Trustworthy assessment methods, systems, equipment and media for the evolution of digital twins

Abstract

The present invention belongs to the field of digital twin technology, and discloses a trusted evaluation method, system, device and medium for the evolution process of digital twin, including: quantitatively analyzing the characteristics of each model parameter in the basic model of digital twin, fitting the performance deviation estimation model based on the analysis results; constructing an evaluation time window, monitoring the evolution process of digital twin in real time, obtaining real-time monitoring data, calculating the credibility value of the digital twin for the current task application in the current evaluation time window based on the real-time monitoring data combined with the performance deviation estimation model; performing weighted summation on the credibility values ​​of various task applications in the current evaluation time window according to the frequency of the corresponding task application, and obtaining the overall credibility value of the digital twin in the current evaluation time window; when the evaluation time window is updated, dynamically updating the corresponding overall credibility value. The technical solution of the present invention can accurately reflect the credibility value of the digital twin in this demand scenario.

Description

Technical Field

[0001] The present invention belongs to the field of digital twin technology, and in particular relates to a trusted assessment method, system, device and medium for a digital twin evolution process. Background Art

[0002] Digital Twin (DT) has been a research hotspot in recent years. It refers to a simulation model that is highly consistent with the physical object. Unlike traditional virtual prototypes, digital twins assimilate the collected real-time data, so that they can maintain consistency with the physical object in real time after the model is established. This allows the digital twin system to simulate and test the twin object that is more in line with the real state in a virtual environment, obtain simulation results with more realistic guiding significance, and then feedback more effective optimization signals to the physical system, promptly solve the current problems of the physical system, avoid hidden risks, and improve overall efficiency.

[0003] Digital twins belong to the field of modeling and simulation (M&S). In the field of M&S, if the model has not been evaluated for credibility, it cannot play its value in practice. Digital twin systems also require a set of credibility evaluation methods. Only when the credibility of the system is higher than the usage threshold can the digital twin truly play its application value.

[0004] Traditional M&S focuses more on offline simulation, and its trusted assessment method is mainly based on VV&A (verification, validation, and confirmation). Digital twins are mainly aimed at online simulation, which expands the real-time, interactive, and dynamic features of traditional offline simulation. Digital twins seem to be an emerging concept that lacks trusted assessment methods. In fact, there are many trusted concepts and assessment methods that can be referenced from the M&S knowledge system that has been developed for many years and has become more mature.

[0005] Existing technical solutions include:

[0006] 1. Only evaluate the results of a single evolution. A single evolution result refers to a new version of the model obtained by correcting a model that has failed to a certain extent using real data collected over a long period of time. This evaluation technology does not consider the time spent on the correction process or the changes that the model has undergone. It only uses traditional model evaluation methods to re-evaluate the latest version of the model. This evaluation is essentially conducted on offline models, which breaks the close connection between digital twin models and real-time applications. The evaluation results obtained only reflect the performance of the latest version of the model, but fail to reflect the impact of continuously changing twin models on online applications.

[0007] 2. Use mean square error, root mean square error, etc. to count the model error at each time point in the evolution process. When the error is less than the set error threshold and the time spent is less than the set time threshold, the twin model is considered to be credible. This evaluation technology solution takes into account every version of the model in the evolution of the twin model, but does not take into account that the actual application is unevenly distributed on the time axis, that is, sometimes the model error is large, but no application will occur at this time. In this demand scenario, the credibility score of the model should not be lowered because of the existence of this error. In addition, the error threshold or time threshold set in this type of technical solution is only subjectively determined by humans, rather than deduced from the actual application requirements. Therefore, this evaluation technology solution is not of much reference value for digital twins, but it is already a better solution in current public research.

[0008] An analysis of existing technical solutions in this field shows the following shortcomings:

[0009] 1. Not considering the time spent on the model calibration process or the changes that the model has undergone, resulting in a large deviation between the credibility value obtained from the evaluation and the actual value.

[0010] 2. Use mean square error, root mean square error, etc. to count the model errors at each time point in the evolution process. This approach gives the same weight to each error, which is inconsistent with actual application requirements, resulting in a large deviation between the credibility value obtained from the evaluation and the actual value.

[0011] 3. The error threshold or time threshold set is only determined by humans subjectively, rather than being deduced from actual application requirements, resulting in the conclusion on whether the model is credible being inconsistent with the actual situation.

[0012] 4. The method of averaging the results of a large number of repeated experiments is costly and time-consuming, and cannot meet the needs of digital twins to quickly obtain credibility and make timely adjustments. Summary of the invention

[0013] The purpose of the present invention is to provide a trusted assessment method, system, device and medium for the digital twin evolution process to solve the problems existing in the above-mentioned prior art.

[0014] To achieve the above object, the present invention provides a trustworthy evaluation method for the digital twin evolution process, comprising: quantitatively analyzing the correlation between each model parameter and the demand index in the model to be analyzed to obtain correlation data; wherein the model to be analyzed is a digital twin basic model;

[0015] Constructing a key parameter set based on the correlation data, and determining a deviation interval of each key parameter in the key parameter set;

[0016] Set sampling rounds and sampling points, and record the performance deviation of the digital twin basic model corresponding to each key parameter with a deviation interval and the key state data of the dual system in the corresponding process point by point in rounds; wherein, the performance deviation is the deviation between the result obtained by executing the control strategy obtained by simulation solution based on the digital twin basic model on the real physical system and the expected performance index; the key state data of the dual system is the real-time state data generated by the observable parameters in the remaining parameters with a high correlation with each key parameter in the key parameter set;

[0017] Taking the key parameters with deviation intervals and the corresponding dual-system key status data as input and the performance deviation as output, a performance deviation estimation model is fitted;

[0018] Constructing an evaluation time window, wherein the evaluation time window includes several types of task applications;

[0019] Monitor the evolution process of the digital twin in real time, obtain real-time monitoring data, and store the real-time monitoring data; wherein the real-time monitoring data includes real-time key status data, evolution real-time result data, current task application data and corresponding time series data;

[0020] The real-time monitoring data is input into the performance deviation estimation model to calculate the credibility value of the digital twin for the current task application in the current evaluation time window; the credibility values ​​of various task applications in the current evaluation time window are weighted and summed according to the frequency of the corresponding task application to obtain the overall credibility value of the digital twin in the current evaluation time window;

[0021] When the evaluation time window is updated, the corresponding overall credibility value is dynamically updated.

[0022] Optionally, a quantitative analysis is performed on the correlation between each model parameter and the demand indicator in the model to be analyzed, specifically including: based on different physical system states of the physical equipment entity, traversing and fine-tuning the parameter values ​​in the digital twin basic model, recording the change data of the demand indicator corresponding to the fine-tuning, and quantitatively analyzing the correlation data between each model parameter and the demand indicator in the digital twin basic model based on the change data; wherein the correlation data includes independent parameter correlation and coupling parameter correlation.

[0023] Optionally, constructing a key parameter set based on the correlation data, and determining a deviation interval of each key parameter in the key parameter set specifically includes:

[0024] Based on the correlation, the model change parts and corresponding change ranges that affect the demand indicators are screened out; a key parameter set corresponding to the model change parts is constructed, and the deviation interval of each key parameter in the key parameter set is determined based on the corresponding change range.

[0025] Optionally, the real-time monitoring data is input into the performance deviation estimation model to calculate the credibility value of the digital twin for the current task application in the current evaluation time window, specifically including:

[0026] Based on the preset evaluation sampling moment, the real-time key status data and the evolving real-time result data at the same moment are retrieved, and the retrieved data is input into the performance deviation estimation model for evaluation to obtain the performance deviation data at the current evaluation sampling moment; the application occurrence probability corresponding to the task application data at the current evaluation sampling moment is calculated; based on the performance deviation data and the application occurrence probability at the current evaluation sampling moment, the credibility value at the current evaluation sampling moment is calculated; the credibility data at each evaluation sampling moment is summed to obtain the credibility value of the digital twin for the current task application in the current evaluation time window.

[0027] Optionally, when the evaluation time window is updated, the corresponding overall credibility value is dynamically updated, specifically including:

[0028] When the last moment of the evaluation time window is aligned with the latest real moment according to the set time interval, an updated evaluation time window is obtained, and the corresponding overall credibility value is calculated in real time based on the updated evaluation time window.

[0029] A credible evaluation system for the digital twin evolution process, comprising:

[0030] A performance deviation estimation model construction module is used to quantitatively analyze the correlation between each model parameter and the demand index in the model to be analyzed to obtain correlation data; wherein, the model to be analyzed is a digital twin basic model; a key parameter set is constructed based on the correlation data, and the deviation interval of each key parameter in the key parameter set is determined; sampling rounds and sampling points are set, and the performance deviation of the digital twin basic model corresponding to each key parameter with a deviation interval and the key state data of the dual system in the corresponding process are recorded point by point in each round; wherein, the performance deviation is the deviation between the control strategy obtained by executing the simulation solution based on the digital twin basic model on the real physical system and the expected performance index; the key state data of the dual system is the real-time state data generated by the observable parameters in the remaining parameters in the key parameter set that have a high correlation with each key parameter; with the key parameters with deviation intervals and the corresponding key state data of the dual system as input, and the performance deviation as output, a performance deviation estimation model is fitted;

[0031] A credibility dynamic update module is used to construct an evaluation time window, which includes several types of task applications; monitor the evolution process of the digital twin in real time to obtain real-time monitoring data, and store the real-time monitoring data; wherein the real-time monitoring data includes real-time key status data, evolution real-time result data, current task application data and corresponding time series data; calculate the credibility value of the digital twin for the current task application in the current evaluation time window based on the real-time monitoring data; perform weighted summation on the credibility values ​​of various types of task applications in the current evaluation time window according to the frequency of the corresponding task application, and obtain the overall credibility value of the digital twin in the current evaluation time window; when the evaluation time window is updated, dynamically update the corresponding overall credibility value.

[0032] An electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a trusted assessment method for a digital twin evolution process.

[0033] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a trusted assessment method for a digital twin evolution process.

[0034] The technical effects of the present invention are:

[0035] The present invention effectively quantifies the association between each updated twin model and the final application effect deviation, which can reflect the actual application effect caused by each version of the model, rather than using artificially roughly set thresholds to measure model performance.

[0036] The present invention considers each update change experienced by the model correction process at each time point, and assigns weights according to the probability of the application time window. The credibility value obtained by evaluation is the mathematical expectation of the actual application effect, and accurately reflects the credibility value of the digital twin in the demand scenario.

[0037] The present invention can achieve the effect of fast real-time updating of credibility values ​​with low time cost by realizing the construction of a performance deviation estimation model and applying a time window real-time updating method.

[0038] The entire method of the present invention adopts a strategy based on a high-credibility basic model and supplemented by physical experiments, and adds an adaptive sampling algorithm, which can significantly reduce economic costs while ensuring evaluation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0041] Figure 1 It is a schematic diagram of the structure of each module in the embodiment of the present invention;

[0042] Figure 2 This is a general flow chart of the dynamic trustworthy assessment method for the non-timely evolution process of the equipment digital twin in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but should be understood as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0044] It should be understood that the terms described in the present invention are only for describing special embodiments and are not intended to limit the present invention. In addition, for the numerical range in the present invention, it should be understood that each intermediate value between the upper and lower limits of the scope is also specifically disclosed. Each smaller range between the intermediate value in any stated value or stated range and any other stated value or intermediate value in the described range is also included in the present invention. The upper and lower limits of these smaller ranges can be independently included or excluded in the scope.

[0045] Unless otherwise indicated, all technical and scientific terms used herein have the same meanings as those of ordinary skill in the art generally understood by the present invention. Although the present invention describes only preferred methods, any method similar or equivalent to that described herein may also be used in the implementation or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods related to the documents. In the event of a conflict with any incorporated document, the content of this specification shall prevail.

[0046] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present invention description without departing from the scope or spirit of the present invention. Other embodiments derived from the present invention description will be apparent to those skilled in the art. The present application description and examples are exemplary only.

[0047] The words “include,” “including,” “have,” “contain,” etc. used in this article are open-ended terms, meaning including but not limited to.

[0048] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0049] Embodiment 1

[0050] like Figure 1 - Figure 2 As shown, in this embodiment, a trusted evaluation method for a digital twin evolution process is provided, including: quantitatively analyzing the correlation between each model parameter and the demand index in the model to be analyzed to obtain correlation data; wherein, the model to be analyzed is a digital twin basic model; constructing a key parameter set based on the correlation data, and determining the deviation interval of each key parameter in the key parameter set; setting sampling rounds and sampling points, and recording the performance deviation of the digital twin basic model corresponding to each key parameter with a deviation interval and the key state data of the dual system in the corresponding process point by point in rounds; wherein, the performance deviation is the deviation between the control strategy obtained by executing a simulation solution based on the digital twin basic model on a real physical system and the expected performance index; the key state data of the dual system is the observable parameters in the remaining parameters in the key parameter set that have a high correlation with each key parameter. The real-time status data generated by the number; taking the key parameters with deviation intervals and the corresponding dual-system key status data as input and the performance deviation as output, a performance deviation estimation model is fitted; an evaluation time window is constructed, and the evaluation time window includes several types of task applications; the evolution process of the digital twin is monitored in real time to obtain real-time monitoring data, and the real-time monitoring data is stored; wherein the real-time monitoring data includes real-time key status data, evolution real-time result data, current task application data and corresponding time series data; based on the real-time monitoring data, the credibility value of the digital twin for the current task application in the current evaluation time window is calculated; the credibility values ​​of various task applications in the current evaluation time window are weighted and summed according to the frequency of the corresponding task applications, to obtain the overall credibility value of the digital twin in the current evaluation time window; when the evaluation time window is updated, the corresponding overall credibility value is dynamically updated.

[0051] This embodiment effectively quantifies and associates each updated twin model with the final application effect deviation, which can reflect the real application effect caused by each version of the model, rather than using a roughly set threshold to measure the model performance. This embodiment considers each update change experienced by the model correction process at each time point, and assigns weights according to the probability of the application time window. The credibility value obtained by the evaluation is the mathematical expectation of the actual application effect, accurately reflecting the credibility value of the digital twin in the demand scenario. This embodiment can achieve the effect of rapid real-time update of the credibility value by realizing the construction of a performance deviation estimation model and a method for real-time update of the application time window, with low time cost. The entire method of this embodiment adopts a strategy based on a high-credibility basic model and supplemented by physical experiments, and adds an adaptive sampling algorithm, which can greatly reduce economic costs while ensuring the accuracy of the evaluation.

[0052] This embodiment is suitable for digital twins applied to discrete processes, such as condition monitoring, discrete control, process planning, phased life prediction, maintenance optimization, etc. It can evaluate general evolution processes including non-timely evolution and timely evolution. The field with a high degree of matching is discrete manufacturing. This embodiment innovatively proposes dual-system sensitivity analysis, digital twin evolution overview analysis, adaptive performance deviation test, performance deviation estimation model construction, application time window fitting, and digital twin credibility value dynamic update algorithm. It can calculate and update the credibility value of the non-timely evolution process of the equipment digital twin in real time, with low cost and high accuracy of the credibility value obtained.

[0053] The dynamic trust evaluation method for the non-timely evolution process of equipment digital twins is mainly composed of a monitoring module, a performance deviation estimation model, an application time window fitting module, and a real-time credibility calculation module. The monitoring module includes three parts: monitoring key state data, monitoring evolution real-time results, and monitoring actual application. Among them, the monitoring key state data and monitoring evolution real-time results parts input the key information of the real-time physical system and the twin model system into the performance deviation estimation model, and then the deviation estimation model gives the performance deviation estimation at each evaluation sampling moment. The monitoring actual application part provides necessary information for the fitting of the application time window, and the application time window fitting module updates the current optimal estimate of the application time window in real time. Finally, the real-time credibility calculation module combines the latest application time window fitting, weights the performance deviation of the latest evolution at each evaluation sampling moment, and updates the credibility of the twin model in the current evolution process in real time.

[0054] The dynamic trustworthy assessment method for the non-real-time evolution process of equipment digital twins includes the following 7 main steps.

[0055] Step 1: Construct a performance deviation estimation model. The function of this model is to input key state data and real-time evolution results to obtain performance deviation estimates. Among them, "performance deviation" refers to the deviation from the expected performance index caused by executing the control strategy obtained by simulation solution based on the twin model on the real physical system. Performance deviation has a direct and important relationship with the credibility of digital twins, but it takes too long to know the deviation through simulation or physical experiments, which cannot meet the needs of digital twins to quickly know the credibility. Therefore, a large number of tests are carried out in advance to obtain sufficient information, and it is unified into a high-precision performance deviation estimation model, so that a high-precision deviation estimate can be obtained at a speed close to the "table lookup method". "Key state data" and "real-time evolution results" will be explained in the fourth step of this step. The performance deviation estimation model of equipment digital twins can be constructed through the following 6 small steps.

[0056] 1) Confirm the high-credibility basic model and application requirements. Digital twins are formed by forming a two-way closed loop between the basic model and the physical system. They are the product of the basic model being able to evolve and update in real time according to the state data of the physical system. The basic model must first pass a comprehensive assessment similar to VV&A (Verification, Validation and Accreditation) in the field of modeling and simulation, and only after it is confirmed to have sufficient credibility can it be connected to the physical system. In this preliminary process, the application requirements of the basic model will be analyzed in detail as formalized quantitative indicators. Therefore, a high-credibility basic model is a prerequisite for the existence of digital twins and a powerful aid to the trustworthy assessment of digital twins. In the first small step, it is necessary to confirm the high-credibility basic model of the digital twin of the equipment and the complete application requirements decomposed into quantitative indicators.

[0057] 2) Dual system sensitivity analysis. Simulate on the high-confidence basic model of the equipment digital twin and quantitatively analyze the correlation between each model parameter and the demand indicators. The specific approach is to replace the real physical system with a highly reliable simulation system, and adjust the simulation system to different states that the real physical system may be in (deviation states different from those when it just leaves the factory). For each possible physical system state, traverse and fine-tune the parameter values ​​of the digital twin system, record the changes in the demand indicators brought about by each fine-tuning, and comprehensively and quantitatively analyze the correlation between each parameter of the digital twin and the demand indicators, including the correlation between independent parameters and the correlation between coupled parameters. Based on dual system sensitivity analysis, the test cost is reduced and the work efficiency of the estimation model is improved.

[0058] 3) Digital twin evolution overview analysis. Based on expert experience and user needs, sort out and list the parts of the digital twin where uncertainty changes may occur, the possible range of changes, and the possible propagation paths of each part during evolution. Combined with the results of the dual-system sensitivity analysis in the second step, screen the parts that have a greater impact on the demand indicators, that is, the model parts involving parameters that have a greater impact on the demand indicators. Based on the digital twin evolution overview analysis, the test cost is reduced and the accuracy of the estimation model is improved.

[0059] 4) Adaptive performance deviation test. It is mainly carried out through high-confidence basic model simulation, and the test results are supplemented or corrected with physical experiments. For the model parts and evolution-affected parts selected in the third step, their key parameter sets are sorted out. Combined with the possible range of changes of each part given in the third step, the possible deviation range of each key parameter in the key parameter set is confirmed. Then, Latin hypercube sampling is used to set the parameters of the high-confidence simulation system that replaces the real physical system according to the sampling points in rounds. After setting the parameters in each round, the sampling points must be traversed, and the parameters of the digital twin system must be modified point by point to test the control strategy given by the digital twin based on different error parameters. How much performance deviation will be generated on the simulation system in this round. Record the performance deviation and the key state data of the dual system in this process point by point in each round. Among them, "key state data" refers to the real-time state data generated by the observable parameters in other parameters that have a high correlation with each key parameter in the key parameter set (refer to the results of the sensitivity analysis of the dual system). "Evolution real-time results" are the updated values ​​of each key parameter in the key parameter set at each latest real moment. Finally, an adaptive test is performed, that is, the intervals with excessive changes in the performance deviation results obtained by the first Latin hypercube sampling are further refined until the gap between all adjacent performance deviation results is less than the set threshold δ. The adaptive performance deviation test reduces the test cost as much as possible while ensuring the accuracy of the estimated model; each updated twin model is effectively quantitatively associated with the final application effect deviation, which can reflect the actual application effect caused by each version of the model.

[0060] 5) Construct a performance deviation estimation model. Linear regression algorithms, nonlinear regression algorithms, machine learning algorithms, etc. can be used to construct an input-output fitting regression model for all deviation test results, with the key parameters with deviations and the corresponding dual-system key state data as input and the performance deviation as output.

[0061] 6) Test the accuracy of the deviation estimation model. Re-sample the possible deviation intervals of all key parameters with Latin hypercube. After setting the dual system parameters according to the sampling points, record the key state data. Use this key state data and the key parameter value of the sampling point as input to obtain the performance deviation estimate from the deviation estimation model. Compare all estimated performance deviations with the actual performance deviation obtained from each simulation, and calculate the accuracy / confidence interval of the estimation model.

[0062] This embodiment realizes the rapid association of twin model deviation to application effect deviation by constructing a performance deviation estimation model and testing the accuracy of the deviation estimation model, providing a strong guarantee for the real-time update of the credibility value.

[0063] Steps 2, 3, and 4 are independent and ongoing processes from the start of the assessment and include:

[0064] Step 2: Monitor key status data. "Key status data" refers to the real-time status data generated by observable parameters in other parameters that are highly correlated with each key parameter in the key parameter set (refer to the results of dual-system sensitivity analysis). Use sensors, software reading, etc. to collect this data in real time, and organize and store it in the database according to certain specifications for subsequent use.

[0065] Step 3: Monitor the real-time results of evolution. The "real-time results of evolution" are the updated values ​​of each key parameter in the key parameter set at each latest real moment. According to the evolutionary algorithm, some key parameters will change due to the integration of new data, thereby changing the parameter set of the twin model. These changed key parameters are read and stored together with other unchanged parameters in the database, and are called as input when the performance deviation is estimated.

[0066] Step 4: Monitor actual applications. The monitoring of actual applications includes: the time point of each application, the application category, the situation of the application calling the model, and the demand for the application in this category. The application categories are divided according to whether the demand indicator set is the same, and the relevant monitoring information is sorted and stored in the database for subsequent calls.

[0067] Step 5: For each evaluation sampling moment, call the key state data and real-time evolution results at the same timestamp in the database, input the performance deviation estimation model, and obtain the performance deviation that will be generated if the twin model is used for application at each evaluation sampling moment. The difference between these deviation estimates and actual deviations is evaluated by the sixth step of step 1. These performance deviation estimates at each evaluation sampling moment are the main components of the digital twin credibility value in this evaluation time window.

[0068] Step 6: Update the application time window fitting in real time. According to step 4, there are one or more types of applications in the evaluation time window, and the length of the time window is T. If there are multiple types of applications, the frequency of occurrence of each type is the weight of each application when it is comprehensively calculated. For each type of application, the GMM algorithm is used to fit the joint Gaussian distribution function of the application occurrence time point. This function can reflect the probability distribution of the application occurrence interval duration t, so that the joint probability can be further used to calculate the time window probability distribution of continuous application of discrete processes to be served by the equipment. This time window probability distribution reflects that if a total of N applications occur in the evaluation time window, what are the probabilities of the application occurring at each time point; N = T / mean(t) + 1. Mean(t) is the average value of t.

[0069] Through steps 4 and 6, the application time window fitting is updated in real time, and the way in which the digital twin is applied in a given demand scenario can be quickly and accurately estimated in a feasible way, thereby effectively giving the mathematical expectation of the digital twin application effect, that is, the credibility value.

[0070] Step 7: Dynamically update the credibility value of the digital twin. Multiply the performance deviation estimate at each evaluation sampling moment obtained in step 5 by the probability of the application occurring at this sampling moment (given by the probability distribution of the time window of continuous application occurrence in step 6), and sum them all up to get the credibility value of the digital twin for this type of application in this evaluation time window. The credibility value of each type of application is weighted and summed according to the frequency of occurrence of this type of application, and the result is the overall credibility value of the digital twin in this evaluation time window. If the last moment of the evaluation time window is aligned with the latest real moment at a small time interval, the credibility value is calculated by fitting the latest application time window after each alignment, that is, the credibility value of the digital twin is updated in real time. The calculation method in this step can effectively calculate a more accurate credibility value of the digital twin through observable data, which is basically consistent with the mathematical expectation of the actual application effect, and supports real-time updates.

[0071] A credible evaluation system for the digital twin evolution process, comprising:

[0072] A performance deviation estimation model construction module is used to quantitatively analyze the correlation between each model parameter and the demand index in the model to be analyzed to obtain correlation data; wherein, the model to be analyzed is a digital twin basic model; a key parameter set is constructed based on the correlation data, and the deviation interval of each key parameter in the key parameter set is determined; sampling rounds and sampling points are set, and the performance deviation of the digital twin basic model corresponding to each key parameter with a deviation interval and the key state data of the dual system in the corresponding process are recorded point by point in each round; wherein, the performance deviation is the deviation between the control strategy obtained by executing the simulation solution based on the digital twin basic model on the real physical system and the expected performance index; the key state data of the dual system is the real-time state data generated by the observable parameters in the remaining parameters in the key parameter set that have a high correlation with each key parameter; with the key parameters with deviation intervals and the corresponding key state data of the dual system as input, and the performance deviation as output, a performance deviation estimation model is fitted;

[0073] A credibility dynamic update module is used to construct an evaluation time window, which includes several types of task applications; monitor the evolution process of the digital twin in real time to obtain real-time monitoring data, and store the real-time monitoring data; wherein the real-time monitoring data includes real-time key status data, evolution real-time result data, current task application data and corresponding time series data; calculate the credibility value of the digital twin for the current task application in the current evaluation time window based on the real-time monitoring data; perform weighted summation on the credibility values ​​of various types of task applications in the current evaluation time window according to the frequency of the corresponding task application, and obtain the overall credibility value of the digital twin in the current evaluation time window; when the evaluation time window is updated, dynamically update the corresponding overall credibility value.

[0074] An electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a trusted assessment method for a digital twin evolution process.

[0075] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a trusted assessment method for a digital twin evolution process.

[0076] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A trustworthy evaluation method for the digital twin evolution process, characterized in that: include: Quantitatively analyze the correlation between each model parameter and the demand index in the model to be analyzed to obtain correlation data; wherein the model to be analyzed is a digital twin basic model; Constructing a key parameter set based on the correlation data, and determining a deviation interval of each key parameter in the key parameter set; Set sampling rounds and sampling points, and record the performance deviation of the digital twin basic model corresponding to each key parameter with a deviation interval and the key state data of the dual system in the corresponding process point by point in rounds; wherein, the performance deviation is the deviation between the result obtained by executing the control strategy obtained by simulation solution based on the digital twin basic model on the real physical system and the expected performance index; the key state data of the dual system is the real-time state data generated by the observable parameters in the remaining parameters with a high correlation with each key parameter in the key parameter set; Taking the key parameters with deviation intervals and the corresponding dual-system key state data as input and the performance deviation as output, a performance deviation estimation model is fitted; Constructing an evaluation time window, wherein the evaluation time window includes several types of task applications; Monitor the evolution process of the digital twin in real time, obtain real-time monitoring data, and store the real-time monitoring data; wherein the real-time monitoring data includes real-time key status data, evolution real-time result data, current task application data and corresponding time series data; The real-time monitoring data is input into the performance deviation estimation model to calculate the credibility value of the digital twin for the current task application in the current evaluation time window; the credibility values ​​of various task applications in the current evaluation time window are weighted and summed according to the frequency of the corresponding task application to obtain the overall credibility value of the digital twin in the current evaluation time window; When the evaluation time window is updated, the corresponding overall credibility value is dynamically updated.

2. According to claim 1, a trustworthy evaluation method for the digital twin evolution process is characterized in that: A quantitative analysis is performed on the correlation between each model parameter and the demand index in the model to be analyzed, specifically including: based on different physical system states of the physical equipment entity, traversing and fine-tuning the parameter values ​​in the digital twin basic model, recording the change data of the demand index corresponding to the fine-tuning, and quantitatively analyzing the correlation data between each model parameter and the demand index in the digital twin basic model based on the change data; wherein the correlation data includes independent parameter correlation and coupling parameter correlation.

3. The trustworthy evaluation method for the digital twin evolution process according to claim 1 is characterized in that: Constructing a key parameter set based on the correlation data, and determining a deviation interval of each key parameter in the key parameter set, specifically includes: Based on the correlation, the model change parts and corresponding change ranges that affect the demand indicators are screened out; a key parameter set corresponding to the model change parts is constructed, and the deviation interval of each key parameter in the key parameter set is determined based on the corresponding change range.

4. The trustworthy assessment method for the digital twin evolution process according to claim 1 is characterized in that: Inputting the real-time monitoring data into the performance deviation estimation model to calculate the credibility value of the digital twin for the current task application in the current evaluation time window specifically includes: Based on the preset evaluation sampling moment, the real-time key status data and the evolving real-time result data at the same moment are retrieved, and the retrieved data is input into the performance deviation estimation model for evaluation to obtain the performance deviation data at the current evaluation sampling moment; the application occurrence probability corresponding to the task application data at the current evaluation sampling moment is calculated; based on the performance deviation data and the application occurrence probability at the current evaluation sampling moment, the credibility value at the current evaluation sampling moment is calculated; the credibility data at each evaluation sampling moment is summed to obtain the credibility value of the digital twin for the current task application in the current evaluation time window.

5. The trustworthy assessment method for the digital twin evolution process according to claim 1 is characterized in that: When the evaluation time window is updated, the corresponding overall credibility value is dynamically updated, including: When the last moment of the evaluation time window is aligned with the latest real moment according to the set time interval, an updated evaluation time window is obtained, and the corresponding overall credibility value is calculated in real time based on the updated evaluation time window.

6. A trusted assessment system for the digital twin evolution process, characterized in that: include: A performance deviation estimation model construction module is used to quantitatively analyze the correlation between each model parameter and the demand index in the model to be analyzed to obtain correlation data; wherein, the model to be analyzed is a digital twin basic model; a key parameter set is constructed based on the correlation data, and the deviation interval of each key parameter in the key parameter set is determined; sampling rounds and sampling points are set, and the performance deviation of the digital twin basic model corresponding to each key parameter with a deviation interval and the key state data of the dual system in the corresponding process are recorded point by point in each round; wherein, the performance deviation is the deviation between the control strategy obtained by executing the simulation solution based on the digital twin basic model on the real physical system and the expected performance index; the key state data of the dual system is the real-time state data generated by the observable parameters in the remaining parameters in the key parameter set that have a high correlation with each key parameter; with the key parameters with deviation intervals and the corresponding key state data of the dual system as input, and the performance deviation as output, a performance deviation estimation model is fitted; A credibility dynamic update module is used to construct an evaluation time window, which includes several types of task applications; monitor the evolution process of the digital twin in real time to obtain real-time monitoring data, and store the real-time monitoring data; wherein the real-time monitoring data includes real-time key status data, evolution real-time result data, current task application data and corresponding time series data; calculate the credibility value of the digital twin for the current task application in the current evaluation time window based on the real-time monitoring data; perform weighted summation on the credibility values ​​of various types of task applications in the current evaluation time window according to the frequency of the corresponding task application, and obtain the overall credibility value of the digital twin in the current evaluation time window; when the evaluation time window is updated, dynamically update the corresponding overall credibility value.

7. An electronic device, characterized in that: It includes a memory and a processor, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform a trusted assessment method for a digital twin evolution process according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements a trusted assessment method for a digital twin evolution process as described in any one of claims 1-5.

Images