Method and system for evaluating the static strength of an airborne external store product structure

Abstract

The application relates to the computer technical field and discloses a method and system for evaluating the static strength of an airborne external store product structure, which improves the response efficiency and accuracy of static strength evaluation through a proxy model. The method comprises the following steps: dividing global characteristic variables into three characteristic variable groups, namely, a base shape, an enhancement and a weakening; constructing an independent sub-proxy model for any characteristic variable group; analyzing the sensitivity of each characteristic variable in the three sub-proxy models and screening out part of the sensitive characteristic variables as target sensitive characteristic variables, then screening out the sensitive regions of each sensitive characteristic variable, then screening out new global characteristic variable sample points composed of characteristic variable sample points within each sensitive region and other characteristic variable groups outside the group, and then obtaining the static strength corresponding to each new global characteristic variable sample point through virtual simulation; then a supplementary sample set is constructed; and finally, the initial global proxy model is optimized based on the supplementary sample set to obtain a global proxy model for evaluating the static strength of the airborne external store product structure.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer technology, and in particular to a method and system for evaluating the static strength of an airborne external object product. BACKGROUND

[0002] An airborne external object can be applied to a use scenario such as a combat mission of a carrier aircraft, and to ensure the structural safety and reliability of the airborne external object under a complex load environment in a service mission profile, the static strength of the structure needs to be evaluated during the development of the airborne external object product.

[0003] The static strength evaluation methods used at different stages of product structure development are usually simulation calculation and physical static force test.

[0004] However, both simulation calculation and physical static force test have their shortcomings: for finite element simulation calculation evaluation methods, the finite element modeling process is relatively tedious and the calculation time is relatively long; physical static force test also has defects such as long test cycle, high test cost, inability to achieve extreme load conditions, and inability to obtain the full-field response of the tested product. SUMMARY

[0005] The present application aims to disclose a method and system for evaluating the static strength of an airborne external object product, to improve the response efficiency and accuracy of static strength evaluation through a proxy model.

[0006] To achieve the above purpose, the method for evaluating the static strength of an airborne external object product disclosed by the present application comprises:

[0007] Step S3: dividing the global feature variables of the target airborne external object product into a base feature variable group, an enhanced feature variable group and a weakened feature variable group, each feature variable group including at least two feature variables and there being no intersection of feature variables between different feature variable groups;

[0008] Step S4: constructing an independent subsample set for each feature variable group, determining the distribution of sample points of each feature variable in the group based on the Latin hypercube sampling method, the values of other feature variables outside the group combined with the same subsample set being consistent, the static strength corresponding to the global feature variable sample points combined by the group and the group outside being obtained through virtual simulation, and then constructing a sub-proxy model of static strength and each feature variable group;

[0009] Step S5, partial differential equations of all feature variables in the group are sequentially constructed based on the three sub-agent models, sensitivities of the feature variables in the three sub-agent models are analyzed according to the partial differential equations, and part of the feature variables are selected as sensitive feature variables of the target, then sensitive regions of the sensitive feature variables are selected according to local extreme values of the partial differential equations, at least one random feature variable sample point in the group and the mean value of other feature variables outside the group within the value constraint range are selected from the sensitive regions to form a new global feature variable sample point, and then the static strength corresponding to each new global feature variable sample point is obtained through virtual simulation;

[0010] Step S6, a supplementary sample set is constructed based on all new global feature variable sample points and corresponding static strengths, and then optimization processing is performed on the initial global agent model based on global feature variables according to the supplementary sample set, and then static strength evaluation of the airborne external store product structure is performed according to the optimized global agent model.

[0011] Preferably, before step S3, it further comprises:

[0012] Step S1, the distribution of global feature variable sample points in the initial sample set is determined by using the optimal Latin hypercube sampling method, and the static strength result of each global feature variable sample point is obtained through virtual simulation, and then an initial global agent model is constructed according to the initial sample set;

[0013] Step S2, it is judged whether the initial global agent model meets the preset precision index, if yes, the subsequent steps are terminated and the initial global agent model is taken as the final global agent model; if not, step S3 is turned to.

[0014] Preferably, the target airborne external store product is a tail cabin.

[0015] The corresponding global feature variables are composed of the tail cabin front end outer diameter, the tail cabin rear end outer diameter, the tail cabin length, the tail cabin body thickness, the tail cabin viewing window width size, the tail cabin viewing window length size, the rudder installation hole diameter, the tail cabin reinforcing rib cross-sectional area and the tail cabin reinforcing rib length.

[0016] Among them, the tail cabin front end outer diameter, the tail cabin rear end outer diameter, the tail cabin length and the tail cabin body thickness are classified as basic shape feature variables; the tail cabin viewing window width size, the tail cabin viewing window length size and the rudder installation hole diameter are classified as weakening feature variables; and the tail cabin reinforcing rib cross-sectional area and the tail cabin reinforcing rib length are classified as enhancing feature variables.

[0017] Preferably, the sub-agent model of the basic shape feature variable adopts a polynomial response surface sub-agent model, and the weakening feature variable and the enhancing feature variable respectively adopt a Kriging sub-agent model.

[0018] Preferably, in step S5, the method for selecting part of the sensitive feature variables as the target is specifically:

[0019] If the absolute value of the partial derivative of any characteristic variable is greater than a set first empirical threshold value and the fluctuation index evaluated by the derivation of the partial differential equation is greater than a set second empirical threshold value, the characteristic variable is screened as a sensitive characteristic variable.

[0020] Preferably, the method for determining the sensitive region is:

[0021] Based on the region where the local extreme value of the partial differential equation of the sensitive characteristic variable appears, the outer boundary of the region is the intersection line of the two-dimensional circle or high-dimensional sphere with the radius , wherein, is the number of sensitive characteristic variables in the partial differential equation, , is the number of sensitive regions of the th characteristic variable.

[0022] To achieve the above purpose, the application further discloses an airborne external object product structure static strength evaluation system, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above method when executing the computer program.

[0023] The application has the following beneficial effects:

[0024] 1. The global characteristic variables are grouped, and a sub-agent model is established based on the grouping. Compared with the global agent model, the variable dimension of the sub-agent model is smaller, and the calculation cost and the number of samples for establishing the sub-agent model are also smaller. Therefore, the variable sensitivity analysis using the sub-agent model has smaller calculation amount when establishing and solving the partial differential equation, and the sensitivity analysis result is more accurate.

[0025] 2. In the sample construction process of the supplementary sample set, the non-sensitive characteristic variables are first filtered out to participate in the sensitive region screening, and then the non-sensitive regions of the sensitive characteristic variables are filtered out. Therefore, the two-stage screening mechanism ensures the high optimization value of each supplementary sample, thereby improving the accuracy of the optimized agent model.

[0026] 3. In the grouping process of the global characteristic variables, each characteristic variable group includes at least two characteristic variables, and there is no intersection between different characteristic variable groups. The grouping of a single characteristic variable can be divided according to its contribution to the overall structure static strength. The relevant experience can be obtained based on virtual simulation, and if necessary, the final global agent model can be determined by comparing the optimization contributions of the supplementary sample sets constructed by two different grouping forms to the initial global agent model.

[0027] 4. The final surrogate model of the application can improve the response efficiency and accuracy of static strength evaluation, and under the effect of post-supplement optimization of the supplement sample set, the accuracy requirement of the initial global surrogate model can be appropriately reduced.

[0028] The application will be described in further detail below with reference to the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0029] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application, and are incorporated in and constitute a part of this application. The embodiments of the application illustrated in the drawings, and their description, serve to explain the application without limiting its scope.

[0030] Figure 1 is a schematic diagram of the method for evaluating the structural static strength of the airborne external object product according to an embodiment of the application.

[0031] Figure 2 is a schematic diagram of the distribution of the basic feature variable group according to the Latin hypercube sampling method.

[0032] Figure 3 is a schematic diagram of the distribution of the enhanced feature variable group according to the Latin hypercube sampling method.

[0033] Figure 4 is a schematic diagram of the distribution of the weakened feature variable group according to the Latin hypercube sampling method.

[0034] Figure 5 is a schematic diagram of the distribution of the sensitive area of the enhanced feature variable according to an embodiment of the application. DETAILED DESCRIPTION

[0035] The embodiments of the application will be described in detail below with reference to the drawings, but the application can be implemented in various different ways limited and covered by the claims.

[0036] Embodiment 1

[0037] This embodiment discloses a method for evaluating the structural static strength of an airborne external object product, and takes a tail cabin as an example of the target airborne external object product to illustrate in detail.

[0038] As shown in Figure 1 , the method of this embodiment includes the following steps:

[0039] Step S1, determining the global feature variables of the target airborne external object product, using the optimal Latin hypercube sampling method to determine the distribution of the global feature variable sample points of the initial sample set, and obtaining the static strength results of each global feature variable sample point through virtual simulation, and then constructing an initial global surrogate model according to the initial sample set.

[0040] In this step, the global characteristic variables corresponding to the tail cabin are composed of the tail cabin front end outer diameter, the tail cabin rear end outer diameter, the tail cabin length, the tail cabin body thickness, the tail cabin inspection window width size, the tail cabin inspection window length size, the rudder installation hole diameter, the tail cabin reinforcing rib cross-sectional area and the tail cabin reinforcing rib length.

[0041] Preferably, the virtual simulation system of the embodiment can fuse the measured test data to ensure the reliability of the simulation results, and the specific fusion method does not belong to the research content of the present application (which can refer to other patents applied by the applicant of the present case), and will not be described here.

[0042] Step S2, judge whether the initial global agent model meets the preset accuracy index, if yes, terminate the subsequent steps and take the initial global agent model as the final global agent model; if not, go to step S3.

[0043] In this step, whether the agent model meets the preset accuracy index can be the root mean square error and the correlation-based accuracy evaluation index, the root mean square "determination coefficient", etc., which are existing technologies in the art and will not be described here.

[0044] Step S3, divide the global characteristic variables into a basic shape characteristic variable group, an enhanced characteristic variable group and a weakened characteristic variable group, each characteristic variable group includes at least two characteristic variables and there is no intersection between different characteristic variable groups.

[0045] In this step, according to the characteristics of the variables, the tail cabin front end outer diameter, the tail cabin rear end outer diameter, the tail cabin length and the tail cabin body thickness are classified as basic shape characteristic variables, which determine the basic shape size and characteristics of the airborne external store tail cabin product; the tail cabin inspection window (hole opened on the tail cabin) width size, the tail cabin inspection window length size and the rudder installation hole diameter are classified as weakened characteristic variables, which weaken the static strength of the airborne external store tail cabin product; the tail cabin reinforcing rib cross-sectional area and the tail cabin reinforcing rib length are classified as enhanced characteristic variables, which enhance the static strength of the airborne external store tail cabin product.

[0046] Step S4, build an independent subsample set for any characteristic variable group, the distribution of each characteristic variable sample point in the group is determined based on the Latin hypercube sampling method, the values of other characteristic variables outside the group combined with the same subsample set are consistent, the static strength corresponding to the global characteristic variable sample points combined by the group and the group outside is obtained through virtual simulation, and then the static strength and the sub-agent model of each characteristic variable group are built respectively.

[0047] The values of the other characteristic variables outside the group of the same sub-sample set are consistent, which aims to avoid affecting the accuracy of the sensitivity analysis results of the sub-agent model on the characteristic variables within the group; and the values of any characteristic variable as the characteristic variable outside the group of the two sub-sample sets can be consistent or inconsistent, and the role is only to constitute the value of the global characteristic variable to obtain the static strength virtual simulation results required for the aforementioned sensitivity analysis.

[0048] In this step, the distribution of the three characteristic variable sample points within the group determined based on the Latin hypercube sampling method (preferably optimal Latin hypercube sampling) is as shown in Figures 2 to 4 . Among them, the values of the other characteristic variables outside the group in the same sub-sample set can be randomly taken, but the consistency of the values of each sample in the other characteristic variables outside the group needs to be ensured, so as to lay a foundation for the sensitivity analysis of the subsequent step S5, and the virtual simulation of this step is consistent with step S1.

[0049] In this step, the PRS (polynomial response surface) sub-agent model can be established for the base characteristic variable; and the Kriging sub-agent model can be established for the weakening characteristic variable and the enhancing characteristic variable.

[0050] The base characteristic sub-agent model structure established can be specifically as follows: ; wherein, ; ; is the coefficient to be solved of the polynomial function; respectively represent the outer diameter of the front end of the tail cabin, the outer diameter of the rear end of the tail cabin, the length of the tail cabin, and the thickness of the tail cabin body; represents the estimated value of the static strength.

[0051] The Kriging sub-agent model can be as follows: ; wherein, is the base function vector, is the base function matrix, is the correlation function (for example: exponential function and Gaussian function, etc.), is the correlation matrix, is the generalized least square estimation; wherein, is the enhancing characteristic variable or weakening characteristic variable matrix, represents the static strength value of the global characteristic variable combined within and outside the group obtained through virtual simulation, represents the response value matrix of the input sample.

[0052] Step S5, constructing partial differential equations of all feature variables in the group in turn based on the three sub-agent models, analyzing the sensitivity of each feature variable in the three sub-agent models according to the partial differential equations and screening out some as the sensitive feature variables of the target, then screening out the sensitive regions of each sensitive feature variable according to the local extreme value of the partial differential equation, and then screening out at least one random in-group feature variable sample point and the mean value of other feature variables outside the group within the value constraint range to form a new global feature variable sample point, and then obtaining the static strength corresponding to each new global feature variable sample point through virtual simulation.

[0053] In this step, within the range of feature variables, if the absolute value of the partial derivative of any feature variable is greater than the set first empirical threshold and the fluctuation index evaluated by the derivative of the partial differential equation is greater than the set second empirical threshold, the feature variable is screened as a sensitive feature variable.

[0054] Wherein, the partial differential equation of the base shape feature variable can be: The partial differential equations of the enhanced feature variable and the weakened feature variable can be: .

[0055] Further, the method for determining the sensitive region is: based on the local extreme value region of the partial differential equation of the sensitive feature variable, taking the extreme value as the center of the circle or sphere, and the outer boundary of the region is the intersection line of the two-dimensional circle or high-dimensional sphere with the radius , wherein is the number of sensitive feature variables in the partial differential equation, , is the number of sensitive regions of the th feature variable.

[0056] As shown in Figure 5 , the sensitive feature variable determined based on the enhanced feature variable is the reinforcement rib cross-sectional area, and based on the sensitive feature variable, four high-sensitivity regions in descending order can be determined, and the circular dashed line represents the boundary of the sensitive region. Within the region, a reasonable number of random in-group feature variables can be reasonably screened, and the specific number can be set according to experience.

[0057] Refer to Figure 5If the partial derivative (absolute value) of a variable in the range of characteristic variables appears a large value and fluctuates greatly (the fluctuation can be evaluated by taking the derivative of the partial differential equation, if the derivative is close to 0, there is no fluctuation), the characteristic variable is defined as a sensitive characteristic variable. If the partial derivative of a variable in the range of characteristic variables tends to a constant without obvious fluctuation (i.e. the second derivative of the variable to the sub-agent model tends to 0), the characteristic variable is defined as a weak sensitive variable, indicating that the variable has an impact on the structural static strength of the product, but the impact on the structural strength is regular, not sudden, and it is easy to construct an accurate mapping relationship.

[0058] Step S6, based on all new global characteristic variable sample points and corresponding static strength, a supplementary sample set is constructed, and then the initial global agent model based on the global characteristic variable is optimized according to the supplementary sample set, and then the static strength of the airborne external store product structure is evaluated according to the optimized global agent model.

[0059] In this step, the specific optimization of the supplementary sample set varies with the agent model, and is not described in detail; in addition, in the above description, the "sample points" in the "global characteristic variable sample points combined by the in-group and out-group" described in step S4 and the "new global characteristic variable sample points composed of the mean value of the in-group characteristic variable sample points and the other characteristic variables within the value constraint range" described in step S5 are all assignment processing of the specific value distribution of the previous characteristic variable combination, and do not represent the addition of new characteristic variables; taking the tail cabin as an example, the global characteristic variable always remains 9 dimensions unchanged, but the specific value changes or is constrained in different steps, and a complete sample is composed of the in-group or global characteristic variable sample points and the corresponding static strength results, which will not be described in detail.

[0060] Embodiment 2

[0061] The embodiment discloses an airborne external store product structure static strength evaluation system, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor implements the method of embodiment 1 when executing the computer program.

[0062] In summary, the method and system disclosed in the above two embodiments of the application have at least the following beneficial effects:

[0063] 1. Grouping the global characteristic variables, establishing a sub-agent model based on grouping, the variable dimension of the sub-agent model is less than that of the global agent model, the calculation cost and sample number of establishing the sub-agent model are also less; thereby making the variable sensitivity analysis using the sub-agent model have smaller calculation amount in establishing and solving the partial differential equation, and the sensitivity analysis result is more accurate.

[0064] 2. In the sample construction process of the supplementary sample set, the non-sensitive feature variables are filtered out to participate in the sensitive region screening, and the non-sensitive regions of the sensitive feature variables are filtered out; thereby, through the two-stage screening mechanism, the high optimization value of each supplementary sample is ensured, and the precision of the optimized agent model is improved.

[0065] 3. In the grouping process of the global feature variables, each feature variable group includes at least two feature variables, and there is no intersection between different feature variable groups, the grouping of a single feature variable can be divided according to its contribution to the overall structural static strength, the relevant experience can be obtained based on virtual simulation, and if necessary, the final global agent model can be determined by comparing the optimization contribution of the supplementary sample sets constructed by two different grouping forms to the initial global agent model.

[0066] 4. The final agent model of the application can improve the response efficiency and precision of static strength evaluation, and under the effect of the post-supplementary optimization of the supplementary sample set, the precision requirement of the initial global agent model can be appropriately reduced.

[0067] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for evaluating the structural strength of an airborne external store product, characterized in that, include: Step S3: Divide the global feature variables of the target airborne external stores into basic feature variable group, enhanced feature variable group and weakened feature variable group. Each feature variable group includes at least two feature variables and there is no overlap of feature variables between different feature variable groups. Step S4: Construct independent sub-sample sets for any feature variable group. The distribution of each feature variable sample point within the group is determined based on the Latin hypercube sampling method and is consistent with the values ​​of other feature variables outside the group in the same sub-sample set combination. The static intensity corresponding to the global feature variable sample points in the group and outside the group combination is obtained through virtual simulation. Then, sub-surrogate models of static intensity and each feature variable group are constructed respectively. Step S5: Based on the three sub-surrogate models, construct partial differential equations for all feature variables within the group in sequence. Analyze the sensitivity of each feature variable in the three sub-surrogate models according to the partial differential equations and select some sensitive feature variables as targets. Then, select the sensitive regions of each sensitive feature variable according to the local extrema of the partial differential equations. Then, select at least one random sample point of the feature variable within the group and the mean of other feature variables outside the group within the value constraint range from each sensitive region to form a new global feature variable sample point. Finally, obtain the static intensity corresponding to each newly added global feature variable sample point through virtual simulation. Step S6: Construct a supplementary sample set based on all new global feature variable sample points and corresponding static strengths, then perform optimization processing on the initial global proxy model based on global feature variables according to the supplementary sample set, and then perform static strength assessment of the airborne external attachment product structure according to the optimized global proxy model.

2. The method of evaluating the structural strength of airborne external store products according to claim 1, characterized in that, The steps preceding step S3 also include: Step S1: The distribution of global feature variable sample points in the initial sample set is determined by using the optimal Latin hypercube sampling method, and the static intensity results of each global feature variable sample point are obtained through virtual simulation. Then, an initial global proxy model is constructed based on the initial sample set. Step S2: Determine whether the initial global proxy model meets the preset accuracy index. If yes, terminate the subsequent steps and use the initial global proxy model as the final global proxy model; otherwise, proceed to step S3.

3. The method of evaluating the structural strength of an airborne external store product according to claim 1 or 2, characterized in that, The target airborne external storage product is a tail compartment; The corresponding global feature variables consist of the outer diameter of the front end of the tail section, the outer diameter of the rear end of the tail section, the length of the tail section, the thickness of the tail section body, the width of the tail section inspection window, the length of the tail section inspection window, the diameter of the servo mounting hole, the cross-sectional area of ​​the tail section stiffener, and the length of the tail section stiffener. Among them, the outer diameter of the front end of the tail section, the outer diameter of the rear end of the tail section, the length of the tail section, and the thickness of the tail section body are classified as basic shape characteristic variables; the width of the tail section inspection window, the length of the tail section inspection window, and the diameter of the servo mounting hole are classified as weakening characteristic variables; and the cross-sectional area of ​​the tail section reinforcing rib and the length of the tail section reinforcing rib are classified as strengthening characteristic variables.

4. The method for evaluating the static strength of airborne external stores according to claim 3, characterized in that, The sub-surrogate model for the basic feature variables adopts the multinomial response surface sub-surrogate model, while the weakening feature variables and the strengthening feature variables adopt the Kriging sub-surrogate model respectively.

5. The method for evaluating the static strength of airborne external stores according to claim 4, characterized in that, In step S5, the method for selecting some sensitive feature variables as targets is as follows: Within the range of characteristic variables, if the absolute value of the partial derivative of any characteristic variable is greater than the set first empirical threshold and the fluctuation index obtained by evaluating the partial differential equation is greater than the set second empirical threshold, then the characteristic variable is selected as a sensitive characteristic variable.

6. The method for evaluating the static strength of airborne external stores according to claim 5, characterized in that, The method for determining sensitive areas is as follows: The region of local extrema of the partial differential equation based on the sensitive characteristic variable is defined as a circle or sphere centered at the extremum, with the outer boundary of the region being the radius. The boundary line between a two-dimensional circle or a higher-dimensional sphere, where... This represents the number of sensitive characteristic variables in the partial differential equation. , For the first The number of sensitive regions for each sensitive feature variable.

7. A static strength assessment system for airborne external stores, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method described in any one of claims 1 to 6.

Images