Intelligent task allocation method and system based on data processing

Abstract

The present invention relates to the field of data processing technology, and in particular to a method and system for intelligent task allocation based on data processing; a comprehensive balance is obtained according to the average level of the quantitative difference characteristics of the total out-degree and total in-degree of all task nodes in a directed acyclic graph and the quantitative difference characteristics of the out-degree and in-degree of the task nodes; a comprehensive sparsity is obtained according to the density of the quantitative difference characteristics of the directed edges and task nodes in the directed acyclic graph and the intersection characteristics of the directed edges; a sorting evaluation degree is obtained according to the comprehensive balance degree and the comprehensive sparsity; and a sorting mode of tasks is judged according to the sorting evaluation degree and a task dependency sequence is obtained. A task complexity model is constructed according to the integration of the task dependency sequence, the hardware required for task processing, the task processing time and the task priority, and tasks are allocated according to the task complexity model, thereby improving the efficiency of task allocation and processing.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for intelligent task allocation based on data processing. Background Art

[0002] Intelligent task allocation is an important part of cloud computing technology. It efficiently allocates tasks submitted by users to computing resources in the cloud platform, thereby maximizing the utilization of computing resources and efficiently completing tasks. Intelligent task allocation mainly uses relevant algorithms and data analysis to allocate tasks to the most suitable computing resources according to the computing resource requirements, priority and task dependencies of tasks. At present, the main task intelligent allocation processing method is based on the task complexity model, which matches tasks and computing resources according to the computing resource requirements, task processing time, priority and task dependencies of tasks. When analyzing the dependencies between different tasks, a directed acyclic graph is usually established, and the dependencies of tasks are obtained through a topological sorting algorithm based on the directed acyclic graph. Since the dependencies between multiple tasks are relatively complex, different topological sorting algorithms will lead to differences in the acquisition speed of task sequences. If the task sequence is acquired slowly, it will make it difficult for the complexity model to quickly match tasks with computing resources, resulting in low efficiency in task allocation and processing in the cloud computing platform. Summary of the invention

[0003] In order to solve the technical problem that the use of different topological sorting algorithms will lead to differences in the speed of obtaining task dependencies, resulting in low efficiency in task allocation and processing in the cloud computing platform, the purpose of the present invention is to provide a method and system for intelligent task allocation based on data processing, and the technical solutions adopted are as follows:

[0004] Get the directed acyclic graph between related tasks;

[0005] A first balance degree is obtained according to the quantitative difference characteristics of the total out-degree and the total in-degree of all task nodes in the directed acyclic graph; a second balance degree is obtained according to the average level of the quantitative difference characteristics of the out-degree and in-degree of the task nodes in the directed acyclic graph; a comprehensive balance degree of the directed acyclic graph is obtained according to the first balance degree and the second balance degree;

[0006] A first sparsity is obtained according to the quantity difference characteristics of the directed edges and task nodes in the directed acyclic graph; a second sparsity is obtained according to the density of the intersection characteristics of the directed edges in the directed acyclic graph; a comprehensive sparsity of the directed acyclic graph is obtained according to the first sparsity and the second sparsity; a sorting evaluation degree of the directed acyclic graph is obtained according to the comprehensive balance degree and the comprehensive sparsity; a sorting method of the tasks is determined according to the sorting evaluation degree and the tasks are sorted to obtain a task dependency sequence;

[0007] A task complexity model is constructed based on the task dependency sequence, the hardware required for task processing, the task processing time and the task priority; and different tasks are allocated and processed based on the task complexity model.

[0008] Furthermore, the step of obtaining a first balance degree according to the quantity difference characteristics of the total out-degree and the total in-degree of all task nodes in the directed acyclic graph includes:

[0009] The absolute value of the difference between the total out-degree and the total in-degree of all task nodes is calculated to obtain the in-degree difference value; the ratio of the in-degree difference value to the total degree is calculated and negatively correlated to obtain the first balance degree of the directed acyclic graph.

[0010] Furthermore, the step of obtaining a second balance degree according to the average level of the quantity difference characteristics of the out-degree and in-degree of the task nodes in the directed acyclic graph comprises:

[0011] The absolute value of the difference between the number of out-degree and in-degree of any task node is calculated to obtain the distribution difference of the arbitrary task node; the average value of the distribution difference of all task nodes is calculated and negatively correlated to obtain the second balance of the directed acyclic graph.

[0012] Furthermore, the step of obtaining a comprehensive balance degree of the directed acyclic graph according to the first balance degree and the second balance degree comprises:

[0013] The product of the first balance degree and a preset first weight is calculated to obtain a first balance factor; the product of the second balance degree and a preset second weight is calculated to obtain a second balance factor; and the sum of the first balance factor and the second balance factor is calculated to obtain a comprehensive balance degree of the directed acyclic graph.

[0014] Furthermore, the step of obtaining the first sparsity according to the quantity difference characteristics of the directed edges and task nodes in the directed acyclic graph includes:

[0015] The ratio of the number of directed edges to the number of task nodes in the directed acyclic graph is calculated and negatively correlated to obtain a first sparsity of the directed acyclic graph.

[0016] Furthermore, the step of obtaining the second sparsity according to the density of the intersection features of the directed edges in the directed acyclic graph comprises:

[0017] The number of intersections of the directed edges intersecting in the directed acyclic graph is calculated to obtain an intersection density eigenvalue; the ratio of the intersection density eigenvalue to the numerical value of the task node is calculated and negatively correlated to obtain a second sparsity of the directed acyclic graph.

[0018] Further, the step of obtaining the comprehensive sparsity of the directed acyclic graph according to the first sparsity and the second sparsity includes:

[0019] Calculate the product of the first sparsity and a preset third weight to obtain a first density factor; calculate the product of the second sparsity and a preset fourth weight to obtain a second density factor; calculate the sum of the first density factor and the second density factor to obtain the comprehensive sparsity of the directed acyclic graph.

[0020] Furthermore, the step of obtaining the ranking evaluation degree of the directed acyclic graph according to the comprehensive balance degree and the comprehensive sparsity degree includes:

[0021] The average values ​​of the comprehensive balance degree and the comprehensive sparsity degree are calculated to obtain the ranking evaluation degree of the directed acyclic graph.

[0022] Furthermore, the step of determining the order of tasks according to the order evaluation degree and performing ordering of tasks to obtain a task dependency sequence includes:

[0023] When the sorting evaluation degree exceeds a preset threshold, a depth-first search algorithm is used to sort the task nodes in the directed acyclic graph; when the sorting evaluation degree does not exceed the preset threshold, a breadth-first search algorithm is used to sort the task nodes in the directed acyclic graph to obtain the task dependency sequence.

[0024] The present invention also proposes a task intelligent allocation method system based on data processing, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any step of a task intelligent allocation method based on data processing.

[0025] The present invention has the following beneficial effects:

[0026] In the present invention, obtaining the first balance degree can reflect the structural balance of the graph and the complexity of the dependency between task nodes according to the difference in the number of out-degrees and in-degrees of the whole graph, and obtaining the second balance degree can reflect the complexity of the dependency between task nodes according to the difference in the number of out-degrees and in-degrees between each task node; and then obtaining the comprehensive balance degree can accurately characterize the complexity of the dependency between task nodes in the graph, thereby improving the accuracy of the final topological sorting algorithm selection. Obtaining the first sparsity degree can reflect the density of the directed acyclic graph according to the number of directed edges in the graph, and obtaining the second sparsity degree can reflect the density according to the number of intersections of directed edges in the graph; and then obtaining the comprehensive sparsity degree can accurately characterize the density of the directed acyclic graph and the complexity of the dependency between task nodes. Obtaining the sorting evaluation degree can more accurately determine the appropriate topological sorting algorithm, so that the acquisition efficiency of the task dependency sequence is higher, thereby improving the efficiency of the task complexity model for task allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0028] Figure 1 A flow chart of a method for intelligent task allocation based on data processing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a task intelligent allocation method and system based on data processing proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0030] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0031] The specific scheme of a task intelligent allocation method and system based on data processing provided by the present invention is described in detail below with reference to the accompanying drawings.

[0032] See also Figure 1, which shows a flow chart of a method for intelligent task allocation based on data processing provided by an embodiment of the present invention, the method comprising the following steps:

[0033] Step S1, obtaining a directed acyclic graph between associated tasks.

[0034] In an embodiment of the present invention, the implementation scenario is to accurately and quickly analyze the dependency relationship between different tasks, so that the task calculation and processing efficiency in the cloud computing platform is higher. Since there are a large number of task instructions that need to be calculated and processed in the cloud computing platform, some tasks have certain associations. Associated tasks refer to some tasks that need to be calculated after other tasks are calculated and processed, or some tasks need to be calculated before other tasks can be calculated and processed. There are dependencies between associated tasks. In order to maximize the resource utilization of the cloud computing platform and complete the tasks efficiently, it is necessary to topologically sort the associated tasks according to the dependency relationship, and assign tasks according to the sorting results and in combination with the calculation time, priority and other characteristics of the tasks. Therefore, firstly, a directed acyclic graph between related tasks is obtained. The cloud computing platform classifies related tasks into one category and constructs a directed acyclic graph of this category of tasks. It should be noted that the construction of a directed acyclic graph belongs to the existing technology, and the specific steps will not be repeated. The directed acyclic graph can reflect the dependency relationship between related tasks. A task node in the graph represents a task instruction that needs to be calculated and processed. If there is a directed connection line between a task node in the graph and other task nodes, it means that there is a requirement for sequential calculation of the two task nodes. For example, a task node points to another task node, which means that the calculation of the task node must be completed before the calculation of the other task node can be performed.

[0035] Furthermore, after obtaining the directed acyclic graph, the existing topological sorting algorithm can be used to sort the task dependencies according to the task pointing relationship in the graph, and the task nodes in the directed acyclic graph are sorted using the topological sorting algorithm to obtain a linear sequence, so that the tasks with the highest order are calculated first in the cloud computing platform; each task node appears only once in the linear sequence, and if a task node points to other task nodes, then the task node must be before the other task nodes in the linear sequence. The existing commonly used topological sorting algorithms include breadth-first search algorithm and depth-first search algorithm. The breadth-first search algorithm and depth-first search algorithm have different sorting efficiencies for different directed acyclic graph structures. The breadth-first search traverses the nodes layer by layer, starting from the source node and gradually expanding to the nodes in each layer. When processing dense graphs, it can explore all nodes and directed edges more evenly, so it is more suitable for situations where the task dependency relationship is complex and the directed acyclic graph is relatively dense, and the sorting acquisition speed is higher. As for depth-first search, a depth-first traversal starting from a node usually reaches the boundary of the graph as early as possible, avoiding unnecessary traversal. Therefore, it is more suitable for situations with fewer directed edges, simpler task dependencies, and sparse directed acyclic graphs, and the sorting acquisition speed is higher. In order to improve the speed of task allocation and the efficiency of task processing, it is necessary to select a topological sorting algorithm to meet the efficiency requirements of task calculation; therefore, it is necessary to select a suitable sorting algorithm based on the structural characteristics of the directed acyclic graph to make task allocation more efficient.

[0036] Step S2, obtaining a first balance degree according to the quantitative difference characteristics of the total out-degree and total in-degree of all task nodes in the directed acyclic graph; obtaining a second balance degree according to the average level of the quantitative difference characteristics of the out-degree and in-degree of the task nodes in the directed acyclic graph; obtaining a comprehensive balance degree of the directed acyclic graph according to the first balance degree and the second balance degree.

[0037] When the structure of the directed acyclic graph is more balanced and the number of out-degrees and in-degrees is closer, the task dependency sequence obtained using the depth-first search algorithm is faster and more efficient, so the first balance degree is obtained according to the difference characteristics of the total out-degrees and total in-degrees of all task nodes in the directed acyclic graph; preferably, in an embodiment of the present invention, the step of obtaining the first balance degree includes: calculating the absolute value of the difference between the total out-degrees and the total in-degrees of all task nodes to obtain the out-degree difference value; the out-degree refers to the number of edges starting from a task node, and the in-degree is the number of edges pointing to a task node; when the out-degree difference value is larger, it means that the dependency relationship between the task nodes in the directed acyclic graph is more chaotic, and some task nodes are more important, involving the calculation process of multiple other task nodes, and some task nodes are less important. Calculate the ratio of the in-degree difference value to the total degree and negatively map it to obtain the first balance degree of the directed acyclic graph; the total degree is the sum of the number of all out-degrees and in-degrees in the graph. When the out-degree difference value is larger, the structure in the directed acyclic graph is more unbalanced, the dependency relationship between the task nodes is more complex, and the first balance degree is smaller.

[0038] The first balance is obtained according to the difference in the number of in-degrees and out-degrees of the whole directed acyclic graph. In order to more accurately reflect whether the dependency relationship between the task nodes in the graph is balanced, the balance characteristics can be further analyzed according to the in-degree relationship of each task node in the graph. Therefore, the second balance is obtained according to the average level of the difference characteristics of the number of in-degrees and out-degrees of the task nodes in the directed acyclic graph; preferably, in an embodiment of the present invention, the step of obtaining the second balance includes: calculating the absolute value of the difference between the number of in-degrees and out-degrees of any task node, and obtaining the distribution difference of the arbitrary task node; when the distribution difference is larger, it means that the in-degrees and out-degrees of the task node are more unbalanced, and it may be that multiple other task nodes point to the task node or the task node points to multiple other task nodes, and the dependency relationship is more chaotic. Calculate the average value of the distribution difference of all task nodes and negatively map them to obtain the second balance of the directed acyclic graph; when the distribution difference of all task nodes is larger, the second balance is smaller, the structure of the directed acyclic graph is more unbalanced, and the dependency relationship of the task nodes in the graph is more complex and chaotic.

[0039] Furthermore, the comprehensive balance of the directed acyclic graph can be obtained according to the first balance and the second balance; preferably, in an embodiment of the present invention, the step of obtaining the comprehensive balance includes: calculating the product of the first balance and the preset first weight to obtain the first balance factor; calculating the product of the second balance and the preset second weight to obtain the second balance factor; in an embodiment of the present invention, the preset first weight is 0.4, and the preset second weight is 0.6, which can be determined by the implementer according to the implementation scenario. Calculate the sum of the first balance factor and the second balance factor to obtain the comprehensive balance of the directed acyclic graph. The larger the comprehensive balance, the more balanced the dependency between the task nodes in the directed acyclic graph, and each task node has a similar number of previous and successor tasks; thus, the task nodes can be sorted more quickly using the depth-first search algorithm. The formula for obtaining the comprehensive balance includes:

[0040]

[0041] In the formula, R represents the comprehensive balance degree, Indicates the preset first weight, Indicates the preset second weight, represents an exponential function with a natural constant as the base, S represents the difference between the in- and out-degrees, and M represents the total degree. represents the first balance degree, N represents the number of task nodes in the directed acyclic graph, Indicates the distribution difference of the nth task node, represents the average value of the distribution difference, represents the second degree of balance, represents the first balance factor, Represents the second balancing factor.

[0042] Step S3, obtaining a first sparsity according to the number difference characteristics of directed edges and task nodes in the directed acyclic graph; obtaining a second sparsity according to the density of the intersection characteristics of the directed edges in the directed acyclic graph; obtaining a comprehensive sparsity of the directed acyclic graph according to the first sparsity and the second sparsity; obtaining a sorting evaluation degree of the directed acyclic graph according to the comprehensive balance degree and the comprehensive sparsity; judging the sorting method of the tasks according to the sorting evaluation degree and performing task sorting to obtain a task dependency sequence.

[0043] Not only does the structural balance of the directed acyclic graph affect the efficiency of the topological sorting, but the density of the directed acyclic graph also affects the efficiency of the topological sorting. The depth-first search algorithm in the topological sorting is faster for sparse graphs with fewer directed edges and simpler dependencies between tasks, and can quickly and accurately find the task dependency order. For directed acyclic graphs with more complex dependencies, it takes a long time to sort. Therefore, the density of the directed acyclic graph can be analyzed to determine the algorithm for topological sorting. When there are more directed edges between task nodes in the graph, it means that the dependencies are more complex and the graph is denser. Therefore, the first sparsity is obtained according to the difference characteristics of the number of directed edges and task nodes in the directed acyclic graph. Preferably, in an embodiment of the present invention, the step of obtaining the first sparsity includes: calculating the ratio of the number of directed edges to task nodes in the directed acyclic graph and negatively mapping them to obtain the first sparsity of the directed acyclic graph; when the number of directed edges is more, the first sparsity is smaller, the directed edges in the directed acyclic graph are denser, and the dependencies of the task nodes are more complex.

[0044] Furthermore, when there are more intersections of directed edges in a directed acyclic graph, it means that the dependencies between task nodes are more complex, the distribution of directed edges in the graph is more chaotic, and the denser the graph is, the more it affects the efficiency of the depth-first search algorithm. Therefore, the second sparsity is obtained according to the density of the intersection features of directed edges in the directed acyclic graph; preferably, in an embodiment of the present invention, the step of obtaining the second sparsity includes: calculating the number of intersections of intersecting directed edges in the directed acyclic graph to obtain an intersection density eigenvalue; when there are more intersections, the larger the intersection density eigenvalue is, which means that the distribution of directed edges in the graph is more chaotic. Calculate the ratio of the intersection density eigenvalue to the numerical value of the task node and negatively map them to obtain the second sparsity of the directed acyclic graph; when the intersection density eigenvalue is larger, the graph is denser and the second sparsity is smaller.

[0045] After obtaining the first sparsity and the second sparsity, the comprehensive sparsity of the directed acyclic graph can be obtained according to the first sparsity and the second sparsity; preferably, in the embodiment of the present invention, the step of obtaining the comprehensive sparsity includes: calculating the product of the first sparsity and the preset third weight to obtain the first density factor; calculating the product of the second sparsity and the preset fourth weight to obtain the second density factor; in the embodiment of the present invention, the preset third weight and the preset fourth weight are 0.5 respectively, and the implementer can determine it according to the implementation scenario. Calculate the sum of the first density factor and the second density factor to obtain the comprehensive sparsity of the directed acyclic graph; when the comprehensive sparsity is larger, it means that the directed edges in the graph are sparser, the dependency relationship of the task nodes is clearer, and the depth-first search algorithm is more accurate and efficient.

[0046] When the comprehensive balance and comprehensive sparsity of the directed acyclic graph are larger, it means that the structure of the graph is more balanced and simple, the dependency relationship between task nodes is clearer and more balanced, and it is more suitable to use the depth-first search algorithm to obtain the task dependency sequence; conversely, it is more suitable to use the breadth-first search algorithm to obtain the task dependency sequence. Therefore, the ranking evaluation degree of the directed acyclic graph is obtained according to the comprehensive balance and comprehensive sparsity, specifically including: calculating the average value of the comprehensive balance and comprehensive sparsity to obtain the ranking evaluation degree of the directed acyclic graph. When the ranking evaluation degree is larger, it means that the structure of the graph is more balanced and simple, the dependency relationship between task nodes is clearer and more balanced, and it is more suitable to use the depth-first search algorithm to obtain the task dependency sequence efficiently and accurately. When the ranking evaluation degree is smaller, it means that the directed acyclic graph is more unbalanced and dense, the dependency relationship between task nodes is more complex and unbalanced, and it is more suitable to use the breadth-first search algorithm to obtain the task dependency sequence efficiently and accurately. Therefore, the sorting method of the tasks can be judged according to the sorting evaluation degree and the tasks can be sorted to obtain the task dependency sequence; preferably, in the embodiment of the present invention, the step of obtaining the task dependency sequence includes: when the sorting evaluation degree exceeds the preset threshold, the task nodes in the directed acyclic graph are sorted using the depth-first search algorithm; when the sorting evaluation degree does not exceed the preset threshold, the task nodes in the directed acyclic graph are sorted using the breadth-first search algorithm to obtain the task dependency sequence. In the embodiment of the present invention, the preset threshold is 0.5, and the implementer can determine it according to the implementation scenario. It should be noted that the depth-first search algorithm and the breadth-first search algorithm belong to the prior art, and the specific sorting steps will not be repeated. The obtained task dependency sequence can reflect the dependency relationship of the tasks. The higher the ranking, the earlier it needs to be calculated and processed in the cloud computing platform.

[0047] Step S4, constructing a task complexity model based on the task dependency sequence, the hardware required for task processing, the task processing time and the task priority; and allocating and processing different tasks according to the task complexity model.

[0048] The task complexity model is constructed by integrating the task dependency sequence, the hardware required for task processing, the task processing time and the task priority, wherein the hardware required for task processing, the task processing time and the task priority are determined by the task publisher and the cloud computing platform, and the hardware required for task processing refers to the memory occupied and the processor requirements, etc. It should be noted that the task complexity model can be constructed by integrating the convolutional neural network, which belongs to the existing technology, and the specific steps will not be repeated. Then, different tasks can be allocated and processed according to the task complexity model, and each task can be matched with a suitable computing unit according to the task complexity model to maximize the utilization of computing resources and improve the efficiency of task processing. Until, by selecting the algorithm for obtaining the dependency relationship of the task, the efficiency of obtaining the task dependency sequence is improved, and finally the efficiency of task allocation and processing of the cloud computing platform is improved.

[0049] In summary, the embodiment of the present invention provides a method for intelligent task allocation based on data processing, which obtains a comprehensive balance degree according to the quantitative difference characteristics of the total out-degree and total in-degree of all task nodes in a directed acyclic graph, and the average level of the quantitative difference characteristics of the out-degree and in-degree of the task nodes; obtains a comprehensive sparsity degree according to the quantitative difference characteristics of the directed edges and task nodes in the directed acyclic graph, and the density of the intersection characteristics of the directed edges; obtains a sorting evaluation degree according to the comprehensive balance degree and the comprehensive sparsity; and determines the sorting method of the tasks according to the sorting evaluation degree and obtains the task dependency sequence. A task complexity model is constructed by integrating the task dependency sequence, the hardware required for task processing, the task processing time, and the task priority, and tasks are allocated according to the task complexity model, thereby improving the efficiency of task allocation and processing.

[0050] The present invention also proposes a task intelligent allocation system based on data processing, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any step of a task intelligent allocation method based on data processing.

[0051] It should be noted that the sequence of the above embodiments of the present invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0052] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A task intelligent allocation method based on data processing, characterized in that: The method comprises the following steps: Get the directed acyclic graph between related tasks; A first balance degree is obtained according to the quantitative difference characteristics of the total out-degree and the total in-degree of all task nodes in the directed acyclic graph; a second balance degree is obtained according to the average level of the quantitative difference characteristics of the out-degree and in-degree of the task nodes in the directed acyclic graph; a comprehensive balance degree of the directed acyclic graph is obtained according to the first balance degree and the second balance degree; A first sparsity is obtained according to the quantity difference characteristics of the directed edges and task nodes in the directed acyclic graph; a second sparsity is obtained according to the density of the intersection characteristics of the directed edges in the directed acyclic graph; a comprehensive sparsity of the directed acyclic graph is obtained according to the first sparsity and the second sparsity; a sorting evaluation degree of the directed acyclic graph is obtained according to the comprehensive balance degree and the comprehensive sparsity; a sorting method of the tasks is determined according to the sorting evaluation degree and the tasks are sorted to obtain a task dependency sequence; Constructing a task complexity model based on the task dependency sequence, the hardware required for task processing, the task processing time and the task priority; allocating and processing different tasks according to the task complexity model; The step of determining the order of tasks according to the order evaluation degree and performing ordering of tasks to obtain a task dependency sequence comprises: When the sorting evaluation degree exceeds a preset threshold, a depth-first search algorithm is used to sort the task nodes in the directed acyclic graph; when the sorting evaluation degree does not exceed the preset threshold, a breadth-first search algorithm is used to sort the task nodes in the directed acyclic graph to obtain the task dependency sequence.

2. The method for intelligent task allocation based on data processing according to claim 1, characterized in that: The step of obtaining a first balance degree according to the quantity difference characteristics of the total out-degree and the total in-degree of all task nodes in the directed acyclic graph comprises: The absolute value of the difference between the total out-degree and the total in-degree of all task nodes is calculated to obtain the in-degree difference value; the ratio of the in-degree difference value to the total degree is calculated and negatively correlated to obtain the first balance degree of the directed acyclic graph.

3. The method for intelligent task allocation based on data processing according to claim 1 is characterized in that: The step of obtaining the second balance degree according to the average level of the quantity difference characteristics of the out-degree and in-degree of the task nodes in the directed acyclic graph comprises: The absolute value of the difference between the number of out-degree and in-degree of any task node is calculated to obtain the distribution difference of the arbitrary task node; the average value of the distribution difference of all task nodes is calculated and negatively correlated to obtain the second balance of the directed acyclic graph.

4. The method for intelligent task allocation based on data processing according to claim 1 is characterized in that: The step of obtaining the comprehensive balance degree of the directed acyclic graph according to the first balance degree and the second balance degree comprises: The product of the first balance degree and a preset first weight is calculated to obtain a first balance factor; the product of the second balance degree and a preset second weight is calculated to obtain a second balance factor; and the sum of the first balance factor and the second balance factor is calculated to obtain a comprehensive balance degree of the directed acyclic graph.

5. The method for intelligent task allocation based on data processing according to claim 1 is characterized in that: The step of obtaining the first sparsity according to the quantity difference characteristics of the directed edges and task nodes in the directed acyclic graph comprises: The ratio of the number of directed edges to the number of task nodes in the directed acyclic graph is calculated and negatively correlated to obtain a first sparsity of the directed acyclic graph.

6. The method for intelligent task allocation based on data processing according to claim 1, characterized in that: The step of obtaining the second sparsity according to the density of the intersection features of the directed edges in the directed acyclic graph comprises: The number of intersections of the directed edges intersecting in the directed acyclic graph is calculated to obtain an intersection density eigenvalue; the ratio of the intersection density eigenvalue to the numerical value of the task node is calculated and negatively correlated to obtain a second sparsity of the directed acyclic graph.

7. The method for intelligent task allocation based on data processing according to claim 1 is characterized in that: The step of obtaining the comprehensive sparsity of the directed acyclic graph according to the first sparsity and the second sparsity comprises: Calculate the product of the first sparsity and a preset third weight to obtain a first density factor; calculate the product of the second sparsity and a preset fourth weight to obtain a second density factor; calculate the sum of the first density factor and the second density factor to obtain the comprehensive sparsity of the directed acyclic graph.

8. The method for intelligent task allocation based on data processing according to claim 1, characterized in that: The step of obtaining the ranking evaluation degree of the directed acyclic graph according to the comprehensive balance degree and the comprehensive sparsity degree comprises: The average values ​​of the comprehensive balance degree and the comprehensive sparsity degree are calculated to obtain the ranking evaluation degree of the directed acyclic graph.

9. A task intelligent allocation system based on data processing, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 8.

Images