76results about How to "Implement dynamic scheduling" patented technology

The invention relates to a parallel computation method and system. The method includes: monitoring each computing node to acquire node monitoring data; calculating load capacity of each computing node according to the node monitoring data; assigning tasks to be assigned according to the load capacities of the computing nodes. Correspondingly, the system comprises a node monitoring unit, a task load management unit and a task scheduling unit. Through the application of the technical scheme, task can be reasonably assigned during parallel computation and parallel computation can be executed more efficiently.

The invention discloses a multimedia service data sending and receiving method, base station equipment and user equipment. The method includes following steps, receiving the information of sending sequences which show a plurality of multimedia service, generating starting index of each multimedia service and scheduling information in the corresponding number of the index; sending the data of each multimedia service based on the sending sequence information and scheduling information. The method and equipment of the invention can realize the dynamic scheduling of MBMS in a monochromatic network so as to save power of user facilities in the condition of discontinuous transmission and make statistic multiplexing processing easy.

The invention discloses a high-efficiency dynamic resource scheduling method based on cloud storage and relates to the technical field of cloud storage. The technical contents of the method consist of a cloud storage resource control node, a cloud storage resource scheduling node, a cloud storage resource monitoring node and various storage nodes, wherein the cloud storage control node acquires a data index and an updating data index according to afferent data and then the cloud storage scheduling node selects storage nodes according to magnitude of index values. The high-efficiency dynamic resource scheduling method based on cloud storage has the beneficial effects that the dynamic scheduling of storage resources is realized by placing high index data at high-speed storage nodes, the high-speed storage nodes are effectively utilized, the hit rate of cloud storage during data reading is improved and the effective utilization ratio of cloud storage resources is greatly improved.

The invention discloses a load balancing method and a load balancing device based on an application proxy server. The method comprises the following steps: receiving a start request carrying the identity information of an application to be started; acquiring the resource demand of the application to be started from a pre-established application historical database according to the received identity information; and determining a target application proxy server used for running the application to be started according to the acquired resource demand and the load information of at least one application proxy server. According to the invention, an application proxy server meeting the resource demand and load demand of the application to be started is assigned for the application to be started.

The invention discloses a scheduling method and a scheduling system for password arithmetic units. The scheduling method comprises the steps of: obtaining an arithmetic request about a password arithmetic task from a user; and selecting the password arithmetic unit matched with the password arithmetic task according to the password arithmetic content of the password arithmetic task and referring to the running status of each password arithmetic unit in a password service pool. The password service pool includes the plurality of password arithmetic units, so that password arithmetic resources are virtualized, the password arithmetic units in the password service pool are dynamically scheduled aiming at the different password arithmetic tasks, and thereby use ratio of the password arithmetic unit is improved.

The invention provides a HRCEA (hybrid of rules and compression-extensive algorithm)-based steel making-continuous casting production scheduling method for a special steel mill. In the scheduling method, model and algorithm parameters are obtained by analyzing a steel making-continuous casting production flow; according to a known production casting times plan, an initial production scheduling plan is worked out by adopting the HRCEA; and all subsequent furnace working plans are adjusted by using the HRCEA according to an established scheduling rule and fed-back real-time data so as to realize dynamic scheduling. According to the HRCEA-based steel making-continuous casting production scheduling method disclosed by the invention, with the aims of reasonably quickening production rhythms and shortening production process time, equipment conditions of the steel mill, produced steel grades and temperature requirements of the produced steel grades are considered; an executable production scheduling plan can be quickly worked out, thus schedulers are facilitated on adjusting the scheduling plan in real time; and the production scheduling of the steel making-continuous casting process is optimized.

The invention discloses a neural network compiling method and a system, and a corresponding heterogeneous computing platform. The method comprises the steps that a trained NN model is acquired; and the trained NN model is input to an NN compiler to generate an NN binary file including graph structure information corresponding to the NN model. Therefore, graph optimization during operation is realized. Furthermore, the graph structure information can be realized as a node resource pool in the NN binary file, together with file headers and free segmentation settings, so that the compiling universality and flexibility for various neural network algorithms can be improved.

The present invention discloses a multi-component self-organizing software-connected cluster computer intelligent resource management method. The method gradually increases the component resources along with the increase of user number and gradually reduces the component resources along with the decrease of user number according to the component resources provided by the user to the dynamic demand of the resources, thus achieving the automatic start-up and the automatic shutdown of the components. Through monitoring the use of the component resources, the method achieves the dynamic scheduling of the component resources in the component cluster, carries out the fault management and safety management of the components, and monitors and maintains the reliable operation of the component resources. The present invention also carries out the partitioning treatment of system storage resources and satisfies the temporary and permanent storage requirements of different user data. The present invention can increase the utilization rate of the component resources and optimizes the resource configuration of the cluster computer, thus further decreasing the investment cost and the operating maintenance and management cost, and increasing the system reliability.

The embodiment of the invention provides a task scheduling method and device and an operating system. The task scheduling method comprises the steps of (1) a task scheduling index list is registered and is generated according to all subtasks contained in a task; (2) subtasks which need to be scheduled to be executed are determined when a user program schedules the task; (3) the task scheduling index list is updated according to the subtasks which need to be scheduled to be executed when the user program schedules the task, so that the task scheduling index list only comprises the subtasks which need to be scheduled to be executed when the user program schedules the task; (4) the subtasks in the updated task scheduling index list are scheduled to be executed to accomplish the operation that the user program schedules the task. According to the task scheduling method and device and the operating system, due to the facts that the subtasks which need to be scheduled to be executed are determined when the user program schedules the task, and the task scheduling index list is updated, dynamic scheduling of the task is achieved; due to the fact that subtasks which do not need to be executed are not scheduled any longer, the purpose of reducing useless expenses and the purpose of reducing the load of a CPU and power consumption of the system are achieved.

The invention discloses a resource scheduling method. The method includes the steps that a service scheduling unit and an environment scheduling unit are set; a service to be executed is divided into one or more sub-services, in the process of executing the sub-services of the service, when corresponding resources need to be used, the service scheduling unit informs the environment scheduling unit to allocate the needed resources. The invention further discloses a resource scheduling device and system. By means of the resource scheduling method, device and system, when real-time signals are processed, the resources can be effectively and dynamically scheduled.

The invention discloses a business hall customer service system, and relates to a computer controlled customer service system. A business hall control unit is separately connected with a plurality of customer service terminals, a plurality of service operation stations and service resource dispatching devices through ports; the system is provided with a service pre-processing and service resource dispatching control program; the service pre-processing and service resource dispatching control program executes the operation process of the whole system; the system automatically recognizes customers, provides categorized services and dynamically dispatches customers and counters; automatic customer recognizing is thus realized, categorized services are provided, and dynamic dispatch of customers and counters is realized, so work strength of the counter personnel is reduced, working efficiency of the counter personnel is increased, useless queuing of customers is eliminated, and waiting time of customers is shortened. The invention can be used as multi-business hall customer service system in banks, public security bureaus, industrial and commercial bureaus, taxation bureaus, transport bureaus and social insurance offices.

The invention relates to a multiple-ring-tangency-type wavelength division multiplexing optical network system based on a single fiber, and belongs to the technical field of fiber communication. The multiple-ring-tangency-type wavelength division multiplexing optical network system based on the single fiber comprises a main ring formed by connecting an optical line terminal OLT with M remote nodes RN through a fiber, and sub-rings formed by taking at least two remote nodes RN on the main ring as tangency points, and connecting each tangency point with N remote nodes RN through a fiber respectively, wherein each remote node RN is connected with an optical network unit ONU through a distribution fiber. When some section of a feeder fiber has fault, through the combination of photoswitches and optical circulators, signals can be transmitted to the optical line terminal of each remote node in a multiple-ring system in an anticlockwise or clockwise manner, thereby finding new paths for signal transmission in time and realizing the functions of system protection, dynamic wave length scheduling and scale expansion.

The embodiment of the invention provides a service routing method based on a client. The method comprises the following steps: obtaining a routing rule according to a user tag, wherein the routing rule is updated in real time according to at least one of the following items: a business operation state, a business busy state, a client area, an IP address, a user serial number and a user role; and determining and accessing a corresponding business system service group according to the routing rule. The embodiment of the invention further provides a service routing apparatus based on the client.

The invention relates to a resource allocation method in a time-sensitive network, which comprises the following steps of: 1, modeling a network topology into a directed graph G (V, E); 2, eliminatingunmatched routes; 3, detecting the traffic forwarding rate of each node of each path; 4, formulating an objective function and an output function, wherein the objective function is the minimized jitter of AVB traffic, and the output function is the traffic quantity allocated to each path; 5, formulating constraint conditions according to the target function in the step 4; 6, searching for a pathwith the largest hop count, and removing the path in a feasible solution, and remaining p-1 paths at the moment; and 7, recording a solution generated by each round of calculation in the dynamic change as a sample space. On the premise of ensuring TT traffic delay, the jitter of AVB traffic is minimized, the audio and video transmission effect is ensured, the AVB traffic delay is reduced to a certain extent, and the multi-path dynamic allocation problem of mixed traffic at a source node is solved.

The invention discloses a dynamic scheduling method and a system for collaborative computing between a CPU and a GPU based on two-level scheduling. The method includes: forecasting the processing capacity of each node in the system, a global scheduling module dynamically distributing data to each node according to the processing capacity of each node according to the batch number of the request ofthe node scheduling module in each node, when the node scheduling module finds that the data queue used to place the processing data is empty, requesting the next batch of data to be processed from the global scheduling module, and dynamically scheduling the tasks according to the CPU and GPU processing capacity in the node. According to the heterogeneity of system resources, the invention allowsweak nodes to share fewer tasks and strong nodes to process more tasks, which can improve the overall concurrency degree of the CPU / GPU heterogeneous hybrid parallel system and reduce the task completion time.

The invention provides an electric automobile and a battery management system thereof. The system comprises single batteries, a mother board and a plurality of daughter boards, the daughter boards corresponding to a plurality of single batteries exchange information with the mother board through CAN (controller area network) buses, acquire temperature information and voltage information of the single batteries and transmit the information to the mother board, the mother board / daughter boards can calculate the score of the daughter boards according to the variance of average temperature and thevariance of average voltage of the daughter boards within a certain time period by score parameters obtained by a ranking algorithm, and the mother board / daughter boards can dynamically adjust the priority of the daughter boards according to the score of the daughter boards. The system solves the problems that a traditional battery management system adopts a static scheduling algorithm in communication, consequently, important data information sometimes cannot be transmitted or timely transferred, and the traditional battery management system cannot timely find battery defects. Daughter boardmessage transmitting reliability and safety are realized, and the performance of the electric automobile is further improved.

The invention provides a timeliness-first scheduling method and system, which is used for drop and pull transport. The method comprises the steps of establishing a task queue for any truck or trailer,and estimating position and time of finishing the final task by the truck or the trailer; giving an order set, and finding out an optimal truck and trailer combination, meeting the following conditions, of each order in the order set: the excepted compartment arrival time of the order is greater than a sum of available time and predicted driving time of the trailer, and the expected departure time of the order is greater than a sum of available time and predicted driving time of the truck; calculating a single order price, wherein the single order price is equal to a sum of a short-distance price and a trunk line price; if the trailer and the truck sent by an optimal solution of each order do not have a conflict, obtaining a global optimal solution, otherwise, searching for the global optimal solution through pruning search, replacing the orders with conflicts, using a next non-conflicting truck and trailer combination with a minimum price to solve an overall price of the order set, and if the overall price is increased, performing pruning and discarding; and if no solution exists after the search, discarding the orders with the expected compartment arrival time farthest from current time, and repeatedly executing the steps.

The invention relates to a wavelength division multiplexing passive optical network system based on a tangent ring structure and belongs to the technical field of fiber communication. The wavelength division multiplexing passive optical network system based on the tangent ring structure comprises a main ring formed by connecting an optical line terminal OLT with M remote nodes RN through double fibers, and a sub-ring formed by taking any one of the remote nodes RN on the main ring as a tangency point and connecting the tangency point with N remote nodes RN through fibers, wherein each remote node RN is connected with an optical network unit ONU through a distribution fiber. When some section of a feeder fiber has fault, through the combination of photoswitches and optical circulators, signals in the main ring and the sub-ring can be transmitted to each remote node RN in an anticlockwise or clockwise manner, thereby finding new paths for signal transmission in time and realizing the functions of system protection, dynamic wave length scheduling and scale expansion.

The invention provides a dynamic distributed scheduling method for double-layer terminal nodes based on a hudson expansion function. The dynamic distributed scheduling method is characterized by comprising the following steps: (a) constructing a double -layer terminal node structure, wherein the bottom-layer terminal node is a mobile terminal, and the upper-layer terminal node is a PC (personal computer) machine; the bottom-layer terminal nod is connected with the upper-layer terminal node; (b) dynamically scheduling the double-layer terminal nodes through an expansion function server based on a hudson system. According to another aspect of the invention, the invention provides a dynamic distributed scheduling system for the double-layer terminal nodes based on the hudson expansion function. The dynamic distributed scheduling system comprises a daemon process module, an agency module and a management module which are used for realizing the functions described by the method. According to the dynamic distributed scheduling method and the dynamic distributed scheduling system, the shortcomings of the hudson server are overcome, and real-time dynamic scheduling of tasks is realized. Moreover, due to the characteristics of the dynamic distributed scheduling method and the dynamic distributed scheduling system, the parallelism of item running is improved by adding terminals.

The invention provides a hybrid cluster task scheduling method based on Monte Carlo tree search in order to overcome the defects that time overhead is large and the method is not suitable for hybrid scenes needing dynamic scheduling; the method comprises the following steps of: acquiring a current system state, a task queue, available resources of all machines and a time sequence to form a Monte Carlo tree by setting a root node; selecting a current optimal node according to a preset in-tree selection strategy, judging whether the current optimal node is a termination node, and if not, directly judging whether a preset search threshold value is reached; if yes, randomly selecting one action from all possible actions in the current optimal node as an expansion node of the current optimal node; on the basis of the state of the expansion node, scheduling the remaining tasks in the task queue to obtain a scheduling plan, and then starting from the termination node, backtracking the nodes on the path upwards to update; and repeatedly executing the steps until a preset search threshold value is reached.

The invention provides a cloud computing resource scheduling method and system for geographic big data. According to the invention, the cloud computing resource scheduling rule base is established offline and the triggering event is monitored online; and determining a resource scheduling action of the trigger event, and determining an optimal scheduling scheme by adopting a ConvLSTM load flow prediction model, thereby realizing dynamic scheduling of cloud computing resources of different task types (monitored trigger events) under the condition of not stopping the cloud application.

The invention provides a dynamic dispatching method for virtual desktop resources for multiple power tenants, comprising the step of starting an overall physical resource monitoring mode, a single-point physical resource monitoring mode and a virtual desktop resource monitoring mode to monitor the running states of virtual desktops until an overall physical server optimally runs, a single-point physical server is balanced in load, and the resource occupancy of a single virtual desktop is reasonable. The dynamic dispatching method for virtual desktop resources for multiple power tenants provided by the invention is completely realized via an independent design. Compared with the prior art, the method gives consideration to dynamic dispatching for both physical resources and virtual desktop resources, and solves the problems of the optimally running of overall physical resources, the load balance of single-point physical resources and the reasonable resource occupancy of the single virtual desktop, thus being capable of effectively increasing the utilization rate of the resources and meeting the national policies of energy conservation and environmental protection.

The invention discloses an energy servitization system of a multi-machine production line and a control method, the system is composed of a driving system, an execution device, an energy supply bus and a controller, and energy is used as a service to be provided for a task requiring energy in the production line with high energy efficiency; all machines in a production line are reorganized to formthree types of controllable entities including a driving system, an energy supply bus and an execution device; a control center is provided; in the working process, when an execution device starts toprocess a corresponding task stage, an energy demand is provided, a control system responds to the demand, a sub-driving system which is in an idle state and can complete the working stage with highenergy efficiency is selected to provide energy service for the corresponding execution device through an energy supply bus, and the energy efficiency of a multi-machine production line is improved.

The invention provides a big data processing method and system, and the method comprises the steps: carrying out the data collection according to an edge data collection rule; carrying out data processing on the collected data by utilizing an artificial intelligence model to obtain processed data; performing dynamic knowledge graph creation and heterogeneous data storage on the processed data; performing visual twinning on the structure, the state and the behavior in the information space; adopting the virtual cloud operation, and carrying out the virtual resource dynamic scheduling, resourcetrusted management and resource running state management and control. The system comprises modules corresponding to the steps: the method. According to the big data processing method provided by the invention, edge computing, artificial intelligence and visualization twinning technologies are adopted, so the problems of data redundancy, missing and conflict can be avoided.

The invention relates to the technical field of cloud computing virtualization, provides a cloud resource dynamic scheduling method, device and equipment, and aims to solve the problem that an existing virtual machine cannot dynamically adjust and schedule cloud resources during operation. The method comprises the following steps: acquiring performance data of the virtual machine before the current time node; calculating load prediction results of the virtual machine and the physical host; according to the load prediction results, making various cloud resource adjustment decisions, such as a virtual machine expansion decision, a virtual machine expansion decision, a virtual machine emigration decision, a virtual machine immigration decision, a physical host awakening decision or a physicalhost standby decision, for the virtual machine or the physical host in combination with decision rules preset by a management end and / or input by a user. According to the method provided by the invention, the performance data is periodically acquired, and the load prediction result is calculated, so that the cloud resource adjustment decision can be periodically made, and the dynamic adjustment of the cloud resource is realized.

The invention discloses a multi-robot cloud control system based on a cloud side end hybrid computing environment. The multi-robot cloud control system comprises an execution module, a communication module, a knowledge base module, an intelligent algorithm module and a general control module; the execution module comprises a robot and a sensing device, and is responsible for acquisition and processing of sensing data, executing a control instruction; the communication module is responsible for forwarding and processing real-time and dynamic data in a cloud side end environment; the scene general knowledge base module is used for storing general knowledge involved in a scene and providing a unified calling interface; the intelligent algorithm module is responsible for providing an intelligent perception and autonomous decision algorithm required in the scene; and the general control module is responsible for arranging and scheduling robots in the scene to cooperatively complete a specified task. The multi-robot cloud control system is stable, efficient and high in fault tolerance.

The embodiment of the invention discloses a dynamic dispatching and displaying method, device and equipment of massive data and a computer readable storage medium. The method comprises the steps of calculating an initial hierarchy of a quad-tree structure within a global range according a vector data hierarchy range of vector data; according to the initial hierarchy range of the quad-tree structure and the preset tile size, determining the number of hierarchies of the quad-tree structure; initializing a tile node of the initial hierarchy; according to a preset rule, processing the tile node ofthe initial hierarchy, determining a target tile node satisfying preset conditions and requesting the vector data crossing the tile range of the target tile node to dispatch the data; according to each piece of requested vector data, building a corresponding drawing node and making all the drawing nodes hung on the target tile node. Massive vector data can be dynamically dispatched in a three-dimensional scene, the data amount in memory is effectively reduced, and the system performance is improved.

The invention provides an electric automobile and a battery management system thereof. The system comprises a mainboard, a plurality of sub boards and a plurality of battery monomers, the mainboard acquires the message quantity and recurrence rate corresponding to each sub board sent by each sub board in the same network in a certain time period so as to calculate a bus load rate of a CAN bus in the time period; the mainboard also obtains the temperature information and voltage information of the battery monomer corresponding to each sub board in the time period, the mainboard / sub board calculates the score of the each board according to the score parameters obtained by using a sorting algorithm according to a variance of an average temperature and a variance of an average voltage of the sub board in a certain time period so as to obtain the priority of the sub board; and the mainboard judges whether the bus load rate is in an expected range or not, and dynamically adjusts the frequency of sending messages by each sub board according to the priority of each sub board when the bus load rate exceeds the expected range. The battery management system can reduce the bus load rate according to the data weight, achieves the reliability and safety of message information transmission of each sub board, and further improves the performance of the electric automobile.

The invention discloses an AGV scheduling path optimization method based on the 5G Internet of Things. The method comprises the steps that firstly, an MES system generates a carrying task according to a production work order and issues the carrying task to a WMS system; the WMS system receives the task instruction and generates an AGV carrying task, and then constructs a mathematical model of a search state of AGV scheduling path planning; and the AGV management and control system solves the scheduling scheme according to a heuristic artificial intelligence algorithm, and outputs an AGV path planning optimization scheme. Aiming at the constraint problem of an existing AGV scheduling system in the aspects of scheduling quantity, data processing and algorithm optimization, the characteristics of large bandwidth, low time delay and multiple connections of the 5G Internet of Things are applied, and AGV path planning, real-time scheduling and position tracking are more flexible through a mode of combining an AGV management and control system and an artificial intelligence algorithm; the transportation efficiency and stability are improved, and the application of 5G enabled intelligent manufacturing is realized.

A server has a Guaranteed-Bandwidth (GB) Array 21 and an Extra-Bandwidth (EB) Array 31 with differing quality of Service requirements for transmission of cells held in queues. The server operating to distribute queues around a circular array until there is, at most, one queue per array location in order to minimise cell delay variation.