Method and device for evaluating quality of engineering task assurance data of aerospace system

Abstract

The invention discloses a spaceflight system engineering task assurance data quality evaluation method and device. The method comprises the following steps: acquiring a spaceflight system engineering task index data set; the aerospace system engineering task index data set comprises a product domain data index, a use domain data index and an ecological domain data index; processing the aerospace system engineering task index data set to obtain a data quality evaluation result; and verifying the data quality evaluation result to obtain a data quality verification result. According to the method, the comprehensive and practical aerospace system engineering task assurance data quality evaluation method is constructed, the aerospace system engineering task index data is processed, and a comprehensive and complete evaluation result is obtained.

Description

Technical Field

[0001] This invention relates to the field of data quality assessment technology, and in particular to a method and apparatus for assessing data quality in aerospace systems engineering mission assurance. Background Technology

[0002] As simulation models become increasingly complex, simulation model verification methods have also developed significantly. These methods can be classified in many ways, such as by their construction principles, applicable scope, and result types. White-box verification verifies each corresponding submodule of the simulation system and the real system individually, verifying the model from a microscopic perspective. Black-box verification checks the overall consistency between the simulation system and the real system, verifying the model from a macroscopic perspective, and is the opposite of white-box verification. Gray-box verification combines white-box and black-box verification, exploring both the microscopic performance of the simulation model and verifying its macroscopic output state. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method and apparatus for assessing the quality of data for aerospace system engineering mission assurance, thereby improving the accuracy and comprehensiveness of the assessment of aerospace system engineering mission indicator data.

[0004] To address the aforementioned technical problems, a first aspect of this invention discloses a method for assessing the quality of data in aerospace systems engineering mission assurance, the method comprising: S1, Obtain the aerospace systems engineering mission indicator dataset; the aerospace systems engineering mission indicator dataset includes product domain data indicators, usage domain data indicators, and ecological domain data indicators; The product domain data indicators include operational data indicators, ground development information indicators, ground monitoring data indicators, ground testing data indicators, and ground development phase quality problem information indicators. The data indicators used in the domain include on-orbit / online fault data indicators, quality problem information indicators, spacecraft single-unit performance evaluation data indicators, and operation and control system and its equipment performance evaluation data indicators. The ecological domain data indicators include space environment data indicators, ground environment data indicators, supplier data, and on-orbit / online data indicators for research and development testing; S2, process the aerospace system engineering mission indicator dataset to obtain data quality assessment results; S3, verify the data quality assessment results to obtain the data quality verification results.

[0005] As an optional implementation, in the first aspect of the present invention, processing the aerospace systems engineering mission indicator dataset to obtain data quality assessment results includes: S21, preprocess the aerospace system engineering mission indicator dataset to obtain preprocessed data information; S22, the preprocessed data information is divided into T training subsets, where T is a positive integer; S23, using a preset T index evaluation model, process the T training subsets to obtain T evaluation results; S24, perform a comprehensive evaluation on the T evaluation results to obtain the data quality assessment results.

[0006] As an optional implementation, in the first aspect of the present invention, the preprocessing of the aerospace system engineering mission indicator dataset to obtain preprocessed data information includes: S211, perform data missing value filling on the aerospace system engineering mission indicator dataset to obtain the first preprocessed data information; S212, the first preprocessed data information is repaired to obtain the second preprocessed data information; S213, perform correlation analysis on the second preprocessed data information to obtain preprocessed data information.

[0007] As an optional implementation, in the first aspect of the present invention, the step of performing correlation analysis on the second preprocessed data information to obtain preprocessed data information includes: S2131, The second preprocessed data information is processed using a correlation calculation model to obtain the correlation coefficient value; The correlation calculation model expression is as follows: In the formula, The correlation coefficient value is the sequence. ,sequence , n Let be the dimension of the sequence. For sequence The mean, For sequence The mean of , where K() is the Gaussian kernel function. The kernel width is used to control the local range; S2132, process the second preprocessed data information, remove the second preprocessed data information whose correlation coefficient value is greater than a preset threshold, and obtain the third preprocessed data information; S2133, Process the third preprocessed data information to obtain data tag information; S2134, Oversample the data label information and the aerospace system engineering mission indicator dataset to obtain preprocessed data information.

[0008] As an optional implementation, in the first aspect of the present invention, processing the third preprocessed data information to obtain data tag information includes: S21331, Process the aerospace system engineering mission indicator dataset to obtain indicator data weight information; S21332, Dimensionality reduction processing is performed on the third preprocessed data information to obtain the fourth preprocessed data information; S21333, perform clustering processing on the fourth preprocessed data information to obtain the fifth preprocessed data information; S21334, integrate the fifth preprocessed data information and the indicator data weight information to obtain data label information.

[0009] As an optional implementation, in the first aspect of the present invention, processing the aerospace system engineering mission indicator dataset to obtain indicator data weight information includes: S213311, Process the aforementioned aerospace system engineering mission indicator dataset to obtain the first... The first expert The first evaluation indicator Ratings for each evaluated individual: in , , m The number of indicators. n For the number of evaluation objects, For the first The first expert The first evaluation indicator The minimum score for each evaluation object For the first The first expert The first evaluation indicator The most likely score value for each evaluation object. For the first The first expert The first evaluation indicator The maximum score of each evaluation object , The number of experts; S213312, regarding the first The first expert The first evaluation indicator The scores of the first evaluation object are processed to obtain the first evaluation object. The first expert The first evaluation indicator The optimized score value of each evaluation object; The expression for the optimized score value is: S213313, Process the optimized score value to obtain indicator data weight information; The expression for the weight information of the indicator data is: in, , , , .

[0010] As an optional implementation, in the first aspect of the present invention, the step of comprehensively evaluating the T evaluation results to obtain a data quality assessment result includes: S241, The T evaluation results are weighted to obtain a weighted evaluation result; S242, using a preset reinforcement learner, the weighted evaluation results are processed to obtain data quality assessment results.

[0011] A second aspect of this invention discloses a data quality assessment device for aerospace systems engineering mission assurance, the device comprising: The data acquisition module is used to acquire a dataset of aerospace system engineering mission indicators; the dataset of aerospace system engineering mission indicators includes product domain data indicators, usage domain data indicators, and ecological domain data indicators. The product domain data indicators include operational data indicators, ground development information indicators, ground monitoring data indicators, ground testing data indicators, and ground development phase quality problem information indicators. The data indicators used in the domain include on-orbit / online fault data indicators, quality problem information indicators, spacecraft single-unit performance evaluation data indicators, and operation and control system and its equipment performance evaluation data indicators. The ecological domain data indicators include space environment data indicators, ground environment data indicators, supplier data, and on-orbit / online data indicators for research and development testing; The quality assessment module is used to process the aerospace system engineering mission indicator dataset to obtain data quality assessment results; The data verification module is used to verify the data quality assessment results and obtain the data quality verification results.

[0012] As an optional implementation, in the second aspect of the present invention, the processing of the aerospace systems engineering mission indicator dataset to obtain data quality assessment results includes: S21, preprocess the aerospace system engineering mission indicator dataset to obtain preprocessed data information; S22, the preprocessed data information is divided into T training subsets, where T is a positive integer; S23, using a preset T index evaluation model, process the T training subsets to obtain T evaluation results; S24, perform a comprehensive evaluation on the T evaluation results to obtain the data quality assessment results.

[0013] As an optional implementation, in the second aspect of the present invention, the preprocessing of the aerospace system engineering mission indicator dataset to obtain preprocessed data information includes: S211, perform data missing value filling on the aerospace system engineering mission indicator dataset to obtain the first preprocessed data information; S212, the first preprocessed data information is repaired to obtain the second preprocessed data information; S213, perform correlation analysis on the second preprocessed data information to obtain preprocessed data information.

[0014] As an optional implementation, in a second aspect of the present invention, the step of removing correlation analysis from the second preprocessed data information to obtain preprocessed data information includes: S2131, The second preprocessed data information is processed using a correlation calculation model to obtain the correlation coefficient value; The correlation calculation model expression is as follows: In the formula, The correlation coefficient value is the sequence. ,sequence , n Let be the dimension of the sequence. For sequence The mean, For sequence The mean of , where K() is the Gaussian kernel function. The kernel width is used to control the local range; S2132, process the second preprocessed data information, remove the second preprocessed data information whose correlation coefficient value is greater than a preset threshold, and obtain the third preprocessed data information; S2133, Process the third preprocessed data information to obtain data tag information; S2134, Oversample the data label information and the aerospace system engineering mission indicator dataset to obtain preprocessed data information.

[0015] As an optional implementation, in the second aspect of the present invention, processing the third preprocessed data information to obtain data tag information includes: S21331, Process the aerospace system engineering mission indicator dataset to obtain indicator data weight information; S21332, Dimensionality reduction processing is performed on the third preprocessed data information to obtain the fourth preprocessed data information; S21333, perform clustering processing on the fourth preprocessed data information to obtain the fifth preprocessed data information; S21334, integrate the fifth preprocessed data information and the indicator data weight information to obtain data label information.

[0016] As an optional implementation, in the second aspect of the present invention, processing the aerospace system engineering mission indicator dataset to obtain indicator data weight information includes: S213311, Process the aforementioned aerospace system engineering mission indicator dataset to obtain the first... The first expert The first evaluation indicator Ratings for each evaluated individual: in , , m The number of indicators. n For the number of evaluation objects, For the first The first expert The first evaluation indicator The minimum score for each evaluation object For the first The first expert The first evaluation indicator The most likely score value for each evaluation object. For the first The first expert The first evaluation indicator The maximum score of each evaluation object , The number of experts; S213312, regarding the first The first expert The first evaluation indicator The scores of the first evaluation object are processed to obtain the first evaluation object. The first expert The first evaluation indicator The optimized score value of each evaluation object; The expression for the optimized score value is: S213313, Process the optimized score value to obtain indicator data weight information; The expression for the weight information of the indicator data is: in, , , , .

[0017] As an optional implementation, in the second aspect of the present invention, the step of comprehensively evaluating the T evaluation results to obtain a data quality assessment result includes: S241, The T evaluation results are weighted to obtain a weighted evaluation result; S242, using a preset reinforcement learner, the weighted evaluation results are processed to obtain data quality assessment results.

[0018] A third aspect of this invention discloses another aerospace systems engineering mission assurance data quality assessment device, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in the aerospace systems engineering mission assurance data quality assessment method disclosed in the first aspect of the present invention.

[0019] The fourth aspect of the present invention discloses a computer-storable medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the aerospace systems engineering mission assurance data quality assessment method disclosed in the first aspect of the present invention.

[0020] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: This invention implements a method for assessing the quality of data for aerospace system engineering mission assurance. After performing operations such as filling missing values, repairing, and removing correlations on the dataset of aerospace system engineering mission indicators, quality assessment and data verification are performed to achieve a comprehensive and accurate assessment of the data for aerospace system engineering mission assurance. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating a method for assessing the quality of data in aerospace systems engineering missions, as disclosed in an embodiment of the present invention. Figure 2 This is a flowchart illustrating another method for assessing the quality of aerospace system engineering mission assurance data, as disclosed in an embodiment of the present invention. Figure 3 This is the result of data cleaning and transformation disclosed in the embodiments of the present invention; Figure 4 These are the verification and validation activities in the modeling and simulation process disclosed in the embodiments of this invention; Figure 5 This is a schematic diagram of multivariate model validation based on kernel principal component analysis disclosed in an embodiment of the present invention; Figure 6 These are the data quality rules disclosed in the embodiments of this invention; Figure 7 These are the big data quality-related issues and scenarios disclosed in the embodiments of this invention; Figure 8 This is the open-source data quality assessment system disclosed in the embodiments of the present invention; Figure 9 This is a flowchart of the ADS-B data quality assessment disclosed in an embodiment of the present invention; Figure 10 This is a schematic diagram of the AE principle disclosed in an embodiment of the present invention; Figure 11 This is a schematic diagram of the structure of a data quality assessment device for ensuring aerospace system engineering missions, as disclosed in an embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention, 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0025] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0026] This invention discloses a method and apparatus for assessing the quality of data for aerospace systems engineering mission assurance. The method includes: acquiring a dataset of aerospace systems engineering mission indicators; the dataset includes product domain data indicators, usage domain data indicators, and ecological domain data indicators; processing the dataset to obtain a data quality assessment result; and verifying the data quality assessment result to obtain a data quality verification result. This invention constructs a comprehensive and practical method for assessing the quality of data for aerospace systems engineering mission assurance, processing aerospace systems engineering mission indicator data to obtain comprehensive and complete evaluation results. These are described in detail below.

[0027] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a data quality assessment method for ensuring aerospace system engineering missions, as disclosed in an embodiment of the present invention. Figure 1 The described aerospace systems engineering mission assurance data quality assessment method is applied in the field of data quality assessment technology, and the embodiments of this invention are not limited thereto. Figure 1 As shown, the data quality assessment method for ensuring the aerospace systems engineering mission may include the following operations: S1, Obtain the aerospace systems engineering mission indicator dataset; the aerospace systems engineering mission indicator dataset includes product domain data indicators, usage domain data indicators, and ecological domain data indicators; The product domain data indicators include operational data indicators, ground development information indicators, ground monitoring data indicators, ground testing data indicators, and ground development phase quality problem information indicators. The data indicators used in the domain include on-orbit / online fault data indicators, quality problem information indicators, spacecraft single-unit performance evaluation data indicators, and operation and control system and its equipment performance evaluation data indicators. The ecological domain data indicators include space environment data indicators, ground environment data indicators, supplier data, and on-orbit / online data indicators for research and development testing; S2, process the aerospace system engineering mission indicator dataset to obtain data quality assessment results; S3, verify the data quality assessment results to obtain the data quality verification results.

[0028] Optionally, the processing of the aerospace systems engineering mission indicator dataset to obtain data quality assessment results includes: S21, preprocess the aerospace system engineering mission indicator dataset to obtain preprocessed data information; S22, the preprocessed data information is divided into T training subsets, where T is a positive integer; S23, using a preset T index evaluation model, process the T training subsets to obtain T evaluation results; S24, perform a comprehensive evaluation on the T evaluation results to obtain the data quality assessment results.

[0029] Optionally, the preprocessing of the aerospace systems engineering mission indicator dataset to obtain preprocessed data information includes: S211, perform data missing value filling on the aerospace system engineering mission indicator dataset to obtain the first preprocessed data information; S212, the first preprocessed data information is repaired to obtain the second preprocessed data information; S213, perform correlation analysis on the second preprocessed data information to obtain preprocessed data information.

[0030] Optionally, the step of performing correlation analysis on the second preprocessed data information to obtain preprocessed data information includes: S2131, The second preprocessed data information is processed using a correlation calculation model to obtain the correlation coefficient value; The correlation calculation model expression is as follows: In the formula, The correlation coefficient value is the sequence. ,sequence , n Let be the dimension of the sequence. For sequence The mean, For sequence The mean of , where K() is the Gaussian kernel function. The kernel width is used to control the local range; S2132, process the second preprocessed data information, remove the second preprocessed data information whose correlation coefficient value is greater than a preset threshold, and obtain the third preprocessed data information; S2133, Process the third preprocessed data information to obtain data tag information; S2134, Oversample the data label information and the aerospace system engineering mission indicator dataset to obtain preprocessed data information.

[0031] Optionally, the step of processing the third preprocessed data information to obtain data tag information includes: S21331, Process the aerospace system engineering mission indicator dataset to obtain indicator data weight information; S21332, Dimensionality reduction processing is performed on the third preprocessed data information to obtain the fourth preprocessed data information; The specific method for dimensionality reduction is as follows: The third preprocessed data is trained using a variational autoencoder. During training, the likelihood of the data is maximized to learn feature representations in the latent space, and the learned latent representations are extracted from the variational autoencoder model to obtain latent variables. Then, the latent variables are dimensionality reduced using the following steps: Calculate the joint probability representing distance similarity using latent variables. : in, The latent variables are , , for dimensionality for The mean, , for The distance to its nearest neighbor, For matching of The local distance of the nearest neighbor, Based on the experimental setup, for and The distance between them , for The distance to its nearest neighbor, For matching of The local distance of the nearest neighbor, This is the global distribution scale parameter.

[0032] Stream learning is used to map latent variables to a low-dimensional embedded space to obtain low-dimensional vectors. Calculate the embedding similarity probability : in, These are preset hyperparameters, set experimentally. , , M for The elements in M for The dimension of.

[0033] Constructing the loss function C : To compare the loss weights, It is the distance threshold between dissimilar samples, minimized. (Achieved through stochastic gradient descent) to obtain low-dimensional parameters. .

[0034] S21333, perform clustering processing on the fourth preprocessed data information to obtain the fifth preprocessed data information; The clustering method is: (1) Randomly in the fourth preprocessed data information Select a point as the cluster center and set a threshold. The number of data; (2) Calculate all points within the threshold distance range of the cluster center, record them in set M, and classify them into cluster c. The distance measurement formula is: For clusters The distance value, For the first Data points, For the first A cluster; (3) Calculate the distance vector between each element in set M and cluster c and sum them to obtain the offset vector. : (4) The cluster center moves along the direction of the offset vector by a distance equal to the magnitude of the offset vector; (5) Repeat steps (2), (3), and (4) until the offset vector value reaches the threshold set by the termination condition, and record the cluster center at this time; (6) Repeat steps (1) to (5) until all data points have been traversed; (7) Finally, calculate the number of times each element is accessed by each class, and classify the element into the cluster class that is accessed the most.

[0035] S21334, integrate the fifth preprocessed data information and the indicator data weight information to obtain data label information.

[0036] Optionally, the processing of the aerospace systems engineering mission indicator dataset to obtain indicator data weight information includes: S213311, Process the aforementioned aerospace system engineering mission indicator dataset to obtain the first... The first expert The first evaluation indicator Ratings for each evaluated individual: in , , m The number of indicators. n For the number of evaluation objects, For the first The first expert The first evaluation indicator The minimum score for each evaluation object For the first The first expert The first evaluation indicator The most likely score value for each evaluation object. For the first The first expert The first evaluation indicator The maximum score of each evaluation object , The number of experts; S213312, regarding the first The first expert The first evaluation indicator The scores of the first evaluation object are processed to obtain the first evaluation object. The first expert The first evaluation indicator The optimized score value of each evaluation object; The expression for the optimized score value is: S213313, Process the optimized score value to obtain indicator data weight information; The expression for the weight information of the indicator data is: in, , , , .

[0037] Optionally, the step of comprehensively evaluating the T evaluation results to obtain the data quality assessment result includes: S241, The T evaluation results are weighted to obtain a weighted evaluation result; The weight values ​​were set experimentally. S242, using a preset reinforcement learner, the weighted evaluation results are processed to obtain the data quality assessment results. The preset reinforcement learner structure is as follows: The system consists of three convolutional modules, each employing different kernel sizes to extract features from the input data across varying receptive fields. Each module contains four convolutional layers. Following these layers, a CBAM module replaces the pooling layers in traditional CNN models, distinguishing feature importance and further enhancing the network's feature extraction capability and quality assessment efficiency. Adaptive pooling unifies the different lengths of features output from the three modules to the same length. Feature weights are calculated for each module to highlight scale features that are more critical to data quality. After fusing multi-scale features according to their weights, a 1×1 convolution is used to enhance channel expressiveness.

[0038] Convolution module 1 includes (64 3×1 convolution kernels, convolution stride of 2), CBAM (64 CBAM modules). (32 3×1 convolution kernels, convolution stride of 2), CBAM (32 CBAM modules). (16 3×1 convolution kernels, convolution stride of 2), CBAM (16 CBAM modules). (8 3×1 convolutional kernels, stride of 2), CBAM (8 CBAM modules); Convolutional module 2 includes (64 5×1 convolution kernels, convolution stride of 2), CBAM (64 CBAM modules). (32 5×1 convolution kernels, convolution stride of 2), CBAM (32 CBAM modules). (16 5×1 convolution kernels, convolution stride of 2), CBAM (16 CBAM modules). (8 5×1 convolutional kernels, stride of 2), CBAM (8 CBAM modules); Convolutional module 3 includes (64 7×1 convolution kernels, convolution stride of 2), CBAM (64 CBAM modules). (32 7×1 convolution kernels, convolution stride of 2), CBAM (32 CBAM modules). (16 7×1 convolution kernels, convolution stride of 2), CBAM (16 CBAM modules). (8 7×1 convolutional kernels, convolution stride of 2), CBAM (8 CBAM modules); feature fusion layer ([186×8,184×8,182×8]), fully connected layer (512 neurons) and output layer (5 neurons, corresponding to data quality assessment results of excellent, good, medium, qualified and unqualified).

[0039] The reinforcement learner output is finally implemented in a fully connected layer with the softmax activation function. In the last layer, the softmax function transforms the feature information into probabilities of the model's evaluation of various data quality outcomes. In the formula The probability of each data quality evaluation result output by the Softmax activation function; For the types of data quality assessment results, Indicates the first i One evaluation result.

[0040] As can be seen, the present invention realizes a method for assessing the quality of aerospace system engineering mission assurance data. After performing operations such as filling missing data values, repairing data, and removing correlations on the aerospace system engineering mission indicator dataset, quality assessment and data verification are performed to achieve a comprehensive and accurate assessment of aerospace system engineering mission assurance data.

[0041] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating another method for assessing the quality of aerospace system engineering mission assurance, as disclosed in an embodiment of the present invention. Figure 2 The described aerospace systems engineering mission assurance data quality assessment method is applied in the field of data quality assessment technology, and the embodiments of this invention are not limited thereto. Figure 2 As shown, the data quality assessment method for ensuring the aerospace systems engineering mission may include the following operations: 1. Data cleaning technology Given the diverse sources, massive volume, and varied formats of aerospace systems engineering data, appropriate technical methods are needed for data cleaning and fusion analysis to achieve physically dispersed yet logically unified big data fusion and provide comprehensive and long-term data resource assurance. Data cleaning techniques primarily include Bayesian probabilistic inference-based missing value imputation, functional dependency-based data inconsistency detection and repair, entity recognition based on similarity probability graphs, and truth discovery based on user feedback. Processing aerospace system MBMA data using data cleaning techniques improves data uniformity, accuracy, and completeness.

[0042] The implementation of MBMA system business management requires high-quality data support that is consistent, accurate, and complete. However, the data results from the integrated MBMA systems are not uniform in format, contain some redundancy, and their accuracy needs further improvement, affecting the quality of the integrated data and hindering its effective support for data analysis and application. Therefore, standardizing the format of MBMA data, eliminating redundancy, and removing false data are crucial to ensuring the quality of MBMA system data and are a prerequisite for accurate data analysis. Data cleaning and transformation can convert raw data into standardized integrated data. The purpose of data governance is to: Improving data consistency is crucial. Different data sources provide varying information about the same entity, leading to redundancy and inconsistencies that can negatively impact the quality of data integration and the effectiveness of analysis and data mining. Data governance, by identifying differences in the same entity, can effectively eliminate redundancy and integrate inconsistent information, thereby improving the consistency and simplicity of integrated data.

[0043] Improving data accuracy is crucial. During data aggregation, due to network issues or other reasons, some data sources may provide incomplete, erroneous, or outdated data. This can lead to contradictions and data conflicts between different descriptions of the same entity from different data sources, making it difficult to guarantee the accuracy of the integrated data. Data governance, by resolving conflicts between data from multiple data sources, can effectively identify and distinguish between true and false data, thereby improving the accuracy of the integrated data.

[0044] Improving data integrity is crucial. Because different systems target different users and have different information focuses, the types of information provided by different data sources vary, and their descriptions of the same entity differ in aspect or perspective. This makes integrated data relatively discrete, hindering the formation of a complete and comprehensive information view of the system. Therefore, data governance can effectively achieve information complementarity and improve the integrity of integrated data.

[0045] 1.1 Structured Data Cleaning Techniques Structured data generally refers to data that can be represented and stored using relational databases and logically expressed using two-dimensional tables. This type of data storage and arrangement conforms to certain rules. In spacecraft fault knowledge, expert rules are generally stored in the form of structured data. Expert rules consist of relevant rules set by domain experts based on design specifications or experience. The results of data cleaning and transformation are as follows... Figure 3 As shown.

[0046] Data governance technologies mainly include data missing value imputation technology based on Bayesian probabilistic inference, data inconsistency detection and repair technology based on functional dependency, entity recognition technology based on similarity probability graphs, and truth discovery technology based on user feedback.

[0047] (1) Data missing value imputation technique based on Bayesian probabilistic inference This technique first generates dependencies between attributes based on statistical relationships between attribute values ​​in the data, then creates a Bayesian network based on these dependencies, and finally fills in the missing values ​​using the Bayesian network. After constructing the Bayesian network, the variables are organized into a network based on their correlations. Missing value imputation is then performed based on the information provided by the network. This technique still treats discrete and continuous variables separately; the former is handled using probabilistic inference, while the latter is handled using the least squares method.

[0048] (2) Data inconsistency detection and repair technology based on function dependency The purpose of this technique is to detect errors in data based on a given set of rules and to correct potential errors. The most commonly used rules, CFD and FD, are chosen as the rule formats.

[0049] (3) Entity recognition technology based on similarity probability graphs The purpose of entity recognition is to quickly and effectively identify entities from large datasets that are frequently updated and have complex structures. Its input is a database file within the system, and its output is a collection of multiple identified entities. The specific operation process is as follows: 1) The system exports massive amounts of data from the database and preprocesses it to ensure that the data format meets the requirements for system recognition.

[0050] 2) Considering that several entities share the same or similar attributes across all data, an attribute index detection model is used. By modifying attribute values ​​and inserting them into the attribute index table, entities with the same attribute value share the same index, thus constructing an attribute index table and forming a preliminary cluster. This method is effective and practically feasible for removing redundancy. In subsequent identification processes, only entities within the same attribute index table need to be identified individually, unlike traditional entity recognition methods that compare all entities one by one, thereby achieving rapid identification.

[0051] 3) Extract the required samples. In the process of threshold learning, it is first necessary to randomly extract a set of samples for learning and comparison from the massive amount of data in the database.

[0052] 4) By analyzing and learning from the extracted data samples, a threshold can be derived to accurately determine the similarity of entities.

[0053] 5) Entity identification is performed by comparing the similarity between entity pairs with a threshold. Entities with similarity greater than the threshold are identified as meeting the criteria.

[0054] 6) Create a public neighbor information pair graph for neighbor information between adjacent entities.

[0055] 7) Evaluate the various metrics of the algorithm used by the system (including time efficiency, accuracy, recall, etc.).

[0056] 1.2 Semi-structured data cleaning technology Semi-structured data does not conform to the data model structure of relational databases. Its characteristic is a mixture of data structure and content, with no strict distinction between the two, but it contains relevant tags that can be used to separate semantic elements and hierarchically categorize records and fields. Spacecraft fault knowledge contains much semi-structured data in tabular form, such as FMEA analysis. This type of data cannot be directly mapped to a knowledge graph, but it is easier to extract structured knowledge than unstructured data. For this type of data, a wrapper is constructed through manual analysis to extract rules, thereby transforming semi-structured data into structured data.

[0057] In relational databases, the knowledge storage schema is a complete set of tables, while in graph databases, it consists of nodes, relations, their labels, and attributes. In a graph database, an entity (including relations) is a basic storage unit. Labels are the basis for distinguishing entity and relation types, containing only one element: the label name. Attributes store node information, containing two elements: the attribute name and the attribute value. Knowledge transformation is achieved by mapping the schema of the relational database to the schema of the graph database.

[0058] (1) Direct transfer For standardized, structured information in tables, direct dumping is possible. Column names in relational databases correspond to labels, attributes, or relations in graph databases. The following points should be considered based on the information and function they contain: 1) Some column names in a table can simultaneously function as attributes, labels, or relations; 2) For ease of querying or reasoning, values ​​in a column can function as nodes and can also be repeatedly used as attribute values ​​for other nodes; 3) The number of relation types used as edges should be minimized to reduce the difficulty of reasoning.

[0059] (2) Standardized transfer Most of the information in the table is standardized and formatted, but some semi-structured data needs cleaning. This is addressed by directly checking for marker characters or stop words, extracting structured data from long clauses, and merging statements. Guided by the ontology, the merged results are summarized and organized, and then the entities, relations, and attributes in the results are matched with the ontology.

[0060] 1.3 Unstructured Data Cleaning Techniques Unstructured data is data without a fixed structure. The large amount of unstructured data generated on spacecraft, primarily in text format, originates from design documents, spacecraft logs, fault records, and anomaly messages. This unstructured data from spacecraft is text-intensive, with fault knowledge implicitly embedded within it. Typically, documents contain possible spacecraft fault phenomena, causes, and maintenance measures. Fault text is recorded according to certain standards, facilitating text preprocessing and knowledge acquisition. Unstructured data will be processed using text processing methods to mine entities, relationships, and attributes.

[0061] While relevant materials in the field of spacecraft fault diagnosis possess standardized expressions and logic, data suitable for training Chinese text processing tools is scarce. Therefore, this study leverages the advantages of standardized expressions to reduce data dependence, employing an ontology-entity bidirectional constraint logic. It proposes a keyword recognition + text classification approach for knowledge extraction from fault texts with a certain structure. At the entity layer, a combination of a classic CNN model and a multi-head attention CNN model is used to extract entities and relationships, with comparative verification. At the ontology layer, the ontology is optimized and improved based on the types corresponding to the extracted entities and relationships. During the knowledge extraction process, existing tools such as jieba word segmentation and pyltp are used for text preprocessing, including word segmentation, stop word removal, and sentence segmentation.

[0062] Entities are extracted using a classic CNN text classification model. In this method, the model first converts text clauses into vector forms represented by IDs based on the corpus, performs sequence padding and alignment, and then connects to a fully connected layer as network input; the entity type is used as the network output. Relations are extracted using a multi-head self-attention CNN classification model. This involves segmenting sentences based on the corpus, converting words into vector forms, performing sequence padding and alignment, and then connecting to a fully connected layer as word embedding features; simultaneously, the relative distance between each word and two entities in the sentence is also converted into embedding vectors as distance features; the two features are concatenated as the overall network input; the relation type is used as the network output. The two methods complement each other. The classic CNN model solves the problem of entity label classification with known relation types, which are obtained from text analysis. The multi-head self-attention CNN classification model solves the problem of relation classification with known entities and labels. Therefore, the methods can mutually validate each other, improving the accuracy of knowledge extraction.

[0063] 2. Model verification and validation technology This technology is an important means to ensure the reliability of digital construction of aerospace systems. Figure 4 Verification and validation activities are conducted during system modeling and simulation. Verification determines whether the digital model is correctly transformed from one form to another during the modeling process, judging the correctness of transformations from "problem definition" to "modeling requirements," from "modeling requirements" to "conceptual model," and from "conceptual model" to "simulation model." Verification is performed before the model is run and is an interpretable process; the model that can be verified must be a white-box model. Validation examines the accuracy with which the digital model represents the real-world prototype system, based on the simulation application purpose. Accreditation, based on verification and validation, is the final determination by an authoritative organization of whether the simulation model is acceptable for the specific application purpose.

[0064] This primarily supports "data information assurance" techniques for task assurance from the perspective of the correctness and consistency of the MBMA model and data. Specifically, digital model verification and data quality assessment techniques include: digital model verification techniques, data cleaning techniques, and data quality assessment techniques.

[0065] As simulation models become increasingly complex, simulation model verification methods have also developed significantly. These methods can be classified in many ways, such as according to their construction principles, applicable scope, and result types. Table 1 shows the classification based on different model construction principles. White-box verification verifies each corresponding sub-module of the simulation system and the real system one by one, verifying the model from a microscopic perspective. Black-box verification checks the overall consistency between the simulation system and the real system, verifying the model from a macroscopic perspective, and is the opposite of white-box verification. Gray-box verification combines white-box and black-box verification, exploring both the microscopic performance of the simulation model and verifying its macroscopic output state.

[0066] Table 1. Simulation Model Verification Methods 2.1 Digital Model Static Data Parameter Verification Techniques Static validation methods focus on the consistency of statistical characteristics of data, such as mean, variance, quantiles, and probability distribution functions, generally requiring samples to be independent and identically distributed. Classical hypothesis testing and Bayesian hypothesis testing methods are suitable for consistency analysis of normally distributed populations. Non-parametric testing methods are not sensitive to data distribution, but their analysis results are more risky than parametric methods. The area method measures the consistency of output results based on the difference in the cumulative distribution function area between the simulation output and the reference output. Bayesian hypothesis testing has been widely used in model validation considering uncertainty. Static data consistency analysis falls under the category of statistical analysis. It is used to verify the similarity between the statistical distribution of random data in the simulation model and the real model. Random data is a time-independent data sequence, such as off-target values. Its analysis methods mainly include hypothesis testing, parameter estimation, and Bayesian parameter estimation. Hypothesis testing can be further divided into parametric and non-parametric hypothesis testing, and parameter estimation can be further divided into point estimation and interval estimation.

[0067] 2.2 Dynamic Data Parameter Verification Technology for Digital Models Some product digital models are stochastic dynamic systems, whose dynamic performance exhibits time-varying characteristics, such as thrust, velocity, and position. These can all be considered as multidimensional stochastic processes, denoted as... The corresponding simulation model's dynamic performance is The problem to be solved is comparing two sampling time series obtained from field experiments and policy experiments. }and{ Consistency between the results can be assessed. Generally, qualitative comparisons can be made first using judgmental comparisons, such as graphical comparisons. This involves first visualizing the observed samples and then judging consistency through observation and subjective analysis, such as the periodicity, phase shift, frequency, and trend term of a random process. Although these methods are subjective, qualitative, rough, and exploratory, they are very practical. Of course, this method cannot provide an objective evaluation of the simulation results and can only serve as an initial verification tool. If significant differences are found, the reasons need to be considered; if no differences are found, quantitative methods should be used for further in-depth comparisons.

[0068] In general, quantitative methods for dynamic parameter consistency verification can be divided into time-domain, frequency-domain, and time-frequency-domain methods. Regardless of whether the analysis is conducted in the time domain or the frequency domain, we can broadly categorize them into two cases based on the different objects being analyzed: The first approach involves directly comparing the features of the two time-series samples {xt} and {yt} themselves. The main idea is to first estimate some typical features of each sample from their observation samples, and then compare them.

[0069] The second scenario involves the error sequence of two time-series samples. Perform "error analysis." Determine consistency based on the characteristics of the error sequence. This method is coarser than the previous one, as using error metrics as the basis for verification is clearly insufficient.

[0070] Commonly used methods for validating dynamic data parameters of digital models include: error analysis, model validation methods based on improved time-domain analysis, model validation methods based on maximum entropy spectral estimation, and multivariate validation methods based on kernel principal component analysis, etc. Figure 5 As shown.

[0071] 3. Digital quality assessment technology Digital quality assessment technologies include using existing rule-based methods for scaled micro-level control and macro-level governance of all aspects of data quality, as well as using big data analytics methods for meso-level analysis and management (a research scale between macro and micro). Through big data analytics and visualization methods, it is possible to quickly switch between macro and micro levels, observe data distribution and anomalies in a human-computer interactive manner, and improve the level and efficiency of data quality management.

[0072] Data quality rules are logical constraints that determine whether data meets data quality requirements, such as... Figure 6As shown. Throughout the entire data quality monitoring process, the quality of data quality rules directly impacts the monitoring effectiveness; therefore, designing data quality rules is crucial. Based on the quality characteristics of data when it is stored in the database and the types of data quality rules, four types of data quality classification frameworks are designed: 1) Single-column data quality rules: focusing on the presence or absence of data attribute values ​​and their compliance with their own specifications; 2) Cross-column data quality rules: focusing on the logical judgment of the relationships between data attributes; 3) Cross-row data quality rules: focusing on the logical judgment of the relationships between data records; 4) Cross-table data quality rules: focusing on the logical judgment of the relationships between datasets. Figure 7 From the perspective of big data quality problem characteristic analysis, this study examines big data granularity scaling, cross-domain correlation, and global perspective. Figure 3 The characteristics of major issues were analyzed. Among them, the quality problems of big data granularity scaling characteristics mainly refer to data quality problems when scaling between different granularity levels, which are related to data perception, connection and collection processes and capabilities; the quality problems of cross-boundary correlation characteristics mainly refer to data quality problems encountered when expanding and integrating spatial boundaries, which are related to the data governance perspective; and the quality problems of global view characteristics mainly emphasize the control of data quality in the overall context.

[0073] A data quality assessment system is a prerequisite for the effective implementation of data quality assessment. From the perspectives of data characteristics, data content, and effectiveness, a data quality detection index and assessment system are constructed to meet diverse data quality assessment needs. The data quality assessment system includes five main types: data confidence assessment, data verification assessment, data value assessment, comprehensive data quality assessment, and special data approval. It comprises ten dimensions: data timeliness, data completeness, data volatility, data uniqueness, data standardization, data usage, data missingness, data similarity, data freshness, and data coverage. The corresponding quality assessment is completed through a combination of these multiple dimensions. Open-source data quality assessment systems, such as... Figure 8 As shown.

[0074] The open-source data quality assessment process, based on the open-source data assessment system, implements the system through processes such as open-source data source configuration, multi-dimensional data quality detection model configuration, data quality assessment model configuration, data quality tracking and closure, and result feedback and optimization. The main processes are as follows: Figure 2 As shown.

[0075] The data quality assessment algorithm based on clustering and ensemble learning follows the process as follows: Figure 9 As shown.

[0076] 1) Pearson correlation analysis The indicators in the indicator system exhibit a certain degree of correlation. However, two highly correlated indicators are prone to information redundancy and overlap, which may reduce the efficiency of data quality assessment. Pearson correlation analysis is performed on the data to selectively remove two indicators with a correlation coefficient greater than 0.8. Based on the Pearson correlation analysis, a formula for calculating the correlation coefficient between sequences is derived.

[0077] 2) Entropy weight method combined with TOPSIS scoring Entropy weighting is an objective weighting method that determines weights based on the amount of information. Compared to methods such as the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation, it reduces bias caused by subjective factors. For the j-th indicator of the i-th evaluation object, the greater the difference between xij values, the greater the degree of change in the indicator and the greater its relative effect. This method first standardizes the data to determine the entropy value.

[0078] 3) Dimensionality reduction of data features Conventional clustering algorithms suffer from high computational costs and poor performance when processing high-dimensional data. Using autoencoders to reduce the dimensionality of the data before clustering can improve the results. An autoencoder (AE) is a neural network model used for feature dimensionality reduction. Compared to principal component analysis, it can learn non-linear relationships and offers better performance. AE dimensionality reduction involves two processes: encoding and decoding. The encoding process learns new features from high-dimensional data and compresses them into low-dimensional data. The decoding process reconstructs the original high-dimensional data using the learned new features. Figure 10 This is a schematic diagram of the principle structure of Advanced Image Processing (AE). When using AE for feature dimensionality reduction, it is necessary to reconstruct high-dimensional data as output data, and use the root mean square error as the reconstruction error to evaluate the dimensionality reduction effect.

[0079] 4) Construct a data quality assessment model The K-Means clustering algorithm is used to cluster the dimensionality-reduced data. Compared with other clustering methods, K-Means clustering is suitable for large amounts of data and has a faster convergence speed. Its core idea is to divide the data into K independent clusters, such that the distance between nodes in each cluster is as small as possible, while the distance between clusters is as large as possible. During clustering, the optimal number of clusters is selected using the evaluation metric CH.

[0080] After clustering, the large difference in the number of samples between different categories leads to an imbalance in the amount of training data, which may cause the decision boundary to shift, resulting in misclassification and affecting the classification performance. Synthetic Minority Oversampling Technique (SMOTE) is used to synthesize new samples to balance the sample size. Specifically, it generates new samples by randomly linearly interpolating between the positive class samples (samples with fewer data points) and their k closest samples of the same class.

[0081] To address the local minimization and overfitting problems of classic logistic regression and machine learning methods, the AdaBoost ensemble learning model is selected as the optimal choice. Ensemble learning performs classification tasks by weighted combination of multiple weak learners. Typically, a set of individual learners is first generated, and then these are combined using a combination strategy to create a strong learner. The AdaBoost algorithm connects multiple homogeneous weak classifiers in a serial structure, improving the accuracy and stability of the classifier. Its final learning result depends on the strategy combination of the classification results of the multiple weak classifiers. The key to AdaBoost lies in how to generate multiple different weak classifiers. Its core idea is to increase the weight of misclassified samples while decreasing the weight of correctly classified samples, thereby updating the data weight distribution of the weak learners in each iteration. After the iteration, the weak classifiers are combined according to the principle of giving greater weight to those with smaller classification error rates to form a strong learner with higher prediction accuracy.

[0082] Data quality assessment problems can be transformed into classification problems by clustering indicator features. After clustering, the entropy weight method combined with TOPSIS scoring can be used to determine the category labels. Then, AdaBoost ensemble learning and decision trees as weak classifiers are used to train the model, ultimately resulting in a data quality assessment model with high accuracy.

[0083] 3.1 Data Confidence Assessment Data confidence assessment mainly addresses issues such as data authenticity and usability caused by factors like multiple sources, diverse types, and conflicting data content in open-source data. It combines dimensions such as data missingness, data standardization, data freshness, and similarity between multiple data sources to construct a data confidence model and calculate data confidence.

[0084] 3.2 Data Value Assessment Data value assessment primarily evaluates data value during the data service selection and data usage phases. The data service selection phase mainly provides users with data comparison tools, intelligently and multidimensionally evaluating the best quality data to provide a basis for data selection decisions and ensure optimal acquisition or purchase of open-source data services. This typically involves mirroring open-source sample data, comparing it with existing data across dimensions such as coverage, freshness, standardization, completeness, timeliness, and uniqueness to comprehensively evaluate value. The data usage phase mainly involves users analyzing the usage of existing data resources to ensure optimal service and support for business analysis. This typically employs interfaces and shared libraries to deeply analyze data usage frequency, scope, and users, constructing a comprehensive evaluation model to generate data usage value assessment results.

[0085] 3.3 Comprehensive Data Quality Assessment The comprehensive data quality assessment provides a general data quality assessment model, which allows you to set the weights of different quality detection indicators for each type of data, generate a comprehensive data quality assessment model, and obtain a comprehensive quality score.

[0086] Example 3 Please see Figure 11 , Figure 11 This is a schematic diagram of the structure of a data quality assessment device for ensuring aerospace system engineering missions, as disclosed in an embodiment of the present invention. Figure 11 The described aerospace systems engineering mission assurance data quality assessment device is applied in the field of data quality assessment technology, and the embodiments of this invention are not limited thereto. Figure 11 As shown, the aerospace systems engineering mission assurance data quality assessment device may include the following operations: S301, Data Acquisition Module, used to acquire a dataset of aerospace system engineering mission indicators; the dataset of aerospace system engineering mission indicators includes product domain data indicators, usage domain data indicators and ecological domain data indicators. The product domain data indicators include operational data indicators, ground development information indicators, ground monitoring data indicators, ground testing data indicators, and ground development phase quality problem information indicators. The data indicators used in the domain include on-orbit / online fault data indicators, quality problem information indicators, spacecraft single-unit performance evaluation data indicators, and operation and control system and its equipment performance evaluation data indicators. The ecological domain data indicators include space environment data indicators, ground environment data indicators, supplier data, and on-orbit / online data indicators for research and development testing; S302, Quality Assessment Module, is used to process the aerospace system engineering mission indicator dataset to obtain data quality assessment results; S303, Data Verification Module, is used to verify the data quality assessment results and obtain data quality verification results.

[0087] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0088] Finally, it should be noted that the aerospace system engineering mission assurance data quality assessment method and apparatus disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assessing the quality of data in aerospace systems engineering mission assurance, characterized in that, The method includes: S1, Obtain the aerospace systems engineering mission indicator dataset; the aerospace systems engineering mission indicator dataset includes product domain data indicators, usage domain data indicators, and ecological domain data indicators; The product domain data indicators include operational data indicators, ground development information indicators, ground monitoring data indicators, ground testing data indicators, and ground development phase quality problem information indicators. The data indicators used in the domain include on-orbit / online fault data indicators, quality problem information indicators, spacecraft single-unit performance evaluation data indicators, and operation and control system and its equipment performance evaluation data indicators. The ecological domain data indicators include space environment data indicators, ground environment data indicators, supplier data, and on-orbit / online data indicators for research and development testing; S2, process the aerospace system engineering mission indicator dataset to obtain data quality assessment results; S3, verify the data quality assessment results to obtain the data quality verification results.

2. The aerospace systems engineering mission assurance data quality assessment method according to claim 1, characterized in that, The process of processing the aerospace system engineering mission indicator dataset to obtain data quality assessment results includes: S21, preprocess the aerospace system engineering mission indicator dataset to obtain preprocessed data information; S22, the preprocessed data information is divided into T training subsets, where T is a positive integer; S23, using a preset T index evaluation model, process the T training subsets to obtain T evaluation results; S24, perform a comprehensive evaluation on the T evaluation results to obtain the data quality assessment results.

3. The aerospace systems engineering mission assurance data quality assessment method according to claim 2, characterized in that, The preprocessing of the aerospace system engineering mission indicator dataset to obtain preprocessed data information includes: S211, perform data missing value filling on the aerospace system engineering mission indicator dataset to obtain the first preprocessed data information; S212, the first preprocessed data information is repaired to obtain the second preprocessed data information; S213, perform correlation analysis on the second preprocessed data information to obtain preprocessed data information.

4. The aerospace systems engineering mission assurance data quality assessment method according to claim 3, characterized in that, The step of performing correlation removal analysis on the second preprocessed data information to obtain preprocessed data information includes: S2131, The second preprocessed data information is processed using a correlation calculation model to obtain the correlation coefficient value; The correlation calculation model expression is as follows: In the formula, The correlation coefficient value is the sequence. ,sequence , n Let be the dimension of the sequence. For sequence The mean, For sequence The mean of , where K() is the Gaussian kernel function. The kernel width is used to control the local range; S2132, process the second preprocessed data information, remove the second preprocessed data information whose correlation coefficient value is greater than a preset threshold, and obtain the third preprocessed data information; S2133, Process the third preprocessed data information to obtain data tag information; S2134, Oversample the data label information and the aerospace system engineering mission indicator dataset to obtain preprocessed data information.

5. The aerospace systems engineering mission assurance data quality assessment method according to claim 4, characterized in that, The process of processing the third preprocessed data information to obtain data tag information includes: S21331, Process the aerospace system engineering mission indicator dataset to obtain indicator data weight information; S21332, Dimensionality reduction processing is performed on the third preprocessed data information to obtain the fourth preprocessed data information; S21333, perform clustering processing on the fourth preprocessed data information to obtain the fifth preprocessed data information; S21334, integrate the fifth preprocessed data information and the indicator data weight information to obtain data label information.

6. The aerospace systems engineering mission assurance data quality assessment method according to claim 5, characterized in that, The process of processing the aerospace system engineering mission indicator dataset to obtain indicator data weight information includes: S213311, Process the aforementioned aerospace system engineering mission indicator dataset to obtain the first... k The first expert The first evaluation indicator Ratings for each evaluated individual: in , , m The number of indicators. n For the number of evaluation objects, For the first The first expert The first evaluation indicator The minimum score for each evaluation object For the first The first expert The first evaluation indicator The most likely value for the rating of each evaluation object. For the first The first expert The first evaluation indicator The maximum score of each evaluation object , The number of experts; S213312, regarding the first The first expert The first evaluation indicator The scores of the first evaluation object are processed to obtain the first evaluation object. The first expert The first evaluation indicator The optimized score value of each evaluation object; The expression for the optimized score value is: S213313, Process the optimized score value to obtain indicator data weight information; The expression for the weight information of the indicator data is: in, , , , .

7. The aerospace systems engineering mission assurance data quality assessment method according to claim 2, characterized in that, The comprehensive evaluation of the T evaluation results to obtain the data quality assessment result includes: S241, The T evaluation results are weighted to obtain a weighted evaluation result; S242, using a preset reinforcement learner, the weighted evaluation results are processed to obtain data quality assessment results.

8. A data quality assessment device for ensuring mission assurance in aerospace systems engineering, characterized in that, The device includes: The data acquisition module is used to acquire a dataset of aerospace system engineering mission indicators; the dataset of aerospace system engineering mission indicators includes product domain data indicators, usage domain data indicators, and ecological domain data indicators. The product domain data indicators include operational data indicators, ground development information indicators, ground monitoring data indicators, ground testing data indicators, and ground development phase quality problem information indicators. The data indicators used in the domain include on-orbit / online fault data indicators, quality problem information indicators, spacecraft single-unit performance evaluation data indicators, and operation and control system and its equipment performance evaluation data indicators. The ecological domain data indicators include space environment data indicators, ground environment data indicators, supplier data, and on-orbit / online data indicators for research and development testing; The quality assessment module is used to process the aerospace system engineering mission indicator dataset to obtain data quality assessment results; The data verification module is used to verify the data quality assessment results and obtain the data quality verification results.

9. A data quality assessment device for ensuring mission assurance in aerospace systems engineering, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the aerospace systems engineering mission assurance data quality assessment method as described in any one of claims 1-7.

10. A computer-storable medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the aerospace systems engineering mission assurance data quality assessment method as described in any one of claims 1-7.

Images