CN-121979849-A - Method and system for storing and calling historical data of fractional order model

CN121979849ACN 121979849 ACN121979849 ACN 121979849ACN-121979849-A

Abstract

The embodiment of the application provides a method and a system for storing and calling historical data of a fractional order model, and relates to the technical field of finite element analysis. The method can establish a historical variable index tag based on the unit number, gao Siji points and time dimension information associated with the fractional order model after constructing the binary direct access external file system. When the current analysis data is obtained, the writing address of the current analysis data in the external file system is calculated according to the historical variable index tag, and the current analysis data is written into the external file system. When the history circulation is started, reversely calculating the reading addresses corresponding to the past time dimensions according to the history variable index tag, and grabbing history data from an external file system to a temporary variable. According to the method, the binary storage is adopted, so that the double-precision floating point number strictly occupies 8 bytes, the utilization rate of the storage space is improved, and the calculation efficiency is improved by establishing a storage and reading path with an external file system.

Inventors

ZHANG YULONG
FENG TAO
LIN GANG
LIN FENG
GUO QIWEN
ZHANG HAIXIA
WANG YUHAN
Bi Xinmao

Assignees

东北大学
中国科学院地理科学与资源研究所

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. A method for calling historical data storage of a fractional order model, the method comprising: constructing an external file system, wherein the external file system is a file system based on binary direct access of a user external data management subprogram, and is used for storing historical data in a fractional model analysis process; Acquiring multidimensional tensor data of the fractional order model, wherein the multidimensional tensor data comprises a unit number, gao Siji points and time dimension information associated with the fractional order model; Establishing a historical variable index tag of the fractional order model based on the multidimensional tensor data; When current analysis data is obtained, calculating a write address of the current analysis data in the external file system according to the historical variable index tag, and writing the current analysis data into the external file system according to the write address, wherein the current analysis data comprises physical quantities of a current increment step; When the history circulation is started, reversely calculating a reading address corresponding to each time dimension in the past according to the history variable index tag, and grabbing history data from the external file system to a temporary variable according to the reading address.
2. The method according to claim 1, wherein the method further comprises: establishing a fractional order model based on a fractional order calculus theory, wherein the fractional order model comprises a fractional order constitutive model; determining a target variable according to the fractional order model, wherein the target variable is a state variable which needs to be stored and called in the writing process of user-defined unit program codes; obtaining modeling parameters, wherein the modeling parameters comprise material parameters, fractional derivative operators, time steps, weight functions and fractional orders; And establishing a control equation corresponding to the fractional order model according to the modeling parameters and the target variables.
3. The method of claim 1, wherein creating the historical variable index tab of the fractional order model based on the multi-dimensional tensor data comprises: Defining a global file establishment module, wherein the global file establishment module is used for setting a file path, a record length and global index parameters; setting global shared constant parameters by using the global file building module, wherein the constant parameters comprise file related parameters and model dimension parameters; And setting the record length of file storage according to the constant parameters and the multidimensional tensor data.
4. The method of claim 1, wherein creating the historical variable index tab of the fractional order model based on the multi-dimensional tensor data comprises: acquiring scale parameters of the fractional order model, wherein the scale parameters comprise unit Gaussian integral points and preset maximum time analysis steps; reading the unit number, the Gao Siji minutes and the time step number from the multidimensional tensor data; And calculating a physical storage address of the current analysis data in the external file system according to the scale parameter, the unit number, the Gaussian integration point and the time step number.
5. The method of claim 4, wherein the physical memory address is calculated as: Wherein, the E is the unit number; the number of points is the unit Gaussian integral; the method is characterized in that the maximum time analysis step number is preset, P is Gao Siji minutes, and T is the time step number.
6. The method according to claim 1, wherein the method further comprises: The method comprises the steps of compiling a subroutine of a user external data management interface in a formula translation language to construct an analysis control module, wherein the user external data management interface is connected with an external file system; when the mark variable is a preset starting value, establishing a file connection channel according to the scale parameter of the fractional order model, wherein the file connection channel is a channel for transmitting binary files between the external file system and the memory; And closing the file connection channel when the mark variable is a preset end value, and releasing system resources used by the fractional order model in the analysis process.
7. The method of claim 1, wherein calculating a write address of the current analysis data in the external file system from the historical variable index tag comprises: reading material parameters of the fractional order model; calculating a state variable according to the material parameter and a variable value at a unit Gaussian integration point, wherein the state variable comprises a physical quantity of a current increment step; and calculating a writing address by utilizing an index algorithm corresponding to the historical variable index tag, wherein the writing address is used for writing the state variable into the external file system.
8. The method of claim 7, wherein reversely calculating a read address corresponding to each time dimension in the past according to the index tag of the history variable, and grabbing history data from the external file system to temporary variables according to the read address, comprises: Reversely calculating the reading addresses corresponding to all time steps by utilizing an index algorithm corresponding to the historical variable index tag; Capturing historical data from the external file system based on the read address; convolving and accumulating the historical data by combining a weight function to obtain fractional derivatives of state variables at the current moment; Substituting the fractional derivative into a residual matrix and a stiffness matrix which are established based on a control equation, and obtaining the current analysis data through a finite element solver.
9. The method according to claim 1, wherein the method further comprises: Acquiring a user-written program, wherein the user-written program comprises a first program, a second program, a third program and a fourth program, the first program is used for parameter reading and geometric calculation, the second program is used for writing in a state variable of a current time point, the third program is used for calculating a fractional derivative of the state variable, and the fourth program is used for executing matrix assembly; Writing a user-defined unit subroutine according to a standard format described in a finite element analysis application help document, wherein the user-defined unit subroutine comprises a user-defined unit code writing area; and adding the user-written program to the user-defined unit code writing area.
10. A fractional order model history data store recall system, the system comprising: The external file module is used for constructing an external file system, and the external file system is a file system based on binary direct access of a user external data management subprogram; the external file system is used for storing historical data in the fractional order model analysis process; The data acquisition module is used for acquiring multidimensional tensor data of the fractional order model, wherein the multidimensional tensor data comprises a unit number, gao Siji points and time dimension information associated with the fractional order model; The index tag module is used for establishing a historical variable index tag of the fractional order model based on the multidimensional tensor data; The data writing module is used for calculating the writing address of the current analysis data in the external file system according to the historical variable index tag when the current analysis data is obtained, and writing the current analysis data into the external file system according to the writing address, wherein the current analysis data comprises the physical quantity of the current increment step; and the data reading module is used for reversely calculating a reading address corresponding to each time dimension in the past according to the history variable index tag when the history cycle is started, and grabbing history data from the external file system to a temporary variable according to the reading address.

Description

Method and system for storing and calling historical data of fractional order model Technical Field The application relates to the technical field of finite element analysis, in particular to a method and a system for storing and calling historical data of a fractional order model. Background In the fields of geotechnical engineering, biomechanics, material mechanics and the like, the medium shows remarkable time memory effect and rheological property. The fractional calculus theory can be used for forming a fractional constitutive model related to mechanics, seepage and other aspects because of the historical dependence of the system can be effectively described, and the fractional constitutive model can be applied to finite element analysis. When the fractional order constitutive model is applied to finite element analysis, the calculation of fractional order derivative is totally history dependent, and when the physical quantity of the current time step is required to be calculated, weighted convolution integration is required to be carried out on all history state variables from the initial moment to the current moment. Currently, ABAQUS is the main implementation platform of fractional order models. The historical variable storage and calling modes of the fractional order model developed based on ABAQUS comprise a memory array and TXT text reading and writing. The memory array scheme is implemented by storing history data in a COMMON area in the Fortran language. Although the scheme has extremely high read-write speed when processing a simple model, the scheme has the limitation of 2GB storage space and cannot be suitable for long-time simulation of a complex model. The TXT text read-write scheme is to write history data into the TXT text file to release the limitation of the data storage space. However, the storage density of the scheme is low, in order to ensure the numerical integrity of the double-precision floating point number, the text file needs to record the numerical value by using a long character format, so that the file volume is far greater than the data volume of the data, and the waste of hard disk resources is caused. In addition, the files need to be frequently switched on and off in the reading and writing of the text files and the cyclic calculation, so that the running resources of the computer are wasted, and the calculation efficiency is reduced. Disclosure of Invention In view of this, the embodiment of the application provides a method and a system for storing and calling historical data of a fractional model, so as to solve the problem of low efficiency of storing and calling the historical data of the fractional model. According to a first aspect of the present application, there is provided a fractional order model history data store calling method, the method comprising: constructing an external file system, wherein the external file system is a file system based on binary direct access of a user external data management subprogram, and is used for storing historical data in a fractional model analysis process; Acquiring multidimensional tensor data of the fractional order model, wherein the multidimensional tensor data comprises a unit number, gao Siji points and time dimension information associated with the fractional order model; Establishing a historical variable index tag of the fractional order model based on the multidimensional tensor data; When current analysis data is obtained, calculating a write address of the current analysis data in the external file system according to the historical variable index tag, and writing the current analysis data into the external file system according to the write address, wherein the current analysis data comprises physical quantities of a current increment step; When the history circulation is started, reversely calculating a reading address corresponding to each time dimension in the past according to the history variable index tag, and grabbing history data from the external file system to a temporary variable according to the reading address. In some embodiments, the method further comprises: establishing a fractional order model based on a fractional order calculus theory, wherein the fractional order model comprises a fractional order constitutive model; determining a target variable according to the fractional order model, wherein the target variable is a state variable which needs to be stored and called in the writing process of user-defined unit program codes; obtaining modeling parameters, wherein the modeling parameters comprise material parameters, fractional derivative operators, time steps, weight functions and fractional orders; And establishing a control equation corresponding to the fractional order model according to the modeling parameters and the target variables. In some embodiments, building a historical variable index tab of the fractional order model based on the multi-dimensional tensor data includes: D