data are obtained, and based on the current power grid operation data and the current coal furnace operation data, a constraint condition is constructed; a target function is constructed based on the coal furnace performance indexes, a mixed coal parameter determination model is constructed based on the target function, the coal power coupling model and the constraint condition, the mixed coal parameter determination model is solved, and the target mixed coal parameter corresponding to the coal furnace on the power plant side is determined; and the target mixed coal parameter is used for representing the mixed coal proportion of different coal types. Through the above method, the mixed coal parameter can be quickly and accurately determined, and the power grid peak regulation demand can be adapted.

The present disclosure relates to a YARN cluster resource scheduling method, apparatus, medium and computer equipment, wherein the method includes: receiving a job resource request of a job task sent by an application manager; when determining that the job resource request is a rigid resource request, adding the rigid resource request to a request queue, the request queue includes multiple request units, and multiple rigid resource requests belonging to different job tasks are added to corresponding different request units; starting multiple first threads, each first thread obtains multiple rigid resource requests in each corresponding request unit from the request queue to perform resource pre-allocation processing; after resource pre-allocation, each first thread determines whether each pre-allocation result meets the resource demand of the corresponding job task; if so, each first thread submits the pre-allocation result to the application manager respectively. The implementation scheme of the present disclosure can improve the resource scheduling performance of the cluster.

The invention relates to the technical field of unmanned collaboration and resource virtualization, and provides a task collaboration-oriented unmanned system virtual resource representation and organization method and device and a storage medium. The method comprises the following steps: firstly, determining physical resource virtualization granularity (platform level, equipment level and function unit level) according to unmanned collaboration styles (task level, resource level and element level); secondly, establishing a resource tree model comprising a resource base class, an attribute resource, a data table resource and a resource sub-tree based on granularity; resource storage is carried out by utilizing an organization structure constructed by a hash table, a list and a set; and finally, forming a resource pool through standardized packaging, providing management interfaces (resource registration, cancellation and state maintenance) and scheduling interfaces (demand response, resource allocation and release), and realizing resource addressing by adopting URI coding. According to the method, the problems of low task execution efficiency, insufficient resource utilization rate and complex management caused by poor coordination task and resource adaptability, non-uniform heterogeneous resource description and non-standard packaging are solved.

The application discloses a novel flexible manufacturing system dynamic scheduling method based on multi-agent reinforcement learning, relates to the technical field of intelligent manufacturing, and comprises the following steps: constructing a cross-domain knowledge graph driven state space, fusing process knowledge and equipment capacity to perform semantic embedding; establishing an adaptive evolution intelligent agent collaborative network of a space-time heterogeneous graph, and dynamically adjusting a topological structure according to task dependence and equipment coupling; adopting a progressive collaborative value decomposition network to optimize a strategy and accurately realize credit distribution; and constructing a multi-scale closed-loop compensation decision mechanism to form deep coupling and closed-loop feedback at the equipment layer, the production line layer and the system layer, so that the scheduling efficiency and robustness of the flexible manufacturing system are significantly improved.

The utility model discloses a kind of feeding device and spraying system, it is related to the technical field of automation equipment.The feeding device includes base, material transmission assembly and feeding back component, material transmission assembly is located in base and has material transmission passage, the width of material transmission passage is adapted to product to make product be sequentially conveyed from the feeding end of material transmission passage to the discharge end of material transmission passage;Feeding back component is located in base and includes first conveyor belt and second conveyor belt arranged side by side, the feeding end of first conveyor belt and the feeding end of second conveyor belt are connected, and the discharge end of second conveyor belt is connected respectively with the feeding end of material transmission passage to make the part of product of the discharge end of second conveyor belt be conveyed to the feeding end of material transmission passage, part of product is conveyed to the discharge end of first conveyor belt and by the feeding end of first conveyor belt be conveyed to the feeding end of second conveyor belt.The technical scheme provided by the utility model improves the scheduling capability of feeding device.

The application relates to an energy data processing method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: obtaining original energy data based on space division multiplexing transmission energy data; sequentially performing denoising processing and data quality evaluation processing on the original energy data to obtain denoised energy data and a quality evaluation result of the denoised energy data; performing denoising processing on the denoised energy data again according to the quality evaluation result to obtain target energy data; and inputting the target energy data into a trained energy scheduling processing model to obtain target scheduling information of the target energy data. The method can improve the energy data processing efficiency, utilize the target scheduling information to improve the scheduling effect of energy equipment, and realize intelligent management of the energy equipment.

This invention belongs to the field of sensor resource allocation and discloses a sensor resource scheduling method and system based on reinforcement learning of combined action space. The method first obtains the system state of the sensor network system to be scheduled, then inputs the system state into a pre-trained sensor resource allocation model to obtain combined actions, and finally outputs the combined actions as the optimal scheduling scheme for flight targets. By designing a combined action space, this method overcomes the limitation of classical single-agent reinforcement learning algorithms that execute one discrete action at a time in discrete action space, enabling the execution of a multi-dimensional action combination at a given moment, and facilitating the scheduling of multiple sensors for tracking multiple flight targets. The sensor network system is trained using reinforcement learning to obtain a sensor scheduling scheme, enabling adaptive resource scheduling, dynamic adjustment of allocation strategies, and optimization of system performance.

The embodiment of the invention provides an end-to-end scheduling method and device of a power system, electronic equipment and a storage medium, a scheduling instruction can be obtained by inputting the first environment data and the current environment data into the pre-trained scheduling model, the conditions of relatively large information transmission loss and relatively large accumulative error can be avoided, and the scheduling efficiency of the power system is improved. And the dispatching effect of the power system can be improved.

The application discloses a power distribution network scheduling method and system based on multi-stage safety reinforcement learning, and belongs to the technical field of power system operation control, which comprises the following steps: constructing a cost reward function with the minimum power distribution network cost as the target, taking the candidate output adjustment amount of the cost scheduling subject as the cost action space, and constructing a cost agent; obtaining the actual output adjustment amount of the cost scheduling subject by using the cost agent according to the output boundary constraint of the cost scheduling subject; constructing an adaptive reward function with the maximum safety scheduling performance of the safety scheduling subject as the target, constructing a power grid safety penalty function with the maximum power distribution network safety as the target, taking the candidate output adjustment amount of the safety scheduling subject as the safety action space, and constructing a safety agent; obtaining the actual output adjustment amount of the safety scheduling subject by using the safety agent; and scheduling the power distribution network based on the actual output adjustment amount of each subject. The technical problem that the prior art is difficult to balance the safety of the power distribution network and the scheduling performance of new energy is solved.

This invention discloses a joint optimization method and apparatus for tandem quay crane loading considering carbon emissions. The method includes: constructing a three-layer coding structure comprising a container allocation layer, a container loading sequence layer, and a truck allocation layer; decoding the three-layer coding structure to generate an executable loading operation plan and calculate the loading operation completion time and total carbon emissions; employing a multi-objective evolutionary mechanism to perform a global search on the coding structure, achieving iterative population updates through crossover, mutation, non-dominated sorting, and reference point selection; performing local search for container allocation and iterative greedy optimization of truck scheduling on candidate plans, updating the non-dominated solution set based on Pareto dominance, and outputting the joint optimization results. This invention achieves synergistic optimization of loading efficiency and carbon emissions by unifying the modeling of container allocation, truck scheduling, and tandem quay crane collaborative operation decisions, solving the problems of insufficient synergy, low operational efficiency, and high carbon emissions caused by the separation of multiple decisions in existing loading operations.

The invention discloses a computing power scheduling method and system based on multi-model cooperation. The method comprises the following steps: acquiring a plurality of computing power requests of a plurality of computing power users; determining a request parameter corresponding to each computing power request according to the request data of each computing power request; respectively inputting request parameters corresponding to all the computing power requests into prediction models corresponding to a plurality of computing power index types to obtain a plurality of corresponding predicted computing power index parameters; and calculating a computing power resource scheduling strategy corresponding to the plurality of computing power requests according to a dynamic programming algorithm based on a target function and a limiting condition associated with the plurality of predicted computing power index parameters. Visibly, the computing power scheduling method and system can realize computing power scheduling for computing power requests of multiple users more efficiently and reasonably, improve the efficiency and effect of computing power resource scheduling, and provide better user experience for computing power leasing users.

This invention discloses a power system dispatch optimization method and system based on deep reinforcement learning, belonging to the field of smart grid optimization and dispatch technology. It includes: receiving and synchronizing real-time operational data, meteorological data, equipment health indicators, and renewable energy output data; constructing a power grid diagram and generating a node time-series matrix, encoding it as a multi-temporal embedding vector; generating short-term output prediction values ​​and uncertainty indicators based on meteorological and renewable energy output data, converting them into compensation factors; calculating risk scores based on equipment health indicators and meteorological data and mapping them to dynamic weights; inputting the multi-temporal embedding vectors, compensation factors, and dynamic weights into a deep reinforcement learning model to generate a dispatch strategy and performing feasibility verification; if the verification passes, the strategy is executed and the results are returned for online model updates. This invention effectively improves the security and robustness of power system dispatch by integrating multi-source data and a risk perception mechanism.

The invention relates to the technical field of commodity supply chain scheduling, in particular to an intelligent commodity supply chain scheduling system based on an internet telemarketing platform, which comprises a supply chain scheduling platform, a data acquisition unit, a supply efficiency evaluation unit, a vehicle scheduling analysis unit, a supply smoothness analysis unit, a scheduling division unit and a scheduling management unit. According to the invention, analysis is carried out from two points of historical supply commodity quality and efficiency of the commodity supply chain of the target telemarketing platform so as to understand the commodity supply condition of the commodity supply chain, and the unqualified supply chain is analyzed from two points of vehicle scheduling and commodity supply. On one hand, whether the abnormity of the unqualified supply chain is caused by vehicle scheduling or commodity supply interference can be known, and on the other hand, the unqualified supply chain can be divided according to the existing defects, and interactive analysis is carried out in a layer-by-layer progressive mode. Reasonable and targeted management and adjustment are carried out for the scheduling problem of the unqualified supply chain.

The application provides a tracking method, device and equipment of a target object and a storage medium. The method comprises the following steps: in the case that a panoramic camera device detects a target object, acquiring target parameters corresponding to each of at least two dynamic point camera devices, wherein the target parameters comprise a service life usage rate of the dynamic point camera device and / or a rotation distance of the dynamic point camera device rotating to a target position, and the target position is a position at which the dynamic point camera device can monitor the target object; determining a target dynamic point camera device based on the target parameters; and controlling the target dynamic point camera device to track the target object. The application can optimize the scheduling of the dynamic point camera device and improve the scheduling performance of the monitoring system.

This invention discloses a hierarchical dynamic scheduling method based on the policy optimization DDQN algorithm for solving the dynamic scheduling problem of a reentrant hybrid pipelined workshop with batch processing machines. First, the objective function and constraints of the scheduling problem are determined. Then, by introducing a self-attention mechanism, a hierarchical structure based on DDQN is proposed, constructing two agents: a batching agent and a scheduling agent, to solve the batching and scheduling subproblems respectively. Furthermore, to address the multi-stage batch processing and reentrant scheduling characteristics of the problem, a Markov decision process considering the characteristics of these two agents is designed, including state, action, and reward. Further, an action selection strategy based on masking combined with an ε-greedy strategy and a soft-start target network update strategy are proposed to improve efficiency and generalization ability. This invention demonstrates significant effectiveness in solving the dynamic scheduling problem of a reentrant hybrid pipelined workshop with batch processing machines.

This utility model discloses a dual GNSS antenna cooperative positioning device for container loading and unloading operations. A first positioning antenna is positioned at the center of the lifting beam, and a second positioning antenna is positioned at the end of the lifting beam, allowing the positioning antennas to move with the lifting device. The distances from the first and second positioning antennas to the center of the lifting beam are configured differently. When the lifting beam lifts a container, the center of gravity detection component in the container overload / off-center load detection device generates rising or falling edge signals due to factors such as changes in the container's weight. This signal triggers the second positioning antenna, allowing the device to determine the container's specific location. When the lifting beam rotates, the position of the second positioning antenna relative to the first positioning antenna changes. Therefore, the positioning host can use the correspondence between the positions of the first and second positioning antennas and the container's center of gravity to obtain the container's position coordinates and center of gravity coordinates.

The invention provides a Pod scheduling method and device, and the method comprises the steps: adjusting Pods in a scheduling queue, guaranteeing the continuous arrangement of the Pods in the same group in the scheduling queue, and obtaining a second scheduling queue; when the Pod in the second scheduling queue enters a scheduling stage, if the current resource does not meet the minimum number of copies required by normal operation of a group to which the Pod belongs, the Pod is placed in a waiting state; for each Pod in the waiting state, determining a processing mode of the Pod in the waiting state according to a user-defined condition preset by a user; and if the processing mode of the Pod in the waiting state is not determined within the timeout time, refusing to schedule the Pod in the waiting state and all Pods in the group to which the Pod in the waiting state belongs. By releasing the resources in time, the resource utilization efficiency of the cluster is improved, and normal scheduling of other groups is ensured. And the utilization efficiency and the scheduling performance of cluster resources are effectively improved.