New energy enterprise master data management method based on artificial intelligence

Abstract

The invention discloses a new energy enterprise master data management method based on artificial intelligence, and relates to the technical field of data management, and the method comprises the steps: collecting new energy enterprise master data; performing data verification on the collected main data, and marking abnormal data in the main data; intelligent error correction is carried out on the data marked as abnormal data, and error data are automatically repaired. According to the new energy enterprise master data management method based on artificial intelligence provided by the invention, the automation and intelligence level of data verification and restoration is improved. The problem that a traditional method depends on manual intervention is solved. The machine learning model is adopted to automatically classify the abnormal data into a rule repairable type or an inference repair type, and repair is performed through the predefined business rule and the similar data inference algorithm, so that various abnormal conditions can be flexibly handled, the repair precision is improved, and the data quality and consistency are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to a master data management method for new energy enterprises based on artificial intelligence. Background Art

[0002] In the new energy industry, enterprise management involves a vast amount of master data, including equipment, project, personnel, and customer and supplier data. With this dramatic increase in data volume, traditional master data management methods struggle to meet data accuracy and consistency requirements. This is especially true during data collection, storage, and processing, where abnormal, erroneous, and inconsistent data are prone to appearing. Existing technologies typically rely on manual intervention or rule-based systems to address these data issues. However, in a large-scale data environment, manual review is not only inefficient but also prone to omissions and errors.

[0003] The existing technology has defects in the following aspects:

[0004] Existing master data management methods rely on manual review and processing of abnormal data and cannot achieve automated processing, resulting in low processing efficiency.

[0005] Traditional data verification methods rely on predefined rules for verification and repair, but this method cannot cope with complex and frequently changing data environments and cannot handle abnormal data outside the rules.

[0006] When data from different sources and formats are managed in a unified manner, data inconsistencies are likely to occur, and existing technologies cannot effectively ensure the logical consistency between data.

[0007] Current methods are difficult to adapt to large-scale data environments, cannot intelligently detect and repair abnormal data, and lack self-optimization capabilities.

[0008] Therefore, existing technologies still have significant limitations in efficiently and accurately processing and repairing abnormal data. Especially in application scenarios such as new energy enterprises that involve a large amount of equipment, project and personnel data, traditional technologies cannot meet the needs of efficient and high-precision master data management. Summary of the Invention

[0009] In view of the above-mentioned problems, the present invention is proposed.

[0010] The technical problem addressed by this invention is that existing technologies often rely on manual rules or traditional data verification methods to detect data anomalies, which are unable to adapt to large-scale, dynamically changing data environments. This is particularly true in new energy companies, which involve a large amount of equipment, sensors, projects, and personnel data. The processing of anomaly data is cumbersome and error-prone, making it difficult to automate and efficiently process it.

[0011] Existing master data management systems typically perform data validation based on simple rules. This approach cannot deeply verify the logical relationships between complex data. For example, data associations between devices and projects, or the logical consistency between device status and temperature data, cannot guarantee data integrity and consistency.

[0012] While traditional methods can repair abnormal data using rules, these methods rely on manual intervention and predefined rule bases, making them ineffective for handling new types of abnormal data or complex anomalies that cannot be defined using rules. As data scale and types grow, the efficiency and accuracy of existing repair methods decline.

[0013] Existing methods rely on fixed rules for data repair, but are unable to adapt to the processing of unknown anomaly data. For example, when faced with new devices, new sensors, or new business scenarios, the rule base may not cover all possible anomaly types, resulting in inaccurate repairs or omissions.

[0014] To solve the above technical problems, the present invention provides the following technical solutions: a new energy enterprise master data management method based on artificial intelligence, comprising:

[0015] Collect master data of new energy enterprises;

[0016] Perform data verification on the collected master data and mark abnormal data in the master data;

[0017] Perform intelligent error correction on data marked as abnormal and automatically repair erroneous data.

[0018] As a preferred solution of the artificial intelligence-based new energy enterprise master data management method of the present invention, wherein: the new energy enterprise master data includes equipment master data, project master data, personnel master data, and customer and supplier data;

[0019] The collection of new energy enterprise master data includes offline import, API interface collection, database extraction and ETL tools.

[0020] As a preferred solution of the master data management method for new energy enterprises based on artificial intelligence described in the present invention, wherein: the data verification includes constructing a data set D = {X1, X2, ..., X n}, where each X i Represents a data item;

[0021] Use deep learning models to train each data item X in the dataset i Perform anomaly detection, mark potential abnormal data, and use the trained deep learning model f dl(X), predict the data set and output the abnormal probability of each data item, expressed as,

[0022] y i =f dl (X i )where y i ∈{0,1}

[0023] Among them, f dl Represents a deep learning model;

[0024] If y i =1 means X i is the potential abnormal data, y i =0 means X i is normal data;

[0025] Potential anomaly dataset D potential Contains all data items marked as abnormal, that is, all y i =1 data item, expressed as,

[0026] D potential ={X i ∣y i =1,i=1,2,…,n}

[0027] In the potential anomaly dataset D potential In the data relationship diagram, node v i Represents data item X i , side e ij Represent relationships between data items;

[0028] The data relationship diagram is represented as follows:

[0029] G=(V,E)

[0030] Where V={v1,v2,…,v m} represents all nodes, each node corresponds to a data item X i ,E={e ij} represents an edge, e ij Represents node v i With v j the relationship between;

[0031] The graph features are updated through the graph convolutional network model GCN. Through GCN, the features of each node are updated to obtain a new node representation. Use the updated node features to perform anomaly judgment.

[0032] As a preferred solution of the artificial intelligence-based new energy enterprise master data management method of the present invention, the graph convolutional network model GCN is represented as follows:

[0033]

[0034] in, is node v i In the feature representation of the kth layer, A ij is an element of the adjacency matrix, representing the node v i and v j The relationship between is node v i The set of neighbor nodes, W (k) is the weight matrix of the kth layer, c ij is the normalization constant, and σ(·) is the activation function.

[0035] As a preferred solution of the artificial intelligence-based new energy enterprise master data management method of the present invention, the abnormality judgment includes applying a classification algorithm to classify each node and outputting whether it is abnormal data, which is expressed as:

[0036]

[0037] Among them, f class Represents a classification model;

[0038] like Represents node v i The corresponding data item X i It is abnormal data. Represents node v i The corresponding data item X i It is normal data;

[0039] Construct anomaly dataset D based on the final anomaly data final_abnormal , which contains all data items that are judged to be abnormal, expressed as,

[0040]

[0041] As a preferred solution of the artificial intelligence-based master data management method for new energy enterprises described in the present invention, the intelligent error correction includes extracting historical data of the master data, training a machine learning model, and classifying abnormal data items as rule-repairable anomalies or inference-repairable anomalies; and using the trained machine learning model to predict data items marked as abnormal.

[0042] As a preferred solution of the artificial intelligence-based master data management method for new energy enterprises described in the present invention, if the abnormal data is determined to be a rule-correctable abnormality, the predefined business rules are applied to perform the repair;

[0043] According to the characteristics of abnormal data items, the specific type of abnormal data is identified, and for each abnormal data item, the appropriate repair method is automatically selected according to business rules.

[0044] As a preferred solution of the artificial intelligence-based master data management method for new energy enterprises described in the present invention, if the abnormal data is determined to be an inferred repair abnormality, the most reasonable repair value is determined by calculating multiple data items in the historical data that are most similar to the data item to be repaired based on the pattern of similar data in the historical data;

[0045] The repair value is calculated by using a K-nearest neighbor algorithm to calculate similarity, and K data items similar to the repair value are selected from the historical data to generate the repair value.

[0046] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned artificial intelligence-based new energy enterprise master data management method are implemented.

[0047] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned artificial intelligence-based new energy enterprise master data management method.

[0048] Beneficial effects of the present invention: The artificial intelligence-based master data management method for new energy enterprises provided by the present invention improves the automation and intelligence level of data verification and repair. Through deep learning and graph neural network technology, efficient abnormal data detection and consistency verification are achieved, solving the problem of traditional methods relying on manual intervention. A machine learning model is used to automatically classify abnormal data into "rule-repairable" or "inference repair" types, and repairs are performed through predefined business rules and similar data inference algorithms respectively. It can flexibly respond to various abnormal situations, improve repair accuracy, significantly improve data quality and consistency, provide a reliable data foundation for new energy enterprises, support more accurate decision-making and operation management, and improve overall efficiency and management capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 This is an overall flow chart of a new energy enterprise master data management method based on artificial intelligence provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0051] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0053] Example 1

[0054] Reference Figure 1 , which is an embodiment of the present invention, provides a new energy enterprise master data management method based on artificial intelligence, including:

[0055] Collect master data of new energy enterprises;

[0056] Perform data verification on the collected master data and mark abnormal data in the master data;

[0057] Perform intelligent error correction on data marked as abnormal and automatically repair erroneous data.

[0058] The new energy enterprise master data includes equipment master data, project master data, personnel master data, and customer and supplier data;

[0059] The collection of new energy enterprise master data includes offline import, API interface collection, database extraction and ETL tools.

[0060] In the method of the present invention, the master data mainly includes the following: Equipment master data: including key information such as the equipment's unique code, model, specifications, installation location, measurement unit, equipment classification, and equipment maintenance cycle.

[0061] Project master data: covers project name, project progress, site information, budget, personnel allocation, project leader and other information.

[0062] Personnel master data: including personnel name, position, department, qualification certificate and other information.

[0063] Customer and supplier data: includes basic information about customers and suppliers, such as company name, contact person, contract records, cooperation history, etc.

[0064] Master data is collected mainly through the following technical means:

[0065] Offline import: Use data import tools to import data from spreadsheets (such as Excel) or CSV files into the master data management system.

[0066] APIs: Master data is retrieved from other systems or databases through standard RESTful APIs, ensuring standardized and efficient data transfer. APIs enable cross-system data integration and automated acquisition of relevant data on equipment, projects, or personnel.

[0067] Database extraction: Extract required master data directly from existing enterprise databases through SQL query statements to ensure data integrity and real-time performance.

[0068] ETL tools: Use ETL tools to extract master data from multiple data sources, and after necessary conversion and cleaning operations, load it into the master data management system to ensure data consistency and accuracy.

[0069] The frequency and cycle of data collection are determined by the nature of the master data. Equipment and project data are typically static and are typically synchronized regularly via batch import or APIs. Personnel data, customer, and supplier information have longer update cycles and can be collected manually or automatically.

[0070] The collected data must follow preset format specifications. For example, the equipment code must be unique and meet format requirements, and personnel information should include key information such as position and qualification certificates.

[0071] During the collection process, the reliability of the data source should be ensured, and API interfaces and database extraction should be used first to avoid data inconsistencies caused by manual import.

[0072] The data verification includes constructing a data set D={X1, X2, ..., X n}, where each X i Represents a data item;

[0073] Use deep learning models to train each data item X in the dataset i Perform anomaly detection, mark potential abnormal data, and use the trained deep learning model f dl (X), predict the data set and output the abnormal probability of each data item, expressed as,

[0074] y i =f dl (X i )where y i ∈{0,1}

[0075] Among them, f dl Represents a deep learning model;

[0076] If y i =1 means X i is the potential abnormal data, y i =0 means X i is normal data;

[0077] Potential anomaly dataset D potential Contains all data items marked as abnormal, that is, all y i =1 data item, expressed as,

[0078] D potential ={X i ∣y i =1,i=1,2,…,n}

[0079] In the potential anomaly dataset D potential In the data relationship diagram, node v i Represents data item X i , side e ij Represents the relationship between data items, such as equipment and projects, equipment and personnel, etc. When constructing a graph, the characteristics between nodes are the characteristics of the data items, and the edges represent the logical relationship between data items;

[0080] The data relationship diagram is represented as follows:

[0081] G=(V,E)

[0082] Where V={v1,v2,…,v m} represents all nodes, each node corresponds to a data item X i ,E={e ij} represents an edge, e ij Represents node v i With v j the relationship between;

[0083] The graph features are updated through the graph convolutional network model GCN. Through GCN, the features of each node are updated to obtain a new node representation. Use the updated node features to perform anomaly judgment.

[0084] The graph convolutional network model GCN is expressed as,

[0085]

[0086] in, is node v i In the feature representation of the kth layer, A ij is an element of the adjacency matrix, representing the node v i and v j The relationship between is node v i The set of neighbor nodes, W (k) is the weight matrix of the kth layer, c ij is the normalization constant, and σ(·) is the activation function.

[0087] The abnormality judgment includes applying a classification algorithm (such as a support vector machine or a neural network) to classify each node and outputting whether it is abnormal data, which is expressed as:

[0088]

[0089] Among them, f class Represents a classification model;

[0090] like Represents node v i The corresponding data item X i It is abnormal data. Represents node v i The corresponding data item X i It is normal data;

[0091] Construct anomaly dataset D based on the final anomaly data final_abnormal , which contains all data items that are judged to be abnormal, expressed as,

[0092]

[0093] The intelligent error correction includes extracting historical data of master data, training a machine learning model, and classifying abnormal data items as rule-repairable anomalies or inference-repairable anomalies. During the training process, the model's structure, parameters, loss function, and algorithm are optimized to ensure that the model can identify and predict whether the abnormal data belongs to the "rule-repairable" or "inference-repairable" type; the trained machine learning model is used to predict data items marked as abnormal. For example, if the temperature data of a device is abnormal, the system can perform inference repair through the temperature data of similar devices.

[0094] By extracting historical data from master data and preprocessing it, missing values, outliers, and noise in the data are addressed to ensure data quality and consistency. Missing values ​​can be filled using appropriate methods, such as mean or median imputation, and outliers are detected and corrected or marked using statistical analysis methods (such as Z-score and IQR). After data standardization or normalization, the data is converted into a format suitable for machine learning model input. Through feature engineering, features related to abnormal data, such as historical equipment usage time, equipment status, environmental data, and time series features, are selected to construct a feature set. During this process, the features most relevant to anomaly detection and data repair are selected, and redundant and irrelevant features are removed.

[0095] The processed dataset is used to train the machine learning model. Appropriate algorithms, such as random forests, support vector machines, or gradient boosting trees, are selected to train the model to classify data items and determine which data items may be "rule-fixable anomalies" or "inference-fixable anomalies." During model training, the model's generalization ability is ensured by dividing the dataset into training and validation sets, and cross-validation is used. The model's hyperparameters are tuned, such as the tree's depth, number of branches, kernel function type, and C parameter, to optimize model performance and ensure it can identify and predict the type of abnormal data. Using the cross-entropy loss function as the optimization objective, the model is trained using methods such as gradient descent until optimal predictive performance is achieved. For the trained model, its performance on the validation set is evaluated, and the optimal model is selected using metrics such as accuracy, F1 value, and ROC curve.

[0096] After training is complete, the model is used to predict data items marked as anomalies. The model outputs whether each data item is a "rule-fixable anomaly" or "inference-fixable anomaly."

[0097] If the abnormal data is judged to be rule-correctable, the predefined business rules are applied to perform the repair;

[0098] According to the characteristics of abnormal data items, the specific type of abnormal data is identified, and for each abnormal data item, the appropriate repair method is automatically selected according to business rules.

[0099] For rule-repairable anomaly data items, a repair method based on predefined business rules is used. First, by analyzing the attributes of the data item (for example, device code, project number, personnel ID, etc.), the anomaly type (such as device code error, missing data, etc.) is identified. If it is determined to be rule-repairable, the system will perform the following repair steps:

[0100] Abnormal data items are flagged using deep learning or other machine learning methods, and the type of anomaly determines whether they are rule-fixable. Rule-fixable data items typically have clear repair criteria, such as device encoding errors that can be repaired by looking up registration information.

[0101] For abnormal data items with device coding errors, the erroneous data can be corrected by querying the device registration information (such as a database or external system). The specific steps are:

[0102] Query the device registration information database to find the device's unique identifier (such as device ID).

[0103] Get the exact encoding information associated with the device ID.

[0104] Correct the abnormal data items with incorrect device codes to the correct device codes.

[0105] The repaired data items are returned to the master data management system and updated with the latest repair data.

[0106] The repaired data items will be verified again to confirm that they comply with predefined rules and data standards to ensure data consistency and correctness.

[0107] If the abnormal data is judged to be an inferred repair anomaly, the most reasonable repair value is determined by calculating the multiple data items in the historical data that are most similar to the data item to be repaired based on the pattern of similar data in the historical data;

[0108] The repair value is calculated by using a K-nearest neighbor algorithm to calculate similarity, and K data items similar to the repair value are selected from the historical data to generate the repair value.

[0109] First, by selecting an appropriate similarity metric (such as Euclidean distance or Manhattan distance), the similarity between the data item to be repaired and other data items in the historical data is calculated. The primary purpose of this step is to determine which historical data items are closest to the abnormal data to be repaired. The choice of similarity metric is closely related to the nature and characteristics of the data. In this invention, calculating the similarity of data items is a key step in ensuring the accuracy of the repair.

[0110] Next, an appropriate K value is selected, i.e., the K data items that are most similar to the data item to be repaired are selected from all historical data items (K is a pre-set parameter, such as K = 3). The choice of K value should be based on the characteristics and distribution of the dataset and can be optimized through cross-validation. The setting of K value directly affects the accuracy of the final repair result. By calculating the similarity of these K data items and selecting the K data items with the smallest distance, these most similar historical data provide a reference for repair.

[0111] Next, the Euclidean distance is calculated to measure the similarity between the data item to be repaired and the historical data item. Specifically, for the data item to be repaired X i =(x i1 ,x i2 ,…,x in ) and historical data item X j =(x j1 ,x j2 ,…,x jn ), the Euclidean distance between them d(X i ,X j ) can be calculated using the following formula:

[0112]

[0113] Among them, X i Indicates the abnormal data item to be repaired, X jRepresents historical data items, x ik Indicates the data item X to be repaired i The value of the kth feature, x jk Represents historical data item X j The value of the kth feature, d(X i ,X j ) represents the Euclidean distance between the data item to be repaired and the historical data item, and n represents the number of features of the data item.

[0114] Next, the weighted average method is used to generate the repair value. The weighted average method considers the contribution of each similar data item to the repair value, which is usually inversely proportional to the similarity between the data items. Specifically, given K data items similar to the data item to be repaired, the repair value is generated by calculating the weighted average of these data items. The formula is as follows:

[0115]

[0116] Among them, R i Indicates the value of the repaired data item, w j Represents data item X j The weight of x j Represents historical data item X j The value of , K represents the number of the K most similar historical data items selected.

[0117] Weight w j This is calculated based on the distance between data items. Smaller distances indicate more similar data items, resulting in a greater weight and contribution. Finally, the repaired data item replaces the erroneous value in the abnormal data item. If the repaired data value meets the data specification and is consistent with the historical data pattern, the repair is considered successful.

[0118] Example 2

[0119] An embodiment of the present invention is different from the previous embodiment in that:

[0120] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0121] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0122] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0123] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0124] Example 3

[0125] As an embodiment of the present invention, a master data management method for new energy enterprises based on artificial intelligence is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0126] The purpose of the experiment was to compare the method presented in this paper with traditional manual rule-based repair methods, evaluating the differences in data repair accuracy, efficiency, and system stability, thereby validating the superiority of the present invention's technical solution. During the experiment, a master dataset of 10,000 data items from a new energy enterprise was selected, encompassing master data on equipment, projects, and personnel. This dataset contained a certain proportion of abnormal data items, simulating common equipment coding errors, project association anomalies, and personnel responsibility assignment errors.

[0127] In the experiment, the dataset was first processed. The traditional method group employed manual rule-based repair, with staff inspecting and repairing data items based on the rules contained in them. Staff manually searched device registration information to correct errors such as device coding errors, and also handled other types of abnormal data based on empirical rules. Each repaired data item required manual intervention, making the repair process time-consuming.

[0128] At the same time, the method group of the present invention uses a deep learning model based on artificial intelligence to detect data anomalies. First, historical data is trained through a deep learning model to learn the characteristics and abnormal patterns of the data and automatically identify abnormal items in the data. The trained model can predict the repair method based on the marked abnormal data items. For abnormal data items determined to be "rule-repairable", the system automatically calls predefined business rules to repair them and quickly correct the errors. For abnormal data items that are "inferred to be repaired", the system selects the historical data that is most similar to the abnormal data through the KNN algorithm or regression model, and generates reasonable repair values ​​based on the patterns of similar data.

[0129] During the experiment, we primarily recorded data on repair accuracy, repair efficiency, and system stability. These data reflect the different performance of each method in handling abnormal data. The experimental results are shown in Table 1.

[0130] Table 1 Comparison of experimental results

[0131] index Traditional method group The method of the present invention Repair accuracy 82％ 96％ Repair efficiency (minutes) 120 40 System stability (fluctuation %) 25％ 5％

[0132] Analysis of experimental data reveals that traditional methods, while capable of repairing data in some cases, suffer from significant shortcomings due to their heavy reliance on manual rules. The manual repair process is complex and time-consuming, requiring manual intervention for each data item, resulting in low overall repair efficiency. Furthermore, due to their reliance on rules and human experience, the accuracy of repair results is insufficient, making it prone to incorrect or missed repairs. System stability is also limited by the volatility of manual operation.

[0133] The method of the present invention significantly improves the accuracy and efficiency of data repair by combining deep learning and graph neural network (GNN) models. The model can quickly detect abnormal data by automatically learning the patterns and characteristics of historical data, and automatically select the appropriate repair method through the classification system. For "rule-repairable" abnormal data, the system automatically corrects the error through predefined business rules, avoiding the complexity and uncertainty of manual intervention. For "inferred repair" abnormal data, the KNN algorithm or regression model is used to infer similar patterns based on historical data to generate reasonable repair values. Through this intelligent repair method, the limitations of manual rules are greatly reduced and the accuracy of repair is improved.

[0134] This improvement not only enhances repair accuracy but also significantly improves repair efficiency, reducing manual operation time. System stability is also ensured, as human interference is significantly reduced, and the data repair process is more standardized and consistent. These technical improvements enable the method of the present invention to repair abnormal data more efficiently and accurately in practical applications, offering significant advantages over traditional methods.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. The master data management method for new energy enterprises based on artificial intelligence is characterized by: include: Collect master data of new energy enterprises; Perform data verification on the collected master data and mark abnormal data in the master data; Perform intelligent error correction on data marked as abnormal and automatically repair erroneous data.

2. The artificial intelligence-based new energy enterprise master data management method according to claim 1, characterized in that: The new energy enterprise master data includes equipment master data, project master data, personnel master data, and customer and supplier data; The collection of new energy enterprise master data includes offline import, API interface collection, database extraction and ETL tools.

3. The artificial intelligence-based new energy enterprise master data management method according to claim 2, characterized in that: The data verification includes constructing a data set D=X1, X2, ..., X n }, where each X i Represents a data item; Use deep learning models to train each data item X in the dataset i Perform anomaly detection, mark potential abnormal data, and use the trained deep learning model f dl (X), predict the data set and output the abnormal probability of each data item, expressed as, y i =f dl (X i )where y i ∈{0,1} Among them, f dl Represents a deep learning model; If y i =1 means X i is the potential abnormal data, y i =0 means X i is normal data; Potential anomaly dataset D potential Contains all data items marked as abnormal, that is, all y i =1 data item, expressed as, D potential ={X i ∣y i =1,i=1,2,…,n} In the potential anomaly dataset D potential In the data relationship diagram, node v i Represents data item X i , side e ij Represent relationships between data items; The data relationship diagram is represented as follows: G=(V,E) Where V={v1,v2,…,v m } represents all nodes, each node corresponds to a data item X i ,E={e ij } represents an edge, e ij Represents node v i With v j the relationship between; The graph features are updated through the graph convolutional network model GCN. Through GCN, the features of each node are updated to obtain a new node representation. Use the updated node features to perform anomaly judgment.

4. The artificial intelligence-based new energy enterprise master data management method according to claim 3, characterized in that: The graph convolutional network model GCN is expressed as, in, is node v i In the feature representation of the kth layer, A ij is an element of the adjacency matrix, representing the node v i and v j The relationship between is node v i The set of neighbor nodes, W (k) is the weight matrix of the kth layer, c ij is the normalization constant, and σ(·) is the activation function.

5. The artificial intelligence-based new energy enterprise master data management method according to claim 4, characterized in that: The abnormality judgment includes applying a classification algorithm to classify each node and outputting whether it is abnormal data, which is expressed as: Among them, f class Represents a classification model; like Represents node v i The corresponding data item X i It is abnormal data. Represents node v i The corresponding data item X i It is normal data; Construct anomaly dataset D based on the final anomaly data final_abnormal , which contains all data items that are judged to be abnormal, expressed as, 6. The artificial intelligence-based new energy enterprise master data management method according to claim 5, characterized in that: The intelligent error correction includes extracting historical data of master data, training a machine learning model, and classifying abnormal data items as rule-correctable anomalies or inference-correctable anomalies; Use the trained machine learning model to make predictions on data items marked as anomalies.

7. The artificial intelligence-based new energy enterprise master data management method according to claim 6, characterized in that: If the abnormal data is judged to be rule-correctable, the predefined business rules are applied to perform the repair; According to the characteristics of abnormal data items, the specific type of abnormal data is identified, and for each abnormal data item, the appropriate repair method is automatically selected according to business rules.

8. The artificial intelligence-based new energy enterprise master data management method according to claim 7, characterized in that: If the abnormal data is judged to be an inferred repair anomaly, the most reasonable repair value is determined by calculating the multiple data items in the historical data that are most similar to the data item to be repaired based on the pattern of similar data in the historical data; The repair value is calculated by using a K-nearest neighbor algorithm to calculate similarity, and K data items similar to the repair value are selected from the historical data to generate the repair value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the new energy enterprise master data management method based on artificial intelligence are implemented as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the new energy enterprise master data management method based on artificial intelligence are implemented as described in any one of claims 1 to 8.

Images