A method for scheduling aerodynamic shape optimization design tasks for supercomputers
By setting up a scheduling and computing core on a supercomputer and decomposing the parallel sample point evaluation process, the problem of low efficiency in aircraft aerodynamic shape optimization design was solved, and efficient optimization design and experimental design sampling were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- AVIC SHENYANG AERODYNAMICS RES INST
- Filing Date
- 2022-11-09
- Publication Date
- 2026-05-05
AI Technical Summary
Currently, the optimization design of aircraft aerodynamic shape is inefficient on supercomputers. Due to the limited number of cores on a single workstation, supercomputing resources cannot be fully utilized, resulting in excessively long optimization design time.
A task scheduling method for aerodynamic shape optimization design using a supercomputer is adopted. By setting up a scheduling core and a computing core on a local workstation, the sample point evaluation process is decomposed into three sub-processes: sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and sample point evaluation postprocessing. These sub-processes are executed in parallel, and simulation analysis is performed using a high-performance computing cluster.
It significantly improves the efficiency of aerodynamic shape optimization design and experimental design sampling for aircraft, makes full use of supercomputing resources, and shortens the optimization design time.
Smart Images
Figure CN115935502B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of aircraft optimization design, specifically relating to a method for scheduling aerodynamic shape optimization design tasks for supercomputers. Background Technology
[0002] With the development of aviation technology, aircraft design is moving towards refinement and rapid iteration. Aircraft aerodynamic shape optimization design no longer considers only aerodynamic characteristics but involves a comprehensive design encompassing multiple disciplines such as aerodynamics, structure, stealth, and flight mechanics. This results in aircraft aerodynamic shape optimization design exhibiting characteristics of multiple parameters, constraints, objectives, and extrema, placing higher demands on the efficiency and effectiveness of optimization design. In engineering, to achieve satisfactory optimization design results, large mesh sizes and high-precision algorithms are required for numerical simulation evaluation of various aircraft disciplines, employing multi-round iterative optimization methods. The numerous numerical simulation evaluations and long individual evaluation times severely impact optimization design efficiency. Therefore, a single-machine workstation combined with a supercomputing system is often used to carry out refined design tasks.
[0003] Currently, when using a single workstation paired with a supercomputing system for aircraft optimization design tasks, a synchronous parallel approach is employed. This means that multiple sample point evaluation processes occur in parallel, while a single sample point evaluation process is sequential. The sample point evaluation process involves obtaining the corresponding optimization objective (response) and constraints from a set of design variables. For example, in aircraft aerodynamic shape optimization design, the sample point evaluation process includes parameterization, mesh generation, numerical simulation analysis, and result extraction. The sequential sample point evaluation process involves a single core on the single workstation performing parameterization, mesh generation, submission of the supercomputing numerical simulation analysis job, (supercomputing) numerical simulation analysis, and result extraction. While the supercomputing is performing numerical simulation analysis, the core on the single workstation submitting the supercomputing numerical simulation analysis job is in a dormant state. When a single workstation uses a 4-core CPU for optimization design tasks, it can only perform a maximum of 4 sample point evaluations simultaneously. Similarly, on a supercomputer, a maximum of 4 computational tasks can be submitted. If an optimization design task requires 500 sample point evaluations (10 million grid cells, 192 supercomputing cores, aerodynamic numerical analysis takes 1 hour), the parallel numerical simulation analysis time on a single 4-core CPU would be 5.2 days, which is unacceptable. Although current domestic supercomputing resources have reached the tens of thousands of cores level, synchronous parallel methods cannot fully utilize supercomputing resources, and the limitation of the number of parallel cores on a single workstation reduces work efficiency. Summary of the Invention
[0004] The problem this invention aims to solve is to improve the efficiency of optimization design tasks on supercomputers, and proposes a scheduling method for aerodynamic shape optimization design tasks on supercomputers.
[0005] To achieve the above objectives, the present invention provides the following technical solution:
[0006] A scheduling method for aerodynamic shape optimization design tasks using a supercomputer is disclosed. A local workstation serves as the master control device for the aerodynamic shape optimization design task, while a high-performance computing cluster acts as the aerodynamic numerical simulation analysis device. The master control device initiates a parallel sample point set evaluation task. The first available CPU core of the master control device is designated as the scheduling core. In the parallel sample point set evaluation task, the scheduling core is responsible for decomposing the aerodynamic evaluation process of a single sample point, constructing a task queue system, controlling the aerodynamic evaluation process of the sample point set, and generating a sample point set evaluation result file recognizable by the optimization design program. The other available CPU cores of the master control device serve as computing cores. These computing cores are responsible for executing various sub-processes of the single sample point evaluation in the parallel sample point set evaluation task, including three sub-processes: sample point evaluation preprocessing, invoking the high-performance computing cluster for aerodynamic numerical simulation analysis, and sample point evaluation postprocessing.
[0007] Further steps include the following:
[0008] S1. Divide all available CPU cores of the main control equipment for the aerodynamic shape optimization design task into scheduling cores and computing cores. The scheduling core is the first available CPU core, and the computing cores are the other available CPU cores. The scheduling core is coded as Core0, and the computing core is coded as CoreX, where X = (1,2,…,N-1) and N is the number of all available CPU cores.
[0009] S2. The scheduling core reads in the information of the sample point set to be evaluated for the parallel sample point set evaluation task, including the number of samples in the parallel sample point set evaluation task, the dimension of a single sample point, the sample point evaluation file, and the sample point evaluation script information.
[0010] S3. Set the initial state of the computing core to idle, and create the file free_id_X.dat. The file free_id_X.dat is empty.
[0011] S4. The scheduling core creates a single sample point aerodynamic evaluation process execution folder based on the index of the single sample point in the parallel sample point set, copies the sample point evaluation file to the single sample point aerodynamic evaluation process execution folder, and then decomposes the single sample point aerodynamic evaluation process script into three sub-process scripts: single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing. The script numbers are set as follows: sub-process 1 represents the single sample point evaluation preprocessing process, sub-process 2 represents the aerodynamic numerical simulation analysis process, and sub-process 3 represents the sample point evaluation postprocessing process.
[0012] S5. The scheduling core checks if there is an idle computing core. If it exists, it writes the GoToWork_ID_X.dat file and deletes the free_id_X.dat file. The content of the GoToWork_ID_x.dat file is the address of the execution folder for the single sample point aerodynamic evaluation process created in step S4 and the corresponding execution sub-process script number. If it does not exist, the scheduling core sleeps for 5 seconds.
[0013] S6. The idle computing core checks if the GoToWork_ID_X.dat file exists. If it exists, the idle computing core acquires the task; if it does not exist, it continues to wait until the GoToWork_ID_X.dat file appears or the optimization design is completed.
[0014] S7. The computing cores of the task execute the sub-process corresponding to the sequence number in the GoToWork_ID_X.dat file at the address shown in the file to perform optimization design. In the initial stage of optimization design, all computing cores execute sub-processes 1 and 2. In the later stage of optimization design, it is adjusted to have 2 computing cores execute sub-process 3. The initial and later stages of the optimization process are determined according to the proportion of the number of sample points of the executed sub-processes 1 and 2 to the sample point set. After the task is completed, the computing cores generate the free_id_X.dat file and return to the idle computing core queue.
[0015] S8. Repeat steps S4 to S7 until all sample points have been evaluated.
[0016] S9. The scheduling core generates a summary file of the evaluation results of the sample point set according to the requirements of the aerodynamic shape optimization design task, and completes the aerodynamic shape optimization design task for the supercomputer.
[0017] Furthermore, the sample point evaluation file in step S2 includes the model generation script, the mesh generation script, and the initial mesh file.
[0018] Furthermore, in step S4, the method for decomposing the single sample point aerodynamic evaluation process script into three sub-process scripts—single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing—is as follows: a fixed-order method is used, setting parameterization and dynamic mesh as the sample point evaluation preprocessing sub-process; cluster task submission as the aerodynamic numerical simulation analysis sub-process; and result processing as the sample point evaluation postprocessing sub-process.
[0019] Furthermore, in step S4, the method for decomposing the single sample point aerodynamic evaluation process script into three sub-process scripts—single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing—is as follows: given a keyword method, the sample point aerodynamic evaluation process is written using a computer scripting language, and sub-process decomposition is performed by adding keywords to the single sample point aerodynamic evaluation process script.
[0020] Furthermore, in step S7, when the number of sample points of sub-process 1 and sub-process 2 executed in the initial stage of the optimization process reaches 95% of the sample point set, the later stage of optimization design begins.
[0021] The beneficial effects of this invention are:
[0022] This invention discloses a task scheduling method for aerodynamic shape optimization design using a supercomputer. This method is an asynchronous parallel task scheduling method, which contrasts with the traditional optimization design process where a single core is responsible for evaluating a complete single sample point. The workflow is as follows: The aerodynamic shape optimization design task is decomposed into sample point evaluation preprocessing (including parameterization, mesh generation, and cluster task submission), aerodynamic numerical simulation analysis (high-performance cluster processing), and sample point evaluation postprocessing (cluster computing status check and result extraction). The aerodynamic numerical simulation analysis process is performed by the cluster and is primarily limited by cluster computing resources. The sample point evaluation preprocessing and postprocessing processes are performed in parallel, using MPI parallelism. The two processes are executed serially and do not interfere with each other; that is, the cluster execution process is performed only after all local execution processes are completed.
[0023] The present invention discloses a task scheduling method for aerodynamic shape optimization design using a supercomputer. This method is applied to the process of aerodynamic shape optimization design and experimental design sampling of aircraft using a single workstation combined with a supercomputer, which greatly improves the efficiency of aerodynamic shape optimization design and experimental design sampling. Attached Figure Description
[0024] Figure 1 This is a flowchart of a task scheduling method for aerodynamic shape optimization design of a supercomputer, as described in this invention.
[0025] Figure 2 This is a demonstration diagram of a task scheduling method for aerodynamic shape optimization design of a supercomputer, as described in this invention. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.
[0027] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.
[0028] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with detailed descriptions in conjunction with the accompanying drawings: Specific implementation method one:
[0030] A scheduling method for aerodynamic shape optimization design tasks using a supercomputer is disclosed. A local workstation serves as the master control device for the aerodynamic shape optimization design task, while a high-performance computing cluster acts as the aerodynamic numerical simulation analysis device. The master control device initiates a parallel sample point set evaluation task. The first available CPU core of the master control device is designated as the scheduling core. In the parallel sample point set evaluation task, the scheduling core is responsible for decomposing the aerodynamic evaluation process of a single sample point, constructing a task queue system, controlling the aerodynamic evaluation process of the sample point set, and generating a sample point set evaluation result file recognizable by the optimization design program. The other available CPU cores of the master control device serve as computing cores. These computing cores are responsible for executing various sub-processes of the single sample point evaluation in the parallel sample point set evaluation task, including three sub-processes: sample point evaluation preprocessing, invoking the high-performance computing cluster for aerodynamic numerical simulation analysis, and sample point evaluation postprocessing.
[0031] Further steps include the following:
[0032] S1. Divide all available CPU cores of the main control equipment for the aerodynamic shape optimization design task into scheduling cores and computing cores. The scheduling core is the first available CPU core, and the computing cores are the other available CPU cores. The scheduling core is coded as Core0, and the computing core is coded as CoreX, where X = (1,2,…,N-1) and N is the number of all available CPU cores.
[0033] S2. The scheduling core reads in the information of the sample point set to be evaluated for the parallel sample point set evaluation task, including the number of samples in the parallel sample point set evaluation task, the dimension of a single sample point, the sample point evaluation file, and the sample point evaluation script information.
[0034] Furthermore, the sample point evaluation file in step S2 includes the model generation script, the mesh generation script, and the initial mesh file;
[0035] S3. Set the initial state of the computing core to idle, and create the file free_id_X.dat. The file free_id_X.dat is empty.
[0036] S4. The scheduling core creates a single sample point aerodynamic evaluation process execution folder based on the index of the single sample point in the parallel sample point set, copies the sample point evaluation file to the single sample point aerodynamic evaluation process execution folder, and then decomposes the single sample point aerodynamic evaluation process script into three sub-process scripts: single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing. The script numbers are set as follows: sub-process 1 represents the single sample point evaluation preprocessing process, sub-process 2 represents the aerodynamic numerical simulation analysis process, and sub-process 3 represents the sample point evaluation postprocessing process.
[0037] Furthermore, in step S4, the method for decomposing the single sample point aerodynamic evaluation process script into three sub-process scripts—single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing—is as follows: a fixed-order method is used, setting parameterization and dynamic mesh as the sample point evaluation preprocessing sub-process; cluster task submission as the aerodynamic numerical simulation analysis sub-process; and result processing as the sample point evaluation postprocessing sub-process.
[0038] S5. The scheduling core checks if there is an idle computing core. If it exists, it writes the GoToWork_ID_X.dat file and deletes the free_id_X.dat file. The content of the GoToWork_ID_x.dat file is the address of the execution folder for the single sample point aerodynamic evaluation process created in step S4 and the corresponding execution sub-process script number. If it does not exist, the scheduling core sleeps for 5 seconds.
[0039] S6. The idle computing core checks if the GoToWork_ID_X.dat file exists. If it exists, the idle computing core acquires the task; if it does not exist, it continues to wait until the GoToWork_ID_X.dat file appears or the optimization design is completed.
[0040] S7. The computing cores of the task execute the sub-process corresponding to the sequence number in the GoToWork_ID_X.dat file at the address shown in the file to perform optimization design. In the initial stage of optimization design, all computing cores execute sub-processes 1 and 2. In the later stage of optimization design, it is adjusted to have 2 computing cores execute sub-process 3. The initial and later stages of the optimization process are determined according to the proportion of the number of sample points of the executed sub-processes 1 and 2 to the sample point set. After the task is completed, the computing cores generate the free_id_X.dat file and return to the idle computing core queue.
[0041] Furthermore, in step S7, when the number of sample points of sub-process 1 and sub-process 2 executed in the initial stage of the optimization process reaches 95% of the sample point set, the later stage of optimization design begins.
[0042] S8. Repeat steps S4 to S7 until all sample points have been evaluated.
[0043] S9. The scheduling core generates a summary file of the evaluation results of the sample point set according to the requirements of the aerodynamic shape optimization design task, and completes the aerodynamic shape optimization design task for the supercomputer. Specific Implementation Method Two:
[0045] A scheduling method for aerodynamic shape optimization design tasks using a supercomputer is disclosed. A local workstation serves as the master control device for the aerodynamic shape optimization design task, while a high-performance computing cluster acts as the aerodynamic numerical simulation analysis device. The master control device initiates a parallel sample point set evaluation task. The first available CPU core of the master control device is designated as the scheduling core. In the parallel sample point set evaluation task, the scheduling core is responsible for decomposing the aerodynamic evaluation process of a single sample point, constructing a task queue system, controlling the aerodynamic evaluation process of the sample point set, and generating a sample point set evaluation result file recognizable by the optimization design program. The other available CPU cores of the master control device serve as computing cores. These computing cores are responsible for executing various sub-processes of the single sample point evaluation in the parallel sample point set evaluation task, including three sub-processes: sample point evaluation preprocessing, invoking the high-performance computing cluster for aerodynamic numerical simulation analysis, and sample point evaluation postprocessing.
[0046] Further steps include the following:
[0047] S1. Divide all available CPU cores of the main control equipment for the aerodynamic shape optimization design task into scheduling cores and computing cores. The scheduling core is the first available CPU core, and the computing cores are the other available CPU cores. The scheduling core is coded as Core0, and the computing core is coded as CoreX, where X = (1,2,…,N-1) and N is the number of all available CPU cores.
[0048] S2. The scheduling core reads in the information of the sample point set to be evaluated for the parallel sample point set evaluation task, including the number of samples in the parallel sample point set evaluation task, the dimension of a single sample point, the sample point evaluation file, and the sample point evaluation script information.
[0049] Furthermore, the sample point evaluation file in step S2 includes the model generation script, the mesh generation script, and the initial mesh file;
[0050] S3. Set the initial state of the computing core to idle, and create the file free_id_X.dat. The file free_id_X.dat is empty.
[0051] S4. The scheduling core creates a single sample point aerodynamic evaluation process execution folder based on the index of the single sample point in the parallel sample point set, copies the sample point evaluation file to the single sample point aerodynamic evaluation process execution folder, and then decomposes the single sample point aerodynamic evaluation process script into three sub-process scripts: single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing. The script numbers are set as follows: sub-process 1 represents the single sample point evaluation preprocessing process, sub-process 2 represents the aerodynamic numerical simulation analysis process, and sub-process 3 represents the sample point evaluation postprocessing process.
[0052] Furthermore, in step S4, the method for decomposing the single sample point aerodynamic evaluation process script into three sub-process scripts—single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing—is as follows: given a keyword method, the sample point aerodynamic evaluation process is written in a computer scripting language, and sub-process decomposition is performed by adding keywords to the single sample point aerodynamic evaluation process script.
[0053] Furthermore, the single sample point evaluation script in aerodynamic shape optimization design includes parameterization, dynamic meshing, aerodynamic numerical simulation analysis, and aerodynamic characteristic extraction. By adding "start CFD" to the next line of the dynamic meshing script statement and "done CFD" to the next line of the aerodynamic numerical simulation analysis script statement, the single sample point evaluation process script can be decomposed into three sub-process scripts based on the above keywords: single sample point evaluation preprocessing, numerical simulation analysis, and single sample point evaluation postprocessing.
[0054] S5. The scheduling core checks if there is an idle computing core. If it exists, it writes the GoToWork_ID_X.dat file and deletes the free_id_X.dat file. The content of the GoToWork_ID_x.dat file is the address of the execution folder for the single sample point aerodynamic evaluation process created in step S4 and the corresponding execution sub-process script number. If it does not exist, the scheduling core sleeps for 5 seconds.
[0055] S6. The idle computing core checks if the GoToWork_ID_X.dat file exists. If it exists, the idle computing core acquires the task; if it does not exist, it continues to wait until the GoToWork_ID_X.dat file appears or the optimization design is completed.
[0056] S7. The computing cores of the task execute the sub-process corresponding to the sequence number in the GoToWork_ID_X.dat file at the address shown in the file to perform optimization design. In the initial stage of optimization design, all computing cores execute sub-processes 1 and 2. In the later stage of optimization design, it is adjusted to have 2 computing cores execute sub-process 3. The initial and later stages of the optimization process are determined according to the proportion of the number of sample points of the executed sub-processes 1 and 2 to the sample point set. After the task is completed, the computing cores generate the free_id_X.dat file and return to the idle computing core queue.
[0057] Furthermore, in step S7, when the number of sample points of sub-process 1 and sub-process 2 executed in the initial stage of the optimization process reaches 95% of the sample point set, the later stage of optimization design begins.
[0058] S8. Repeat steps S4 to S7 until all sample points have been evaluated.
[0059] S9. The scheduling core generates a summary file of the evaluation results of the sample point set according to the requirements of the aerodynamic shape optimization design task, and completes the aerodynamic shape optimization design task for the supercomputer.
[0060] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0061] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.
Claims
1. A task scheduling method for aerodynamic shape optimization design of a supercomputer, characterized in that: The local workstation serves as the main control device for the aerodynamic shape optimization design task, while the high-performance computing cluster acts as the aerodynamic numerical simulation analysis device. The main control device initiates a parallel sample point set evaluation task, setting the first available CPU core as the scheduling core. In the parallel sample point set evaluation task, the scheduling core is responsible for decomposing the aerodynamic evaluation process of a single sample point, building a task queue system, controlling the aerodynamic evaluation process of the sample point set, and generating a sample point set evaluation result file that can be recognized by the optimization design program. The other available CPU cores of the main control device serve as computing cores. In the parallel sample point set evaluation task, the computing cores are responsible for executing each sub-process of the single sample point evaluation in the aerodynamic shape optimization design task, including three sub-processes: sample point evaluation preprocessing, calling the high-performance computing cluster for aerodynamic numerical simulation analysis, and sample point evaluation postprocessing. Includes the following steps: S1. Divide all available CPU cores of the main control equipment for the aerodynamic shape optimization design task into scheduling cores and computing cores. The scheduling core is the first available CPU core, and the computing cores are the other available CPU cores. The scheduling core is coded as Core0, and the computing core is coded as CoreX. X=(1,2,…,N-1), where N is the number of all available CPU cores. S2. The scheduling core reads in the information of the sample point set to be evaluated for the parallel sample point set evaluation task, including the number of samples in the parallel sample point set evaluation task, the dimension of a single sample point, the sample point evaluation file, and the sample point evaluation script information. S3. Set the initial state of the computing core to idle, and create the file free_id_X.dat. The file free_id_X.dat is empty. S4. The scheduling core creates a single sample point aerodynamic evaluation process execution folder based on the index of the single sample point in the parallel sample point set, copies the sample point evaluation file to the single sample point aerodynamic evaluation process execution folder, and then decomposes the single sample point aerodynamic evaluation process script into three sub-process scripts: single sample point evaluation preprocessing, aerodynamic numerical simulation analysis, and single sample point evaluation postprocessing. The script numbers are set as follows: sub-process 1 represents the single sample point evaluation preprocessing process, sub-process 2 represents the aerodynamic numerical simulation analysis process, and sub-process 3 represents the sample point evaluation postprocessing process. S5. The scheduling core checks if there is an idle computing core. If it exists, it writes the GoToWork_ID_X.dat file and deletes the free_id_X.dat file. The content of the GoToWork_ID_x.dat file is the address of the execution folder for the single sample point aerodynamic evaluation process created in step S4 and the corresponding execution sub-process script number. If it does not exist, the scheduling core sleeps for 5 seconds. S6. The idle computing core checks if the GoToWork_ID_X.dat file exists. If it exists, the idle computing core acquires the task; if it does not exist, it continues to wait until the GoToWork_ID_X.dat file appears or the optimization design is completed. S7. The computing cores of the task execute the sub-process corresponding to the sequence number in the GoToWork_ID_X.dat file at the address shown in the file to perform optimization design. In the initial stage of optimization design, all computing cores execute sub-processes 1 and 2. In the later stage of optimization design, it is adjusted to have 2 computing cores execute sub-process 3. The initial and later stages of the optimization process are determined according to the proportion of the number of sample points of the executed sub-processes 1 and 2 to the sample point set. After the task is completed, the computing cores generate the free_id_X.dat file and return to the idle computing core queue. S8. Repeat steps S4 to S7 until all sample points have been evaluated. S9. The scheduling core generates a summary file of the evaluation results of the sample point set according to the requirements of the aerodynamic shape optimization design task, and completes the aerodynamic shape optimization design task for the supercomputer.
2. The method for scheduling aerodynamic shape optimization design tasks for supercomputers according to claim 1, characterized in that: The sample point evaluation files in step S2 include the model generation script, the mesh generation script, and the initial mesh file.
3. The method for scheduling aerodynamic shape optimization design tasks for supercomputers according to claim 2, characterized in that: The method for decomposing the single sample point aerodynamic evaluation process script into three sub-process scripts in step S4 is as follows: fixed order method, setting parameterization and dynamic mesh as the sample point evaluation preprocessing sub-process. Cluster task submission is a sub-process for aerodynamic numerical simulation analysis, and result processing is a sub-process for sample point evaluation and post-processing.
4. The method for scheduling aerodynamic shape optimization design tasks for supercomputers according to claim 3, characterized in that: The method for decomposing the single sample point aerodynamic evaluation process script into three sub-process scripts in step S4 is as follows: given the keyword method, the sample point aerodynamic evaluation process is written in a computer scripting language, and the sub-process is decomposed by adding keywords to the single sample point aerodynamic evaluation process script.
5. A task scheduling method for aerodynamic shape optimization design of a supercomputer according to claim 3 or 4, characterized in that: In step S7, when the number of sample points of sub-process 1 and sub-process 2 that have been executed in the initial stage of the optimization process reaches 95% of the sample point set, the later stage of optimization design begins.