CN113283692B - Intelligent man-machine cooperation scheduling method and system for supervising resource allocation in bulk commodity trade market - Google Patents
Intelligent man-machine cooperation scheduling method and system for supervising resource allocation in bulk commodity trade market Download PDFInfo
- Publication number
- CN113283692B CN113283692B CN202110296445.9A CN202110296445A CN113283692B CN 113283692 B CN113283692 B CN 113283692B CN 202110296445 A CN202110296445 A CN 202110296445A CN 113283692 B CN113283692 B CN 113283692B
- Authority
- CN
- China
- Prior art keywords
- task
- tasks
- time
- scheduling
- resource
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000013468 resource allocation Methods 0.000 title claims description 19
- 238000012545 processing Methods 0.000 claims abstract description 5
- 239000002243 precursor Substances 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 5
- 230000003111 delayed effect Effects 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000009123 feedback regulation Effects 0.000 claims description 2
- 230000001960 triggered effect Effects 0.000 claims description 2
- 230000001105 regulatory effect Effects 0.000 abstract description 5
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 229920000742 Cotton Polymers 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- -1 energy Substances 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Data Mining & Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an intelligent man-machine cooperation scheduling method and system for regulating and managing resources in a commodity trade market, which are used for intelligently scheduling and distributing man-machine resources in real time according to task requirements so as to improve the task completion rate, wherein when a task arrives, a task processing module is used for solving part of a critical path set and distributing and calculating the deadline according to the topological graph relationship formed by different links of the task so as to obtain a task queue to be executed and a task pool; then, intelligent real-time allocation of human and machine resources is carried out through a resource scheduling module, and the execution sequence and specific starting time of the tasks are output; and automatically adjusting a scheduling and distributing strategy according to the execution information fed back by the tasks in real time, so that as many tasks as possible are completed before the deadline. The scheduling method gives consideration to the influence of task execution uncertainty, and has the advantages of high robustness of scheduling plan and low task delay risk compared with the prior method of 'first come first serve' and single fixed allocation.
Description
Technical Field
The invention relates to an intelligent man-machine cooperation scheduling technology, and belongs to an intelligent man-machine cooperation scheduling method and system for monitoring resource allocation in a commodity trade market.
Background
With the rapid development of internet technology, electronic commerce market development in bulk commodity fields is rapid, and transaction platforms are increasing. The commodity trade market involves national strategic materials such as energy, mineral products, cotton, grain and oil, and has the characteristics of large trade quantity, large price fluctuation, large trade risk, large influence on radiation, and the like. In recent years, a series of risk events and industry mess of a large number of commodity electronic commerce markets reflect the general current situation and serious problems of difficult market supervision and poor platform service in the current stage.
The current commodity electronic commerce platform mainly provides commodity transaction related services for clients and cooperates with the supervision functions of transaction data reporting and the like provided by a clearing house. Although many functions have been basically intelligent and automated, some processes remain semi-automated, requiring the participation of specialized personnel. Professionals utilize professional background knowledge in conjunction with the efficient performance of computing resources to co-accomplish tasks such as wind control intervention, end of day settlement, reconciliation, etc. Based on the current nationwide scene of a large number of supervision tasks, the number of workers of a supervision department or a platform is limited, and the method for improving the efficiency of a human-computer cooperation mode (shown in figure 1) has a research significance in order to achieve the aim of improving the quality of the completion of the supervision tasks and reduce the possible loss caused by task lag.
Generally, a supervisory task is composed of a plurality of different links with precedence relationships, and involves personnel of different departments and platforms to participate in execution, and each task has its own start time, deadline, estimated execution time and required attributes of the executives. All tasks form a task flow graph, each task schedule needs to be distributed to different people for execution, as shown in fig. 1, the tasks have complex topological structures, and the matching relationship of the people and the tasks, the task execution sequence and the specific starting time need to be clear. But the completion time of the task is uncertain and the time follows a certain normal distribution (which can be fitted by historical data). As shown in example fig. 2, each task has an average completion time μ, a completion time variance σ, and a required resource r, nodes 1, 2, 3 are different links of the same task, nodes 4, 5 are different links of the same task, node 0 and node 6 are added start and end nodes, and all attribute values are 0. It can be seen that in the formed task topology, the task represented by each node must be completed when all of its predecessor tasks have been completed. However, uncertainty in task completion time and limited resources can result in delayed completion of the task, which can result in significant economic losses for the bulk commodity trade market. Therefore, the scheduling plan is expected to have good robustness, the risk brought by completion of task delay can be dealt with, task scheduling is allocated to suitable personnel to be executed as far as possible before the deadline comes, and the efficiency and quality of monitoring task scheduling execution are improved.
Disclosure of Invention
Technical problems: the invention aims to provide an intelligent man-machine cooperation scheduling method and system for regulating and managing resources in a commodity trade market, which are based on the characteristic of limited and heterogeneous supervisory resources in a regulating system, and perform resource scheduling and distribution on supervisory tasks in the system so as to improve task completion efficiency and resource utilization rate. The method and the system give consideration to the influence of task execution uncertainty, and avoid the problems of low scheduling plan robustness and high task delay risk caused by the prior 'first come first serve'.
The technical scheme is as follows: in a bulk commodity trade market supervision resource allocation system, a dispatching center node and a plurality of man-machine resources exist. Each task has a different topological structure, each node in the topological structure represents a different subtask and has corresponding resource requirements, so that how to dispatch and allocate different human and machine resources to participate in executing the supervisory task is a key of the technical scheme. The main technical scheme of the scheduling method is as follows:
the resources existing in the supervisory resource allocation system can be roughly divided into two types, a heterogeneous human resource with limited amount of expertise and a high-performance computing resource which can be regarded as unlimited. When a task of supervising a large commodity trade market occurs, the task execution time and efficiency are often determined by human resources, because the time efficiency of computing resource processing data is far higher than the working time efficiency of people, and many links require the participation of professional skills of operators, such as wind control intervention, day end settlement and the like. Therefore, the cooperative scheduling of human and machine resources is the key of the technical scheme. The execution of the task by personnel can bring uncertainty in execution time, and how to schedule limited personnel in real time can ensure the completion of the task on time to the greatest extent, so that the method becomes an important index for measuring the rationality of the method.
An intelligent man-machine cooperation scheduling method and system for supervising resource allocation in a commodity trade market, the method comprises the following steps:
(1) When the task reaches the dispatching center, the task is disassembled into a topological structure according to the supervision flow, and a source point and a sink point are added to form a directed acyclic graph formed by all the tasks;
(2) Judging whether available resources related to tasks exist or not according to all available human and machine resources in the system, and fitting the average value and variance of the completion time of different types of tasks and the resource demand according to historical data;
(3) Establishing optimization targets, time, resources and other constraints of the task delay risk according to the task topological graph, and judging whether the task arrives in real time;
(4) When the task reaches in real time, solving part of the critical path set, calculating the deadline of the child node, adding the executable task to a waiting queue, and placing the rest in a task pool; calculating the task priority of the waiting queue, and carrying out scheduling and distribution according to the attributes of the idle people and the machine resources; then according to feedback information of task execution conditions, releasing occupied person and machine resources, updating task deadlines and priorities, and moving executable tasks from a task pool to a waiting queue;
(5) When the task does not reach the real-time state, judging whether the task scale and the constraint variable exceed the threshold value, if so, rapidly solving according to a preset heuristic algorithm or evolutionary algorithm, if not, calculating an accurate solution by using a branch-and-bound or integer programming solver, and then obtaining a scheduling scheme with low delay risk and high robustness, sequentially executing the tasks according to the scheduling scheme and feeding back the execution condition.
An intelligent man-machine cooperation scheduling method and system for supervising resource allocation in a commodity trade market, wherein the system comprises:
when a task reaches a dispatching center, the task processing module firstly decomposes the task into subtasks, and then solves part of critical path sets and calculates the deadline of the subtasks according to the topological graph relation formed by the subtasks, so as to obtain a task to-be-executed queue and a task pool (the subtasks which cannot be immediately executed);
the resource scheduling module is used for carrying out intelligent real-time allocation of idle human and machine resources along with execution of the tasks, and determining the matching relation between operators and the tasks, the execution sequence of the tasks and the specific starting time;
and the feedback adjustment module is used for automatically adjusting the scheduling allocation strategy according to the specific execution information (the completion time, the execution condition and the like) fed back by the task in real time. By adjusting the scheduling strategy in real time, as many tasks as possible are allocated to proper resources, and the tasks can be completed before the deadline, so that risks and losses brought to the trade market by task delay are avoided.
As a preferable technical scheme of the invention, the invention is characterized in that: in the system, n supervision tasks are totally added, and each supervision task can be represented by one five-tuple:V i ，E i g for directed acyclic graph composed of n supervision tasks respectively representing subtask point and edge set in directed acyclic graph<V，E>Representing task T i At A i Time of arrival, W i Representing the set of predicted working times for each subtask, D i Represents the latest completion time of the task, B i ＝num(V i ) Representing task T i Is a subtask number of (1). Assume that there are multiple types R of supervisory resources i Each subtask nodeAll require some type of resource r ij Estimated working time w ij Standard deviation is sigma ij . Each policing resource may be represented by a triplet: r is R i ＝{TYPE i ，NUM i ，OFFLOAD i Respectively representing the type of resource (human resources representing its work skills, computing resources representing its computing power), quantity and execution time load. In the directed acyclic graph formed by all tasks, only after all the precursor nodes (tasks) are completed, the nodes (tasks) can be continuously completed.
Under uncertain conditions, assume a task completion time d i，j Is subject to a mean and standard deviation of mu (d i，j ) Sum sigma (d) i，j ) Is a normal distribution of tasks σ (d i，j ) The larger it is, the higher the instability of its completion time, and the greater the risk of delay. Defining the deferred risk weight of a task as
Since the actual start time and the completion time of any one of the preceding tasks are directly affected after the completion time of the task is delayed, i.e. the delay risk of the current task is increased due to the preceding activity, the accumulated delay risk of the task is
The edges (i, n) belong to the task topological graph set G < V, E >, and the robustness optimization index is
S i，j Representing task v i，j Is s i，j Representative is the start time of the task reference dispatch plan, N i，j Representing task v i，j Is a precursor task number of (1).
As a preferred technical scheme of the invention, all tasks in the step (1) form a directed acyclic graph, part of critical paths in the directed acyclic graph are marked reversely from a sink through a part of critical path set solving algorithm until no unallocated precursor nodes exist, a part of critical paths are obtained, and the process is repeated until all nodes finish marking, so that a critical path set is obtained. Then traversing all subtask nodes in the critical path set, wherein the current node is v i，j When according to the formula
The calculation of the deadline is performed for each node. M is M est ，M eft ，M lct The earliest start time, the earliest completion time and the latest completion time are calculated according to the topological relation of each node of the directed acyclic graph. v i，1 Representing the earliest task in a critical path, v i，k Representing the task that starts the latest. And after the calculation of the deadline is completed, adding all subtasks to a task pool to be allocated, and waiting for resource allocation.
As a preferred technical scheme of the present invention, the tasks that can be allocated in the step (4) exist in a task queue, according to the calculated critical degree of the task, the scheduling center matches the task with a proper high priority for each idle human resource, and assists the computing resource to assist the operator in performing task execution, when one task is completed, the executable subsequent task will be added into the waiting queue from the task pool.
As a preferable technical scheme of the invention, the task in the waiting queue is required to be subjected to priority calculation before the resource scheduling process in the step (4). All the time stored in the waiting queue is the task which can be allocated to the resource to start execution, namely the precursor task is completed. Assume that the current system time is t system ，ω(w i，j ) Representing the initial estimated execution time, sigma being the standard deviation of the average execution time, subtask v i，j Priority P of (2) i，j Definition:
sequencing the emergency degree of the tasks according to the priority, scheduling a resource scheduling algorithm by a system scheduling center, distributing real-time idle resources in the system according to the priority, and obtaining task nodes v of the resources i，j Removed from the queue, the corresponding operator (human resource) is assigned to task v i，j Subsequent execution is performed while the current resource is placed in an occupied state.
As a preferable technical scheme of the invention, after each round of resource scheduling is completed, a corresponding operator performs specific execution of the task, and when the task is actually completed, the actual task completion time is needed to readjust d sub (v i，j ) And adding the new tasks of which all the precursor tasks have been completed to the waiting queue. Assume the current task v i，j The actual moment of completion of execution is λ (v i，j ) The maximum execution time of the subtasks reserved before is adjusted to
δ(v i，j )＝d sub (v i，j )-M est (v i，j )，
Corresponding updated earliest start timev i，p Representing v i，j Is a precursor node of (c).
The task pool and the subsequent subtask deadlines in the wait queue then also need to be recalculated,
task v completed for current execution i，j Current task v of any subsequent task of (c) i，k If task v i，k All the precursor tasks of (a) have been completed except that a process using lambda (v i，j ) To update the deadlines of other subtasks, also requiring v i，k Removed from the task pool and added to the wait queue, v at that time i，k Resource allocation may be performed.
As a preferable technical scheme of the invention, the invention is characterized in that: every time a task is completed, the resource is idle, and the feedback regulation and resource scheduling functions are sequentially called until all tasks are allocated with corresponding resources and are completed. When a new task arrives, the corresponding data needs to be decomposed and calculated for the task, and subtasks are added to a waiting queue or a task pool. And when the task is completed, the supervision resource occupation state is changed into an idle state, the signal is given to trigger feedback adjustment to update and adjust the task state of the task pool and the waiting queue, and then the resource allocation is triggered and invoked to allocate the idle state resources.
Assuming that the initial task number in the system is n, each task flow can be disassembled into subtask flows with topological structures, different subtasks need different resources, and all task links can form a directed acyclic graph. The completion time and the characteristics and the number of the required human and machine resources can be estimated for each subtask link according to the historical data.
The beneficial effects are that:
(1) The task completion rate is improved: the man-machine cooperation scheduling method can dynamically adjust resources in the whole system according to the characteristics and the requirements of the tasks, and the task completion rate is improved by timely feedback adjustment of the tasks and reasonable reservation of the task delay time.
(2) The resource utilization efficiency is improved: the method has the advantages that the quantity of the supervision resources (mainly human resources, relatively abundant high-performance computing resources) in the commodity trade market is limited, and the commodity trade market has a certain skill characteristic, plays a key role in task execution, and can better utilize the resources and avoid idle waste of the resources due to the fact that the resources are dynamically scheduled and allocated to a large number of supervision tasks all the time.
(3) High robustness and low risk of delay: in a bulk commodity trade market supervision resource allocation system, the supervision task flow is complex, a plurality of links are needed, different personnel use professional skills, and high-performance computing resources are assisted to supervise and examine the trade. The man-machine cooperation scheduling method effectively reduces task execution uncertainty caused by personnel participation, and reduces delay risk through online scheduling.
Drawings
FIG. 1 is a human-machine collaboration mode
FIG. 2 is a schematic diagram of a task topology
FIG. 3 is a schematic diagram of the main functions of a dispatch system
FIG. 4 is a principal sketch of the method of the invention
Detailed Description
The invention discloses an intelligent man-machine cooperation scheduling method and system for regulating and managing resources in a commodity trade market, which are used for carrying out resource scheduling and distribution on supervisory tasks in a system based on the characteristic of limited and heterogeneous supervisory resources in a regulating and managing system so as to improve task completion efficiency and resource utilization rate. As shown in fig. 2-4. The method comprises the following steps:
(1) Assuming that n supervision tasks are totally carried out in the system, after the task arrives, task decomposition is carried outEach administrative task may be represented by a five-tuple:V i ，E i g for directed acyclic graph composed of n supervision tasks respectively representing subtask point and edge set in directed acyclic graph<V，E>Representing task T i At A i Time of arrival, W i Representing the set of predicted working times for each subtask, D i Represents the latest completion time of the task, B i ＝num(V i ) Representing task T i Is a subtask number of (1). Assume that there are multiple types R of supervisory resources i Each subtask node->The part needs some kind of resource r ij Estimated working time w ij Standard deviation is sigma ij . Each policing resource may be represented by a triplet: r is R i ＝{TYPE i ，NUM i ，OFFLOAD i Respectively representing the type of resource (human resources representing its work skills, computing resources representing its computing power), quantity and execution time load. In the directed acyclic graph formed by all tasks, only after all the precursor nodes (tasks) are completed, the nodes (tasks) can be continuously completed. As shown in fig. 2, one task is composed of three subtasks of nodes 1, 2 and 3, the other task is composed of nodes 4 and 5, nodes 0 and 6 are added source points and sink points, arrows between tasks represent a sequential dependency relationship, and three numerical values of each node represent an average value, a standard deviation and required resource attributes of estimated working time of the task respectively.
Under uncertain conditions, assume a task completion time d i，j Is subject to a mean and standard deviation of mu (d i，j ) Sum sigma (d) i，j ) Is a normal distribution of tasks σ (d i，j ) The larger it is, the higher the instability of its completion time, and the greater the risk of delay. Defining the deferred risk weight of a task as
Since the actual start time and the completion time of any one of the preceding tasks are directly affected after the completion time of the task is delayed, i.e. the delay risk of the current task is increased due to the preceding activity, the accumulated delay risk of the task is
The edges (i, n) belong to the task topological graph set G < V, E >, and the robustness optimization index is
S i，j Representing task v i，j Is s i，j Representative is the start time of the task reference dispatch plan, N i，j Representing task v i，j Is a precursor task number of (1).
(2) A certain deferrable time is reserved considering that a task deadline is assigned to each subtask in the topology. Since all tasks constitute one directed acyclic graph, it is first considered how to reasonably assign deadlines to different subtasks. And (3) reversely marking part of the critical paths in the directed acyclic graph from the sink through a part of critical path set solving algorithm until no unallocated precursor nodes exist to obtain a part of critical paths, and repeating the process until all nodes finish marking to obtain a critical path set. Then traversing all subtask nodes in the critical path set, wherein the current node is v i，j When the time of the cut-off is calculated as
M est ，M eft ，M lct The earliest starting time calculated according to the topological relation of each node of the directed acyclic graph,Earliest completion time and latest completion time. v i，1 Representing the earliest task in a critical path, v i，k Representing the task that starts the latest. And after the calculation of the deadline is completed, adding all subtasks to a task pool to be allocated, and waiting for resource allocation. As in fig. 2, the first addition to the waiting queue is task nodes 1 and 4 (node 0 is the source point of the active addition), and the task pool is nodes 2, 3, 5, and 6.
(3) Tasks in the waiting queue need to be subjected to priority calculation before the resource scheduling process. All the time stored in the waiting queue is the task which can be allocated to the resource to start execution, namely the precursor task is completed. Assume that the current system time is t system ，ω(w i，j ) Representing the initial estimated execution time, sigma being the standard deviation of the average execution time, subtask v i，j Priority P of (2) i，j Definition:
sequencing the emergency degree of the tasks according to the priority, scheduling a resource scheduling algorithm by a system scheduling center, distributing real-time idle resources in the system according to the priority, and obtaining task nodes v of the resources i，j Removed from the queue, the corresponding operator (human resource) is assigned to task v i，j Subsequent execution is performed while the current resource is placed in an occupied state. As shown in fig. 3, the waiting queue (i.e. the preparation task queue) is a task set that is obtained after the task processing and can be executed in real time, the emergency degree of the task is determined according to the priority, and then the real-time available resources (operator resources and computing resources) are allocated to the task with the highest emergency degree through the resource scheduling, and the corresponding tasks and the resources form a corresponding relationship, so that the resource state is occupied (green) as shown in the figure.
(4) After each round of resource scheduling is completed, the corresponding operator performs specific execution of the task, however, the specific completion condition and completion time of the task cannot be estimated accurately. When the task is actually completed, d needs to be readjusted with the actual task completion time sub (v i，j ) And adding the new tasks of which all the precursor tasks have been completed to the waiting queue. Assume the current task v i，j The actual moment of completion of execution is λ (v i，j ) The maximum execution time of the subtasks reserved before is adjusted to
δ(v i，j )＝d sub (v i，j )-M est (v i，j )，
Corresponding updated earliest start timev i，p Representing v i，j Is a precursor node of (c). The task pool and the subsequent subtask deadlines in the wait queue also need to be re-updated for computation,
task v completed for current execution i，j Current task v of any subsequent task of (c) i，k If task v i，k All the precursor tasks of (a) have been completed except that a process using lambda (v i，j ) To update the deadlines of other subtasks, also requiring v i，k Removed from the task pool and added to the wait queue, v at that time i，k Resource allocation may be performed. As shown in fig. 2, after the subtask nodes 1 and 4 schedule and allocate to the corresponding operators for execution, if the node 1 actually completes execution, the corresponding operators can allocate, the estimated deadlines of the subtask nodes 2 and 3 need to be recalculated according to the actual completion time of the node 1, and at the same time, the nodes 2 and 3 need to be removed from the task pool and added to a waiting queue to wait for resource scheduling allocation.
(5) And (3) when the tasks are completed, the resources are idle, and the feedback adjustment in the step (4) and the resource scheduling in the step (3) are sequentially invoked until all the tasks are allocated with corresponding resources and are executed. And when a new task arrives, the step (1) is required to be called, corresponding data are decomposed and calculated for the task, and subtasks are added to a waiting queue or a task pool. As shown in fig. 3, each time a task is completed, the supervision resource occupancy state (green) is changed to an idle state (white), and a signal is given to trigger feedback adjustment to update and adjust the task state of the task pool and the waiting queue, and then trigger the allocation of the calling resource, and the resource in the idle state is allocated.
Claims (3)
1. An intelligent man-machine cooperation scheduling method for supervising resource allocation in a commodity trade market is characterized in that: comprising a scheduling system, the system comprising:
when a task reaches a dispatching center, the task processing module firstly decomposes the task into subtasks, and then solves part of critical path sets and calculates the deadline of the subtasks according to the topological graph relationship formed by the subtasks, so as to obtain a task queue to be executed and a task pool; the task pool represents subtasks which cannot be immediately executed;
the resource scheduling module is used for carrying out intelligent real-time allocation of idle human and machine resources along with execution of the tasks, and determining the matching relation between operators and the tasks, the execution sequence of the tasks and the specific starting time;
the feedback adjustment module feeds back specific execution information in real time according to the task; scheduling allocation strategy after time and execution condition are automatically adjusted; the scheduling strategy is adjusted in real time, so that the tasks are distributed to proper resources and can be completed before the deadline, and the risk and loss caused by the delay of the tasks to the trading market are avoided;
in the system, n supervision tasks are totally added, and each supervision task can be represented by one five-tuple: t (T) i ＝{V i ，E i ，A i ，W i ，D i }，V i ，E i G for directed acyclic graph composed of n supervision tasks respectively representing subtask point and edge set in directed acyclic graph<V，E>Representing task T i At A i Time of arrival, W i Representing each subtaskPredicted working time set, D i Represents the latest completion time of the task, B i ＝num(V i ) The method comprises the steps of carrying out a first treatment on the surface of the Representing task T i Is the number of subtasks; assume that there are multiple types R of supervisory resources i Each subtask node->All require some type of resource r ij Estimated working time w ij Standard deviation is sigma ij The method comprises the steps of carrying out a first treatment on the surface of the Each policing resource may be represented by a triplet: r is R i ＝{TYPE i ，NUM i ，OFFLOAD i Respectively representing the types of resources, wherein the resources represent the work skills and the computing resources represent the computing power; number and execution time load; in the directed acyclic graph formed by all tasks, the tasks of all the precursor nodes can be continuously completed only after the tasks of the nodes are completed;
under uncertain conditions, assume a task completion time d i，j Is subject to a mean and standard deviation of mu (d i，j ) Sum sigma (d) i，j ) Is a normal distribution of tasks σ (d i，j ) The larger it is, the higher the instability of its completion time, the greater the risk of delay; defining the deferred risk weight of a task as
Since the actual start time and the completion time of any one of the preceding tasks are directly affected after the completion time of the task is delayed, i.e. the delay risk of the current task is increased due to the preceding activity, the accumulated delay risk of the task is
The edges (i, n) belong to the task topological graph set G < V, E >, and the robustness optimization index is
S i，j Representing task v i，j Is s i，j Representative is the start time of the task reference dispatch plan, N i，j Representing task v i，j Precursor task number of (2);
the method comprises the following steps:
(1) When the task reaches the dispatching center, the task is disassembled into a topological structure according to the supervision flow, and a source point and a sink point are added to form a directed acyclic graph formed by all the tasks; all tasks in the step (1) form a directed acyclic graph, part of the critical paths in the directed acyclic graph are marked reversely from a sink through a part of critical path set solving algorithm until unallocated precursor nodes do not exist, a part of critical paths are obtained, and the process is repeated until all nodes finish marking, so that a critical path set is obtained; then traversing all subtask nodes in the critical path set, wherein the current node is v i，j When according to the formula
Calculating the deadline of each node; m is M est ，M eft ，M lct The earliest starting time, the earliest finishing time and the latest finishing time are calculated according to the topological relation of each node of the directed acyclic graph; v i，1 Representing the earliest task in a critical path, v i，k Representing the task that started the latest; after the calculation of the deadline is completed, adding all subtasks to a task pool to be allocated, and waiting for resource allocation
(2) Judging whether available resources related to tasks exist or not according to all available human and machine resources in the system, and fitting the average value and variance of the completion time of different types of tasks and the resource demand according to historical data;
(3) Establishing an optimization target and time and resource constraint of a task delay risk according to a task topological graph, and judging whether a task arrives in real time;
(4) When the task reaches in real time, solving part of the critical path set, calculating the deadline of the child node, adding the executable task to a waiting queue, and placing the rest in a task pool; calculating the task priority of the waiting queue, and carrying out scheduling and distribution according to the attributes of the idle people and the machine resources; then according to feedback information of task execution conditions, releasing occupied person and machine resources, updating task deadlines and priorities, and moving executable tasks from a task pool to a waiting queue;
the tasks which can be executed and allocated in the step (4) exist in a task queue, the scheduling center matches the tasks with proper priority for each idle human resource according to the calculated critical degree of the tasks and assists the computing resource to assist an operator in executing the tasks, and when one task is completed, the executable subsequent tasks are added into a waiting queue from a task pool; the tasks in the waiting queue are required to be subjected to priority calculation before the resource scheduling process in the step (4); all the time, the task which can be allocated to the resource to start execution is stored in the waiting queue, namely the precursor task is completed; assume that the current system time is t system ，ω(w i，j ) Representing the initial estimated execution time, sigma being the standard deviation of the average execution time, subtask v i，j Priority P of (2) i，j Definition:
sequencing the emergency degree of the tasks according to the priority, scheduling a resource scheduling algorithm by a system scheduling center, distributing real-time idle resources in the system according to the priority, and obtaining task nodes v of the resources i，j Removed from the queue, the corresponding operator is assigned to task v i，j Performing subsequent execution while the current resource is set to an occupied state;
(5) When the task does not reach the real-time state, judging whether the task scale and the constraint variable exceed the threshold value, if so, rapidly solving according to a preset heuristic algorithm or evolutionary algorithm, if not, calculating an accurate solution by using a branch-and-bound or integer programming solver, and then obtaining a scheduling scheme with low delay risk and high robustness, sequentially executing the tasks according to the scheduling scheme and feeding back the execution condition.
2. The intelligent man-machine cooperation scheduling method for bulk commodity transaction market supervision resource allocation according to claim 1, wherein the method is characterized in that: after each round of resource scheduling is completed, the corresponding operator performs specific execution of the task, and when the task is actually completed, the actual task completion time is needed to readjust d sub (v i，j ) Adding the task which is completed by all the new precursor tasks to a waiting queue; assume the current task v i，j The actual moment of completion of execution is λ (v i，j ) The maximum execution time of the subtasks reserved before is adjusted to delta (v i，j )＝d sub (v i，j )-M est (v i，j )，
Corresponding updated earliest start timev i，p Representing v i，j Is a precursor node of (2);
the task pool and the subsequent subtask deadlines in the wait queue then also need to be recalculated,
task v completed for current execution i，j Current task v of any subsequent task of (c) i，k If task v i，k All of the precursor tasks of (c) have been completed except that the use of (v i，j ) To update the deadlines of other subtasks, also requiring v i，k Removed from the task pool and added to the wait queue, v at that time i，k Resource allocation can be performed。
3. The intelligent man-machine cooperation scheduling method for bulk commodity transaction market supervision resource allocation according to claim 1, wherein the method is characterized in that: every time a task is completed, the resource is idle, and the feedback regulation and resource scheduling functions are sequentially called until all tasks are allocated with corresponding resources and are executed; when a new task arrives, the corresponding data are needed to be decomposed and calculated for the task, and subtasks are added into a waiting queue or a task pool; and when the task is completed, the supervision resource occupation state is changed into an idle state, the signal is given to trigger feedback adjustment to update and adjust the task state of the task pool and the waiting queue, and then the resource allocation is triggered and invoked to allocate the idle state resources.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110296445.9A CN113283692B (en) | 2021-03-19 | 2021-03-19 | Intelligent man-machine cooperation scheduling method and system for supervising resource allocation in bulk commodity trade market |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110296445.9A CN113283692B (en) | 2021-03-19 | 2021-03-19 | Intelligent man-machine cooperation scheduling method and system for supervising resource allocation in bulk commodity trade market |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113283692A CN113283692A (en) | 2021-08-20 |
| CN113283692B true CN113283692B (en) | 2024-04-16 |
Family
ID=77275950
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110296445.9A Active CN113283692B (en) | 2021-03-19 | 2021-03-19 | Intelligent man-machine cooperation scheduling method and system for supervising resource allocation in bulk commodity trade market |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113283692B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113918323B (en) * | 2021-09-17 | 2022-10-21 | 中标慧安信息技术股份有限公司 | High-energy-efficiency computing task allocation method and device in edge computing |
| WO2023242942A1 (en) * | 2022-06-14 | 2023-12-21 | 日本電信電話株式会社 | Schedule execution assistance device, method, and program |
| CN115423603B (en) * | 2022-08-31 | 2023-05-23 | 厦门国际银行股份有限公司 | Wind control model building method, system and storage medium based on machine learning |
| CN115936331B (en) * | 2022-11-01 | 2023-10-27 | 中科雨辰科技有限公司 | Data processing system for acquiring target task object based on distance |
| CN115392814A (en) * | 2022-11-01 | 2022-11-25 | 中科雨辰科技有限公司 | Data processing system for configuration task |
| CN117095804B (en) * | 2023-10-19 | 2023-12-29 | 江苏健康无忧网络科技有限公司 | Medical examination idle period scheduling method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345501A (en) * | 2017-01-24 | 2018-07-31 | 全球能源互联网研究院 | A kind of distributed resource scheduling method and system |
| CN108920261A (en) * | 2018-05-23 | 2018-11-30 | 中国航天***科学与工程研究院 | A kind of two-stage self-adapting dispatching method suitable for large-scale parallel data processing task |
| CN108958919A (en) * | 2018-07-13 | 2018-12-07 | 湘潭大学 | More DAG task schedule expense fairness assessment models of limited constraint in a kind of cloud computing |
-
2021
- 2021-03-19 CN CN202110296445.9A patent/CN113283692B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345501A (en) * | 2017-01-24 | 2018-07-31 | 全球能源互联网研究院 | A kind of distributed resource scheduling method and system |
| CN108920261A (en) * | 2018-05-23 | 2018-11-30 | 中国航天***科学与工程研究院 | A kind of two-stage self-adapting dispatching method suitable for large-scale parallel data processing task |
| CN108958919A (en) * | 2018-07-13 | 2018-12-07 | 湘潭大学 | More DAG task schedule expense fairness assessment models of limited constraint in a kind of cloud computing |
Non-Patent Citations (1)
| Title |
|---|
| 基于网格资源超图模型的可信任务调度;郭美云;杨博;陈志刚;;计算机工程(第13期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113283692A (en) | 2021-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113283692B (en) | Intelligent man-machine cooperation scheduling method and system for supervising resource allocation in bulk commodity trade market | |
| Márkus et al. | A market approach to holonic manufacturing | |
| Krothapalli et al. | Design of negotiation protocols for multi-agent manufacturing systems | |
| Saad et al. | Performance evaluation of contract net-based heterarchical scheduling for flexible manufacturing systems | |
| CN101237469B (en) | Method for optimizing multi-QoS grid workflow based on ant group algorithm | |
| CN113705866B (en) | Scheduling optimization method and system based on resource-constrained project scheduling problem model | |
| CN103279818A (en) | Method for cloud workflow scheduling based on heuristic genetic algorithm | |
| CN108711007A (en) | A kind of multitask real-time scheduling method of energy integration collection system | |
| Wang et al. | Synchronizing production scheduling with resources allocation for precast components in a multi-agent system environment | |
| CN106327053B (en) | Construction method of weaving process recommendation model based on multi-mode set | |
| US6731999B1 (en) | Wafer start order release algorithm in a foundry fab | |
| CN116258337A (en) | Industry chain collaborative management system based on enterprise manufacturing operation | |
| Biswas et al. | Predictive auto-scaling techniques for clouds subjected to requests with service level agreements | |
| Zhou et al. | DPS: Dynamic pricing and scheduling for distributed machine learning jobs in edge-cloud networks | |
| WO2023202326A1 (en) | Data processing method and related apparatus | |
| CN109784543B (en) | Balance scheduling method based on weighted round robin scheduling | |
| CN110969351A (en) | Production task scheduling method for multi-task cross-factory cooperation | |
| Koruca et al. | A priority rule based production scheduling module on faborg-sim simulation tool | |
| CN114862133A (en) | Knowledge-fused manufacturing service network resource scheduling method and system | |
| CN113723936A (en) | Power engineering quality supervision and management method and system | |
| CN114493199B (en) | Intelligent cloud scheduling method based on small-sized cross-medium enterprises | |
| Shao et al. | Fairness scheduling for tasks with different real-time level on heterogeneous systems | |
| JP2003022119A (en) | Multiproduct production system, its design/operation method, its design/operation program and recording medium recorded with the program | |
| CN113635302B (en) | Integrated mobile cooperative robot control system based on field bus | |
| Salmani et al. | Multi-criteria Scheduling of Soft Real-time Tasks on Uniform Multiprocessors Using Fuzzy Inference |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |