An evaluation method for target vulnerability model confidence based on analytic hierarchy process

Abstract

A confidence evaluation method for target vulnerability models based on the analytic hierarchy process (AHP) is proposed, which involves four layers of evaluation: First, intelligence data is scored and weighted across six dimensions, including accuracy and authority, to evaluate confidence. Next, the confidence of damage calculation data is obtained through parametric simulation, distribution fitting, and hypothesis testing. Experimental data is evaluated using either the damage calculation data evaluation process or a small sample distribution fitting method, depending on the sample size. Finally, based on the data layer results, the confidence of the sub-model layers—comprising the digital model, key component damage criteria, and damage tree—is calculated using weighted averaging, AHP, and Boolean operations. These results are then integrated using AHP to obtain the overall model confidence, which is combined with the confidence of the appropriate munition's power field to obtain the application confidence. This method achieves end-to-end confidence quantification from data to model application, making the evaluation scientific and operable, and providing a reliable basis for the practical application of target vulnerability models.

Description

Technical Field

[0001] This invention belongs to the field of damage assessment and target vulnerability analysis technology, specifically involving a confidence evaluation method for target vulnerability models based on the analytic hierarchy process. Background Technology

[0002] Target vulnerability models are a crucial tool for assessing damage effectiveness, and their accuracy directly impacts the formulation of strike plans and the reliability of assessment results. However, the construction of vulnerability models relies on multi-source data (such as intelligence data, simulation calculation data, and experimental data) and multiple sub-models (such as digital models, damage criteria, and damage trees). These data and models themselves contain uncertainties, making it difficult to quantify the reliability of the final model. Currently, there is a lack of a systematic, hierarchical, and quantifiable method to comprehensively evaluate the overall confidence level of target vulnerability models. Summary of the Invention

[0003] To address the aforementioned problems, the purpose of this invention is to provide a confidence evaluation method for target vulnerability models based on the analytic hierarchy process (AHP), thereby solving the problem of objectively evaluating model confidence in existing technologies.

[0004] To achieve the above objectives, the technical solution adopted by the present invention includes:

[0005] A confidence evaluation method for a target vulnerability model based on the analytic hierarchy process (AHP) includes the following steps: S1, Intelligence Data Confidence Assessment The target intelligence data is scored based on six dimensions: accuracy, authority, traceability, standardization, completeness, and usefulness. The weight of each dimension is determined and the overall score is calculated. The confidence level of the target intelligence data is evaluated based on the overall score. S2, Confidence evaluation of damage calculation data Under the same working conditions, sufficient damage calculation data are obtained. The probability density distribution model of the damage calculation data is obtained by fitting the data dispersion analysis method. The probability density distribution model is used to describe the statistical distribution characteristics of the damage calculation data. The validity of the probability density distribution model is judged by the hypothesis testing method. Then, the confidence interval is calculated based on the z value obtained from the Z table and the statistical parameters of the probability density distribution model. The confidence level of the damage calculation data is obtained according to the ratio of the sample size in the confidence interval to the total sample size. S3, Confidence Evaluation of Experimental Data If the sample size of the experimental results is greater than the preset threshold, the confidence level is evaluated using the same procedure as S2 to obtain the confidence level of the experimental data; if the sample size of the experimental results is not greater than the preset threshold, a data distribution model is established through experimental data acquisition and preprocessing and a small sample distribution fitting method. The data distribution model is used to describe the statistical distribution characteristics of the small sample experimental data. The characteristic value is calculated by combining the preset confidence level and the data distribution model. The characteristic value is compared with the experimental results to obtain the confidence level of the experimental data. S4, Confidence Evaluation of Target Vulnerability Model Using the confidence levels of the target intelligence data obtained in S1, the damage calculation data obtained in S2, and the test data obtained in S3 as the confidence evaluation results of the data layer, a sub-model layer is constructed. This sub-model layer is an intermediate level of the target vulnerability model, consisting of three independent and complementary sub-models: a digital model, a key component damage criterion, and a damage tree. A comprehensive confidence evaluation index is calculated for each sub-model. The comprehensive confidence evaluation index for the digital model is obtained by weighted averaging of the confidence levels of the target intelligence data obtained in S1. The key component damage criterion... The comprehensive confidence evaluation index is calculated using the confidence scores of the damage calculation data obtained in S2 and the test data obtained in S3, respectively, through the analytic hierarchy process (AHP). The comprehensive confidence evaluation index of the damage tree is calculated using a combination of logical Boolean operations and weighted average. Then, the comprehensive confidence evaluation index of the digital model, the comprehensive confidence evaluation index of the damage criteria for key components, and the comprehensive confidence evaluation index of the damage tree are integrated using the AHP to obtain the confidence evaluation index of the target vulnerability model. Finally, combined with the confidence evaluation index of the effective munition's power field, the confidence result for the target vulnerability model evaluation application is obtained.

[0006] Preferably, the six scoring indicators in S1 are as follows: Accuracy: Precision, clarity, and matching; Authority: Reliability, professionalism, historical record, consistency, transparency, and compliance; Traceability: Source tracing, data flow recording, version control, audit logs, responsibility attribution, data correlation, and lifecycle management; Standardization: Data format standardization, data naming rules, data classification, data quality standards, data security standards, and data storage standards; Completeness: Target characteristic name, target characteristic definition, target characteristic classification, target characteristic value or level, target characteristic measurement method, target characteristic weight, and target characteristic related requirements; Usefulness: Accuracy of intelligence data and suitability for application scenarios.

[0007] Preferably, the weight allocation of each dimension in S1 is determined by the analytic hierarchy process, the overall score is set to 100 points, and the standard score of each dimension and scoring index is the product of the total score and the corresponding weight coefficient.

[0008] Preferably, the data dispersion analysis method in S2 is one or more of normal distribution fitting and Weibull distribution fitting; the hypothesis testing method is one of KS test and chi-square test.

[0009] Preferably, the small sample distribution fitting method in S3 is one or more of the small sample Bayes estimation method and the small sample Bootstrap estimation method, with the preset confidence level ranging from 80% to 99%, and the feature values ​​being one or more of the feature lifetime, damage probability threshold, and performance parameter critical value.

[0010] Preferably, the comprehensive evaluation index of confidence level of the digital model in S4 Calculated using the following formula: ; in, This represents the confidence level evaluation result of the i-th target intelligence data, where n represents the number of target intelligence data.

[0011] Preferably, the comprehensive evaluation index of confidence level for the damage criterion of key components in S4 Calculated using the following formula: ; The weighting coefficients of the calculated data representing the damage criterion for the j-th critical component are as follows: This represents the confidence level of the damage calculation data for the j-th critical component. The test data weighting coefficients representing the damage criteria for the j-th critical component. represents the confidence level of the damage test data of the j-th critical component, and m represents the number of components.

[0012] Preferred, the comprehensive confidence evaluation index for damaged trees in S4 Calculated using the following formula:

[0013] in, This represents a comprehensive confidence evaluation index for the destruction of trees in a series-parallel system. ; Indicates the confidence level of the series connection. ; Indicates the confidence level of the parallel component. , represents the confidence level assessment result of the k-th damage event, and l represents the total number of damage events.

[0014] Preferably, the confidence evaluation index of the target vulnerability model in S4 Calculated using the following formula:

[0015] in, The weighting coefficients represent the overall evaluation index of the confidence level of the digital model. The weighting coefficients represent the comprehensive evaluation index of the confidence level of the critical component damage criterion. The weighting coefficients represent the overall confidence level evaluation index for damaged trees. + + =1 Preferably, the confidence score of the target vulnerability model assessment in S4 is calculated using the following formula:

[0016] in, The confidence evaluation index for the power field of suitable ammunition is determined by referring to the calculation method of the comprehensive confidence evaluation index for the damage criterion of key components.

[0017] Compared with the prior art, the advantages of the present invention are: (1) The target vulnerability model confidence evaluation method based on the analytic hierarchy process of the present invention solves the technical problem that the confidence of the target vulnerability model cannot be systematically and accurately quantified in traditional methods. It realizes the full-link confidence quantification from basic multi-source data to the model as a whole and then to actual engineering applications, providing a scientific and reliable confidence basis for the actual use of the model.

[0018] (2) The target vulnerability model confidence evaluation method based on the analytic hierarchy process of the present invention designs differentiated evaluation methods for three types of multi-source data: intelligence, damage calculation and test. It also distinguishes between large and small sample sizes for test data to adapt to different evaluation processes. This not only achieves accurate quantification of uncertainty of multi-source data, but also breaks through the technical bottleneck of small sample test data being difficult to effectively evaluate confidence due to insufficient sample size.

[0019] (3) The target vulnerability model confidence evaluation method based on the analytic hierarchy process of the present invention establishes a standardized evaluation system for intelligence data, including six dimensions such as accuracy and subdivided indicators. It combines the analytic hierarchy process to complete scientific weighting and formulate a 100-point quantitative standard, replacing the subjective evaluation method and greatly improving the objectivity, standardization and repeatability of the confidence evaluation of intelligence data.

[0020] (4) The target vulnerability model confidence evaluation method based on the analytic hierarchy process of the present invention combines the analytic hierarchy process, logical Boolean operation, weighted average and other methods. According to the construction characteristics of different sub-models of digital model, key component damage criteria and damage tree, the corresponding confidence calculation method is adapted to realize the scientific calculation of the confidence of each sub-model. Moreover, it fits the actual construction logic of the vulnerability model to complete the reasonable integration of the overall confidence, and ensures the scientificity and fit of the model confidence evaluation.

[0021] (5) The target vulnerability model confidence evaluation method based on the analytic hierarchy process of the present invention calculates the evaluation application confidence based on the overall confidence of the model and the confidence of the power field of the suitable ammunition. It comprehensively considers the dual uncertainties of the model itself and the application load, making the evaluation results more in line with the actual engineering scenario of damage assessment and significantly improving the practical application value of the evaluation results. Attached Figure Description

[0022] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the following detailed description to explain the invention, but do not constitute a limitation thereof. In the drawings: Figure 1 : Confidence evaluation process for typical artillery target intelligence data; Figure 2 Damage calculation data confidence evaluation process; Figure 3 Experimental data confidence evaluation process Figure 4 : Confidence evaluation process for vulnerability models. Detailed Implementation

[0023] The invention is not limited to the specific embodiments described below. All equivalent modifications made based on the technical solutions of this application fall within the protection scope of this invention. Unless otherwise specified, all components and devices in this invention utilize components and devices known in the prior art.

[0024] Example A confidence evaluation method for a target vulnerability model based on the analytic hierarchy process (AHP) includes the following steps: S1, Intelligence Data Confidence Assessment The target intelligence data is scored based on six dimensions: accuracy, authority, traceability, standardization, completeness, and usefulness. The weight of each dimension is determined and the overall score is calculated. The confidence level of the target intelligence data is evaluated based on the overall score.

[0025] This embodiment takes the confidence evaluation of intelligence data for a certain type of target vulnerability model as an example. The intelligence data scheme includes component types, component functions, component dimensions, component materials, etc. During the target vulnerability analysis, online materials, professional literature, and internal documents are investigated to establish an intelligence data scheme, and this method is used to evaluate its confidence. Specifically, the following steps are included: First, based on the requirements of missile target vulnerability analysis, the quality evaluation criteria for missile target intelligence data are divided into six aspects: accuracy, authority, traceability, standardization, completeness, and usefulness, and their connotations are defined for each. Then, based on the connotations of the missile target intelligence data quality evaluation criteria, the scoring index requirements for each evaluation criterion are given. The specific scoring indexes for the six dimensions in this embodiment are as follows: Accuracy: Precision, clarity, and matching; Authority: Reliability, professionalism, historical record, consistency, transparency, and compliance; Traceability: source tracking, data flow recording, version control, audit logs, accountability, data correlation, and lifecycle management; Standardization: data format standardization, data naming rules, data classification, data quality standards, data security standards, and data storage standards; Completeness: Target characteristic name, target characteristic definition, target characteristic classification, target characteristic value or level, target characteristic measurement method, target characteristic weight, and related requirements for the target characteristic; Usefulness: accuracy of intelligence data and suitability for application scenarios.

[0026] In this embodiment, the weight allocation of each dimension is determined by the analytic hierarchy process. The overall score is set to 100 points, and the standard score of each dimension and scoring indicator is the product of the total score and the corresponding weight coefficient.

[0027] In this embodiment, the relative weight coefficients of the missile target intelligence data quality evaluation criteria and scoring indicators with respect to the overall target of missile target intelligence data quality evaluation are calculated. The total score of the target intelligence data quality evaluation is set to 100 points. After multiplying by 100 points, the corresponding scores of each indicator are obtained, and a target intelligence data quality evaluation scoring form is established, as shown in Table 1.

[0028] Table 1 Intelligence Data Scoring Table

[0029] Six experts in the field were invited to give scores for a missile target intelligence data scheme based on the missile target intelligence data quality evaluation scoring form, and the total score for each expert was calculated.

[0030] Table 2 Expert Scoring Results

[0031] Establish a correlation between the confidence level of intelligence data for a certain missile target and the scoring range of the intelligence data, and give the confidence level of the intelligence data scheme for a certain missile target based on this correlation.

[0032] S2, Confidence evaluation of damage calculation data Under the same working conditions, sufficient damage calculation data are obtained, and a probability density distribution model of the damage calculation data is obtained by fitting the data dispersion analysis method. The probability density distribution model is used to describe the statistical distribution characteristics of the damage calculation data. The validity of the probability density distribution model is determined by combining hypothesis testing methods. Then, the confidence interval is calculated based on the z-value obtained from the Z-table and the statistical parameters of the probability density distribution model. The confidence level of the damage calculation data is obtained based on the ratio of the sample size to the total sample size within the confidence interval. The data dispersion analysis method described in S2 is one or more of normal distribution fitting and Weibull distribution fitting; the hypothesis testing method is one of KS test and chi-square test.

[0033] Based on the calculation results, this embodiment calculates the probability distribution characteristics of the prediction results, such as the mean μ and standard deviation σ, obtains the z value by looking up the Z table, calculates the confidence interval, and counts the number of samples n that fall within the confidence interval. Then, n / N is the confidence level.

[0034] Projectile-target intersection calculations were performed by randomly selecting fragment diameter, velocity, angle of incidence, elevation angle at intersection, and intersection position. The calculation results are as follows: Table 3 Target destruction probability under different projectile-target encounter conditions

[0035] Given that the above calculated data conforms to a normal distribution, the mean μ and standard deviation σ of the above 30 groups of failure probabilities are calculated to be 23.71 and 0.52, respectively.

[0036] Assuming a significance level of 0.2 and a significance level of 0.1, we can obtain the results by consulting the Z-table. Z 0.1 =1.28, based on sample values =23.85, N =30, ,but: Confidence lower bound: =23.54 Confidence ceiling: =23.88 so μ The confidence interval is (23.54, 23.88). The number of samples falling within the confidence interval is n=25, and n / N=83.33%, so the confidence level is 83.33%.

[0037] S3, Confidence Evaluation of Experimental Data If the sample size of the experimental results is large, the same process as S2 is used to evaluate the confidence level and obtain the confidence level of the experimental data; if the sample size of the experimental results is small, a data distribution model is established through experimental data acquisition and preprocessing and a small sample distribution fitting method. The data distribution model is used to describe the statistical distribution characteristics of the small sample experimental data. The feature value is calculated by combining the preset confidence level and the data distribution model. The feature value is compared with the experimental results to obtain the confidence level of the experimental data. The small sample distribution fitting method described in S3 is one or more of the small sample Bayes estimation method and the small sample Bootstrap estimation method. The preset confidence level ranges from 80% to 99%. The feature value is one or more of the feature lifetime, damage probability threshold, and performance parameter critical value.

[0038] This embodiment first collects and statistically analyzes experimental data obtained from typical experiments. Then, it calculates the experimental data distribution model using small sample or extremely small sample evaluation methods. Finally, it calculates eigenvalues ​​at a certain confidence level, compares the eigenvalues ​​with the experimental data, and clarifies the confidence level of the experimental data.

[0039] The experimental data with a small sample size are shown in the table below. The fifth column is the mean estimate of the characteristic lifetime; the sixth column is the estimated characteristic lifetime at a confidence level of c, where c=95% and α=3.

[0040] Table 4. Small sample Bayes estimation of small sample size experimental data.

[0041] In comparison, the confidence levels of lifetime data such as 42.37 million, 20.39 million, and 17.03 million cycles are above 95%, while the confidence level of lifetime data of 104.41 million cycles is far below 95%.

[0042] S4, Confidence Evaluation of Target Vulnerability Model The confidence levels of the target intelligence data obtained in S1, the damage calculation data obtained in S2, and the test data obtained in S3 are used as the confidence evaluation results of the data layer. A sub-model layer is constructed. The sub-model layer is an intermediate level of the target vulnerability model and consists of three independent and complementary sub-models. The three sub-models are a digital model, a damage criterion for key components, and a damage tree. The confidence comprehensive evaluation index is calculated for each sub-model. The confidence comprehensive evaluation index of the digital model is obtained by weighted average of the confidence of the target intelligence data obtained in S1. The confidence comprehensive evaluation index of the damage criterion of the key component is obtained by using the analytic hierarchy process (AHP) to calculate the confidence of the damage calculation data obtained in S2 and the confidence of the test data obtained in S3. The confidence comprehensive evaluation index of the damage tree is obtained by combining logical Boolean operations with weighted average. Then, by integrating the comprehensive evaluation index of the confidence level of the digital model, the comprehensive evaluation index of the confidence level of the damage criterion of key components, and the comprehensive evaluation index of the damage tree through the analytic hierarchy process, the confidence evaluation index of the target vulnerability model is obtained. Combined with the confidence evaluation index of the power field of the suitable ammunition, the final confidence result of the target vulnerability model evaluation application is obtained.

[0043] In this embodiment, during the target reverse design and digital modeling process, the structural and functional parameters and values ​​are mainly determined based on target intelligence data. Given the known target structural and functional data, human error in the digital modeling process can be reduced through iterative correction. Furthermore, current 3D digital modeling software boasts high precision, making both objective and cognitive uncertainties in the 3D digital modeling process highly controllable and relatively small. Therefore, the introduction of uncertainty in the 3D modeling process is ignored, and the uncertainty in the intelligence data is the primary consideration. Based on the confidence evaluation results of the intelligence data, a weighted average method is used to construct a comprehensive evaluation index for the confidence of the digital model. The comprehensive evaluation index for the confidence of the digital model described in S4... Calculated using the following formula: ; in, This represents the confidence level evaluation result of the i-th target intelligence data, where n represents the number of target intelligence data.

[0044] A series of damage simulation calculations and experiments were conducted, and damage criteria and judgments for relevant components were given based on damage modes and damage data. In addition to introducing uncertainties from the 3D digital model constructed based on intelligence data, multi-source mixed uncertainties, including simulation model uncertainty and experimental result uncertainty, were introduced during the complex damage calculation and experiment process. Based on the damage judgment form, the mapping relationship between input and output damage parameters was clarified. The uncertainty evaluation or confidence evaluation of the input parameters referenced the confidence evaluation results of vulnerability data. The confidence level of the output result under the corresponding damage judgment was calculated based on the uncertainty propagation relationship. Based on the confidence evaluation results of the damage judgments for each component, the weight coefficients of different component damage criteria were calculated using the analytic hierarchy process (AHP). Finally, a comprehensive evaluation index for the confidence of the damage judgments of key components was established. The comprehensive evaluation index for the confidence of the damage judgments of key components described in S4 is also included. Calculated using the following formula:

[0045] The weighting coefficients of the calculated data representing the damage criterion for the j-th critical component are as follows: This represents the confidence level of the damage calculation data for the j-th critical component. The test data weighting coefficients representing the damage criteria for the j-th critical component. represents the confidence level of the damage test data for the j-th critical component, and m represents the number of components. Test data and computational data are assigned different weights, determined according to the criterion construction method. For criterions obtained solely through computational data, the weight of test data is 0; for criterions obtained solely through experiments, the weight of computational data is 0; for criterions obtained through a combination of experiments and computation, the weight of computational data is 0.4, and the weight of test data is 0.6.

[0046] This embodiment introduces unquantifiable cognitive uncertainties, such as expert experience, into the process of classifying target damage levels and constructing damage trees. Therefore, referring to the confidence evaluation method for target intelligence data, a confidence evaluation result for the target damage level classification is provided. The base events of the damage tree consist of damage events of critical components. Based on the logical Boolean operation relationships and the confidence levels of the damage criteria for each critical component, a comprehensive confidence evaluation index for the damage tree is calculated. The comprehensive confidence evaluation index for the damage tree of a series-parallel system is as follows: The comprehensive evaluation index of confidence level for tree destruction in a cascaded system is: ,in, The confidence level representing the series connection is , The confidence level representing the parallel part is , represents the confidence evaluation result of the k-th event, and l represents the number of events. Since the damage tree is constructed with reference to the target damage level classification results, a weighted average method is used to construct the final comprehensive confidence evaluation index for the damage tree. The comprehensive confidence evaluation index for the damage tree described in S4... Calculated using the following formula:

[0047] in, This represents a comprehensive confidence evaluation index for the destruction of trees in a series-parallel system. ; Indicates the confidence level of the series connection. ; Indicates the confidence level of the parallel component. , represents the confidence level assessment result of the k-th damage event, and l represents the total number of damage events.

[0048] This embodiment, based on the comprehensive evaluation index of confidence in the digital model, the comprehensive evaluation index of confidence in the damage criterion for key components, and the comprehensive evaluation index of confidence in the damage tree, uses the analytic hierarchy process (AHP) to calculate the weights of different indices, ultimately constructing the confidence evaluation index of the target vulnerability model as described in S4. Calculated using the following formula:

[0049] in, The weighting coefficients represent the overall evaluation index of the confidence level of the digital model. The weighting coefficients represent the comprehensive evaluation index of the confidence level of the critical component damage criterion. The weighting coefficients represent the overall confidence level evaluation index for damaged trees. + + =1 Finally, in the application process, the effective field of the appropriate ammunition is introduced as the load input. The modeling of the ammunition effective field will utilize corresponding simulation calculations and experimental data. The confidence evaluation index for the ammunition effective field can be provided by referring to the damage criterion confidence evaluation method. And the results. Based on the confidence evaluation results of the target vulnerability model and the confidence evaluation results of the ammunition power field, the product of the two is taken as the confidence evaluation result of the target vulnerability model assessment application. The confidence result of the target vulnerability model assessment application described in S4 is calculated by the following formula:

[0050] in, The confidence evaluation index for the power field of suitable ammunition is determined by referring to the calculation method of the comprehensive confidence evaluation index for the damage criterion of key components.

[0051] This embodiment verifies the feasibility of the target vulnerability model confidence evaluation method based on the analytic hierarchy process by evaluating the full-link confidence of a missile target vulnerability model. It achieves accurate quantification of uncertainties in multi-source data and multi-sub-models, and solves the problem of difficulty in evaluating the confidence of target vulnerability models in traditional methods. The evaluation results can directly provide data support for damage assessment engineering practice.

[0052] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure.

[0053] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.

[0054] Furthermore, the various implementation methods disclosed in this solution can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content invented by this disclosure.

Claims

1. A confidence evaluation method for a target vulnerability model based on the analytic hierarchy process (AHP), characterized in that, Includes the following steps: S1, Intelligence Data Confidence Assessment The target intelligence data is scored based on six dimensions: accuracy, authority, traceability, standardization, completeness, and usefulness. The weight of each dimension is determined and the overall score is calculated. The confidence level of the target intelligence data is evaluated based on the overall score. S2, Confidence evaluation of damage calculation data Under the same working conditions, sufficient damage calculation data are obtained, and a probability density distribution model of the damage calculation data is obtained by fitting the data dispersion analysis method. The probability density distribution model is used to describe the statistical distribution characteristics of the damage calculation data. The validity of the probability density distribution model is determined by combining hypothesis testing methods. Then, the confidence interval is calculated based on the z-value obtained from the Z-table and the statistical parameters of the probability density distribution model. The confidence level of the damage calculation data is obtained based on the ratio of the sample size to the total sample size within the confidence interval. S3, Confidence Evaluation of Experimental Data If the sample size of the experimental results is greater than the preset threshold, the confidence level is evaluated using the same procedure as S2 to obtain the confidence level of the experimental data; If the sample size of the experimental results is not greater than the preset threshold, a data distribution model is established by preprocessing experimental data and fitting a small sample distribution. The data distribution model is used to describe the statistical distribution characteristics of small sample experimental data. The feature value is calculated by combining the preset confidence level and the data distribution model. The feature value is compared with the experimental results to obtain the confidence level of the experimental data. S4, Confidence Evaluation of Target Vulnerability Model The confidence levels of the target intelligence data obtained in S1, the damage calculation data obtained in S2, and the test data obtained in S3 are used as the confidence evaluation results of the data layer. A sub-model layer is constructed. The sub-model layer is an intermediate level of the target vulnerability model and consists of three independent and complementary sub-models. The three sub-models are a digital model, a damage criterion for key components, and a damage tree. The confidence comprehensive evaluation index is calculated for each sub-model. The confidence comprehensive evaluation index of the digital model is obtained by weighted average of the confidence of the target intelligence data obtained in S1. The confidence comprehensive evaluation index of the damage criterion of the key component is obtained by using the analytic hierarchy process (AHP) to calculate the confidence of the damage calculation data obtained in S2 and the confidence of the test data obtained in S3. The confidence comprehensive evaluation index of the damage tree is obtained by combining logical Boolean operations with weighted average. Then, by integrating the comprehensive evaluation index of the confidence level of the digital model, the comprehensive evaluation index of the confidence level of the damage criterion of key components, and the comprehensive evaluation index of the damage tree through the analytic hierarchy process, the confidence evaluation index of the target vulnerability model is obtained. Combined with the confidence evaluation index of the power field of the suitable ammunition, the final confidence result of the target vulnerability model evaluation application is obtained.

2. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The scoring indicators for the six dimensions in S1 are as follows: Accuracy: Precision, clarity, and matching; Authority: Reliability, professionalism, historical record, consistency, transparency, and compliance; Traceability: source tracking, data flow recording, version control, audit logs, accountability, data correlation, and lifecycle management; Standardization: data format standardization, data naming rules, data classification, data quality standards, data security standards, and data storage standards; Completeness: Target characteristic name, target characteristic definition, target characteristic classification, target characteristic value or level, target characteristic measurement method, target characteristic weight, and related requirements for the target characteristic; Usefulness: accuracy of intelligence data and suitability for application scenarios.

3. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1 or 2, characterized in that, The weight allocation of each dimension in S1 is determined by the analytic hierarchy process. The overall score is set to 100 points, and the standard score of each dimension and scoring indicator is the product of the total score and the corresponding weight coefficient.

4. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The data dispersion analysis method described in S2 is one or more of normal distribution fitting and Weibull distribution fitting; the hypothesis testing method is one of KS test and chi-square test.

5. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The small sample distribution fitting method described in S3 is one or more of the small sample Bayes estimation method and the small sample Bootstrap estimation method. The preset confidence level ranges from 80% to 99%. The feature value is one or more of the feature lifetime, damage probability threshold, and performance parameter critical value.

6. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The comprehensive confidence evaluation index of the digital model described in S4 Calculated using the following formula: ; in, This represents the confidence level evaluation result of the i-th target intelligence data, where n represents the number of target intelligence data.

7. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The comprehensive confidence evaluation index of the critical component damage criterion described in S4 Calculated using the following formula: ； The weighting coefficients of the calculated data representing the damage criterion for the j-th critical component are as follows: This represents the confidence level of the damage calculation data for the j-th critical component. The test data weighting coefficients representing the damage criteria for the j-th critical component. represents the confidence level of the damage test data of the j-th critical component, and m represents the number of components.

8. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The confidence index of the damaged tree described in S4 Calculated using the following formula: in, This represents a comprehensive confidence evaluation index for the destruction of trees in a series-parallel system. ; Indicates the confidence level of the series connection. ; Indicates the confidence level of the parallel component. , represents the confidence level assessment result of the k-th damage event, and l represents the total number of damage events.

9. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, Confidence evaluation index of the target vulnerability model described in S4 Calculated using the following formula: in, The weighting coefficients represent the overall evaluation index of the confidence level of the digital model. The weighting coefficients represent the comprehensive evaluation index of the confidence level of the critical component damage criterion. The weighting coefficients represent the overall confidence level evaluation index for damaged trees. + + =1.

10. The confidence evaluation method for target vulnerability models based on the analytic hierarchy process as described in claim 1, characterized in that, The confidence score for the target vulnerability model assessment described in S4 is calculated using the following formula: in, The confidence evaluation index for the power field of suitable ammunition is determined by referring to the calculation method of the comprehensive confidence evaluation index for the damage criterion of key components.

Images