CN113283151A - Method for optimizing design by using remote server - Google Patents

Abstract

The invention discloses a method for carrying out optimization design by using a remote server. The method comprises the following steps: setting parameters of the basic data at the Client; uploading the parameter setting file to a local server LS, and performing centralized management and lightweight optimization calculation; finite element meshing of ANSYS computing resources and part of simulation data are transmitted to a remote high-performance server end RHPS through a preset path to complete optimization design; returning the result data of the optimized design to a local server LS for extraction, arrangement and storage; the local server LS automatically generates a design report and a structure digital-analog file and returns the design report and the structure digital-analog file to the Client. According to the method, an ANSYS simulation calculation link is deployed on a remote large-scale server, the fat node calculation capacity of the large-scale server is far greater than that of a local workstation, the grid number can be increased to 100 ten thousand orders of magnitude compared with the original grid number of about 20 ten thousand, and meanwhile, the large memory and hard disk capacity is enough to support multiple users to use on line at the same time.

Description

Method for optimizing design by using remote server

Technical Field

The invention belongs to the technical field of wind tunnel tests, and particularly relates to a method for optimizing design by using a remote server.

Background

The high-precision balance optimization design platform is a multidisciplinary optimization balance design platform, integrates various software such as EASA, Isight, EXCEL, ANSYS, CATIA and the like, can realize simultaneous optimization design of multiple target values (micro-strain, relevant interference, heat output, strength, rigidity and the like) in the conventional structural design of the wind tunnel balance, has a plurality of balance functional modules such as design task setting, submission, analog calculation, automatic generation of later reports and the like, and is a complete wind tunnel full-automatic optimization design platform.

However, since the high-precision balance optimization design platform is deployed and installed on the local workstation and is limited by the capability of the local workstation, the finite element meshing which can be completed is not fine enough, and better analog simulation calculation cannot be performed, and meanwhile, the limited memory of the local workstation also restricts multiple users from using the platform on line at the same time.

Currently, it is urgently needed to develop a method for optimizing design by using a remote server to realize high-precision balance optimization design.

Disclosure of Invention

The invention aims to provide a method for carrying out optimization design by using a remote server.

The device used by the method for carrying out optimization design by utilizing the remote server comprises the following steps: the Client, the local server LS and the remote high-performance server RHPS are positioned in the same local area network;

the Client side has an optimized design display interface; a user sets relevant parameters of an optimization task and browses and downloads results through an optimization platform parameter setting and task submitting interface of a Client;

the local server LS comprises an input module, a result module and an output module; the user manages all information of the optimization platform through the local server LS, wherein the information comprises basic data, task data submitted by the user, a task process and the like, and communication between the local server LS and a Client side, and communication between the local server LS and a remote high-performance server RHPS are realized;

the RHPS at the remote high-performance server end comprises an ANSYS-input module and a RUN-ANSYS module; a user provides ANSYS computing resources to a local server LS through a remote high-performance server RHPS, calls the ANSYS computing resources for computing, and returns result data to the local server LS after computing;

basic data are transmitted from a local server side LS to a remote high-performance server side RHPS and sequentially pass through an input module, an ANSYS-input module, a RUN-ANSYS module, a result module and an output module;

the method comprises the following steps:

s1, setting parameters of basic data at a Client to obtain a parameter setting file;

s2, uploading the parameter setting file to a local server side LS, and carrying out centralized management on the parameter setting file by the local server side LS to carry out light-weight optimization calculation;

s3, after the optimization calculation of light weight is completed, the finite element grid division and part of simulation data of ANSYS calculation resources are transmitted to a remote high-performance server end RHPS through a preset path, and the RHPS completes the optimization design;

s4, returning result data of the optimized design to a local server side LS, and extracting, sorting and storing the result data by the local server side LS;

and S5, the local server LS automatically generates a design report and a structure digital-analog file and returns the design report and the structure digital-analog file to the Client.

The method for optimizing design by using the Remote Server is characterized in that the process links of task submission, process management, data processing and the like are placed on a Local Server (Local Server, Local Server side LS), and only the ANSYS solution calculation part which consumes resources very much is transplanted to a High Performance Server (Remote High Performance Server, Remote High Performance Server side RHPS) for calculation. There are three objects to accomplish this task: the Client (Client), the local server LS and the remote high-performance server RHPS are all distributed in the same local area network to realize data and information transmission, and the system belongs to a distributed management type. Therefore, the high-performance computing capability of the RHPS at the remote high-performance server end can be fully utilized to divide finer grids, and finite element computation with better convergence is carried out.

The Client in the method for optimizing the design by using the remote server has an optimization design display interface, the local server LS is provided with an input module, a result module and an output module, and the remote high-performance server RHPS is provided with an ANSYS-input module and a RUN-ANSYS module. The local server side LS is used as a relay station, the input module receives input parameters input by a user through an optimization design display interface of the Client side RHPS and transmits the input parameters to the ANSYS-input module of the remote high-performance server side RHPS, the ANSYS-input module carries out optimization calculation through the RUN-ANSYS module, the result obtained by the optimization calculation is the optimization design, the optimization design is transmitted to the result module, the user arranges the data of the result module into a required format through the output module and displays the data through the optimization design display interface of the Client side.

The method for optimizing the design by utilizing the remote server deploys an ANSYS simulation calculation link to the remote large-scale server, the fat node calculation capacity of the large-scale server is far greater than that of a local workstation, the grid number can be increased to 100 ten thousand orders of magnitude compared with the original grid number of about 20 ten thousand, and simultaneously the large memory and the hard disk capacity are enough to support multiple users (more than or equal to 20) and simultaneously use a high-precision balance optimization design platform on line.

Drawings

FIG. 1 is a flow chart of wind tunnel balance optimization design using the method of the present invention for optimization design using a remote server;

FIG. 2 is an optimization result (rod balance profile) of wind tunnel balance optimization design using the method of the present invention for optimization design using a remote server;

fig. 3a is an optimization result of wind tunnel balance optimization design by using the method for optimization design by using a remote server according to the present invention (a first rectangular measuring beam of a rod balance is a cross-sectional view);

FIG. 3b is a diagram showing the optimization result of wind tunnel balance optimization design using the method of the present invention for optimization design using a remote server (a rod balance rectangular measuring beam cross-sectional view II);

fig. 3c is an optimization result of wind tunnel balance optimization design using the method of the present invention for optimization design using a remote server (bar balance rectangular measuring beam section three);

FIG. 3d is the optimization result of wind tunnel balance optimization design using the method of the present invention for optimization design using a remote server (bar balance rectangular measuring beam cross-section view four);

fig. 3e shows the optimization result of the wind tunnel balance optimization design using the method for optimization design using a remote server according to the present invention (cross-sectional view of rectangular measuring beam of rod balance five).

In the figure, l is the total length of the balance; D. the diameter of the balance; I3. a free end transition section; I8. a fixed end transition section;

I4. length of rectangular beam A; I7. length B of the rectangular beam; br. rectangular beam width; hr. rectangular beam height;

is. height of the single piece of elastic support sheet; bs. single sheet support sheet width; hs. thickness of the single piece of flexible support sheet; ss. single sheet support piece spacing; ns. total number of single support pieces; I6. the length of the resistance transition section; ang. cutting angle;

depth of T-shaped beam holes; bm. width of t-beam; a gap between the Sdm. T-shaped beam and the bottom of the hole; h, thickness of T-shaped beam longitudinal beam; lzm. T-beam stringer height; h m. t beam cross beam thickness; height of a beam of the lhm.t beam; the gap between the Shm.T beam cross beam and the hole wall.

Detailed Description

The present invention will be described in detail below with reference to the accompanying drawings and examples.

Example 1

In the embodiment, the rod balance structure optimization design is performed by using the high-precision balance optimization design platform by using the eosin large server as the remote server.

The high-precision balance optimization design platform is developed based on EASA, and comprises task management functions including functions of task submission queue management, result management and the like, and the optimization quantity can be understood as the quantity of tasks which are simultaneously calculated in parallel, so that the adjustment of the quantity of the parallel tasks can be realized by modifying the setting of a calculation server in the EASA, and 20 or more parallel tasks can be simultaneously executed as long as the calculation resources are sufficient.

For this specific embodiment, the balance optimization design flow shown in fig. 1 is designed first, and in the steps of "calling ANSYS simulation optimization and generating optimization digifax", and "transferring machine content: the method for optimizing design by using the remote server is applied in the process of invoking ANSYS simulation optimization.

The calling and setting steps of this embodiment are as follows:

remote server arrangement

After a user opens a setting interface of a balance detailed design optimization task, a calculation resource setting option is provided in the optimization task setting, and the setting options comprise 4 setting parameters including the core number, the memory, the remote submission type and the execution node type, wherein:

1. the core count and memory are the computing resources that are used to specify the ANSYS to compute on the high performance server.

2. The remote submission type is selected to Ssh, namely PuTTY, execution is carried out, a multitask concurrent process can be seen in an LS local task manager at a local server end, and a specific calculation task list can be seen on an RHPS remote server at a remote high-performance server end;

3. the execution node type includes two types of computing nodes and login nodes, and the user should generally select the computing node to compute.

(II) wind tunnel balance detailed design application

1. Selecting a 'rod balance' type template in ANSYS detailed design from the 'design optimization template';

2. selecting a task execution type as 'structural optimization design', and completing design load setting and structural type selection;

3. structural parameter definition, including selection of design variables, upper and lower limit range setting, discrete value increment and the like;

4. on the premise of not considering the influence factor of the impact load, only setting information such as material parameters, grid size, computing resources, constraint conditions, objective functions, optimization algorithms and the like of the balance;

5. setting the optimization parameters of each balance structure type, clicking a task submission button, adding submitted task information into a balance design task list, and displaying the execution conditions of all tasks in a task operation status bar;

6. meanwhile, the task execution state can be checked in a result management page, and when the task state is displayed as 'calculation completion', the result report of the corresponding task can be checked through the check result report on the right side.

The method comprises the following specific steps:

a. performing PuTTY setting on an input module;

isight remote calling is completed on a local server side LS, applicability setting is carried out on PuTTY of an input module, and secret-free login of a Linux account is realized through public key/private key pairing between the local server side LS and a remote high-performance server side RHPS; the method comprises the following specific steps:

a1. preparing in an early stage;

a11. create a button.exe program in PuTTY: the method comprises the steps that a user configures and opens a remote high-performance server end RHPS through a remote login client SSH interface of the remote high-performance server end RHPS at a local server end LS;

a12. establish the puttygen. exe program in PuTTY: establishing a secret-free login key pair, and setting a Generator button;

a13. build plink. exe program in PuTTY: executing a remote command through a remote login client SSH;

a14. establish pscp. exe program in PuTTY: copying files from Windows to Linux;

a2. generating a key pair;

a21. exe program is operated to generate a secret login-free public key/private key pair;

a22. click the Generate button of the puttygen. exe program interface, move the mouse to Generate a random sequence;

a23. the private key file is saved from ppk to a position where Windows is not easy to be modified;

a24. pasting the public key content to an authorized _ keys file under an SSH directory of a remote login client SSH;

a3. setting and saving PuTTY Session of PuTTY;

a31. inputting an IP address in a Host Name of a Session page of PuTTY, and inputting a Session Name to be Saved in Saved Sessions;

a32. adding a Linux login user name in a Connection-Data page of PuTTY;

a33. adding a previously stored private key into a Connection-SSH-Auth page of PuTTY, wherein the file type is x.ppk;

a34. starting PuTTY Session to obtain PuTTY Session data;

b. establishing Windows and Linux communication based on the SSH and PuTTY of the remote login client;

b1. establish pscp. exe program: the method comprises the following steps that a local server LS uploads a file to a remote high-performance server RHPS and downloads the file from the remote high-performance server RHPS to the local server LS;

b2. build plink. exe program: the method comprises the steps that a simulation execution command is sent to the RHPS, and the step of calling an ANSYS program of the RHPS to perform solution calculation is carried out;

b3. uploading a file to a remote high-performance server end RHPS by a local server end LS:

pscp.exe -l soyotec -i C:\soyotec\soyotec_priv.ppk E:\soyotec\1.txt 192.168.88.128:/home/soyotec/1.txt；

b4. executing a first copying command on the remote high-performance server terminal RHPS:

plink.exe -l soyotec -i c:\soyotec\soyotec_priv.ppk 192.168.88.128 cp /home/soyotec/1.txt /home/soyotec/2.txt；

b5. downloading the file to a local server LS by a remote high-performance server RHPS:

pscp.exe -1 soyotec -i C:\soyotec\soyotec_priv.ppk192.168.88.128:/home/

soyotec/2.txt E:\soyotec\2.txt；

b6. uploading a file to a remote high-performance server terminal RHPS:

pscp.exe -l bal_opt -i D:\EASA\bal_opt_priv_ssh.ppk D:\SOYOTEC\q.txt 7.31.130.28:/vol6/home/bal_opt/BalOptPF_Workdirectory/q.txt；

b7. executing a second copying command on the remote high-performance server terminal RHPS:

plink.exe -l bal_opt -i D:\EASA\bal_opt_priv_ssh.ppk 7.31.130.28 cp /vol6/home/bal_opt/BalOptPF_Workdirectory/q.txt；

plink.exe -l bal_opt -i D:\EASA\bal_opt_priv_ssh.ppk 7.31.130.28 cp /vol6/home/bal_opt/BalOptPF_Workdirectory/p.txt；

b8. downloading the file to a local server LS:

pscp.exe -1 bal_opt -i D:\EASA\bal_opt_priv_ssh.ppk 7.31.130.28: /vol6/home/bal_opt/BalOptPF_Workdirectory/p.txt D:\SOYOTEC\；

b9, after the PutTTY environment configuration is completed, performing attribute configuration on Grid of an OS Command component of integrated optimization software Isight of the remote high-performance server end RHPS, selecting Components-OS Commands-Grid in Preferences, and checking 'Enable use of the Grid Plug-in';

b10. adding paths of plink.exe and pscp.exe in Preferences of the integrated optimization software Isight, adding a path of a "-l user name-i secret key to the pscp.exe, and configuring and executing the integrated optimization software Isight according to a PutTTY environment after submitting task calculation;

b11. the settings for the OS Command component in the integrated optimization software Isight to invoke the remote ANSYS calculation are as follows:

grid setting: selecting a PuTTY remote type of a remote login client SSH, and setting a PuTTY Session name which is configured and stored in a registry in a Romote Host;

b112. inputting an ANSYS calling Command of a remote high-performance server end RHPS in a Command Line of integrated optimization software Isight;

b113. the APDL command file executed by ANSYS is added to Input Files of ANSYS and generated by an ANSYS _ Input component, and the ANSYS calculation result file is added to the Output Files, so that the result file can be transmitted back to the local for processing after calculation is completed.

Finally, the optimization results of the rod balance are shown in fig. 2, fig. 3a to fig. 3e, tables 1a to 1d, tables 2a to 2b and tables 3a to 3 b. Tables 1a to 1d are rod balance characteristic parameters, tables 2a to 2b are rod balance optimization schemes, and tables 3a to 3b are rod balance simulation verification results.

As can be seen from the figures and tables:

1. the whole balance structure design process can be completed to obtain an optimized structure;

2. the division of grids can be more refined by fully utilizing the remote high-performance server, and the calculation quality of simulation is improved;

3. the high-performance remote server can support simultaneous online design of a plurality of users, and operation efficiency is remarkably improved.

The method for optimizing the design by utilizing the remote server can strip a large part of simulation operation from the remote server, can be popularized to other fields, such as a CFD fluid simulation calculation part in aircraft layout optimization, an electromagnetic interference calculation part in multi-physical-field coupling comprehensive design and the like, can be calculated by the remote high-performance server, and returns to the local for final comprehensive optimization evaluation after calculation.

Although the embodiments of the present invention have been disclosed above, it is not limited to the applications listed in the description and the embodiments, but it can be applied to various fields suitable for the present invention. Additional modifications and refinements of the present invention will readily occur to those skilled in the art without departing from the principles of the present invention, and therefore the present invention is not limited to the specific details and illustrations shown and described herein without departing from the general concept defined by the claims and their equivalents.

Claims (1)

1. A method for performing an optimization design using a remote server, the method comprising: the Client, the local server LS and the remote high-performance server RHPS are positioned in the same local area network;

the Client side has an optimized design display interface; a user sets relevant parameters of an optimization task and browses and downloads results through an optimization platform parameter setting and task submitting interface of a Client;

the local server LS comprises an input module, a result module and an output module; the user manages all information of the optimization platform through the local server LS, including basic data, task data submitted by the user and task processes, and communication between the local server LS and the Client, and between the local server LS and the remote high-performance server RHPS is realized;

the RHPS at the remote high-performance server end comprises an ANSYS-input module and a RUN-ANSYS module; a user provides ANSYS computing resources to a local server LS through a remote high-performance server RHPS, calls the ANSYS computing resources for computing, and returns result data to the local server LS after computing;

basic data are transmitted from a local server side LS to a remote high-performance server side RHPS and sequentially pass through an input module, an ANSYS-input module, a RUN-ANSYS module, a result module and an output module;

the method comprises the following steps:

s1, setting parameters of basic data at a Client to obtain a parameter setting file;

s2, uploading the parameter setting file to a local server side LS, and carrying out centralized management on the parameter setting file by the local server side LS to carry out light-weight optimization calculation;

s3, after the optimization calculation of light weight is completed, the finite element grid division and part of simulation data of ANSYS calculation resources are transmitted to a remote high-performance server end RHPS through a preset path, and the RHPS completes the optimization design;

s4, returning result data of the optimized design to a local server side LS, and extracting, sorting and storing the result data by the local server side LS;

and S5, the local server LS automatically generates a design report and a structure digital-analog file and returns the design report and the structure digital-analog file to the Client.

Images