Digital twin operation and maintenance data quality inspection method

Abstract

The application discloses a kind of digital twin operation and maintenance data quality inspection methods, comprising: digital twin operation and maintenance data preprocessing module, grouping, transformation and elimination are carried out to data;Digital twin operation and maintenance data integrity inspection module, data integrity is inspected from data acquisition integrity, coverage integrity and composition integrity angle;Digital twin operation and maintenance data consistency inspection module, data consistency is inspected from data statistical consistency, trend consistency and correlation consistency angle;Digital twin operation and maintenance data feature inspection module, whether data has required feature is inspected from data distribution and constraint feature, correlation feature and deep feature angle;Digital twin operation and maintenance data quality comprehensive evaluation module, based on user-defined setting weight, data quality is comprehensively evaluated.The application can inspect and evaluate the quality of digital twin data, provide reference and guidance for the quality optimization of data and the implementation of data-driven operation and maintenance control, optimization, decision-making and other methods.

Description

Technical Field

[0001] This invention belongs to the fields of electronic engineering and computer science, and specifically relates to a method for inspecting the quality of digital twin operation and maintenance data. Background Technology

[0002] With the continuous development and application of emerging information technologies in the manufacturing industry, the digitalization level of manufacturing workshops has gradually improved, and the types and quantities of data that workshops can acquire have significantly increased. To improve the operational efficiency of manufacturing workshops and enhance product quality, technologies and concepts such as digital twins have become a hot topic and goal in the intelligent workshop operation and maintenance process, further increasing user demand for accurate models, real-time data, and intelligent operation and maintenance. In applying cutting-edge technologies such as digital twins, users often face a variety of types and large volumes of operation and maintenance data. This data not only differs significantly in content and form but also in quality. Unprocessed, low-quality raw data cannot support the implementation of optimization algorithms and data models; even partially processed historical data suffers from inconsistent data quality due to differences in processing methods and means, indirectly affecting the performance of algorithms and models; for intelligent operation and maintenance algorithms and models with high real-time requirements, high-quality real-time data is essential for support. Summary of the Invention

[0003] To address the challenges of uncertain data quality and difficulty in selecting data ranges during the implementation and verification of algorithms and models in intelligent operations and maintenance (O&M) processes, this invention proposes a digital twin O&M data quality inspection method. First, the data is grouped multiple times based on its source and content, and its format is standardized to remove invalid data. The method verifies data integrity from the perspectives of data collection, coverage, and composition, determining detailed weights for integrity indicators. It also verifies data consistency from the perspectives of data statistics, trends, and correlations, determining detailed weights for consistency indicators. Finally, it verifies data characteristics from the perspectives of data distribution and constraints, correlation, and deep features, determining detailed weights for data feature indicators. Finally, the three inspection indicators are combined with user requirements to comprehensively evaluate data quality. This invention enables data quality assessment based on user needs, thus providing support and guidance for subsequent data selection and data quality optimization.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A method for inspecting the quality of digital twin operation and maintenance data, characterized by comprising the following steps:

[0006] Step 1, digital twin operation and maintenance data preprocessing, is implemented as follows:

[0007] (1) Based on the actual situation of the current database system and data acquisition system, the data is grouped according to the data source, which includes acquired data, simulation data, and mixed data.

[0008] (2) Based on data usage requirements and user definitions, the data is grouped in a second way according to the data range. The granularity of the data range can be freely changed, including larger granularity workshop data, environmental data, and logistics data; smaller granularity production line data, equipment data, and personnel data; and even smaller granularity component parameters, fault parameters, and operating parameters. The division results will be used in subsequent data quality inspection.

[0009] (3) Based on the data content, set a unified data display format and unify the data display format of each group;

[0010] (4) Based on the physical meaning of the data, obviously invalid data is removed;

[0011] Step 2, Digital Twin Operation and Maintenance Data Integrity Verification, is implemented as follows:

[0012] (1) Data collection integrity check: This check is performed to check whether there are any omissions or losses during the data collection and transmission process. Specifically, this is manifested as missing data points or invalid data points. The check result is calculated based on user definition, namely the data collection integrity level A1, with a value range of [0,100].

[0013] (2) Data coverage integrity test, which tests whether a certain part of the data is missing in the required scope of the scenario, task, and object. The specific manifestation of this situation is that some data is collectable but has not been obtained. The test result is calculated based on the user definition, that is, the data coverage integrity degree A2, with a value range of [0,100].

[0014] (3) Data integrity check: This check is performed to determine whether there is a lack of certain data that is not collectable or difficult to collect within the scope required by the scenario, task, or object, but such data has a function and characteristics. The check result is calculated based on the user definition, i.e., the data integrity level A3, with a value range of [0,100].

[0015] (4) Based on the data usage and user needs, determine the detailed weights i of the data integrity indicators. 1,2,3 ∈[0,1], where i1+i2+i3=1;

[0016] Step 3, Digital Twin Operation and Maintenance Data Consistency Verification, is implemented as follows:

[0017] (1) Data statistical consistency test: The similarity and fitting degree of the generated data and the collected data are tested. The same scene, object and input are selected, and the output under a single time and task sequence is compared and fitted. Statistical indicators are calculated, and the test results are calculated based on user definition, namely the data statistical consistency degree B1, with a value range of [0,100].

[0018] (2) Data trend consistency test: The test is performed on the trend of the same group of data. The same scenario, task and object are selected, and the deviation and trend indicators of multiple groups of data are compared. The test result is calculated based on the user definition, namely the data trend consistency degree B2, with a value range of [0,100].

[0019] (3) Data association consistency test: Based on human experience and mechanism, some relationships between data can be obtained. The relationships include data content association, constraint association and subordinate association. The consistency of each association relationship between twin data is tested. The test result is calculated based on user definition, i.e., the data association consistency degree B3, with a value range of [0,100].

[0020] (4) Based on the data usage and user needs, determine the detailed weights j of the data consistency index. 1,2,3 ∈[0,1], where j1+j2+j3=1;

[0021] Step 4, digital twin operation and maintenance data feature verification, is implemented as follows:

[0022] (1) Data distribution and constraint characteristics test: whether twin data has domain, scenario and principle characteristics related to content and use, including data density, extreme values, mean and direction of change. The test results are calculated based on user definition, namely data distribution and operation and maintenance characteristic index C1, with a value range of [0,100].

[0023] (2) Data correlation feature test: The test is conducted to check whether there are closely related variable elements in each group of data. The correlation features include Pearson correlation coefficient, Spearman rank correlation coefficient and rank correlation coefficient. The test results are calculated based on user definition, namely the data correlation feature index C2, with a value range of [0,100].

[0024] (3) Data deep feature test: Some data are used in situations where deep learning and other algorithms are used to mine deep features. The test results are calculated based on user definition, namely the data deep feature index C3, with a value range of [0,100].

[0025] (4) Determine the detailed weights k of the data feature indicators based on the data usage and user needs. 1,2,3 ∈[0,1], where k1+k2+k3=1;

[0026] Step 5: Comprehensive evaluation of the quality of digital twin operation and maintenance data, which is implemented as follows:

[0027] (1) First, based on user needs, determine the purpose of the data. Different purposes correspond to different weights of various indicators. The weights a, b, and c range from [0, 1]. When the user chooses not to perform a certain type of test, the weight of that type of indicator is set to 0.

[0028] (2) Calculate the test results for each type of indicator and the data integrity test results. Data consistency test results Data feature test results

[0029] (3) Calculate the comprehensive evaluation results of data quality inspection

[0030] The advantages of this invention compared to the prior art are:

[0031] (1) This invention evaluates data from the perspective of data quality, from three dimensions: data integrity, consistency and characteristics. The evaluation results have significant guiding significance for the selection, use and quality optimization of subsequent data.

[0032] (2) The weights set by this invention can be customized and adjusted according to the user's needs, so that the user can better judge and select the high-quality data that best meets their expectations. Attached Figure Description

[0033] Figure 1 This is a structural block diagram of the digital twin operation and maintenance data quality verification method of the present invention. Detailed Implementation

[0034] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0035] This invention relates to a method for inspecting the quality of digital twin operation and maintenance data. The method groups data multiple times based on its source and content, standardizes the data format, and removes invalid data. It verifies data integrity from the perspectives of data collection, coverage, and composition, determining detailed weights for integrity indicators; it verifies data consistency from the perspectives of data statistics, trends, and correlations, determining detailed weights for consistency indicators; and it verifies data characteristics from the perspectives of data distribution and constraints, correlation, and deep features, determining detailed weights for data feature indicators. Finally, it comprehensively evaluates data quality by integrating these three inspection indicators with user needs. This invention enables data quality assessment based on user requirements, thereby providing support and guidance for subsequent data selection and data quality optimization.

[0036] like Figure 1 As shown, the present invention provides a digital twin operation and maintenance data quality inspection method, which is implemented through a digital twin operation and maintenance data preprocessing module 1, a digital twin operation and maintenance data integrity inspection module 2, a digital twin operation and maintenance data consistency inspection module 3, a digital twin operation and maintenance data feature inspection module 4, and a digital twin operation and maintenance data quality comprehensive evaluation module 5. The specific steps are as follows:

[0037] (1) The specific implementation of the digital twin operation and maintenance data preprocessing module 1 is as follows:

[0038] ① Based on the current situation of the database system and data acquisition system, the data is grouped according to its source, which includes "acquired data, simulation data, and mixed data (i.e., both acquired data and simulation data exist)".

[0039] ② Based on data usage requirements and user definitions, the data is grouped a second time according to the data range. The granularity of the data range can be freely changed. For example, the granularity is larger, such as "workshop data, environmental data, logistics data, etc."; the granularity is smaller, such as "production line data, equipment data, personnel data, etc."; and the granularity is even smaller, such as "component parameters, fault parameters, operating parameters, etc." The division results will be used in subsequent data quality inspection.

[0040] ③ Based on the data content, set a unified data display format and unify the data display format of each group. For example, the data representing "time", "2022-01-01" and "2022 / 01 / 01" have the same data content;

[0041] ④ Based on the physical meaning of the data, obviously invalid data is removed. For example, "the temperature data in K (Kelvin) is -9" is obviously outside the meaningful range, so such obviously invalid data is removed.

[0042] (2) The specific implementation of the digital twin operation and maintenance data integrity verification module 2 is as follows:

[0043] ① Data collection integrity check: This check is performed to check whether there are any omissions or losses during the data collection and transmission process. Specifically, this is manifested as missing data points or invalid data points. The check result is calculated based on user definition, which is the data collection integrity level A1, with a value range of [0,100].

[0044] ② Data coverage integrity test: This test checks whether a certain part of the data is missing from the scope required by the scenario, task, or object (research, analysis, evaluation, decision-making, etc.). This situation is specifically manifested in the fact that some data is collectable but has not been obtained. For example, in the production logistics and distribution process of operation and maintenance, "production line data" and "equipment data" have been obtained, but "personnel data" has not been obtained. Obviously, "personnel data" is within the research scope of "production logistics and distribution" and can be obtained through manual input, image recognition, etc. The test result is calculated based on user definition, that is, the data coverage integrity level A2, with a value range of [0,100].

[0045] ③ Data integrity check: This check examines whether certain types of data are missing from the scope required by the scenario, task, or object (research, analysis, evaluation, decision-making, etc.), but such data has a function and characteristics. For example, "operating condition data" under extreme conditions, "fault data" with high experimental costs and extremely low frequency, etc. The check result is calculated based on user definition, namely the data integrity level A3, with a value range of [0,100].

[0046] ④ Based on the data usage and user needs, determine the detailed weights i for the data integrity indicators. 1,2,3 ∈[0,1], where i1+i2+i3=1;

[0047] (3) The specific implementation of the digital twin operation and maintenance data consistency verification module 3 is as follows:

[0048] ① Data statistical consistency test: This test examines the similarity and fit between the generated data and the collected data. It selects the same scene, object and input, compares and fits the output under a single time and task sequence, calculates statistical indicators, and calculates the test results based on user definition, namely the data statistical consistency degree B1, with a value range of [0,100].

[0049] ② Data trend consistency test: Test the trend of data in the same group. Select the same scenario, task and object, and compare the deviation and trend of multiple groups of data. For example, the overall trend of the "temperature data" of the autoclave equipment under the curing task is "heating-heating-cooling", with the temperature change range within 2 degrees Celsius. Each group of temperature data of this type should meet the overall trend and range constraints. Calculate the test result based on user definition, that is, the data trend consistency degree B2, with a value range of [0,100].

[0050] ③ Data association consistency test: Based on human experience and mechanisms, some relationships between data can be obtained, such as data content association, constraint association, and subordinate association. The test is conducted to check whether the association relationship between twin data is consistent. The test result is calculated based on user definition, that is, the data association consistency degree B3, with a value range of [0,100].

[0051] ④ Based on data usage and user needs, determine the detailed weights j of the data consistency index. 1,2,3 ∈[0,1], where j1+j2+j3=1;

[0052] (4) The specific implementation of the digital twin operation and maintenance data feature verification module 4 is as follows:

[0053] ① Data distribution and constraint characteristics test: The test is conducted to check whether the twin data has domain, scenario and principle characteristics related to the content and use, such as data density, extreme values, mean, and direction of change. The test results are calculated based on user definition, namely the data distribution and operation and maintenance characteristic index C1, with a value range of [0,100].

[0054] ② Data correlation feature test: This test examines whether there are closely related variable elements in each group of data. Correlation features include Pearson correlation coefficient, Spearman rank correlation coefficient, rank correlation coefficient, etc. The test results are calculated based on user definition, namely the data correlation feature index C2, with a value range of [0,100].

[0055] ③ Data deep feature inspection: Some data are used in situations where deep learning and other algorithms are used to mine deep features. For example, the method of generating simulation data is used to supplement data samples. The test is conducted to see if the "simulation data" has the same or similar deep features as the "physical data". The test result is calculated based on user definition, namely the data deep feature index C3, with a value range of [0,100].

[0056] ④ Based on the data usage and user needs, determine the detailed weights k of the data feature indicators. 1,2,3 ∈[0,1], where k1+k2+k3=1;

[0057] (5) The specific implementation of the digital twin operation and maintenance data quality comprehensive evaluation module 5 is as follows:

[0058] ① First, based on user needs, determine the purpose of the data. Different purposes will correspond to different weights of various indicators. For example, if the purpose of the data is monitoring, the weight of the data integrity indicator will increase. The value range of weights a, b, and c is [0,1]. In particular, users can choose not to perform a certain type of inspection, and the weight of that type of indicator will be set to 0.

[0059] ② Calculate the test results for each type of indicator, and the data integrity test results. Data consistency test results Data feature test results

[0060] ③ Calculate the comprehensive evaluation results of data quality inspection

[0061] In summary, this invention discloses a method for verifying the quality of digital twin operation and maintenance data, comprising: a digital twin operation and maintenance data preprocessing module, a digital twin operation and maintenance data integrity verification module, a digital twin operation and maintenance data consistency verification module, a digital twin operation and maintenance data feature verification module, and a digital twin operation and maintenance data quality comprehensive evaluation module. The method disclosed in this invention can solve the problems of data quality uncertainty and difficulty in selecting data ranges faced by algorithms and models during the implementation and verification process in intelligent operation and maintenance, providing support and guidance for subsequent data selection and data quality optimization.

[0062] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0063] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for inspecting the quality of digital twin operation and maintenance data, characterized in that, Includes the following steps: Step 1, digital twin operation and maintenance data preprocessing, is implemented as follows: (1) Based on the actual situation of the current database system and data acquisition system, the data is grouped according to the data source, which includes acquired data, simulation data, and mixed data; (2) Based on data usage requirements and user definitions, the data is grouped in a second way according to the data range. The granularity of the data range can be freely changed, including larger granularity workshop data, environmental data, and logistics data; smaller granularity production line data, equipment data, and personnel data; and even smaller granularity component parameters, fault parameters, and operating parameters. The division results will be used in subsequent data quality inspection. (3) Based on the data content, set a unified data display format and unify the data display format of each group; (4) Based on the physical meaning of the data, obviously invalid data is removed; Step 2, Digital Twin Operation and Maintenance Data Integrity Verification, is implemented as follows: (1) Data collection integrity check, which checks whether there are any omissions or losses in the data collection and transmission process. The omissions or losses are specifically manifested as missing data points or invalid data points. The check result is calculated based on user definition, namely the data collection integrity level A1, with a value range of [0,100]. (2) Data coverage integrity test, which tests whether a certain part of the data is missing in the required scope of the scenario, task, or object. The specific manifestation of this situation is that some data is collectable but not obtained. The test result is calculated based on the user definition, that is, the data coverage integrity degree A2, with a value range of [0,100]. (3) Data integrity check: This check is performed to determine whether there is a lack of certain data that is not collectable or difficult to collect within the scope required by the scenario, task, or object, but such data has a function and characteristics. The check result is calculated based on the user definition, i.e., the data integrity level A3, with a value range of [0,100]. (4) Based on the data usage and user needs, determine the detailed weights i1 for the data integrity index. 2,3 ∈[0,1], where i1+i2+i3=1; Step 3, Digital Twin Operation and Maintenance Data Consistency Verification, is implemented as follows: (1) Data statistical consistency test: The similarity and fitting degree of the generated data and the collected data are tested. The same scene, object and input are selected, the output under a single time and task sequence is compared and fitted, the statistical index is calculated, and the test result is calculated based on the user definition, that is, the data statistical consistency degree B1, with a value range of [0,100]. (2) Data trend consistency test: The trend of the same group of data is tested. The same scenario, task and object are selected, and the deviation and trend indicators of multiple groups of data are compared. The test result is calculated based on the user definition, that is, the data trend consistency degree B2, with a value range of [0,100]. (3) Data association consistency test: Based on human experience and mechanism, the relationship between some data is obtained. The relationship includes data content association, constraint association and subordinate association. The consistency of each association relationship between twin data is tested. The test result is calculated based on user definition, that is, the data association consistency degree B3, with a value range of [0,100]. (4) Based on the data usage and user needs, determine the detailed weights j1 for the data consistency index. 2,3 ∈[0,1], where j1+j2+j3=1; Step 4, digital twin operation and maintenance data feature verification, is implemented as follows: (1) Data distribution and constraint characteristics test: whether twin data has domain, scenario and principle characteristics related to content and use, including data density, extreme values, mean and direction of change. The test results are calculated based on user definition, namely data distribution and operation and maintenance characteristic index C1, with a value range of [0,100]. (2) Data correlation feature test: Test whether there are closely related variable elements in each group of data. The correlation features include Pearson correlation coefficient, Spearman rank correlation coefficient and rank correlation coefficient. The test results are calculated based on user definition, namely the data correlation feature index C2, with a value range of [0,100]. (3) Data deep feature test. Some data are used in situations where deep learning algorithms are used to mine deep features. The test results are calculated based on user definition, namely the data deep feature index C3, with a value range of [0,100]. (4) Based on the data usage and user needs, determine the detailed weights k1 for the data feature indicators. 2,3 ∈[0,1], where k1+k2+k3=1; Step 5: Comprehensive evaluation of the quality of digital twin operation and maintenance data, which is implemented as follows: (1) First, based on user needs, determine the purpose of the data. Different purposes correspond to different weights of various indicators. The range of weights a, b, c is [0,1]. When the user chooses not to perform a certain type of test, the weight of that type of indicator is set to 0. (2) Calculate the test results for each type of indicator and the data integrity test results. Data consistency test results Data feature test results ; (3) Calculate the comprehensive evaluation results of data quality inspection .

Images