An artificial intelligence-based data governance and material intelligent management method

Abstract

The application discloses a data governance and material intelligent management method based on artificial intelligence, and belongs to the technical field of intelligent data governance and material management; the method comprises the following steps: collecting multi-source heterogeneous material data and dividing the material data into structured and unstructured data; analyzing the unstructured data, standardizing all the data, and generating standardized material data conforming to a master data model; extracting historical behavior change features from the standardized material data to construct a historical feature vector set, so as to train a machine learning anomaly detection model; obtaining current behavior change features of material to be detected to construct a feature vector to be detected, inputting the feature vector to the model to obtain an anomaly score; when the anomaly score exceeds a threshold value and the state change satisfies a preset logical conflict judgment condition based on the standardized material data, it is determined that the data is abnormal, and an alarm is generated; the application breaks through the limitation of traditional static rule verification, and realizes intelligent early warning and accurate risk control on the dynamic flow track and business irregularity of the material.

Description

Technical Field

[0001] This invention belongs to the field of intelligent data governance and materials management technology, specifically relating to an artificial intelligence-based data governance and intelligent materials management method. Background Technology

[0002] In recent years, with the continuous expansion of the scale of materials and the increasing complexity of application scenarios, materials have become more diverse, numerous, and dispersed. Traditional material management models relying on manual statistics or single information systems are no longer sufficient to meet the needs of refined and intelligent management. Especially in cross-departmental, cross-regional, or emergency support scenarios, problems such as untimely updates of material information, inaccurate data, and unclear management responsibilities have become increasingly prominent, severely restricting the unified allocation and efficient utilization of materials. Therefore, researching a material data governance and intelligent management technology based on artificial intelligence algorithms, and introducing intelligent analysis methods such as deep learning and machine learning to achieve automatic governance, precise positioning, and dynamic updating of material data, has become a key technological direction for improving the level of intelligent material management and supporting unified material allocation and scientific decision-making.

[0003] Currently, some scholars have proposed intelligent data governance and management methods, among which the technical solutions that are relatively close to this invention include: Patent application (application number: 202511293585.5, title: A ship inspection data intelligent management system and method based on structured storage) provides a ship inspection data intelligent management system and method based on structured storage, which obtains inspection documents based on information such as ship number, parses the inspection documents, and constructs structured data. However, the data format processed by this method is singular and does not target multimodal or other multi-source heterogeneous data. The patent application (application number: 202511051357.7, title: A Rule-Based Dynamic Collection and Intelligent Management System for E-commerce Data) proposes a rule-based dynamic collection and intelligent management system for e-commerce data, including rule subscription, a collection layer, and a governance layer. However, it does not provide intelligent analysis of the data after it is entered into the database, or timely monitoring of the product status.

[0004] Patent application (202411323677.9, title: A data governance management system) discloses a data governance management system, including a data acquisition module, an analysis module, an encryption module and a storage management module, but it does not automatically identify and correct abnormal data, and cannot guarantee the accuracy of the data; The literature (Wang Zhuang, Zhou Wenli, Li Chenhao, et al. Design and application of master data management platform for construction enterprises [J]. Science and Technology Innovation and Application, 2026, 3: 114-117.) proposes a master data management platform based on service-oriented architecture to realize core data interaction and sharing across systems, ensuring data accuracy and consistency. However, it does not mention the inspection and handling of abnormal data.

[0005] In summary, current data governance and management methods have the following shortcomings: (1) lack of unified governance of multi-source heterogeneous data, and cannot handle the situation of scattered data sources; (2) lack of automatic identification and correction mechanism for abnormal data, and data updates are lagging behind; (3) difficulty in quickly locating the status of materials and grasping the total amount, location and responsible person of materials. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide a data governance and intelligent material management method based on artificial intelligence, thereby solving the problems in existing technologies.

[0007] The objective of this invention can be achieved through the following technical solutions: A data governance and intelligent materials management method based on artificial intelligence includes: Collect heterogeneous material data from multiple sources, and divide the material data into structured data and unstructured data according to the data structure; The unstructured data is parsed, and the parsed data and the structured data are standardized according to a preset standard to generate standardized material data that conforms to the material master data model. The standardized material data is used to extract behavioral change features over a historical period, construct a historical feature vector set, and train a machine learning anomaly detection model. The current behavioral change features of the material to be detected are obtained to construct a feature vector to be detected. The feature vector to be detected is input into the trained anomaly detection model to obtain the anomaly score output by the anomaly detection model. When the abnormal score exceeds the preset score threshold, and the status change of the material to be tested is verified based on the standardized material data to meet the preset logical conflict judgment condition, it is determined that the data of the material to be tested is abnormal and an alarm message is generated.

[0008] Furthermore, the unstructured data includes text description data, log data, and image data; The process of parsing the unstructured data includes: Extract keywords from the text description data and log data, and map the extracted keywords to the corresponding fields of the material master data model; An image recognition model is used to identify material identifiers, text information, or location markers in the image data and convert them into corresponding structured field information.

[0009] Furthermore, the material master data model includes the following feature fields: unique material identifier, material type, material quantity, material status, material location information, management unit, and responsible person.

[0010] Furthermore, the behavioral change characteristics include: quantity change characteristics, update frequency characteristics, transfer behavior characteristics, and location change characteristics; The behavioral change characteristics are obtained in the following ways: The quantity change characteristics are obtained by calculating the difference in the quantity of materials within adjacent time periods; The update frequency feature is obtained by calculating the number of information updates per unit time. The characteristics of the allocation behavior are obtained based on statistics of historical allocation records in the standardized material data; The location change characteristics are obtained by calculating the number of times the material's location changes or the distance of the change.

[0011] Furthermore, the anomaly detection model uses the mean squared error loss function as the reconstruction loss function: in, Represents the original sample input features. This represents the reconstructed feature vector output by the output layer. N This indicates the number of samples; after the model training is complete, the anomaly score is calculated based on the reconstruction error between the input features and the reconstructed features. in, This indicates an abnormal score.

[0012] Furthermore, the preset logical conflict determination conditions include any one or more of the following: If there is no corresponding allocation or usage record in the standardized material data, the quantity of the material to be tested may increase or decrease. The data update time of the materials to be tested does not match the time of the corresponding allocation record or usage record; The location of the materials to be tested does not match the corresponding allocation record information.

[0013] Furthermore, the method also includes: Location information of similar materials is extracted from the standardized material data, and the location information is mapped to a unified spatial coordinate system to form material spatial distribution data; The spatial region is divided into multiple preset grid units, with each grid unit corresponding to a spatial region; Count the quantity of materials in each grid cell and calculate the material distribution density of the corresponding grid area; The material distribution density is expressed as: in, Indicates the first k Material distribution density in each grid area Indicates the first k The quantity of materials in each grid area Indicates the first k The spatial area of ​​each grid region; By calculating the material distribution density in each grid region, a spatial density matrix is ​​formed: in, M Represents the thermal matrix of spatial distribution of materials. n Indicates the number of grid regions; The spatial density matrix is ​​analyzed using clustering analysis algorithms to obtain the spatial distribution status of materials. The output results of the spatial distribution status include: region label, region material density value, and region status category.

[0014] Furthermore, after determining that the data of the material to be tested is abnormal and generating an alarm message, the method further includes: synchronously and associatedly storing the standardized material data, the generated alarm message, and the historical behavior status change records of the material into a preset material status data pool.

[0015] An artificial intelligence-based data governance and intelligent materials management system, which executes the above method, includes: Data acquisition module: Collects heterogeneous material data from multiple sources, and divides the material data into structured data and unstructured data according to the data structure; Standardization processing module: parses the unstructured data and standardizes the parsed data and the structured data according to preset standards to generate standardized material data that conforms to the material master data model; Anomaly handling module: Extracts behavioral change features within historical time periods from the standardized material data, constructs a historical feature vector set, and trains a machine learning anomaly detection model; obtains the current behavioral change features of the material to be detected to construct a feature vector to be detected, inputs the feature vector to be detected into the trained anomaly detection model, and obtains the anomaly score output by the anomaly detection model; when the anomaly score exceeds a preset score threshold, and the state change of the material to be detected is verified based on the standardized material data to meet a preset logical conflict judgment condition, it determines that the data of the material to be detected is abnormal and generates an alarm message.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for data governance and intelligent management of materials based on artificial intelligence.

[0017] The beneficial effects of this invention are: 1. This invention constructs a multi-source material data acquisition module to classify data from different sources such as business systems, manual forms, system images, and logs. Based on this, a data standardization processing module is used to transform unstructured data into structured fields, and a material master data model is constructed according to unified measurement, coding, and naming specifications. This enables the transformation of originally messy multimodal data into a standardized and consistent expression, breaking down information silos and providing standardized and reliable underlying data input for subsequent intelligent scheduling.

[0018] 2. This invention introduces a spatial distribution density analysis technique. By mapping the location information of similar materials to a spatial coordinate system and combining spatial algorithms such as cluster analysis and density analysis, it can transform planar tabular data into a spatial distribution state with geographical dimensions. This helps managers to clearly grasp the specific storage location, distribution density, and ownership of materials, and realizes dynamic perception of the material status.

[0019] 3. This invention constructs a material status data pool at the end, which not only integrates basic material information and abnormal alarm data, but also specifically includes historical change data on material quantity, location, status, and management entity changes. This data structure, which combines static slices with dynamic change trajectories, endows the system with historical traceability capabilities, thereby providing comprehensive status data support with high consistency and reliability for unified allocation and decision-making applications at the upper level. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a framework diagram of the data governance and intelligent materials management system based on artificial intelligence, as presented in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0023] Example 1 A data governance and intelligent materials management method based on artificial intelligence includes the following steps: Step 1: First, construct a multi-source material data acquisition module. This module collects material data from multiple different sources through interface calls, file uploads, manual data entry, and automatic system collection. Specifically, this includes: manually filled-out form data, structured file data uploaded by business or management systems, image data uploaded by the system, system-generated log data, and historical material ledger data. The collected data then undergoes preliminary classification and processing.

[0024] The preliminary classification of the collected data refers to classifying the data according to its data structure, dividing the data into structured data and unstructured data. Structured data includes tabular data and coded data, while unstructured data includes text description data, log data, and image data.

[0025] Step 2: Construct a data standardization processing module to perform structured processing on the data classified in Step 1, and standardize the material data field information according to the pre-defined unified standards to generate standardized structured material data and construct the material master data model.

[0026] The data structuring process in step 2 refers to the structuring of unstructured data, specifically including: keyword extraction and field mapping for text description data and log data, and information recognition and conversion of image data into structured field information.

[0027] Keyword extraction and field mapping for text description data and log data involves using keyword extraction methods based on statistical features, rule matching, or machine learning models to identify keywords related to material attributes in the text. In this embodiment, this includes information such as material name, quantity, status, location, and management unit. The identified keywords are then mapped to predefined material fields, thereby converting the text description data and log data into corresponding structured field data.

[0028] Information recognition of image data includes: using an image recognition model to recognize information in the image data, wherein the image recognition model is used to identify material identification, text information or location markers in the image and output the corresponding recognition results.

[0029] Step 2's image recognition model includes a feature extraction unit, a target detection unit, and an information recognition unit. The feature extraction unit uses a convolutional neural network to extract convolutional features from the image data, obtaining image feature information. The target detection unit uses a target detection network to locate material identification areas, text areas, or location marker areas in the image based on the image feature information. The information recognition unit uses an OCR text recognition network to recognize text in the target areas, outputting the corresponding material identification information, text information, or location information recognition results.

[0030] The unified standards in step 2 include: material classification and coding rules, quantity measurement unit specifications, field naming rules, and data format specifications.

[0031] The standardization process in step 2 includes: standardizing material codes, standardizing quantity measurement units, and standardizing field naming and data formats.

[0032] The material master data model constructed in step 2 is represented as follows: Where ID represents the unique identifier of the material, T represents the material type, Q represents the material quantity, S represents the material status, L represents the material location information, U represents the management unit, and R represents the responsible person. The material status S includes in-stock status, usage status, allocation status, and scrapped status.

[0033] Step 3: Construct a data anomaly processing module, train an anomaly detection model using historical material data, use the anomaly detection model to detect anomalies in the material data, output anomaly scores, classify anomalies according to the score thresholds, and issue alarm information for abnormal data. Step 3, training the anomaly detection model using historical material data, requires first constructing a historical dataset based on that data. This historical dataset consists of multiple samples, each corresponding to the status information of a material within a specific time period. The sample features are represented as follows: Where i represents the i-th sample, Indicates the characteristics of quantity changes. Indicates the update frequency. This indicates the characteristics of the transfer behavior. It indicates the characteristics of positional changes.

[0034] The quantity change feature is obtained by calculating the quantity difference between adjacent time periods, the update frequency feature is obtained by calculating the number of updates per unit time, the transfer behavior feature is obtained by statistical analysis of transfer records, and the location change feature is obtained by calculating the number of location changes or the distance of the change.

[0035] In step 3, when constructing the historical dataset, it is preferable to extract historical material data without anomalies that have been manually verified or validated by business rules as a normal sample set for training the anomaly detection model.

[0036] In step 3, the samples in the historical dataset are normalized using Min-Max normalization before being input into the anomaly detection model. The normalization process can be represented as follows: in Represents the features of the original sample. This represents the minimum value of the corresponding eigenvalue. This represents the maximum value of the corresponding feature. This represents the normalized eigenvalues.

[0037] The anomaly detection model used in step 3 is a deep autoencoder network based on a multilayer perceptron.

[0038] The deep autoencoder network comprises an input layer, an encoding layer, a hidden layer, a decoding layer, and an output layer. The input layer receives the four-dimensional sample feature vector; the encoding layer performs low-dimensional compression representation of the input features; the hidden layer extracts the potential correlations between resource state features; the decoding layer reconstructs the low-dimensional features; and the output layer outputs the reconstructed feature vector. During model training, the mean squared error loss function is used as the reconstruction loss function. in Represents the original sample input features. This represents the reconstructed feature vector output by the output layer. N Indicates the number of samples.

[0039] After the model training is completed, the anomaly score is calculated based on the reconstruction error between the input features and the reconstructed features: in, This indicates an abnormal score.

[0040] In step 3, the scoring threshold is set to Th. When an abnormal score exceeds the scoring threshold, the corresponding data is judged as abnormal data. Based on the specific manifestations of the abnormal characteristics, the abnormal data is divided into: quantity abnormality, update abnormality, and logical conflict abnormality.

[0041] The criteria for determining abnormal quantities are: 3.1) The change in the number of samples exceeds the preset threshold Tnum; 3.2) The number of samples increased or decreased in the absence of corresponding allocation or usage records.

[0042] The update exception determination condition is: 3.3) The difference between the sample data update frequency and the historical average update frequency is greater than the threshold Tfreq; 3.4) The sample data update time does not match the corresponding allocation or usage record time.

[0043] The conditions for determining logical conflict exceptions are: 3.5) The changes in sample quantity are inconsistent with the allocation or usage records; 3.6) The change in sample location does not match the corresponding transfer record.

[0044] In step 3, alarm information for abnormal data is generated mainly by the data anomaly processing module, and then sent to the management system and relevant responsible parties.

[0045] Step 4: Build a data analysis module to perform spatial distribution analysis on material data and obtain information on the quantity, location and status of materials in a timely manner.

[0046] Spatial distribution analysis of material data includes the following steps: 4.1) Obtain the location information of all materials of the same type, and map the location information to a unified spatial coordinate system to form material spatial distribution data; 4.2) Divide the spatial region into multiple preset grid units, with each grid unit corresponding to one spatial region; 4.2) Based on the quantity of materials and the spatial distribution data of the materials, count the quantity of materials in each grid cell and calculate the material distribution density of the corresponding grid area; The material distribution density can be expressed as: in, Indicates the first k Material distribution density in each grid area Indicates the first k The quantity of materials in each grid area Indicates the first kThe spatial area of ​​each grid region.

[0047] 4.3) By calculating the material distribution density of each grid region, a spatial density matrix is ​​formed: in, M Represents the thermal matrix of spatial distribution of materials. n Indicates the number of grid regions.

[0048] 4.4) The spatial density matrix is ​​analyzed using clustering analysis algorithms to obtain the spatial distribution status of materials. Based on the material distribution density in each region, the spatial distribution status of materials can be divided into: high-density idle areas, high-density reserve areas, low-density shortage areas, and normal distribution areas.

[0049] The output results of spatial distribution status include: region label, region material density value, and region status category.

[0050] Step 5: Construct a material status data pool, including basic material information data, anomaly and alarm data, and historical change data.

[0051] The basic information data of materials in step 5 includes: unique material identifier ID, material type T, material quantity Q, material status S, material location information L, management unit U, and responsible person R; Anomaly and alarm data include: anomaly type information, anomaly score information, anomaly occurrence time information, and alarm status information; Historical change data includes: records of changes in the quantity of materials, records of changes in the location of materials, records of changes in the status of materials, and records of changes in the management entity of materials.

[0052] Example 2 This embodiment proposes an artificial intelligence-based data governance and intelligent materials management system, such as... Figure 1 As shown, it specifically includes: Data acquisition module: Collects heterogeneous material data from multiple sources, and divides the material data into structured data and unstructured data according to the data structure; Standardization processing module: parses the unstructured data and standardizes the parsed data and the structured data according to preset standards to generate standardized material data that conforms to the material master data model; Anomaly handling module: Extracts behavioral change features within historical time periods from the standardized material data, constructs a historical feature vector set, and trains a machine learning anomaly detection model; obtains the current behavioral change features of the material to be detected to construct a feature vector to be detected, inputs the feature vector to be detected into the trained anomaly detection model, and obtains the anomaly score output by the anomaly detection model; when the anomaly score exceeds a preset score threshold, and the state change of the material to be detected is verified based on the standardized material data to meet a preset logical conflict judgment condition, it determines that the data of the material to be detected is abnormal and generates an alarm message.

[0053] Data Analysis Module: Performs spatial distribution analysis on material data to obtain timely information on material quantity, location, and status. Material Status Data Pool: This includes basic material information data, anomaly and alarm data, and historical change data.

[0054] Based on a similar inventive concept, embodiments of the present invention also provide a computer storage medium storing a readable program, which, when run by a processor, can execute the aforementioned artificial intelligence-based data governance and intelligent material management methods.

[0055] Based on a similar inventive concept, this invention provides an electronic device, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus; The memory is used to store at least one executable instruction, which causes the processor to perform the operations corresponding to the above-described artificial intelligence-based data governance and intelligent material management methods.

[0056] Based on a similar inventive concept, embodiments of the present invention also provide a computer program product, including computer instructions, which instruct a computing device to perform the operations corresponding to the above-described artificial intelligence-based data governance and intelligent material management methods.

[0057] The methods of the present invention can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD-ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or as computer code originally stored on a remote recording medium or a non-transitory machine-readable medium and subsequently stored on a local recording medium, downloaded via a network. Thus, the methods described herein can be processed by software stored on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components (e.g., RAM, ROM, flash memory, etc.) capable of storing or receiving software or computer code that, when accessed and executed by the computer, processor, or hardware, implements the methods described herein. Furthermore, when a general-purpose computer accesses the code used to implement the methods shown herein, the execution of the code transforms the general-purpose computer into a dedicated computer for performing the methods shown herein.

[0058] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A data governance and intelligent material management method based on artificial intelligence, characterized in that, include: Collect heterogeneous material data from multiple sources, and divide the material data into structured data and unstructured data according to the data structure; The unstructured data is parsed, and the parsed data and the structured data are standardized according to a preset standard to generate standardized material data that conforms to the material master data model. The standardized material data is used to extract behavioral change features over a historical period, construct a historical feature vector set, and train a machine learning anomaly detection model. The current behavioral change features of the material to be detected are obtained to construct a feature vector to be detected. The feature vector to be detected is input into the trained anomaly detection model to obtain the anomaly score output by the anomaly detection model. When the abnormal score exceeds the preset score threshold, and the status change of the material to be tested is verified based on the standardized material data to meet the preset logical conflict judgment condition, it is determined that the data of the material to be tested is abnormal and an alarm message is generated.

2. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, The unstructured data includes text description data, log data, and image data; The process of parsing the unstructured data includes: Extract keywords from the text description data and log data, and map the extracted keywords to the corresponding fields of the material master data model; An image recognition model is used to identify material identifiers, text information, or location markers in the image data and convert them into corresponding structured field information.

3. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, The material master data model includes the following feature fields: material unique identifier, material type, material quantity, material status, material location information, management unit, and responsible person.

4. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, The behavioral change characteristics include: quantity change characteristics, update frequency characteristics, transfer behavior characteristics, and location change characteristics; The behavioral change characteristics are obtained in the following ways: The quantity change characteristics are obtained by calculating the difference in the quantity of materials within adjacent time periods; The update frequency feature is obtained by calculating the number of information updates per unit time. The characteristics of the allocation behavior are obtained based on statistics of historical allocation records in the standardized material data; The location change characteristics are obtained by calculating the number of times the material's location changes or the distance of the change.

5. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, The anomaly detection model uses the mean squared error loss function as the reconstruction loss function: in, Represents the original sample input features. This represents the reconstructed feature vector output by the output layer. N This indicates the number of samples; after the model training is complete, the anomaly score is calculated based on the reconstruction error between the input features and the reconstructed features. in, This indicates an abnormal score.

6. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, The preset logical conflict determination conditions include any one or more of the following: If there is no corresponding allocation or usage record in the standardized material data, the quantity of the material to be tested may increase or decrease. The data update time of the materials to be tested does not match the time of the corresponding allocation record or usage record; The location of the materials to be tested does not match the corresponding allocation record information.

7. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, Also includes: Location information of similar materials is extracted from the standardized material data, and the location information is mapped to a unified spatial coordinate system to form material spatial distribution data; The spatial region is divided into multiple preset grid units, with each grid unit corresponding to a spatial region; Count the quantity of materials in each grid cell and calculate the material distribution density of the corresponding grid area; The material distribution density is expressed as: in, Indicates the first k Material distribution density in each grid area Indicates the first k The quantity of materials in each grid area Indicates the first k The spatial area of ​​each grid region; By calculating the material distribution density in each grid region, a spatial density matrix is ​​formed: in, M Represents the thermal matrix of spatial distribution of materials. n Indicates the number of grid regions; The spatial density matrix is ​​analyzed using clustering analysis algorithms to obtain the spatial distribution status of materials. Based on the material distribution density in each region, the spatial distribution status of materials is divided into: high-density idle area, high-density reserve area, low-density shortage area, and normal distribution area. The output results of spatial distribution status include: region label, region material density value, and region status category.

8. The data governance and intelligent material management method based on artificial intelligence according to claim 1, characterized in that, After determining that the data of the material to be tested is abnormal and generating an alarm message, the method further includes: synchronously storing the standardized material data, the generated alarm message, and the historical behavior status change records of the material in a preset material status data pool.

9. A data governance and intelligent materials management system based on artificial intelligence, comprising the method described in any one of claims 1-8, characterized in that, include: Data acquisition module: Collects heterogeneous material data from multiple sources, and divides the material data into structured data and unstructured data according to the data structure; Standardization processing module: parses the unstructured data and standardizes the parsed data and the structured data according to preset standards to generate standardized material data that conforms to the material master data model; Anomaly handling module: Extracts behavioral change features within historical time periods from the standardized material data, constructs a historical feature vector set, and trains a machine learning anomaly detection model; obtains the current behavioral change features of the material to be detected to construct a feature vector to be detected, inputs the feature vector to be detected into the trained anomaly detection model, and obtains the anomaly score output by the anomaly detection model; when the anomaly score exceeds a preset score threshold, and the state change of the material to be detected is verified based on the standardized material data to meet a preset logical conflict judgment condition, it determines that the data of the material to be detected is abnormal and generates an alarm message.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a data governance and intelligent material management method based on artificial intelligence as described in any one of claims 1-8.

Images