CN-121998383-A - Underwater sound task intelligent scheduling method based on multi-objective optimization

CN121998383ACN 121998383 ACN121998383 ACN 121998383ACN-121998383-A

Abstract

The invention provides an underwater sound task intelligent scheduling method based on multi-objective optimization, which comprises the steps of determining data dependency relations among all sub-tasks in an underwater sound task and resource requirements of all the sub-tasks when the underwater sound task is received, inquiring resources meeting requirements based on the resource requirements of all the sub-tasks to generate an available resource list of all the sub-tasks, reading physical connection relations and available resource performance parameters of corresponding available resources based on the available resource list of all the sub-tasks, and obtaining a resource scheduling strategy meeting the resource requirements by adopting a multi-objective optimization algorithm based on the data dependency relations among all the sub-tasks, the available resource list of all the sub-tasks, the physical connection relations of available resources and the available resource performance parameters. By implementing the method, the probability of the task execution blocking caused by insufficient consideration of data dependence can be reduced.

Inventors

Chen Shuaike
XIAO ZHUAN
XU JIALIN
LIU SHUYANG

Assignees

中国舰船研究设计中心

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (10)

1. An underwater sound task intelligent scheduling method based on multi-objective optimization is characterized by comprising the following steps: when the underwater sound task is received, determining the data dependency relationship among all the subtasks in the underwater sound task and the resource requirement of all the subtasks; inquiring resources meeting requirements based on the resource requirements of each subtask, and generating an available resource list of each subtask; Based on the available resource list of each subtask, reading the physical connection relation of the corresponding available resources and the performance parameters of the available resources; and obtaining a resource scheduling strategy meeting the resource requirement by adopting a multi-objective optimization algorithm based on the data dependency relationship among the subtasks, the available resource list of the subtasks, the physical connection relationship of the available resources and the performance parameters of the available resources.
2. The method for intelligently scheduling the underwater sound task based on the multi-objective optimization according to claim 1, wherein the data dependency relationship comprises communication relationship data, and the resource scheduling strategy meeting the resource requirement is obtained by adopting a multi-objective optimization algorithm based on the data dependency relationship among the subtasks, the available resource list of the subtasks, the physical connection relationship of the available resources and the performance parameters of the available resources, and comprises the following steps: dividing each subtask based on communication relation data to obtain a plurality of task groups, wherein each task group at least comprises one subtask; Determining a critical path based on the data dependency relationship among the subtasks, wherein the critical path represents a path with the longest total execution time in all paths from a starting point to an end point in the data dependency relationship; performing resource pre-allocation for each subtask in the critical path based on the available resource performance parameters and the critical path to obtain a pre-allocation scheme; Based on the physical connection relation of the selectable resources, the pre-allocation scheme, the available resource list of each subtask and the task grouping, a multi-objective optimization algorithm is adopted to obtain a resource scheduling strategy meeting the resource requirements.
3. The underwater sound task intelligent scheduling method based on multi-objective optimization according to claim 2, wherein the method is characterized in that a multi-objective optimization algorithm is adopted to obtain a resource scheduling strategy meeting resource requirements based on physical connection relation of selectable resources, a pre-allocation scheme, available resource lists of all sub-tasks and task grouping, and comprises the following steps: S1, initializing a population, wherein the population comprises a plurality of individuals, each individual represents a resource allocation scheme, the resource allocation scheme reserves a pre-allocation scheme of each subtask in a critical path, and non-critical path subtasks randomly select resources from an available resource list of the corresponding subtasks; s2, simulating the underwater sound task execution process based on the resource allocation scheme of each individual and the physical connection relation of the optional resources, and extracting target index data; S3, determining the fitness of each individual based on target index data and a preset fitness function, wherein the fitness function is constructed by communication overhead, resource utilization rate and task grouping standard rate, and the task grouping standard rate represents the compactness among the allocated resources of the same task grouping in the allocation scheme; S4, selecting at least one first target individual and a plurality of second target individuals based on fitness of each individual; S5, performing cross mutation on a plurality of second target individuals to generate new child individuals, taking the first target individuals and the new child individuals as total child individuals, and reserving pre-allocation resources bound by each subtask in a critical path in the cross mutation process; And repeating the steps S2-S5 until the termination condition is reached, and obtaining the resource scheduling strategy meeting the resource requirement.
4. The method for intelligently scheduling underwater sound tasks based on multi-objective optimization according to claim 1, wherein the communication relation data comprises data exchange amount and communication frequency, each sub task is divided according to the communication relation data to obtain a plurality of task groups, and the method comprises the following steps: Calculating a communication intensity score between subtasks based on the communication relationship data; Determining a preliminary task group based on a communication density score between subtasks and a preset scoring threshold; when the number of subtasks in the primary task group exceeds the target number, analyzing the overlapping degree of the demand resources of the subtasks in the primary task group; on the premise of meeting the target number, task groups are constructed based on subtasks with smaller overlapping degree of the required resources.
5. The method for intelligently scheduling underwater sound tasks based on multi-objective optimization according to claim 1, wherein the steps of inquiring resources meeting requirements based on resource requirements and generating an available resource list of each subtask include: When the resources meeting the resource requirements of any subtask are inquired to be in an occupied state, acquiring the priority of the occupied task; And when the occupied task priority is lower than that of the subtask, the resource is included in the available resource list of the corresponding subtask.
6. The method for intelligently scheduling underwater sound tasks based on multi-objective optimization according to claim 5, further comprising: splitting the subtask to obtain a plurality of secondary tasks when the occupied task priority is higher than that of the subtask; inquiring resources meeting the requirements of each secondary task to obtain an available resource list corresponding to each secondary task; and selecting one available resource from the available resource list corresponding to each secondary task, and incorporating the available resource list of the corresponding subtask.
7. The underwater sound task intelligent scheduling method based on multi-objective optimization according to claim 3, wherein the fitness function is: Wherein, the The value range of the individual fitness value is 0,1, The larger the value is, the better the resource allocation scheme corresponding to the individual is represented; 、、 weight coefficients of communication overhead, resource utilization rate and task grouping standard reaching rate respectively, and ; As a communication overhead for the current individual scheme, Maximum value of communication overhead in all individual schemes; For the resource utilization of the current individual scheme, The maximum value of the resource utilization rate in all individual schemes, G is the task grouping standard rate of the current individual scheme, The maximum of the standard rate is grouped for the tasks in all individual schemes.
8. An underwater sound task intelligent scheduling device based on multi-objective optimization is characterized by comprising: The task splitting module is used for determining the data dependency relationship among all the subtasks in the underwater sound task and the resource requirement of each subtask when the underwater sound task is received; The resource determining module is used for inquiring the resources meeting the requirements based on the resource requirements of the subtasks and generating an available resource list of the subtasks; the reading module is used for reading the physical connection relation of the corresponding available resources and the performance parameters of the available resources based on the available resource list of each subtask; The scheduling policy decision module is used for obtaining a resource scheduling policy meeting the resource requirements by adopting a multi-objective optimization algorithm based on the data dependency relationship among all the subtasks, the available resource list of all the subtasks, the physical connection relationship of available resources and the performance parameters of the available resources.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor performs the steps of a method for intelligent scheduling of underwater acoustic tasks based on multi-objective optimization as claimed in any of the preceding claims 1-7.
10. A computer storage medium having stored thereon computer instructions which, when executed by a processor, implement the steps of a multi-objective optimization based underwater acoustic task intelligent scheduling method of any of claims 1-7.

Description

Underwater sound task intelligent scheduling method based on multi-objective optimization Technical Field The invention belongs to the technical field of resource scheduling, and particularly relates to an underwater sound task intelligent scheduling method based on multi-objective optimization. Background With the deep development of applications such as underwater acoustic signal processing, target identification and tracking, the computing tasks of the underwater acoustic signal processing and target identification and tracking show high heterogeneity and data dependence. A typical underwater acoustic processing task chain is generally composed of multiple stages of signal preprocessing, beam forming, feature extraction and classification. Not only are these tasks differentially demanded for computing resources (CPU, memory, GPU), there is more frequent and massive data interaction between tasks, which is extremely sensitive to network communication bandwidth and latency. The traditional scheduling strategy only focuses on the static allocation of computing resources, ignores network overhead caused by data flow among tasks, and is very easy to cause that a data processing pipeline becomes a performance bottleneck due to network congestion. Disclosure of Invention Therefore, the invention aims to provide an underwater sound task intelligent scheduling method based on multi-objective optimization so as to meet the requirement of reducing the task execution blocking probability. In order to achieve the above purpose, the present invention provides the following technical solutions: According to a first aspect, the invention provides an underwater sound task intelligent scheduling method based on multi-objective optimization, which comprises the following steps: The method comprises the steps of receiving underwater sound tasks, determining data dependency relations among all sub-tasks in the underwater sound tasks and resource requirements of all the sub-tasks, inquiring resources meeting the requirements based on the resource requirements of all the sub-tasks to generate available resource lists of all the sub-tasks, reading physical connection relations and available resource performance parameters of corresponding available resources based on the available resource lists of all the sub-tasks, and obtaining resource scheduling strategies meeting the resource requirements by adopting a multi-objective optimization algorithm based on the data dependency relations among all the sub-tasks, the available resource lists of all the sub-tasks, the physical connection relations of available resources and the available resource performance parameters. According to a second aspect, the present invention provides an underwater acoustic task intelligent scheduling device based on multi-objective optimization, comprising: The system comprises a task splitting module, a resource determining module, a reading module and a scheduling policy decision module, wherein the task splitting module is used for determining the data dependency relationship among all sub-tasks and the resource requirement of all the sub-tasks in the underwater sound task when the underwater sound task is received, the resource determining module is used for inquiring resources meeting requirements based on the resource requirement of all the sub-tasks to generate an available resource list of all the sub-tasks, the reading module is used for reading the physical connection relationship and the available resource performance parameter of corresponding available resources based on the available resource list of all the sub-tasks, and the scheduling policy decision module is used for obtaining a resource scheduling policy meeting the resource requirement by adopting a multi-objective optimization algorithm based on the data dependency relationship among all the sub-tasks, the available resource list of all the sub-tasks and the available resource performance parameter. According to a third aspect, an embodiment of the present invention provides an electronic device, where the device includes a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the steps of the method for intelligent scheduling of underwater sound tasks based on multi-objective optimization according to the first aspect or any implementation of the first aspect. According to a fourth aspect, an embodiment of the present invention provides a computer storage medium, on which computer instructions are stored, which when executed by a processor implement the steps of the method for intelligent scheduling of underwater acoustic tasks based on multi-objective optimization according to the first aspect or any implementation manner of the first aspect. The embodiment provides an underwater sound task intelligent scheduling method based on multi-objective optimization, which synthesizes a sub-task data dependency relat