CN105160130A - Method for predicting material thermal conductivity on the basis of finite difference method of three-dimensional image - Google Patents

Abstract

The invention discloses a method for predicting material thermal conductivity on the basis of finite difference method of three-dimensional image, and belongs to the method for predicting the material thermal conductivity on the basis of finite difference method. The method has the basic principle: 1) an image analysis method is used for distinguishing different phases or components in the image; 2) a computer language program is used for reading the position and color information of all pixels into a computer memory; 3) each pixel is constructed into a cell element, and the cell element is endowed with the thermal conductivity according to the component to which the cell element belongs; 4) a stable thermal conductivity equation is dispersed, calculating a heat transfer coefficient matrix is calculated, and a calculation equation set is constructed; 5) a temperature field is obtained by the calculation equation set; and 6) the material equivalent thermal conductivity is calculated. The three-dimensional model is adopted to accurately obtain a thermal conductivity result, and the finite difference method is adopted to shorten calculation time and save memory consumption. The whole process is automatically finished, and the method is very practical for the model with a huge node amount.

Description

A kind of method of the prediction of the method for finite difference based on 3-D view material thermal conductivity

Technical field

The present invention relates to a kind of method of numerical Simulation Prediction material thermal conductivity, particularly a kind of method of the prediction of the method for finite difference based on 3-D view material thermal conductivity.

Background technology

The performance of material has conclusive effect for its effective utilization, and therefore the performance of test material is a homework that must do before using it.But for structured material design, the structure of material usually to be changed frequently to the performance wanted can be obtained, this makes to need a large amount of test jobs.By the performance of material structure prediction material, the cycle of design of material can greatly be shortened, therefore widely apply in fields such as composite Materials Design.Traditional Forecasting Methodology generally all uses two kinds of methods: 1) based on the theory calculate and 2 of ideal hypothesis) based on the experimental formula of testing, or both are combined.For first method, actual conditions are often more complicated than perfect condition many, and therefore the result of theory calculate and virtual condition may have relatively large deviation.And for second method, due to the difference of experiment condition, the value calculated by experimental formula also may have larger difference with legitimate reading.This makes to be which kind of method above is all difficult to be applicable to general occasion.

In recent years along with the development of computer technology, people can utilize computer program to carry out virtual structure for simple material structure, then the numerical modelings such as finite element are carried out, rational boundary condition is set again and starting condition calculates, can analyze the performance of material, this method is computer aided design cad (ComputerAidedDesign).Because the method utilizes the real structure of material to carry out modeling, therefore result of calculation has good accuracy rate.At present, this method is widely used in composite structure design field.But, periodic structure, heterogeneous compound substance be there is no for some, such as hot spraying WC-Co coating, because its structure is too complicated, is difficult to build, therefore the method also difficulty have place to show one's prowess.

Numerical modeling based on image refers to the two dimension or the three-dimensional digital photo that utilize and can represent material structure, sets up a kind of method of mathematical calculation model.The method without the need to building material structure in advance, but directly using the image of material as its structure, therefore, it is possible to be applicable to the material structure of various complexity.Nearly ten years, scholar is had to utilize Finite Element Method to carry out finite element modeling to mineral material, building materials and coating material, the success prediction performances such as the mechanics of these materials, calorifics.But, when object is some 3-D views that are larger or that have very fine structure, because node and cell cube quantity exponentially increase, analysis time and required computer resource are increased greatly, and this may lose the simple and efficient advantage of numerical simulation.

Summary of the invention

In order to adapt to the huge three-dimensional model of number of nodes, improve it and take the drawback that computer resource is many, counting yield is low, keep the advantage of numerical simulation on the Optimization of Material Property cycle, the invention provides the method for a kind of method of finite difference based on 3-D view prediction material thermal conductivity, can computer resource be saved, improve arithmetic speed.

The technical solution adopted for the present invention to solve the technical problems is: this Forecasting Methodology is using the pixel of numerical imaging as cell element, and the center of pixel, as node, based on heat transfer basic equation, then replaces finite element method to calculate by difference equation;

Specifically comprise the following steps:

(1) 3-D view that can represent material structure is obtained;

(2) with image analytical method by components different in image or separate mutually;

(3) utilize computer language procedure, the image information processed is read in internal memory, stores in the matrix form;

(4) build cell element with pixel, according to component belonging to it or phase, give thermal conductivity;

(5) calculate the heat-conduction coefficient of each node to all mid-side nodes, value is stored into heat conduction coefficient matrix;

(6) discrete steady state heat transfer equation, is applied to each node, builds accounting equation;

(7) boundary condition and starting condition are set, solve the temperature field in whole territory;

(8) conversion overall thermal conductivity.

In described step (2), different component or be rely on its different color or pattern to realize mutually in segmentation image; Concrete steps are as follows:

A (), for gray level image, adopt suitable threshold values gray-scale map to be converted into the color range with the kind equal number of component or phase, each color range represents a component or phase;

B (), for coloured image, adopts and first image is converted into gray-scale map, or directly according to RGB primary display channels, adopt threshold values to carry out Iamge Segmentation;

C () or the specific form according to some component, coat a kind of color by the region shared by this component, with other components or distinguish mutually.

In described step (4), build cell element, solid model need not be set up in advance, directly according to ownership and the positional information structure cell element of pixel, the center of the position respective pixel of node.

In described step (5), calculate heat-conduction coefficient, adopt 7 point type models, only calculate the direct heat transfer between cell element and coplanar cell element, the direct heat transfer between the cell element not considering only conllinear or only concurrent.

In described step (6), build accounting equation, with the discrete heat transfer equation of the form of finite difference, adopt direct solving method or process of iteration two kinds of method result of calculations.

Beneficial effect, owing to have employed such scheme, pixel direct construction, by the discrete feature of image pixel, is become hexahedron cell element by the present invention, and the 3-D view of material is built into finite difference simulator, calculates the equivalent thermal conductivity of multicomponent or phase structure material.Be compared with the prior art, present invention, avoiding and set up solid model and carry out these two steps of stress and strain model again, save system resource and computing time.

Advantage: adopt three-dimensional model, thermal conductivity result can be drawn more accurately; Adopt method of finite difference, computing time and memory consumption can be saved.Overall process completes automatically, and the model huge for number of nodes is very practical.

1) three-dimensional model more can close to the true Heat transfer of material internal than two dimensional model, the credible result Du Genggao thus calculated.

2) method of finite difference adopts difference equation to carry out discrete to non-individual body, and the temperature value of a computing node, can save system resource than finite element method, improve arithmetic speed greatly.

3) need not solid modelling be carried out, with pixel direct construction cell element, can system resource be saved, avoid irrational stress and strain model.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of the method for finite difference prediction material thermal conductivity that the present invention is based on image.

Fig. 2 is the cell element figure of three-dimensional finite difference model of the present invention.

Fig. 3 is the three-dimensional reconstruction figure of plasma spraying yttria-stabilized zirconia coating of the present invention.

Fig. 4 take Fig. 3 as the temperature field that the finite difference simulator of object calculates.

Embodiment

This Forecasting Methodology: using the pixel of numerical imaging as unit, the summit of pixel, as node, based on heat transfer basic equation, then replaces finite element method to calculate by difference equation; Concrete operation step is as follows:

The first step, picked-up can represent the 3-D view of material structure;

Second step, with image analytical method by components different in image or separate mutually;

3rd step, utilizes computer language procedure, reads in internal memory by the image processed, and stores in the matrix form:

4th step, builds cell element with pixel, according to component belonging to it or phase, gives thermal conductivity;

5th step, calculate each node on rectangular coordinate system three directions with the heat-conduction coefficient of adjacent node, value is stored into heat conduction coefficient matrix;

6th step, discrete steady state heat transfer equation, is applied to each node, builds accounting equation;

7th step, arranges boundary condition and starting condition, solves the temperature field in whole territory;

8th step, conversion overall thermal conductivity.

The described first step, the area size of picked-up is moderate, can represent the structure of material; The too small meeting in region causes the inaccurate problem of result of calculation, crosses conference waste system resource, increases computing time.

Second step, different component or be rely on its different color or pattern to realize mutually in segmentation image; Concrete step is as follows:

(1) for gray level image, adopt suitable threshold values gray-scale map to be converted into the color range with the kind equal number of component or phase, each color range represents a component or phase;

(2) for coloured image, adopt and first image is converted into gray-scale map, or directly according to RGB primary display channels, adopt threshold values to carry out Iamge Segmentation.

(3) according to the specific form of some component, a kind of color is coated in the region shared by this component, with other components or mutually distinguish mutually.

In 3rd step, think component or be homogeneous body mutually, character is same.In the image processed, same component or have identical color attribute mutually, also has identical thermal conductivity.Hole and crackle are as same component, and its thermal conductivity is determined by the gas property of being captured by it.

In 4th step, the territory at each pixel place is the elementary cell of a heat transfer, and be called cell element, its thermal conductivity depends on the material properties of affiliated component.

In 5th step, the position of node is the central point of each cell element, is called center discrete model, the temperature representative of the node medial temperature of whole cell element.

Fig. 2 is cell element and the node schematic diagram of three-dimensional model, and i, j, k represent the position footmark in three-dimensional matrice, and Kx, Ky and Kz are the heat-conduction coefficient in rectangular coordinate system three directions; Heat-conduction coefficient Kx between node (i, j, k) and adjacent node _{(i, j, k)}, Ky _{(i, j, k)}and Kz _{(i, j, k)}be respectively:

${\mathrm{Kx}}_{(i,j,k)}=\frac{2}{\frac{1}{{\mathrm{\λ}}_{(i,j,k)}}+\frac{1}{{\mathrm{\λ}}_{(i+1,j,k)}}}---\left(1.1\right)$

${\mathrm{Ky}}_{(i,j,k)}=\frac{2}{\frac{1}{{\mathrm{\λ}}_{(i,j,k)}}+\frac{1}{{\mathrm{\λ}}_{(i,j+1,k)}}}---\left(1.2\right)$

${\mathrm{Kz}}_{(i,j,k)}=\frac{2}{\frac{1}{{\mathrm{\λ}}_{(i,j,k)}}+\frac{1}{{\mathrm{\λ}}_{(i,j,k+1)}}}---\left(1.3\right)$

In 6th step, the present invention uses the thermal conductivity of steady state heat transfer equation Calculating material, and the expression formula of this equation is:

$-\stackrel{\→}{\▿}.\left(\mathrm{\λ}\stackrel{\→}{\▿}T\right)=0---\left(1.4\right)$

Wherein λ is thermal conductivity, and T is temperature.When this heat transfer equation discrete, the present invention only considers direct heat transfer between territory, neighbor place, i.e. seven point type Three-dimensional Heat-transfer (Fig. 2).For Three-dimensional Heat-transfer, the departure process of equation (1.4) is:

$\begin{array}{c}-\stackrel{\→}{\▿}\·\left(\mathrm{\λ}\stackrel{\→}{\▿}T\right)=-\frac{\∂}{\∂x}\left(\mathrm{\λ}\frac{\∂T}{\∂x}\right)-\frac{\∂}{\∂y}\left(\mathrm{\λ}\frac{\∂T}{\∂y}\right)-\frac{\∂}{\∂z}\left(\mathrm{\λ}\frac{\∂T}{\∂z}\right)\\ =-\frac{{\mathrm{Kx}}_{\left(i,j,k\right)}\left(T_{\left(i+1,j,k\right)}-T_{\left(i,j,k\right)}\right)-{\mathrm{Kx}}_{\left(i-1,j,k\right)}\left(T_{\left(i,j,k\right)}-T_{\left(i-1,j,k\right)}\right)}{{\mathrm{\Δx}}^2}\\ -\frac{{\mathrm{Ky}}_{\left(i,j,k\right)}\left(T_{\left(i,j+1,k\right)}-T_{\left(i,j,k\right)}\right)-{\mathrm{Ky}}_{\left(i,j+1,k\right)}\left(T_{\left(i,j,k\right)}-T_{\left(i,j+1,k\right)}\right)}{{\mathrm{\Δy}}^2}\\ -\frac{{\mathrm{Kz}}_{\left(i,j,k\right)}\left(T_{\left(i,j,k+1\right)}-T_{\left(i,j,k\right)}\right)-{\mathrm{Kz}}_{\left(i,j,k+1\right)}\left(T_{\left(i,j,k\right)}-T_{\left(i,j,k+1\right)}\right)}{{\mathrm{\Δz}}^2}=0\end{array}---\left(1.5\right)$

Due to each limit length of side of pixel equal (Δ x=Δ y=Δ z), can cancellation length of side Δ x, Δ y, Δ z, after abbreviation, equation (1.5) is just rewritten into:

Kx _(i,j,k)T _(i+1,j,k)+Kx _(i-1,j,k)T _(i-1,j,k)+Ky _(i,j,k)T _(i,j+1,k)+Ky _(i,j-1,k)T _(i,j-1,k)(1.6)

+Kz _(i,j,k)T _(i,j,k+1)+Kz _(i,j,k-1)T _(i,j,k-1)-K _(i,j,k)T _(i,j,k)＝0

Wherein

K _(i,j,k)＝Kx _(i,j,k)+Kx _(i-1,j,k)+Ky _(i,j,k)+Ky _(i,j-1,k)+Kz _(i,j,k)+Kz _(i,j,k-1)(1.7)

In 7th step, fixing temperature T is set to two borders in direction to be calculated ₁, T ₂(T ₁≠ T ₂), fixing thermograde is set to internal node, makes temperature along the distribution of heat transfer dimension linear.Other borders arrange adiabatic boundary condition.

In 7th step, all modal equations form linear homogeneous equa tion set:

[K]·[T]＝[C](1.8)

Its matrix of coefficients [K] is spdiags.

1., for less territory, following four kinds of modes can be adopted to solve [T]:

(1) method of elimination;

(2) diagonalization of matrix method;

(3) Gramer's method:

$T_{(i,j,k)}=\frac{\[K_{i,j,k}\]}{\[K\]}---\left(1.9\right)$

(4) inverse matrix of sparse matrix is then obtained, then direct solution:

[T]＝[K] ^-1·[C](1.10)

2. for larger territory, solve with iterative algorithm, as Gauss-Saden that iteration, the iteration form of its three-dimensional problem can be write as:

$T_{(i,j,k)}^{m+1}=\frac{{\mathrm{Kx}}_{(i-1,j,k)}{T}_{\left(i-1,j,k\right)}^{m+1}+{\mathrm{Ky}}_{(i,j-1,k)}{T}_{(i,j-1,k)}^{m+1}+{\mathrm{Kz}}_{\left(i,j,k-1\right)}{T}_{\left(i,j,k-1\right)}^{m+1}+{\mathrm{Kx}}_{\left(i,j,k\right)}{T}_{(i+1,j,k)}^m+{\mathrm{Ky}}_{\left(i,j,k\right)}{T}_{\left(i,j+1,k\right)}^m+{\mathrm{Kz}}_{(i,j,k)}{T}_{\left(i,j,k+1\right)}^m}{K_{(i,j,k)}}---\left(1.11\right)$

Wherein, m is the number of times of iteration; But your iteration convergence of Gauss-Saden is slow, and therefore can adopt continuous over relaxation method, its iteration form is:

$T_{(i,j,k)}^{m+1}=T_{(i,j,k)}^m+\mathrm{\ω}(T_{(i,j,k)}^{m+1,GS}-T_{(i,j,k)}^m)---\left(1.12\right)$

Wherein refer to by Gauss-Seidel method, i.e. formula (1.11), the temperature value of the m+1 time iteration posterior nodal point (i, j, k) of calculating; ω is relaxation factor, and its span is (1,2); The value of ω can affect the speed of convergence of the method, makes the ω that iterative convergence speed is the fastest _optbe worth desirable:

${\mathrm{\ω}}_{opt}=2/\left(1+sin\left(\frac{\mathrm{\π}}{N+1}\right)\right)---\left(1.13\right)$

Wherein N is the nodes in thermograde direction.

In 8th step, whole territory effective thermal expansion coefficient is rate of heat flow q by border and thermograde ratio.For Three Dimensional Thermal conducting problem, temperature loading is in x direction, and the thermal conductivity computing formula in x direction is:

$K_{eff}=\frac{q}{\▿T}=\frac{\[\underset{k=1}{\overset{Nz}{\Σ}}\underset{j=1}{\overset{Ny}{\Σ}}{\mathrm{Kx}}_{(1,j,k)}(T_{2,j,k}-T_2)+\underset{k=1}{\overset{Nz}{\Σ}}\underset{j=1}{\overset{Ny}{\Σ}}{\mathrm{Kx}}_{(Nx-1,j,k)}(T_1-T_{Nx-1,j,k})\]/(2Ny\·Nz)}{(T_1-T_2)/Nx}---\left(1.14\right)$

Nx, Ny, Nz are respectively the number of cells in x, y, z direction.

Temperature loading is in y direction, and the thermal conductivity computing formula in y direction is:

$K_{eff}=\frac{q}{\▿T}=\frac{\[\underset{k=1}{\overset{Nz}{\Σ}}\underset{i=1}{\overset{Nx}{\Σ}}{\mathrm{Ky}}_{(i,1,k)}(T_{i,2,k}-T_2)+\underset{k=1}{\overset{Nz}{\Σ}}\underset{i=1}{\overset{Nx}{\Σ}}{\mathrm{Ky}}_{(i,Ny-1,k)}(T_1-T_{i,Ny-1,k})\]/(2Nx\·Nz)}{(T_1-T_2)/Ny}---\left(1.15\right)$

Temperature loading is in z direction, and the thermal conductivity computing formula in z direction is:

$K_{eff}=\frac{q}{\▿T}=\frac{\[\underset{i=1}{\overset{Nx}{\Σ}}\underset{j=1}{\overset{Ny}{\Σ}}{\mathrm{Kz}}_{(i,j,1)}(T_{i,j,2}-T_2)+\underset{i=1}{\overset{Nx}{\Σ}}\underset{j=1}{\overset{Ny}{\Σ}}{\mathrm{Kz}}_{(i,j,Nz-1)}(T_1-T_{i,j,Nz-1})\]/\left(2Nx\·Ny\right)}{(T_1-T_2)/Nz}---\left(1.16\right).$

Embodiment 1: predicting along spraying direction thermal conductivity of microstructure of plasma sprayed yttria-stabilized zirconia (YSZ) coating.

This example, for plasma spraying YSZ coating, illustrates the forecasting process of the present invention for porosint thermal conductivity.

Adopt the thermal conductivity of the present invention to plasma spray yttria-stabilized zirconia (YSZ) coating to predict, comprise the following steps:

(1) 3-D view of plasma spraying YSZ coating structure is reconstructed, as shown in Figure 3; Fig. 3 is the three-dimensional reconstruction image (300 × 300 × 300 pixel) of plasma spraying yttria-stabilized zirconia (YSZ) coating of the present invention.In figure, black is hole and crackle, and white is YSZ.

(2) utilize computer language procedure, image shown in Fig. 3 is read in internal memory, store [A with three-dimensional matrice form _{(i, j, k)}];

(3) build cell element with pixel, according to material belonging to it, give thermal conductivity [λ _{(i, j, k)}];

(4) calculate the heat-conduction coefficient of each node to three directions under rectangular coordinate system, value is stored into heat conduction coefficient matrix [Kx _{(i, j, k)}], [Ky _{(i, j, k)}], [Kz _{(i, j, k)}];

(5) discrete steady state heat transfer equation, is applied to each node, builds accounting equation;

(6) boundary condition is set to: the temperature of all cell elements of upper surface is set to 100 DEG C, and the temperature of lower surface cell element is set to 0 DEG C, and four side surfaces are set to adiabatic condition; Starting condition is set to: temperature linearity distribution from top to bottom.

(7) temperature field in whole territory is solved with continuous over-relaxation iteration algorithm;

(8) equivalent thermal conductivity in coating y direction is calculated with formula (1.15).

Plasma spraying YSZ coating is a kind of coating of porous, therefore can think and comprise two kinds of components in coating: YSZ solid material and hole, the thermal conductivity of two kinds of components is as shown in table 1.Because the structure of this coating porosity is very complicated, include the blind hole crackle of various spatial form and non-combination interface etc., therefore adopt two dimensional model to be difficult to the Heat transfer of simulating its inside, three-dimensional model need be adopted to calculate.For the three-dimensional reconstruction image (300 × 300 × 300 pixel) of the plasma spraying YSZ coating shown in Fig. 3, containing 27,000 in its model, 000 cell element.Illustrate in table 2 and adopt method of finite difference and finite element method (ANSYS12.1) to carry out internal memory that Modeling Calculation consumes and the time compares to carrying out Fig. 3.If adopt FEM-software ANSYS to calculate, estimation about needs the internal memory more than 70GB, and therefore general computing machine cannot complete calculating at all, adopts finite difference simulator then can address this problem.In order to adapt to the computing power of computing machine, image shown in Fig. 3 is divided into respectively 150 × 150 × 150 pixels and 100 × 100 × 100 pixels.Can find, 1.2% of ANSYS computing time and consume 1.6% of internal memory is only needed for 150 × 150 × 150 pixel finite difference simulators; And for 100 × 100 × 100 pixels, finite difference simulator only needs 1.5% of ANSYS computing time and consume 1.5% of internal memory.This describes three-dimensional model counting yield of the present invention fully higher than Finite Element Method.

Table 3 illustrates the comparison (being standard deviation in bracket) adopting method of finite difference and finite element method (ANSYS12.1) image shown in Fig. 3 to be carried out to the thermal conductivity of analog computation, similarly, 150 × 150 × 150 pixels and 100 × 100 × 100 pixels split gained by original image.Can find, the heat conductivity value that method of finite difference calculates is less than finite element method, little by 17.9% for the image of 150 × 150 × 150 pixels, and the image of 100 × 100 × 100 pixels is little by 16.9%.This is because the interpolation equation difference of the two causes.And with reference to the thermal conductivity 0.99Wm that this coating is surveyed ^-1k ^-1, method of finite difference result of calculation more meets measured value.

Table 4 illustrates and adopts method of finite difference to carry out the thermal conductivity results contrast (being standard deviation in bracket) of two and three dimensions analog computation respectively to image shown in Fig. 3, and the object of two-dimensional analog is the cross section of this 3-D view along z-axis different layers.Can find, two dimension dimension analog result, far below three-dimensional result, is only the latter's 65.2%, also far below measured value.

The each component thermal conductivity of table 1 plasma spraying YSZ coating

Table 2: adopt method of finite difference and finite element method (ANSYS12.1) to carry out internal memory that Modeling Calculation consumes and the time compares to carrying out Fig. 3.

Table 3: adopt method of finite difference and finite element method (ANSYS12.1) to carry out the thermal conductivity results contrast (being standard deviation in bracket) of three-dimensional simulation calculating to image shown in Fig. 3, little image splits gained by original image.

Table 4: adopt method of finite difference to carry out the thermal conductivity results contrast (being standard deviation in bracket) of two and three dimensions analog computation respectively to image shown in Fig. 3, the object of two-dimensional analog is the cross section of this 3-D view along z-axis different depth.

Can be proved by above-mentioned example, the present invention can predict the thermal conductivity of material by 3-D view, forecasting process is comparatively simple, convenient and rapid, and consume system resources is few, and result comparatively two dimensional model is more accurate.

Claims (5)

1., based on the method for the method of finite difference prediction material thermal conductivity of 3-D view, it is characterized in that: specifically comprise the following steps:

(1) 3-D view that can represent material structure is obtained;

(2) with image analytical method by components different in image or separate mutually;

(3) utilize computer language procedure, the image information processed is read in internal memory, stores in the matrix form;

(4) build cell element with pixel, according to component belonging to it or phase, give thermal conductivity;

(5) calculate the heat-conduction coefficient of each node to all mid-side nodes, value is stored into heat conduction coefficient matrix;

(6) discrete steady state heat transfer equation, is applied to each node, builds accounting equation;

(7) boundary condition and starting condition are set, solve the temperature field in whole territory;

(8) conversion overall thermal conductivity.

2. the method for the method of finite difference based on 3-D view according to claim 1 prediction material thermal conductivity, is characterized in that, in step (2), and different component or be rely on its different color or pattern to realize mutually in segmentation image; Concrete steps are as follows:

A (), for gray level image, adopt suitable threshold values gray-scale map to be converted into the color range with the kind equal number of component or phase, each color range represents a component or phase;

B (), for coloured image, adopts and first image is converted into gray-scale map, or directly according to RGB primary display channels, adopt threshold values to carry out Iamge Segmentation;

C () or the specific form according to some component, coat a kind of color by the region shared by this component, with other components or distinguish mutually.

3. the method for the prediction of the method for finite difference based on 3-D view material thermal conductivity according to claim 1, it is characterized in that, in step (4), build cell element, solid model need not be set up in advance, directly according to ownership and the positional information structure cell element of pixel, the center of the position respective pixel of node.

4. the method for the prediction of the method for finite difference based on 3-D view material thermal conductivity according to claim 1, it is characterized in that, in step (5), calculate heat-conduction coefficient, adopt 7 point type models, only calculate the direct heat transfer between cell element and coplanar cell element, the direct heat transfer between the cell element not considering only conllinear or only concurrent.

5. the method for the prediction of the method for finite difference based on 3-D view material thermal conductivity according to claim 1, it is characterized in that, in step (6), build accounting equation, with the discrete heat transfer equation of the form of finite difference, adopt direct solving method or process of iteration two kinds of method result of calculations.