Confidence evaluation method and device based on strong robust grey correlation analysis

Abstract

The invention provides a confidence evaluation method and device based on strong-robustness grey correlation analysis, and relates to the technical field of data processing. Original test data corresponding to a target research object and original simulation data corresponding to a simulation model to be evaluated can be acquired; according to the original test data and the original simulation data, calculating to obtain an expected valve penetration rate corresponding to the simulation model; according to the expected valve penetration rate corresponding to the simulation model, performing interpolation updating on the original test data and the original simulation data to obtain target test data and target simulation data; and according to the target test data, the target simulation data and a preset deviation acceptability, determining a target confidence coefficient of the simulation model through grey correlation analysis. Therefore, by introducing the expected valve penetration rate to assist interpolation calculation and introducing the deviation acceptability as a distance threshold index, the anti-noise performance and the anti-disturbance performance of the evaluation result are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a confidence evaluation method and device based on strong robust grey relational analysis. Background Art

[0002] The development process of equipment (such as weapons and equipment) is usually characterized by high cost, small sample size, and short development cycle. As one of the indispensable test methods, simulation test runs through the entire development process from scheme design to system finalization. As an approximate representation of the physical model, the simulation model has a certain similarity with the physical model. In order to effectively verify and use the simulation test results, it is necessary to conduct a confidence quantification evaluation of the simulation model.

[0003] Confidence assessment is an important part of the construction of simulation models and simulation systems. Different simulation model indicators correspond to different confidence assessment methods. For dynamic data, if the simulation model confidence is analyzed by time series characteristics, the grey correlation analysis method can be used. This method uses grey correlation to quantify the consistency of the geometric shape of the data sequence in space to analyze the correlation of the data sequence, which is a powerful tool for simulation confidence verification.

[0004] In the study of grey correlation analysis, it was found that the grey correlation degree has the limitation of being unable to guarantee its order preservation, which leads to the poor robustness of the grey correlation analysis method in terms of noise and disturbance resistance, and then causes the confidence of the obtained simulation model to be inconsistent with the actual situation; only by adjusting parameters such as correlation coefficient and resolution coefficient in order to achieve the approximation between the simulation results and the actual test results, it may have an adverse effect on the reliability assessment and applicability of the model, resulting in prediction bias and application limitations in subsequent model evaluation and practical applications. Summary of the invention

[0005] The object of the present invention is to provide a confidence evaluation method and device based on strong robust grey relational analysis to improve the anti-noise performance and anti-disturbance performance.

[0006] In a first aspect, the present invention provides a confidence assessment method based on strong robust grey relational analysis, comprising: Obtaining original test data corresponding to the target research object and original simulation data corresponding to the simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing the physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating the corresponding simulation model of the target research object; According to the original test data and the original simulation data, the expected valve penetration rate corresponding to the simulation model is calculated; According to the expected valve penetration rate corresponding to the simulation model, the original test data and the original simulation data are interpolated and updated to obtain the target test data and the target simulation data; According to the target test data, target simulation data and the preset deviation acceptability, the target confidence of the simulation model is determined through grey correlation analysis.

[0007] Furthermore, the expected valve penetration rate corresponding to the simulation model is calculated based on the original test data and the original simulation data, including: Perform curve fitting on the original test data to obtain the threshold curve; Perform curve fitting on the original simulation data to obtain the disturbance sequence curve; According to the threshold curve and the disturbance sequence curve, the expected valve penetration rate corresponding to the simulation model is calculated.

[0008] Furthermore, the expected valve penetration rate corresponding to the simulation model is calculated based on the threshold curve and the disturbance sequence curve, including: The expected valve penetration rate is calculated by the following formula v a : ; in, y represents the disturbance sequence curve, express y The derivative of a represents the threshold curve, t Indicates time, Represents three variables a , , t The probability density of .

[0009] Furthermore, the original test data and the original simulation data are interpolated and updated according to the expected valve penetration rate corresponding to the simulation model to obtain the target test data and the target simulation data, including: When the goal is to evaluate the absolute confidence of the simulation model, the valve penetration point is determined according to the difference data corresponding to the original simulation data; wherein the difference data is obtained by subtracting the original simulation data from the original test data, and the valve penetration point is the midpoint of adjacent data points with different positive and negative signs in the difference data; When there are no less than 2 simulation models and the goal is to determine the relative confidence between the simulation models, determine whether the expected valve penetration rates corresponding to the simulation models are equal; if not, determine the valve penetration point based on the difference data corresponding to each group of original simulation data; According to the expected valve penetration rate corresponding to the simulation model, the neighborhood spatial position range of the valve penetration point is interpolated for the original test data and the original simulation data respectively to obtain the corresponding interpolation data sequence; wherein the neighborhood spatial position range of the valve penetration point is the range between two adjacent data points corresponding to the valve penetration point; The original test data and the original simulation data are respectively merged with the corresponding interpolation data sequences to obtain the target test data and the target simulation data.

[0010] Furthermore, the above method also includes: If so, the original test data and each group of original simulation data are determined as target test data and each group of target simulation data respectively.

[0011] Furthermore, the target confidence of the simulation model is determined by grey correlation analysis based on the target test data, the target simulation data and the preset deviation acceptability, including: Based on the target test data, target simulation data and deviation acceptability, the grey relational degree corresponding to the simulation model is calculated; Determine the mean square error distance corresponding to the simulation model according to the grey relational degree corresponding to the simulation model and the preset grey relational degree-mean square error distance model; The target confidence of the simulation model is determined according to the mean square error distance corresponding to the simulation model and the preset mean square error distance-confidence model.

[0012] Furthermore, the grey relational degree corresponding to the simulation model is calculated based on the target test data, the target simulation data and the deviation acceptability, including: According to the deviation acceptability and target test data, the disturbance tolerance data is calculated; The grey relational degree is calculated by the following formula: ; in, Represents target test data X With target simulation data Y The grey correlation between , , Indicates the target test data i Of the data points k data values, Indicates the target simulation data i Of the data points k data values, n represents the number of data points, τ Indicates the preset resolution factor, r i Indicates the disturbance tolerance data r i The disturbance tolerance, , Represents the weight function, satisfying: .

[0013] In a second aspect, the present invention further provides a confidence evaluation device based on strong robust grey relational analysis, comprising: An acquisition module is used to acquire original test data corresponding to the target research object and original simulation data corresponding to the simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing the physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating the corresponding simulation model of the target research object; A calculation module, used for calculating an expected valve penetration rate corresponding to a simulation model according to original test data and original simulation data; An interpolation module is used to interpolate and update the original test data and the original simulation data according to the expected valve penetration rate corresponding to the simulation model to obtain target test data and target simulation data; The determination module is used to determine the target confidence of the simulation model through grey correlation analysis according to the target test data, the target simulation data and the preset deviation acceptability.

[0014] In a third aspect, the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the confidence assessment method based on strong robust grey correlation analysis of the first aspect is implemented.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the confidence assessment method based on strong robust grey relational analysis of the first aspect is executed.

[0016] The confidence evaluation method and device based on strong robust grey correlation analysis provided by the present invention can obtain original test data corresponding to a target research object and original simulation data corresponding to a simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing a physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating a corresponding simulation model of the target research object; based on the original test data and the original simulation data, an expected valve penetration rate corresponding to the simulation model is calculated; based on the expected valve penetration rate corresponding to the simulation model, the original test data and the original simulation data are interpolated and updated to obtain target test data and target simulation data; based on the target test data, the target simulation data and the preset deviation acceptability, the target confidence of the simulation model is determined through grey correlation analysis. In this way, by introducing the expected penetration rate to assist the interpolation calculation, the influence weight of distance on the grey correlation degree is indirectly increased, which effectively makes up for the distance similarity ignored by the traditional grey correlation analysis method when judging the similarity of sequence curves. At the same time, by introducing the deviation acceptability as the distance threshold indicator, the grey correlation coefficient of each data point within a certain deviation range is increased. Combined with the grey correlation analysis, a more stable confidence assessment is achieved, thereby effectively improving the anti-noise and anti-disturbance performance of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A flowchart of a confidence evaluation method based on strong robust grey relational analysis provided by an embodiment of the present invention; Figure 2 A schematic diagram of a grey relational degree-mean square error distance-confidence conversion model provided in an embodiment of the present invention; Figure 3 Examples of multiple sets of data sequences provided by embodiments of the present invention; Figure 4 A schematic diagram for comparing random data sequences of the Gramacy&Lee function provided in an embodiment of the present invention; Figure 5 A schematic diagram of an example data sequence provided for an embodiment of the present invention; Figure 6 A global grey correlation coefficient distribution diagram of data points under the traditional grey correlation analysis method provided by an embodiment of the present invention; Figure 7A local grey correlation coefficient distribution diagram of data points under the traditional grey correlation analysis method provided by an embodiment of the present invention; Figure 8 A grey correlation coefficient diagram of local data points after linear interpolation provided by an embodiment of the present invention; Fig. 9 A global grey correlation coefficient distribution diagram of data points after adding deviation acceptability provided by an embodiment of the present invention; Fig.10 A local grey correlation coefficient distribution diagram of data points after adding deviation acceptability provided by an embodiment of the present invention; Fig.11 Another schematic flow chart of a confidence evaluation method based on strong robust grey relational analysis provided by an embodiment of the present invention; Fig.12 A schematic diagram of the structure of a confidence evaluation device based on strong robust grey relational analysis provided by an embodiment of the present invention; Fig.13 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] At present, the traditional grey correlation analysis method does not consider the geometric similarity between dynamic indicators and the proximity of data curve distances when evaluating dynamic simulation data, and has the defect of poor robustness in anti-noise and anti-interference capabilities, which leads to the model confidence not being consistent with reality. Based on this, the embodiment of the present invention provides a confidence assessment method and device based on strong robust grey correlation analysis, introduces the expected valve penetration rate to assist interpolation calculation, takes the deviation acceptability as the distance threshold indicator, and combines the grey correlation degree-mean square error distance-confidence model to achieve a more stable confidence assessment, improves the defect of the traditional grey correlation analysis method that pays more attention to the similarity of sequence curves, and enhances the anti-noise and anti-interference capabilities of the grey correlation coefficient.

[0021] To facilitate understanding of this embodiment, a confidence evaluation method based on strong robust grey relational analysis disclosed in an embodiment of the present invention is first introduced in detail.

[0022] The embodiment of the present invention provides a confidence evaluation method based on strong robust grey relational analysis, which can be executed by an electronic device with data processing capability. Figure 1The flowchart of a confidence evaluation method based on strong robust grey relational analysis is shown, and the method mainly includes the following steps S110 to S140: Step S110, obtaining original test data corresponding to the target research object and original simulation data corresponding to the simulation model to be evaluated.

[0023] The original test data is a set of time series data obtained by testing the physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating the corresponding simulation model of the target research object.

[0024] The target research object can be any equipment, such as missiles and other weapons and equipment. In the future, the target research object will be a solid-liquid rocket engine as an example, and a certain type of solid-liquid rocket engine internal ballistic experiment-simulation dynamic data will be used to verify the strong robustness of this method. The simulation model to be evaluated can be one or more, and the absolute confidence of the simulation model can be evaluated, or the relative confidence between different simulation models can be evaluated. In the original test data and the original simulation data, each set of time series data includes data values ​​of multiple data points (that is, time points), and each data point can include one or more data values.

[0025] Step S120, calculating the expected valve penetration rate corresponding to the simulation model according to the original test data and the original simulation data.

[0026] The expected crossing rate (ECR) refers to the frequency at which a data point crosses a set threshold in a given time series. Incorporating ECR into the calculation can more accurately reflect the data fluctuation characteristics and be used to assist in interpolation calculations. In this embodiment, the expected crossing rate will be used to calculate the number of times the spatial curve formed by the perturbation sequence data crosses the reference sequence curve and the number of interpolation points within the spatial position range of the crossing point neighborhood, providing a position reference for the subsequent construction of interpolation points.

[0027] In some possible embodiments, the above step S120 may include: performing curve fitting on the original test data to obtain a threshold curve; performing curve fitting on the original simulation data to obtain a disturbance sequence curve; and calculating the expected valve penetration rate corresponding to the simulation model based on the threshold curve and the disturbance sequence curve.

[0028] It should be noted that there is no order of execution between the step of acquiring the threshold curve and the step of acquiring the disturbance sequence curve. In other embodiments, the disturbance sequence curve may be acquired first and then the threshold curve.

[0029] Optionally, the expected valve penetration rate can be calculated by the following formula: v a : ; in, y represents the disturbance sequence curve, express y The derivative of a represents the threshold curve, t Indicates time, Represents three variables a , , t The probability density of .

[0030] The above probability density It can be obtained in the following way: first, the original test data and the original simulation data are subtracted to obtain a difference sequence; then, according to the positive and negative transformations between adjacent differences in the difference sequence, the number of different output values ​​is counted for each sub-interval of the time axis to obtain the statistical data of each sub-interval; wherein the output values ​​include positive valve penetration, negative valve penetration and no valve penetration, and the statistical data include the number of positive valve penetration, the number of negative valve penetration and the number of no valve penetration; finally, according to the statistical data of each sub-interval, the preset estimation algorithm is used to estimate the probability density of different output values ​​in each sub-interval, and the probability density can be obtained. ; Among them, the preset estimation algorithm can adopt algorithms such as maximum likelihood estimation or kernel density estimation.

[0031] The time axis is pre-divided into several non-overlapping time sub-intervals. The function value (i.e., the above-mentioned output value) needs to be counted in each sub-interval, and then the probability density function of the corresponding output value in each sub-interval is estimated using methods such as maximum likelihood estimation or kernel density estimation. The above-mentioned positive penetration valve, negative penetration valve, and no penetration valve are explained as follows: if the previous difference is positive and the next difference is negative, there is a positive penetration valve between the two data points; if the previous difference is negative and the next difference is positive, there is a negative penetration valve between the two data points; if the positive and negative signs of the two differences are the same (for example, both are positive, both are negative, or both are 0), there is no penetration valve between the two data points.

[0032] It should be noted that the embodiment of the present invention does not limit the preset estimation algorithm, and in other embodiments, other types of estimation algorithms may also be used.

[0033] Step S130 , interpolating and updating the original test data and the original simulation data according to the expected valve penetration rate corresponding to the simulation model to obtain target test data and target simulation data.

[0034] This embodiment provides different interpolation update methods for the following two goals: Interpolation update method 1: When the goal is to evaluate the absolute confidence of the simulation model, the valve penetration point is determined according to the difference data corresponding to the original simulation data; wherein the difference data is obtained by subtracting the original simulation data from the original test data, and the valve penetration point is the midpoint of adjacent data points with different positive and negative signs of the data values ​​in the difference data; according to the expected valve penetration rate corresponding to the simulation model, the neighborhood spatial position range of the valve penetration point is interpolated for the original test data and the original simulation data respectively to obtain a corresponding interpolation data sequence; wherein the neighborhood spatial position range of the valve penetration point is the range between two adjacent data points corresponding to the valve penetration point, and the expected valve penetration rate is used to determine the number of interpolation points within the neighborhood spatial position range of each valve penetration point; the original test data and the original simulation data are respectively merged with the corresponding interpolation data sequence to obtain the target test data and the target simulation data.

[0035] Interpolation update method 2: When there are no less than two simulation models and the goal is to determine the relative confidence between the simulation models, determine whether the expected valve penetration rates corresponding to the simulation models are all equal; if so, do not perform interpolation operations, and directly determine the original test data and each group of original simulation data as the target test data and each group of target simulation data, respectively; if not, determine the valve penetration point according to the difference data corresponding to each group of original simulation data; according to the expected valve penetration rate corresponding to the simulation model, interpolate the neighborhood spatial position range of the valve penetration point for the original test data and the original simulation data, respectively, to obtain the corresponding interpolation data sequence; merge the original test data and the original simulation data with the corresponding interpolation data sequence, respectively, to obtain the target test data and the target simulation data.

[0036] Optionally, when interpolating the neighborhood spatial position range of the valve penetration point, the number of interpolation points within the neighborhood spatial position range of each valve penetration point can be the average of the expected valve penetration rate and the number of valve penetration points, that is, the number of interpolation points = (expected valve penetration rate + number of valve penetration points) / 2. This method of calculating the number of interpolation points is relatively simple and has high calculation efficiency.

[0037] It should be noted that the embodiment of the present invention does not limit the method for calculating the number of interpolation points, and in other embodiments, other calculation methods may also be used.

[0038] Step S140, determining the target confidence of the simulation model through grey correlation analysis according to the target test data, the target simulation data and the preset deviation acceptability.

[0039] The confidence evaluation principle based on grey relational analysis is to construct a grey relational degree-mean square error distance model and a mean square error distance-confidence model, substitute the simulation test data (such as target simulation data) and the physical test data (such as target test data) into the grey relational degree-mean square error distance model, and then convert the obtained mean square error distance into confidence through the mean square error distance-confidence model. Among them, the grey relational degree is used to measure the similarity between two sequences; the mean square error (MSE) distance reflects the difference between the actual test data and the simulation data predicted by the simulation model.

[0040] In this embodiment, in order to solve the problem that the confidence evaluation result is low, a grey relational degree calculation method based on deviation acceptability is proposed. Based on this, the above step S140 may include the following steps S141 to S143: Step S141, based on the target test data, the target simulation data and the deviation acceptability, the grey relational degree corresponding to the simulation model is calculated.

[0041] In specific implementation, the disturbance tolerance data can be calculated based on the deviation acceptability and the target test data. Specifically, the target test data is used as the reference sequence b, and the deviation acceptability is recorded as p , the disturbance tolerance data r is calculated by the following formula: ; Then the grey relational degree is calculated by the following formula: ; in, Represents target test data X With target simulation data Y The grey correlation between , , Indicates the target test data i Of the data points k data values, Indicates the target simulation data i Of the data points k data values, n represents the number of data points, τ Indicates the preset resolution factor, r i Indicates the disturbance tolerance data r i The disturbance tolerance, , Represents the weight function, satisfying: .

[0042] Optionally, the weight function can be based on the Sigmoid function and fitted with the above formula satisfied by the weight function as the boundary condition. Based on this, the calculation formula of the grey relational degree can be as follows: ; in, .

[0043] Step S142, determining the mean square error distance corresponding to the simulation model according to the grey relational degree corresponding to the simulation model and a preset grey relational degree-mean square error distance model.

[0044] The grey relational degree-mean square error distance model characterizes the conversion relationship between the grey relational degree and the mean square error distance. The grey relational degree corresponding to the simulation model can be brought into the grey relational degree-mean square error distance model to obtain the mean square error distance corresponding to the simulation model.

[0045] Step S143, determining the target confidence of the simulation model according to the mean square error distance corresponding to the simulation model and a preset mean square error distance-confidence model.

[0046] The mean square error distance-confidence model characterizes the conversion relationship between the mean square error distance and the confidence. The mean square error distance corresponding to the simulation model can be brought into the mean square error distance-confidence model to obtain the target confidence of the simulation model.

[0047] The confidence evaluation method based on strong robust grey correlation analysis provided by the present invention can obtain original test data corresponding to a target research object and original simulation data corresponding to a simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing a physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating a corresponding simulation model of the target research object; based on the original test data and the original simulation data, an expected valve penetration rate corresponding to the simulation model is calculated; based on the expected valve penetration rate corresponding to the simulation model, the original test data and the original simulation data are interpolated and updated to obtain target test data and target simulation data; based on the target test data, the target simulation data and the preset deviation acceptability, the target confidence of the simulation model is determined through grey correlation analysis. In this way, by introducing the expected penetration rate to assist the interpolation calculation, the influence weight of distance on the grey correlation degree is indirectly increased, which effectively makes up for the distance similarity ignored by the traditional grey correlation analysis method when judging the similarity of sequence curves. At the same time, by introducing the deviation acceptability as the distance threshold indicator, the grey correlation coefficient of each data point within a certain deviation range is increased. Combined with the grey correlation analysis, a more stable confidence assessment is achieved, thereby effectively improving the anti-noise and anti-disturbance performance of the evaluation results.

[0048] For ease of understanding, the above confidence assessment method based on strong robust grey relational analysis is introduced in detail below.

[0049] In order to solve the shortcoming of poor robustness of grey relational analysis method, this embodiment introduces the expected valve penetration rate and combines it with the interpolation function to calculate the sequence distance correction point; proposes a deviation acceptability control index, filters and screens the time series data, and constructs a distance weight function, which realizes confidence assessment by combining the grey relational degree-mean square error distance-confidence model. Finally, it is verified by using an experimental-simulation dynamic data example.

[0050] 1. Confidence assessment based on grey relational analysis.

[0051] The confidence evaluation principle based on grey relational analysis is to construct a grey relational degree-mean square error distance model and a mean square error distance-confidence model, substitute the simulation test data (corresponding to the simulation model) and the physical test data (corresponding to the physical model) into the grey relational degree-mean square error distance model, and then convert the obtained mean square error distance into confidence through the mean square error distance-confidence model.

[0052] 1.1 Grey relational degree and mean square error distance solution: Assume that the reference time series data measured by the physical test at equal time intervals is , the time series data generated by the simulation model is , the calculation formula of the grey relational degree of two time series is: (1); In formula (1), Representing time series data X and Y Grey correlation between τ represents the resolution coefficient, , determines the maximum distance Regarding the impact on grey correlation, according to the grey correlation space theory proposed by Deng Julong, considering the integrity and anti-interference effect of the correlation, τ is usually taken as 0.5.

[0053] According to the characteristics of the grey correlation coefficient, the larger the grey correlation coefficient is, the higher the consistency between the simulation output time series and the reference data is; conversely, the smaller the grey correlation coefficient is, the lower the consistency between the simulation output time series and the reference data is.

[0054] The mean square error distance between two time series is: (2).

[0055] According to Sun Yugang's research on grey relational analysis, a threshold of mean square error distance is introduced, which is denoted as d ref, called the reference mean square error distance. hour, X and Y The consistency between them is low and the confidence is low; when hour, X and Y The degree of consistency and confidence are high.

[0056] definition d max At any time t, X and Y The maximum mean square error distance between is called the upper limit of the mean square error distance; d min At any time t, X and Y The minimum mean square error distance between is called the mean square error distance lower limit.

[0057] 1.2 Grey relational degree-mean square error distance model: Reference grey relational degree γ ref It is determined by the selected correlation model and actual needs. The higher the value, the higher the requirement for model consistency. When converting from grey correlation to confidence, the change of confidence is smoother near the reference grey correlation. The grey correlation corresponding to the upper limit of the mean square error distance is defined as γ min , which is called the minimum grey relational degree; the grey relational degree corresponding to the lower limit of the mean square error distance is γ max , which is called the maximum grey relational degree.

[0058] In order to express the relationship between grey relational degree and mean square error distance, the distance from the reference mean square error is defined as d ref The corresponding reference grey relational degree is γ ref .when hour, X and Y The degree of consistency between hour, X and Y According to Li Yuning’s research, the ideal model γ ref =0.5.

[0059] Therefore, when When ;when When ,and γ ( X ,Y )and d ( X , Y ) has a linear relationship, and the model curve is as follows Figure 2 The grey relational degree-mean square error distance model is shown in . Based on this, the grey relational degree-mean square error distance model is: (3).

[0060] 1.3 Mean square error distance-confidence model: The function for converting mean square error distance into confidence can be selected according to the specific situation. Generally, mean square error distance d ( X , Y ) converted to the standard model (i.e. normalized to the standard model) d 0( X , Y ) is used for calculation, and the conversion relationship is: (4); In formula (4), d max0 is the maximum mean square error distance under the standard model; d min0 is the minimum mean square error distance under the standard model. At the same time, define the median mean square error distance under the standard model d mean0 for: (5).

[0061] The corresponding confidence level is C mean =0.5; at the same time, the upper limit of the distance from the mean square error d max The corresponding confidence level is called the lower confidence limit and is denoted by C min ; Lower limit of distance from mean square error d min The corresponding confidence level is called the upper confidence limit and is denoted by C max .

[0062] Distance from reference mean square error d ref The corresponding confidence level is called the reference confidence level C ref , which is the passing line of simulation confidence. In engineering, the reference confidence is often taken as 0.6.

[0063] Then, the standard model of mean square error distance-confidence is: (6); In the formula, α is the shape coefficient of the curve, by changing α To change the shape of the curve. The shape coefficient can be 0.5, which has good versatility. The standard model curve is as follows Figure 2 As shown in the mean square error distance-confidence model in.

[0064] By using the above two models, the grey correlation degree is converted into mean square error distance, and then further converted into confidence, so as to complete the confidence evaluation based on time domain analysis in the dynamic time series of the simulation model.

[0065] One of the reasons why the grey correlation cannot determine the geometric similarity of time series data is that the grey correlation cannot guarantee its order preservation during calculation, that is, it cannot determine the positive or negative deviation of the simulation and test data, and thus it is difficult to determine the consistency of the geometric shape on the continuous time-varying data. When applying the grey correlation analysis method to evaluate the confidence of the model, potential disturbance factors must be considered. If the disturbance factor is within the error tolerance threshold, or an acceptable disturbance is inevitably introduced in the actual engineering test, the robustness of the traditional grey correlation analysis method is low and cannot provide ideal evaluation results. Therefore, in order to ensure the accuracy and effectiveness of the evaluation, additional measures need to be taken to control or correct these disturbance factors.

[0066] 2. Confidence assessment based on strong robust grey relational analysis.

[0067] The essence of the traditional grey correlation analysis method is to judge the similarity of the spatial curve shape constructed by the time series data according to the distance between the corresponding data points of the time series. The problem with this method is that it does not consider the relative distance between the two sets of sequences, and thus has the defect that the calculation results cannot accurately evaluate the correlation of the sequences. Figure 3 The case shown, Figure 3 Where X0, X1, X2, and X3 are different data sequences, it can be concluded that the traditional grey correlation analysis method has the following problems: (1) By intuitively judging the consistency of sequences X1 and X2 compared with sequence X0, it can be concluded that the consistency of sequence X2 with sequence X0 should be better than the consistency of sequence X1 with sequence X0, and then infer that However, the difference between the sequences X1 and X2 at each data point corresponding to the reference sequence X0 is equal or opposite, resulting in the calculation result of the traditional grey correlation analysis being , which is inconsistent with the actual situation.

[0068] (2) For sequences X1 and X3, which are parallel, the displacement difference of each sampling point is a constant. At this time, the minimum difference between the two levels is The maximum difference between the two levels Equal, then the calculation result of the traditional grey correlation analysis method is , ignoring the influence of the relative distance of the sequences, which is inconsistent with the actual situation.

[0069] (3) There are sequence data points that are relatively close in distance, but the grey correlation coefficient is low due to the difference between the two levels of extreme values, which in turn makes the overall grey correlation degree low.

[0070] In this embodiment, for problem (1), the expected valve penetration rate is introduced in combination with the interpolation function, the sequence distance correction point is calculated, and the grey correlation coefficient is increased to calculate the weight in the neighborhood of the valve penetration point. For problems (2) and (3), a deviation acceptability control index is proposed, the time series data is filtered and screened, the distance similarity of the data sequence is judged, and a distance weight function is constructed. The time series data distance and geometric morphology are corrected in a small scale range to enhance the robustness of the grey correlation evaluation result, and the grey correlation-mean square error distance-confidence model is combined to realize confidence evaluation.

[0071] 2.1 Expected valve penetration rate: In this embodiment, the expected valve penetration rate is used to calculate the number of times the spatial curve formed by the disturbance sequence data crosses the reference sequence curve and the spatial position range of the valve penetration point neighborhood, providing a position reference for the subsequent construction of interpolation points.

[0072] Assume that Y(t) is a random sequence of continuous time and continuous state, and is at least once mean square differentiable, then the number of times the sequence Y(t) crosses valve a in the time [t1, t2] is called the number of valve crossings, and its expected value can be constructed with the help of the unit step function u to construct a 0-1 process function Z(t): (7); The formal derivative of Z(t) is: (8); In formula (8) δ is the Dirac function.

[0073] Then the number of times the random sequence Y(t) crosses valve a in the time interval [t1, t2] is for: (9); The expected number of valve penetrations is: (10); In formula (10), y yes Y The function expression of .

[0074] The expected number of valve penetrations per unit time, that is, the expected valve penetration rate, is: (11).

[0075] The higher the expected valve penetration rate is, the more times the disturbance sequence curve crosses the reference sequence curve in unit time. Within a certain disturbance threshold range of the valve penetration point, the relative position distance between the disturbance sequence and the reference sequence is small, and the grey correlation degree should be given a positive gain, such as Figure 3 In the reference example, the X1 and X2 curves are within the defined range. , but the grey correlation calculation results of X1 and X2 are Therefore, the weight of the influence of distance on the grey correlation degree can be indirectly increased by adding calculated data points to the neighborhood of the valve point. In order to ensure that the spatial morphology of the curve constituting the data sequence does not change and to avoid the influence of the local extreme value in the neighborhood of the valve point on the distribution of the inserted data points, the data points are created using single linear interpolation, bilinear interpolation or trilinear interpolation according to different data dimensions.

[0076] Figure 4 The example shown uses Gramacy&Lee Function as the basis function (i.e. as a reference sequence) for verification: (12).

[0077] On the basis of this function, random disturbance is added to construct the disturbance sequence data. The disturbance sequence one is the random disturbance result under the positive proportional weight coefficient. The difference between the corresponding data points of the reference sequence and the disturbance sequence one is taken as the benchmark, and the disturbance sequence two is constructed with a random ±1 times gain to satisfy that the difference between the corresponding data points of the disturbance sequence one and the disturbance sequence two and the reference sequence are equal or opposite.

[0078] The grey correlation degree of the two sets of sequence data is calculated, and the results are shown in Table 1 (Table 1 is a comparison of the evaluation results of the two grey correlation analysis methods under the condition of equal distance difference). It is found that under the traditional grey correlation analysis method, the grey correlation degrees of the two sets of sequence data are equal, both 87.3134%, and the confidence after model mapping is 97.1851%, which is inconsistent with the actual situation. By judging the expected valve penetration rate of disturbance sequence 1 and disturbance sequence 2 respectively, we get After calculating the valve point sequence respectively, the sequences are merged. The merged sequence index is used to perform single linear difference on the original disturbance sequence 1 and the disturbance sequence 2, and the grey correlation degree corresponding to the interpolated sequence 1 and the interpolated sequence 2 is calculated. The grey correlation degrees of the interpolated sequence 1 and the interpolated sequence 2 are 87.1090% and 89.7850% respectively. After model mapping, the sequence confidences are 97.1377% and 97.7010% respectively. The TIC (Theil's Inequality Coefficient, also known as Theil's U, is a statistic used to measure the degree of inequality between the predicted value and the actual value) inequality coefficient method is used for calculation and verification, and the result is , that is, the fitting accuracy of the perturbation sequence 2 for the reference sequence is greater than that of the perturbation sequence 1, which is consistent with the evaluation result of the improved method of this embodiment.

[0079] Table 1

[0080] 2.2 Deviation Acceptability: Figure 5 Take the sine function as the basis function (i.e., the reference sequence) and add random perturbations to construct the perturbation sequence as an example, Figure 5 It can be seen that when there is a disturbance, at a larger geometric scale, the disturbance sequence is similar to the test data sequence (i.e., the reference sequence) in geometry, and it can be seen that the simulation model is acceptable, but at a smaller geometric scale, the simulation confidence of the grey relational analysis is unacceptable. Therefore, in the face of overly "harsh" geometric similarity evaluation conditions, qualified simulation models often get unqualified evaluation results in engineering.

[0081] Figure 5 The grey correlation result shown in is about 54.315%, and the corresponding evaluation confidence result is only about 68.312%. However, this type of disturbance is completely acceptable for practical applications.

[0082] Figure 5 In the example, the global grey correlation coefficient distribution of data points is as follows Figure 6 As shown, the local grey correlation coefficient distribution of data points is as follows Figure 7 As shown, from Figure 6 and Figure 7 It can be found that due to the large range of the two levels and the error in selecting the resolution coefficient, there are data points with very small distance differences from the reference sequence, but the grey correlation coefficient is small, which in turn affects the judgment of the overall grey correlation degree.

[0083] Use expected valve perforation rate assessment Figure 5 After the sequence curve in the valve is interpolated and linear interpolation is performed, the estimated grey correlation result is about 62.048%, and the corresponding confidence result is about 80.504%, which is higher than the traditional grey correlation analysis method. Figure 8 As shown ( Figure 8 Shows the Figure 5 The grey correlation coefficient of local data points is obtained after linear interpolation of the mid-sequence). There are still data points with a 1% deviation from the original data that have slightly lower evaluation results.

[0084] It can be concluded that the use of expected valve penetration rate to evaluate the valve penetration of sequence curves and then perform linear interpolation before grey correlation analysis leads to an increase in the grey correlation degree and confidence results. The essence is that adding valve penetration interpolation points increases the number of evaluation data points in a certain neighborhood of the reference sequence data, which increases the distance weight when calculating the grey correlation degree, and effectively makes up for the distance proximity ignored when judging the similarity of sequence curves by the traditional grey correlation analysis method. However, it still belongs to a macro improvement method, and the optimization object is the data sequence as a whole, without improving the evaluation results of a single data point.

[0085] Therefore, this embodiment further proposes a dimensionless disturbance tolerance based on the valve interpolation method: , , n It is the number of data points in the sequence, which means the acceptable error size for a simulation result, which is equivalent to adding an error tolerance band. The simulation results within the error tolerance band are all credible, and the grey correlation coefficient of a single data point is corrected. After adding the disturbance tolerance index, the original grey correlation analysis evaluation formula becomes: (13); in, Represents the weight function, satisfying: (14).

[0086] In this embodiment, the Sigmoid function is used as the basis, and equation (14) is used as the boundary condition for function fitting to achieve a small slope and smooth transformation near the boundary of the error tolerance band, which is used to construct a weight S-shaped distribution so that the output is limited to a specific range, which is beneficial to the stability and interpretability of the model weight update.

[0087] Taking the experimental time series data as the benchmark sequence b, the dimensionless parameter deviation acceptability of disturbance tolerance is p , satisfying formula (15): (15); The calculation formula of grey relational degree after adding the fitted weight function is shown in formula (16): ; in, .

[0088] In summary, for different examples, the corresponding p Value, (100± p )% times the benchmark value is taken as the error tolerance boundary, combined with the linear difference method after the expected valve penetration rate is determined, the grey correlation coefficient of each point in the data sequence is optimized, and a confidence evaluation method based on strong robust grey correlation analysis is constructed.

[0089] The valve interpolation method corrects the distance weight of the grey correlation calculation formula, and the deviation acceptability index filters and corrects the grey correlation coefficient. Figure 5 The example uses the global average deviation of 2% as the deviation acceptance selection standard. The grey correlation coefficient of the data points with 2% deviation acceptance added after the valve interpolation is as follows: Fig. 9 and Fig.10 The calculation results are shown in Table 2. Fig. 9 contrast Figure 6 It is found that the method of this embodiment increases the grey correlation coefficient of each data point within a certain deviation range of the reference sequence, and enhances the weight coefficient of distance in calculating the grey correlation coefficient. By calculating the TIC inequality coefficient, TIC =0.01032, which proves that the fitting accuracy of the sequence is high; when the valve interpolation method and the deviation acceptability are used at the same time, the overall grey correlation degree of the sequence reaches 68.439%, and the confidence level reaches 87.602%. Compared with the traditional grey correlation analysis method, the method in this embodiment is more consistent with the TIC judgment conclusion.

[0090] Table 2

[0091] 2.3 Confidence evaluation process based on strong robust grey relational analysis: The confidence evaluation steps based on strong robust grey relational analysis are summarized as follows: Fig.11 As shown, first, the expected valve penetration rate is calculated based on the test data and simulation data. v a And the valve-penetrating interpolation point data sequence N, after merging the interpolation point data sequence with the original data sequence; select the disturbance tolerance r and the tolerance reference value b according to the actual engineering application, and convert them into dimensionless deviation acceptability p; construct the grey correlation degree-mean square error distance model and mean square error distance-confidence model according to the simulation-experimental time series data, and calculate the grey correlation coefficient (i.e., weight function) of the simulation-experimental time series data (sequence data merged with the interpolation point data sequence) based on the deviation acceptability ), and then calculate the grey relational degree, combine the grey relational degree-mean square error distance model to map the grey relational degree to mean square error distance, and combine the mean square error distance-confidence model to map the mean square error distance to the confidence evaluation result.

[0092] In view of the poor robustness of the grey correlation analysis method when processing the confidence assessment of noisy data, this embodiment introduces an interpolation method guided by the expected penetration rate. On the basis of not changing the curve morphology of the original data sequence, the weight of the data distance for the grey correlation analysis calculation is indirectly increased by increasing the number of data points within the small distance difference range. A deviation acceptability index p is proposed as the data distance threshold. The grey correlation analysis interpolation is corrected by adding the weight after judging the disturbance tolerance on the basis of taking the absolute value of the difference. The grey correlation degree is mapped to the confidence degree by constructing a grey correlation degree-mean square error distance-confidence model. The grey correlation coefficient of a single data point is corrected at the micro level, and the weight of the data point distance is increased at the macro level, which effectively improves the noise resistance and anti-disturbance performance of the evaluation result, and helps to accurately and clearly reflect the confidence of the simulation model.

[0093] Corresponding to the above-mentioned confidence assessment method based on strong robust grey relational analysis, the embodiment of the present invention further provides a confidence assessment device based on strong robust grey relational analysis. Fig.12 A structural schematic diagram of a confidence evaluation device based on strong robust grey relational analysis is shown, the device comprising: The acquisition module 1201 is used to acquire original test data corresponding to the target research object and original simulation data corresponding to the simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing the physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating the corresponding simulation model of the target research object; A calculation module 1202 is used to calculate an expected valve penetration rate corresponding to the simulation model according to the original test data and the original simulation data; An interpolation module 1203 is used to interpolate and update the original test data and the original simulation data according to the expected valve penetration rate corresponding to the simulation model to obtain target test data and target simulation data; The determination module 1204 is used to determine the target confidence of the simulation model through grey correlation analysis according to the target test data, the target simulation data and the preset deviation acceptability.

[0094] The confidence evaluation device based on strong robust grey correlation analysis provided by the present invention can obtain original test data corresponding to a target research object and original simulation data corresponding to a simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing a physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating a corresponding simulation model of the target research object; based on the original test data and the original simulation data, an expected valve penetration rate corresponding to the simulation model is calculated; based on the expected valve penetration rate corresponding to the simulation model, the original test data and the original simulation data are interpolated and updated to obtain target test data and target simulation data; based on the target test data, the target simulation data and the preset deviation acceptability, the target confidence of the simulation model is determined through grey correlation analysis. In this way, by introducing the expected penetration rate to assist the interpolation calculation, the influence weight of distance on the grey correlation degree is indirectly increased, which effectively makes up for the distance similarity ignored by the traditional grey correlation analysis method when judging the similarity of sequence curves. At the same time, by introducing the deviation acceptability as the distance threshold indicator, the grey correlation coefficient of each data point within a certain deviation range is increased. Combined with the grey correlation analysis, a more stable confidence assessment is achieved, thereby effectively improving the anti-noise and anti-disturbance performance of the evaluation results.

[0095] Furthermore, the calculation module 1202 is specifically used for: performing curve fitting on the original test data to obtain a threshold curve; performing curve fitting on the original simulation data to obtain a disturbance sequence curve; and calculating the expected valve penetration rate corresponding to the simulation model based on the threshold curve and the disturbance sequence curve.

[0096] Furthermore, the calculation module 1202 is also used to calculate the expected valve penetration rate by the following formula: v a : ; in, y represents the disturbance sequence curve, express y The derivative of a represents the threshold curve, t Indicates time, Represents three variables a , , t The probability density of .

[0097] Furthermore, the interpolation module 1203 is specifically used to: when the goal is to evaluate the absolute confidence of the simulation model, determine the valve penetration point according to the difference data corresponding to the original simulation data; wherein the difference data is obtained by subtracting the original simulation data from the original test data, and the valve penetration point is the midpoint of adjacent data points with different positive and negative signs of data values ​​in the difference data; When there are no less than 2 simulation models and the goal is to determine the relative confidence between the simulation models, determine whether the expected valve penetration rates corresponding to the simulation models are equal; if not, determine the valve penetration point based on the difference data corresponding to each group of original simulation data; According to the expected valve penetration rate corresponding to the simulation model, the neighborhood spatial position range of the valve penetration point is interpolated for the original test data and the original simulation data respectively to obtain the corresponding interpolation data sequence; wherein the neighborhood spatial position range of the valve penetration point is the range between two adjacent data points corresponding to the valve penetration point; The original test data and the original simulation data are respectively merged with the corresponding interpolation data sequences to obtain the target test data and the target simulation data.

[0098] Furthermore, the interpolation module 1203 is further configured to: if yes, determine the original test data and each group of original simulation data as target test data and each group of target simulation data respectively.

[0099] Furthermore, the above-mentioned determination module 1204 is specifically used to: calculate the grey correlation degree corresponding to the simulation model based on the target test data, the target simulation data and the deviation acceptability; determine the mean square error distance corresponding to the simulation model according to the grey correlation degree corresponding to the simulation model and the preset grey correlation degree-mean square error distance model; determine the target confidence of the simulation model according to the mean square error distance corresponding to the simulation model and the preset mean square error distance-confidence model.

[0100] Furthermore, the determination module 1204 is further used to: calculate disturbance tolerance data according to the deviation acceptability and the target test data; The grey relational degree is calculated by the following formula: ; in, Represents target test data X With target simulation data Y The grey correlation between , , Indicates the target test data i Of the data points k data values, Indicates the target simulation data i Of the data points k data values, n represents the number of data points, τ Indicates the preset resolution factor, r i Indicates the disturbance tolerance data r i The disturbance tolerance, , Represents the weight function, satisfying: .

[0101] The implementation principle and technical effects of the device provided in this embodiment are the same as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0102] like Fig.13 As shown, an electronic device 1300 provided by an embodiment of the present invention includes: a processor 1301, a memory 1302 and a bus, the memory 1302 stores a computer program that can be run on the processor 1301, when the electronic device 1300 is running, the processor 1301 and the memory 1302 communicate through the bus, and the processor 1301 executes the computer program to implement the above-mentioned confidence assessment method based on strong robust grey correlation analysis.

[0103] Specifically, the memory 1302 and the processor 1301 can be general-purpose memories and processors, which are not specifically limited here.

[0104] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the confidence assessment method based on strong robust grey relational analysis in the previous method embodiment is executed. The computer-readable storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a RAM, a magnetic disk or an optical disk, and other media that can store program codes.

[0105] The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0106] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limiting, and thus other examples of the exemplary embodiments may have different values.

[0107] The flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and a part of a module, a program segment or a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A confidence assessment method based on strong robust grey relational analysis, characterized in that: include: Obtaining original test data corresponding to the target research object and original simulation data corresponding to the simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing the physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating the corresponding simulation model of the target research object; Calculating an expected valve penetration rate corresponding to the simulation model according to the original test data and the original simulation data; According to the expected valve penetration rate corresponding to the simulation model, the original test data and the original simulation data are interpolated and updated to obtain target test data and target simulation data; The target confidence of the simulation model is determined by grey correlation analysis according to the target test data, the target simulation data and the preset deviation acceptability.

2. The method according to claim 1, characterized in that The step of calculating the expected valve penetration rate corresponding to the simulation model according to the original test data and the original simulation data includes: Performing curve fitting on the original test data to obtain a threshold curve; Performing curve fitting on the original simulation data to obtain a disturbance sequence curve; According to the threshold curve and the disturbance sequence curve, an expected valve penetration rate corresponding to the simulation model is calculated.

3. The method according to claim 2, characterized in that The step of calculating the expected valve penetration rate corresponding to the simulation model according to the threshold curve and the disturbance sequence curve includes: The expected valve penetration rate is calculated by the following formula v a : ； in, y represents the disturbance sequence curve, express y The derivative of a represents the threshold curve, t Indicates time, Represents three variables a , , t The probability density of .

4. The method according to claim 1, characterized in that: The interpolation and updating of the original test data and the original simulation data according to the expected valve penetration rate corresponding to the simulation model to obtain target test data and target simulation data includes: When the goal is to evaluate the absolute confidence of the simulation model, the valve penetration point is determined according to the difference data corresponding to the original simulation data; wherein the difference data is obtained by subtracting the original simulation data from the original test data, and the valve penetration point is the midpoint of adjacent data points with different positive and negative signs of data values ​​in the difference data; When there are no less than two simulation models and the goal is to determine the relative confidence between the simulation models, determine whether the expected valve penetration rates corresponding to the simulation models are all equal; if not, determine the valve penetration point according to the difference data corresponding to each group of original simulation data; According to the expected valve penetration rate corresponding to the simulation model, the neighborhood spatial position range of the valve penetration point is interpolated for the original test data and the original simulation data respectively to obtain a corresponding interpolation data sequence; wherein the neighborhood spatial position range of the valve penetration point is the range between two adjacent data points corresponding to the valve penetration point; The original test data and the original simulation data are respectively merged with corresponding interpolation data sequences to obtain the target test data and the target simulation data.

5. The method according to claim 4, characterized in that The method further comprises: If yes, the original test data and each group of the original simulation data are respectively determined as the target test data and each group of the target simulation data.

6. The method according to claim 1, characterized in that Determining the target confidence of the simulation model by grey correlation analysis according to the target test data, the target simulation data and the preset deviation acceptability includes: Based on the target test data, the target simulation data and the deviation acceptability, a grey relational degree corresponding to the simulation model is calculated; Determine the mean square error distance corresponding to the simulation model according to the grey relational degree corresponding to the simulation model and a preset grey relational degree-mean square error distance model; The target confidence of the simulation model is determined according to the mean square error distance corresponding to the simulation model and a preset mean square error distance-confidence model.

7. The method according to claim 6, characterized in that The step of calculating the grey relational degree corresponding to the simulation model based on the target test data, the target simulation data and the deviation acceptability includes: Calculating disturbance tolerance data according to the deviation acceptability and the target test data; The grey relational degree is calculated by the following formula: ； in, Represents target test data X With target simulation data Y The grey correlation between , , Indicates the target test data i Of the data points k data values, Indicates the target simulation data i Of the data points k data values, n represents the number of data points, τ Indicates the preset resolution factor, r i Indicates the disturbance tolerance data r i The disturbance tolerance, , Represents the weight function, satisfying: 。 8. A confidence evaluation device based on strong robust grey relational analysis, characterized in that: include: An acquisition module, used to acquire original test data corresponding to the target research object and original simulation data corresponding to the simulation model to be evaluated; wherein the original test data is a set of time series data obtained by testing the physical model of the target research object, and the original simulation data is a set of time series data obtained by simulating the corresponding simulation model of the target research object; A calculation module, used for calculating an expected valve penetration rate corresponding to the simulation model according to the original test data and the original simulation data; An interpolation module, used to interpolate and update the original test data and the original simulation data according to the expected valve penetration rate corresponding to the simulation model, so as to obtain target test data and target simulation data; A determination module is used to determine the target confidence of the simulation model through grey correlation analysis based on the target test data, the target simulation data and a preset deviation acceptability.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When the processor executes the computer program, the confidence assessment method based on strong robust grey relational analysis described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the confidence assessment method based on strong robust grey relational analysis according to any one of claims 1 to 7 is executed.

Images