An efficient and robust simulation management platform

Abstract

The present invention belongs to the field of computer technology, and in particular, relates to an efficient and robust simulation management platform, including a user end, a management end and an HPC execution end, wherein the user end and the HPC execution end are connected via the management end; the user end is used for users to upload and download simulation files and parameter settings; the management end is used to package the simulation files uploaded by the user end, and generate task requirements of the operation stage according to the parameter settings to send different operation stage call commands to the HPC execution end; the HPC execution end enables the corresponding node to execute the operation according to the operation stage call command, and returns the execution results to the management end stage by stage. The present invention can solve the problems of long time consumption and high cost in existing wind tunnel and CFD test operations.

Description

Technical Field

[0001] The invention belongs to the field of computer technology, and in particular relates to an efficient and robust simulation management platform. Background Art

[0002] At present, aerodynamics has received more and more attention in the development of new energy vehicles, and it has an important impact on the endurance of vehicles under high-speed conditions. In the process of automobile aerodynamics development, a combination of wind tunnel tests and CFD (computational fluid dynamics) simulation is generally used. Due to the scarcity of resources and high costs of wind tunnel tests, CFD plays a greater role in the entire development process. It is easy to use, flexible, and cost-effective, but it has high requirements on the user's experience and theoretical level of fluid mechanics. Therefore, how to reduce the error between the calculation results and the wind tunnel test results has become a problem that industry professionals in the field of automobile aerodynamics are more concerned about.

[0003] The first virtual automobile wind tunnel built and debugged in a 1:1 ratio in the central and western regions based on the construction of China Automotive Research Institute has a smaller error than the existing wind tunnel test. However, it still follows the original operation method, submitting jobs at one time and calling hundreds of CPU cores to complete meshing, solution calculation and post-processing jobs in sequence. The following problems exist during use:

[0004] 1. The test process shows that the probability of error in multi-node parallel meshing is higher than that in single-node meshing, so there is a problem that multiple nodes are prone to errors in the meshing stage;

[0005] 2. Low CPU utilization during the grid division phase;

[0006] 3. The post-processing stage takes too long to process each scene one by one, and the time consumed by post-processing a single scene using 24 cores is similar to that using hundreds of cores, resulting in low CPU utilization and high costs in the post-processing stage;

[0007] 4. There is currently no ready-made software available on the market that can solve the above pain points. Summary of the invention

[0008] The technical problem solved by the present invention is to provide an efficient and robust simulation management platform to solve the problems of long time consumption and high cost in existing wind tunnel and CFD test operations.

[0009] The basic solution provided by the present invention is: an efficient and robust simulation management platform, including a user end, a management end and an HPC execution end, wherein the user end and the HPC execution end are connected via the management end;

[0010] The user terminal is used for users to upload and download simulation files and parameter settings;

[0011] The management end is used to package the simulation files uploaded by the user end, and generate the task requirements of the operation stage according to the parameter settings to send different operation stage call commands to the HPC execution end;

[0012] The HPC execution end enables the corresponding node to execute the job according to the job stage call command, and returns the execution result to the management end stage by stage.

[0013] Furthermore, the user terminal includes a user login module, a data acquisition module, a parameter setting module and a data transmission module;

[0014] The user login module is used to log in to the user terminal according to the user's unique ID;

[0015] The data acquisition module is used to acquire the simulation file uploaded by the user, wherein the simulation file includes wind tunnel and CFD test data and file information;

[0016] The parameter setting module is used to obtain parameters set by the user;

[0017] The data transmission module is used to transmit wind tunnel and CFD test data, file information and parameters to the management end.

[0018] Furthermore, the management end includes a job generation module, a packaging and sending module, and a command generation module;

[0019] The job generation module is used to generate a job file of a preset stage and a preset format according to wind tunnel and CFD test data;

[0020] The packaging and sending module is used to package the job files and file information of the preset stage and preset format and send them to the HPC execution end;

[0021] The command generation module is used to generate a call command according to the parameters and information of the preset stage, and transmit the call command to the HPC execution end.

[0022] Furthermore, the preset stage includes grid division, solution calculation and post-processing.

[0023] Furthermore, the management end also includes a judgment module, which is used to judge in turn whether the execution results returned by the HPC execution end in the preset stage are completed. If completed, the management end transmits a control command for executing the next stage to the HPC execution end.

[0024] Furthermore, the HPC execution end enables the corresponding node to execute the job according to the job stage call command, and returns the execution results to the management end stage by stage, specifically:

[0025] In the grid division phase, according to the computing power required for the grid division phase job calculation, nodes with the corresponding number of cores are configured for processing, and the grid division phase job results are transmitted to the judgment module for judgment. After the judgment is completed, the solution calculation phase begins;

[0026] For the solution calculation stage, according to the computing power required for the calculation of the solution calculation stage, the nodes with the corresponding number of cores are configured for processing; and the results of the solution calculation stage are transmitted to the judgment module for judgment; after the judgment is completed, the post-processing stage begins

[0027] For the post-processing stage, based on the number of post-processing scenarios, the required computing power is calculated for each post-processing scenario, and nodes with the corresponding number of cores are configured to execute simultaneously, generate the execution results of the post-processing stage, and transmit them back to the management end.

[0028] Furthermore, the management end is also used to transmit the job execution result back to the user end for the user to view and download.

[0029] The principles and advantages of the present invention are as follows: the efficient and robust simulation management platform of the present application first obtains the simulation file uploaded by the user at the user end, and provides the user with the function of parameter setting, so that the user can adjust and set the parameters of the data in the simulation file when calculating; then the simulation file received by the user end is transmitted to the management end to be packaged into a file format that can be processed by the HPC execution end, and after the packaging is completed, it is sent to the HPC execution end for calculation according to the node calling strategy. Specifically, the node calling strategy makes different calls for the calculation stage of the data in the simulation file. When calculating the grid division stage, because the computing power required for the grid division stage is not large, the corresponding number of core nodes is called according to the computing power required for the grid division stage. Selection; when calculating the solution calculation stage, because the computing power required for the solution calculation is large, the corresponding nodes are selected and calculated according to the number of cores required by the computing power. When calculating the post-processing stage, because the post-processing stage contains multiple scenes, the existing post-processing stage adopts the same method as the solution calculation, and calculates the scenes of the entire post-processing stage together. At this time, the computing power required is large, and the number of cores of the corresponding nodes is also large, resulting in a long time and high cost. For this reason, the present application allocates a node with a matching number of cores to each scene according to the number of scenes in the post-processing stage for simultaneous calculation, so as to effectively utilize the processing nodes in the post-processing stage, while reducing the time consumed in the entire data packaging processing in the post-processing stage and saving costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a functional block diagram of an embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a flow chart of an embodiment of the present invention;

[0032] Figure 3A schematic diagram of CPU utilization of an existing HPC execution end submitting a job once to complete all tasks in an embodiment of the present invention;

[0033] Figure 4 This is a schematic diagram of CPU utilization of the present application adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following is further described in detail through specific implementation methods:

[0035] The embodiment is basically as follows Figure 1 and Figure 2 As shown: An efficient and robust simulation management platform includes a user end, a management end and an HPC execution end, wherein the HPC execution end is a high-performance computing system that can use aggregated computing power to process data-intensive computing tasks that standard workstations cannot complete, and the user end and the HPC execution end establish a connection and communication through the management end.

[0036] The user end is used by users to upload and download simulation files and set parameters. In wind tunnel and CFD tests, the aerodynamic simulation calculation of the test vehicle is performed by given wind speed, yaw angle, and five-belt system speed, so that the simulation result file can be obtained. Specifically, the user end includes a user login module, a data acquisition module, a parameter setting module, and a data transmission module;

[0037] The user login module is used to log in to the user terminal according to the user's unique ID;

[0038] The data acquisition module is used to acquire the simulation file uploaded by the user, wherein the simulation file includes wind tunnel and CFD test data and file information;

[0039] The parameter setting module is used to obtain the parameters set by the user;

[0040] The data transmission module is used to transmit wind tunnel and CFD test data, file information and parameters to the management end.

[0041] The wind tunnel and CFD test data and file information in the simulation file include the simulation condition number, simulation condition name, environmental parameters (temperature, relative humidity, atmospheric pressure and corresponding air density in the test section), and operating parameters (test wind speed, boundary layer control system BLRS operating status, moving road system RRS, wheel rotation unit WRU, yaw angle). For the parameters set by the user, they can be set according to the required parameters in the test result calculation process, such as the stage division of the test data, the evaluation of the required computing power, etc.

[0042] The management end receives the simulation files uploaded by the user and packages them, and sends different job stage call commands to the HPC execution end according to the task requirements of the job stage generated by the parameter settings; the management end of this application includes a job generation module, a package sending module and a command generation module. The job generation module is used to generate job files of preset stages and preset formats according to wind tunnel and CFD test data, and the preset formats are such as ***.txt format, ***.json format, and ***.xml format; the preset stages include network division stage, solution calculation stage and post-processing stage.

[0043] The package sending module is used to package the job files and file information of the preset stage and preset format and send them to the HPC execution end;

[0044] The command generation module is used to generate a call command according to the parameters and the information of the preset stage, and transmit the call command to the HPC execution end;

[0045] The HPC execution end enables the corresponding node to execute the job according to the job stage call command, and returns the execution results to the management end stage by stage.

[0046] In addition, the management end also includes a judgment module, which is used to judge in turn whether the execution results returned by the HPC execution end in the preset stage are completed. If completed, the management end transmits the control command for executing the next stage to the HPC execution end.

[0047] Therefore, the HPC execution end enables the corresponding node to execute the job according to the job stage call command, and returns the execution results to the management end stage by stage. Specifically:

[0048] In the grid division phase, according to the computing power required for the grid division phase job calculation, nodes with the corresponding number of cores are configured for processing, and the grid division phase job results are transmitted to the judgment module for judgment. After the judgment is completed, the solution calculation phase begins;

[0049] For the solution calculation stage, according to the computing power required for the calculation of the solution calculation stage, the nodes with the corresponding number of cores are configured for processing; and the results of the solution calculation stage are transmitted to the judgment module for judgment; after the judgment is completed, the post-processing stage begins

[0050] For the post-processing stage, based on the number of post-processing scenarios, the required computing power is calculated for each post-processing scenario, and nodes with the corresponding number of cores are configured to execute simultaneously, generate the execution results of the post-processing stage, and transmit them back to the management end.

[0051] In another embodiment of the present application, after the current post-processing stage is completed, if post-processing calculations are required later, the HPC execution end is directly started through the management end to continue execution according to the end of the previous post-processing stage.

[0052] To better explain the execution logic of the HPC execution end, such as Figure 3 and Figure 4 As shown in Figure 2, all tasks are executed under the same computing power conditions, where Figure 3 The figure is a schematic diagram of the CPU utilization of the current simulation management platform. Taking 640 cores and 10 nodes as an example (10 nodes * 64 cores / node = 640 cores), in the grid division stage, the solution calculation stage and the post-processing stage, the number of cores required for the solution calculation is relatively large. Therefore, the existing technology uses nodes with 640 cores to process the three stages. The defect caused by this is that since the amount of calculation required in the grid division stage is relatively small, a lower number of cores, such as 64 cores, can be used for processing. However, the existing technology still uses 640 cores in the grid division stage, and the execution logic of the HPC execution end is processed in stages, so there will be a waste of resources in the grid division stage; for example, in the post-processing stage, 640 cores are used to calculate each scene in the post-processing scene in turn, and the amount of calculation in the post-processing stage is lower than that in the solution calculation stage, so there will be a waste of resources, and processing in sequence will cause a long time.

[0053] To this end, the improvement of the present application is that, Figure 4 As shown, for the CPU utilization rate when using the solution of the present application to perform operations at each stage, in the grid division stage, nodes with lower core counts are selected for calculation according to their computing power, such as 64-core nodes. The computing power required for the solution calculation stage is larger, so 640-core nodes are used. In the post-processing stage, 24-core nodes are selected for synchronous calculation of multiple scenes based on the number of scenes and the number of core nodes required for each scene. In this way, not only are the resources of each operation stage reasonably allocated, but also the calculations of multiple scene types are processed synchronously, which saves time.

[0054] The above are only embodiments of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme is not described in detail here. Ordinary technicians in the relevant field know all the common technical knowledge in the technical field to which the invention belongs before the application date or priority date, can obtain all the existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the scope of protection of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. An efficient and robust simulation management platform, characterized by: It includes a user end, a management end and an HPC execution end, wherein the user end and the HPC execution end are connected via the management end; The user terminal is used for users to upload and download simulation files and parameter settings; The management end is used to package the simulation files uploaded by the user end, and generate the task requirements of the operation stage according to the parameter settings to send different operation stage call commands to the HPC execution end; The HPC execution end enables the corresponding node to execute the job according to the job stage call command, and returns the execution result to the management end stage by stage; The management end includes a job generation module, a packaging and sending module, and a command generation module; The job generation module is used to generate a job file of a preset stage and a preset format according to wind tunnel and CFD test data; The packaging and sending module is used to package the job files and file information of the preset stage and preset format and send them to the HPC execution end; The command generation module is used to generate a call command according to the parameters and the information of the preset stage, and transmit the call command to the HPC execution end; The preset stage includes meshing, solution calculation and post-processing; The management end further includes a judgment module, which is used to sequentially judge whether the execution result returned by the HPC execution end in a preset stage is completed, and if completed, the management end transmits a control command for executing the next stage to the HPC execution end; The HPC execution end enables the corresponding node to execute the job according to the job stage call command, and returns the execution results to the management end stage by stage. Specifically: For the grid division stage, select nodes with lower core numbers for calculation according to the amount of calculation in the grid division stage. Configure nodes with corresponding core numbers for processing according to the computing power required for the grid division stage job calculation. Transmit the grid division stage job results to the judgment module for judgment. After the judgment is completed, start the solution calculation stage. In the solution calculation stage, according to the computing power required for the calculation of the solution calculation stage job, nodes with the corresponding number of cores are configured for processing; The results of the calculation phase are transmitted to the judgment module for judgment. After the judgment is completed, the post-processing phase begins. For the post-processing stage, based on the number of post-processing scenarios, the required computing power is calculated for each post-processing scenario, and nodes with the corresponding number of cores are configured to execute simultaneously, generate the execution results of the post-processing stage, and transmit them back to the management end.

2. An efficient and robust simulation management platform according to claim 1, characterized in that: The user terminal includes a user login module, a data acquisition module, a parameter setting module and a data transmission module; The user login module is used to log in to the user terminal according to the user's unique ID; The data acquisition module is used to acquire the simulation file uploaded by the user, wherein the simulation file includes wind tunnel and CFD test data and file information; The parameter setting module is used to obtain parameters set by the user; The data transmission module is used to transmit wind tunnel and CFD test data, file information and parameters to the management end.

3. An efficient and robust simulation management platform according to claim 1, characterized in that: The management end is also used to transmit the job execution results back to the user end for the user to view and download.

Images