Electric power personalized master data management system combined with multi-tenant management mode

Abstract

The invention relates to an electric power personalized master data management system combined with a multi-tenant management mode. The electric power personalized master data management system comprises a tenant management layer, an intelligent master data service layer, an intelligent storage layer and an interface layer, the tenant management layer is used for registering, distributing and authenticating tenants, providing unique identifiers of the tenants, and supporting tenant-level permission and personalized configuration management; the intelligent master data service layer comprises an intelligent data cleaning module, a master data intelligent modeling module and an anomaly detection module; the intelligent storage layer is used for storing main data by adopting a relational database MYSQL; the MongoDB is used for storing unstructured or semi-structured data; and the interface layer provides a standardized API (Application Program Interface) to support integration of multiple systems, ensures identity verification and authorization of all requests through a zero-trust method, and minimizes a trust radius. Efficient storage, cleaning, analysis and application of the master data are achieved, and the master data management quality and efficiency are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of data management, and in particular to an electric power personalized master data management system combined with a multi-tenant management mode. Background Art

[0002] In the modern power industry, with the advancement of digital transformation and intelligent management, master data management has become increasingly important. As the basic data of the entire power system, master data (such as customer information, equipment data, asset data, etc.) not only needs to be accurate and unified, but also needs to meet the personalized needs of different roles such as multi-regional power grid companies, business departments, and operators. In addition, in order to support complex business scenarios of the power system (such as equipment operation monitoring, customer power consumption analysis, etc.) and large-scale concurrent access, the master data management system needs to have excellent data storage performance, high scalability, and intelligent analysis capabilities. However, due to the isolation of business logic, security requirements, and data customization between tenants, the multi-tenant management model has become an important choice to solve this challenge. Summary of the invention

[0003] In order to solve the above problems, the purpose of the present invention is to provide a personalized power master data management system combined with a multi-tenant management mode, so as to realize efficient storage, cleaning, analysis and application of master data and effectively improve the quality and efficiency of master data management.

[0004] To achieve the above object, the present invention adopts the following technical solutions: A power personalized master data management system combined with a multi-tenant management mode, including a tenant management layer, an intelligent master data service layer, an intelligent storage layer and an interface layer; The tenant management layer is used for tenant registration, allocation, and authentication, provides unique identification for tenants, and supports tenant-level permissions and personalized configuration management. The intelligent master data service layer includes an intelligent data cleaning module, a master data intelligent modeling module and an anomaly detection module; the intelligent data cleaning module performs field standardization, duplicate record matching and anomaly correction through NLP and cluster analysis; the master data intelligent modeling module generates a personalized data model using a recommendation algorithm in combination with tenant preferences; the anomaly detection module uses time series analysis and supervised learning methods to detect customer data anomalies; The intelligent storage layer uses the relational database MYSQL for primary data storage and MongoDB for storing unstructured or semi-structured data. The interface layer provides standardized API interfaces to support the integration of multiple systems and ensures that all requests are authenticated and authorized through a zero-trust approach, minimizing the trust radius.

[0005] Furthermore, the tenant management layer includes a tenant registration and allocation module and a tenant authentication module, as follows: The tenant registration and allocation module is used for tenant account registration, including tenant basic information entry, business scenario selection, power service scope and storage resource allocation, and allocates tenant unique identification for data isolation and identification in master data and system management; The tenant authentication module, when called by a user or system, authenticates and authorizes the tenant request according to the tenant identification and authentication credentials, ensuring that only verified tenants can access the specified data.

[0006] Furthermore, field standardization uses NPL to process unstructured or semi-structured fields and convert them into standardized data formats that conform to unified specifications, as follows: First, use regular expressions to clean up redundant symbols, numbers, or meaningless strings, and correct the format of address or device model data based on domain rules; Then use the NLP model to extract the key entity values ​​in the field (for example, extract "province / city / district" when normalizing the address), and use the classification probability distribution to estimate The optimal criteria for extracting entities; ; in, For a given input text x field belongs to entity category e k The probability of; h is the feature vector generated by the NLP model; W l ,b l are the weight and bias of the NLP model respectively; N is the total number of entity categories; l is the category index; W k 、b k are the weight and bias of the kth class respectively, and T is the transpose; Finally, the best standardization result is determined by converting the field text into a vector representation and calculating the similarity between the field and the records in the standard library.

[0007] Furthermore, duplicate records are matched to find and merge duplicate records in the master data, as follows: Use NLP technology to vectorize text fields and generate semantic representations; normalize numeric fields to a unified range; For multiple fields of two records i and j, calculate their similarity, and comprehensively weight the multiple fields to obtain the overall similarity of the two records. : ; in, , Represents the gth field of record i and j respectively; wg is the weight of field g; m is the number of fields; Based on the similarity matrix, density clustering is used to identify clusters of duplicate records. The most comprehensive records are retained in each cluster, and other redundant records are deleted.

[0008] Furthermore, the master data intelligent modeling module includes a tenant behavior data collection submodule, a preference analysis and recommendation submodule, and a dynamic data model generation submodule; the tenant behavior data collection submodule records the tenant's access frequency to the field and model configuration preference data; the preference analysis and recommendation submodule implements collaborative filtering or matrix decomposition to generate personalized field recommendations; the dynamic data model generation submodule uses the field recommendation results to dynamically create a data model and apply it to the master data storage.

[0009] Furthermore, the preference analysis and recommendation submodule implements collaborative filtering or matrix decomposition to generate personalized field recommendations, as follows: Calculate the similarity between fields, find the set of fields that match the tenant's preferences, and calculate the similarity between the two fields using cosine similarity: ; in: For tenants k For Field access behavior; For tenants k For Field access behavior; For Field and Fields similarity; n is the number of tenants; For the fields; Recommend fields similar to the fields that tenants use most frequently: ; in, For tenants k The priority score of the recommended field f; A collection of fields that tenants have previously accessed frequently; For problems with sparse behavior matrices, matrix decomposition techniques are used to predict tenant preferences: Decompose the tenant behavior matrix R into a latent vector matrix: ; Among them: U is the mapping matrix between tenants and potential features; S is a diagonal matrix representing feature importance; V is the mapping matrix between fields and potential features; Combine the mapping matrix between tenants and latent features with the latent feature vector of the field to predict the tenant's preference score for a field. : Score based on predicted preference , generates the top several most likely to be used field sets for the tenant as part of the personalized data model.

[0010] Furthermore, the anomaly detection module extracts long-term trends and periodicities through time series decomposition, and detects obvious anomalies through the remaining residuals; at the same time, combined with the time series prediction model, it further monitors the anomalies that are not captured by the decomposition through the residuals between the predicted value and the actual value.

[0011] Furthermore, the intelligent storage layer shards the database by tenant or business module based on business volume and tenant size to improve read and write performance and horizontal expansion capabilities; each tenant has an independent schema and a master data shared table structure to ensure isolation between data; a tenant_id field is added to each data table to filter data based on tenants.

[0012] Furthermore, add the tenant_id field to each data table and filter data based on tenants, as follows: In all data tables that need to distinguish tenants, add a tenant_id field to identify the tenant to which each piece of data belongs; and use tenant_id as part of the composite primary key; each time a data operation is performed, add the tenant_id condition to the SQL query based on the current tenant context; this ensures that each tenant can only access and operate the data that belongs to him / her, and holds the context information of tenant_id in the application layer session, which is usually stored after the user is authenticated successfully and used in subsequent data operations; In database operations, tenant_id is automatically used as a condition for data filtering. Database query is a filtering process with the formula: Q filtered =Q base ∩{x∣x.tenant_id=current_tenant_id} Among them, Q base Represents the original database query result, Q filtered is the result set after applying the tenant_id filter condition; x is the input data, x.tenant_id is the tenant_id and current_tenant_id of x is the identifier of the current tenant; The composite primary key of the data is composed of tenant_id and other fields to ensure that the data of each tenant is unique and complete. For each x, there is:

[0013] in, Represents the joint primary key space of the entire data set; is the user id of x; Indicates user id; Create a view to display data specifically for a specific tenant. Embed the WHERE tenant_id = :tenant_id condition in the view, eliminating the need to repeat the condition in every query.

[0014] Furthermore, through the zero-trust approach, all requests are authenticated and authorized to minimize the trust radius, as follows: Based on the dynamic authorization mechanism, it uses real-time context to decide whether to grant access rights, integrates identity and access management services, and implements fine-grained role permission control according to predefined policies; Fine-grained network segmentation, limiting east-west traffic within each segment, using network virtualization technology to authorize network connections on data paths and prevent unauthorized access; Set the default policy to deny any unauthenticated traffic, ensure that requests comply with security policies before they are allowed to pass, and that every request must provide authentication credentials, regardless of whether the request comes from the external or internal network.

[0015] The present invention has the following beneficial effects: 1. The present invention not only meets the complex and diverse data management needs of the power industry, but also improves data quality, analysis capabilities and system efficiency through intelligent functions, while taking into account scalability and security; 2. The present invention uses NLP and cluster analysis algorithms to perform field standardization, duplicate record matching, and anomaly correction on the original data in the master data to improve data quality and provide a high-quality data foundation for data modeling and data analysis; 3. The present invention combines tenant preferences and uses recommendation algorithms (such as collaborative filtering and matrix decomposition) to dynamically generate personalized data models, support tenant-level customized services, effectively improve the matching degree between data models and tenant business needs, and achieve precise management. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a system architecture diagram of the present invention. DETAILED DESCRIPTION

[0017] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments: refer to Figure 1 ,In this implementation, a power personalized master data management system in combination with a multi-tenant management mode is provided, including a tenant management layer, an intelligent master data service layer, an intelligent storage layer, and an interface layer; The tenant management layer is used for tenant registration, allocation, and authentication, provides unique identification for tenants, and supports tenant-level permissions and personalized configuration management. The intelligent master data service layer includes an intelligent data cleaning module, a master data intelligent modeling module and an anomaly detection module; the intelligent data cleaning module performs field standardization, duplicate record matching and anomaly correction through NLP and cluster analysis; the master data intelligent modeling module generates a personalized data model using a recommendation algorithm in combination with tenant preferences; the anomaly detection module uses time series analysis and supervised learning methods to detect customer data anomalies; The intelligent storage layer uses the relational database MYSQL for primary data storage and MongoDB for storing unstructured or semi-structured data. The interface layer provides standardized API interfaces (such as RESTful API and GraphQL) to support the integration of multiple systems, and through the zero-trust approach, ensures that all requests are authenticated and authorized to minimize the trust radius.

[0018] In this embodiment, the tenant management layer includes a tenant registration and allocation module and a tenant authentication module, as follows: The tenant registration and allocation module is used for tenant account registration, including tenant basic information entry, business scenario selection, power service scope and storage resource allocation, and allocates tenant unique identification for data isolation and identification in master data and system management; The tenant authentication module, when called by a user or system, authenticates and authorizes the tenant request according to the tenant identifier and authentication credentials (such as OAuth, API Key or JWT), ensuring that only verified tenants can access the specified data.

[0019] In this embodiment, the field standardization adopts NPL, which processes unstructured or semi-structured fields (such as address, customer name, and device model) and converts them into standardized data formats that comply with unified specifications to avoid the influence of spelling and format differences on the uniformity of data. The details are as follows: First, use regular expressions to clean up redundant symbols, numbers, or meaningless strings, and correct the format of address or device model data based on domain rules; Then use the NLP model to extract the key entity values ​​in the field (for example, extract "province / city / district" when normalizing the address), and use the classification probability distribution to estimate The optimal criteria for extracting entities; ; in, For a given input text x field belongs to entity category e kThe probability of; h is the feature vector generated by the NLP model; W l ,b l are the weight and bias of the NLP model respectively; N is the total number of entity categories (such as place names, equipment fields, customer names, etc.); l is the category index; W k 、b k are the weight and bias of the kth class respectively, and T is the transpose; Finally, the best standardization result is determined by converting the field text into a vector representation and calculating the similarity between the field and the records in the standard library.

[0020] In this embodiment, the duplicate record matching is performed to find and merge duplicate records (such as customer information and equipment information) in the master data, avoid data redundancy and conflict, and maintain data uniqueness, as follows: Use NLP technology to vectorize text fields (such as customer names) to generate semantic representations; normalize numeric fields (such as equipment numbers) to a unified range; For multiple fields of two records i and j, calculate their similarity, and comprehensively weight the multiple fields to obtain the overall similarity of the two records. : ; in, , Represents the gth field of record i and j respectively; w g is the weight of field g, m is the number of fields; Based on the similarity matrix, density clustering (such as DBSCAN) is used to identify clusters of duplicate records. The most comprehensive records are retained in each cluster, and other redundant records are deleted.

[0021] In this embodiment, the master data intelligent modeling module includes a tenant behavior data collection submodule, a preference analysis and recommendation submodule, and a dynamic data model generation submodule; the tenant behavior data collection submodule records the tenant's access frequency to the field and the model configuration preference data; the preference analysis and recommendation submodule implements collaborative filtering or matrix decomposition to generate personalized field recommendations; the dynamic data model generation submodule uses the field recommendation results to dynamically create a data model (table structure) and apply it to the master data storage.

[0022] In this embodiment, the preference analysis and recommendation submodule implements collaborative filtering or matrix decomposition to generate personalized field recommendations, as follows: Calculate the similarity between fields, find the set of fields that match the tenant's preferences, and calculate the similarity between the two fields using cosine similarity: ; in: For tenants k For Field access behavior; For tenants k For Field access behavior; For Field and Fields similarity; n is the number of tenants; For the fields; Recommend fields similar to the fields that tenants use most frequently: ; in, For tenants k The priority score of the recommended field f; A collection of fields that tenants have previously accessed frequently; For problems with sparse behavior matrices (some tenants do not have complete usage history), matrix decomposition technology is used to predict tenant preferences: Decompose the tenant behavior matrix R into a latent vector matrix: ; Among them: U is the mapping matrix between tenants and potential features; S is a diagonal matrix representing feature importance; V is the mapping matrix between fields and potential features; Combine the mapping matrix between tenants and latent features with the latent feature vector of the field to predict the tenant's preference score for a field. : Score based on predicted preference , generates the top several most likely to be used field sets for the tenant as part of the personalized data model.

[0023] In this embodiment, the anomaly detection module extracts long-term trends and periodicity through time series decomposition (STL), and detects obvious anomalies through residual errors. At the same time, it combines the time series prediction model (ARIMA / SARIMA) to further monitor anomalies that are not captured by the decomposition through the residuals between the predicted value and the actual value.

[0024] In this embodiment, the intelligent storage layer shards the database by tenant or business module according to the business volume and tenant scale to improve read and write performance and horizontal expansion capabilities; each tenant has an independent schema and a master data shared table structure to ensure isolation between data; a tenant_id field is added to each data table to filter data according to the tenant.

[0025] In this embodiment, the tenant_id field is added to each data table, and data is filtered according to the tenant, as follows: In all data tables that need to distinguish tenants, add a tenant_id field to identify the tenant to which each piece of data belongs; and use tenant_id as part of the composite primary key; each time a data operation is performed, add the tenant_id condition to the SQL query based on the current tenant context; this ensures that each tenant can only access and operate the data that belongs to him / her, and holds the context information of tenant_id in the application layer session, which is usually stored after the user is authenticated successfully and used in subsequent data operations; In database operations, tenant_id is automatically used as a condition for data filtering. Database query is a filtering process with the formula: Q filtered =Q base ∩{x∣x.tenant_id=current_tenant_id} Among them, Q base Represents the original database query result, Q filtered is the result set after applying the tenant_id filter condition; x is the input data, x.tenant_id is the tenant_id and current_tenant_id of x is the identifier of the current tenant; The composite primary key of the data is composed of tenant_id and other fields to ensure that the data of each tenant is unique and complete. For each x, there is:

[0026] in, Represents the joint primary key space of the entire data set; is the user id of x; Indicates user id; Create a view to display data specifically for a specific tenant. Embed the WHERE tenant_id = :tenant_id condition in the view, eliminating the need to repeat the condition in every query.

[0027] In this embodiment, a zero-trust approach is used to ensure that all requests are authenticated and authorized, minimizing the trust radius, as follows: Based on the dynamic authorization mechanism, real-time context is used to decide whether to grant access rights, and identity and access management services are integrated to implement fine-grained role-based permission control according to predefined policies; the details are as follows: Use IP address geolocation analysis or GPS data to obtain the user's geographic location, use device fingerprint recognition technology to evaluate the security status of the device, record and analyze the timestamp of the request, and detect abnormal access time periods; Deploy a rule engine to handle logical decisions, analyze and match real-time context according to predefined policies, and make authorization decisions; Define a policy language to describe and implement access control policies, and use the policy decision point and policy enforcement point architecture to separate decision-making and enforcement; Use IAM services to manage user identities, roles, and permissions, define role and permission models, assign permissions based on roles, and ensure the principle of least privilege; A hybrid model based on role and context, combining ABAC and RBAC, implements fine-grained permission policies and dynamically adjusts access rights based on user attributes, environment context, and request content; Fine-grained network segmentation is performed to limit east-west traffic within each segment. Network virtualization technology is used to implement network connections for authorized data paths to prevent unauthorized access. The details are as follows: Create logical segments in the network, where the resources in each segment are a single application, service, or instance, ensuring that the segments can only communicate with each other in a pre-defined manner; Use VMware NSX to create virtual network topologies, including virtual switches, routers, and firewalls. Provides isolated management of network and security policies for each workload; Configure distributed firewall rules in VMware NSX and apply them to each VM or container, defining security policies to isolate based on workload tags, application type or environment; Use VMware NSX-based automated security policy tools to dynamically adjust based on traffic patterns, threat intelligence and behavioral analysis; Configure VMware NSX network policies to define and restrict authorized communication paths, so that only necessary network ports are open between the source and the destination; synchronize policies with logical firewalls and integrate with IAM to ensure that only authorized users have access to specific paths through authentication and role permissions; Set the default policy to deny any unauthenticated traffic, ensure that requests comply with security policies before they are allowed to pass, and that every request must provide authentication credentials, regardless of whether the request comes from the external or internal network.

[0028] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0029] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0030] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0031] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0032] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A personalized power master data management system combined with a multi-tenant management mode, characterized in that: It includes tenant management layer, intelligent master data service layer, intelligent storage layer and interface layer; The tenant management layer is used for tenant registration, allocation, and authentication, provides unique identification for tenants, and supports tenant-level permissions and personalized configuration management. The intelligent master data service layer includes an intelligent data cleaning module, a master data intelligent modeling module and an anomaly detection module; the intelligent data cleaning module performs field standardization, duplicate record matching and anomaly correction through NLP and cluster analysis; the master data intelligent modeling module generates a personalized data model using a recommendation algorithm in combination with tenant preferences; the anomaly detection module uses time series analysis and supervised learning methods to detect customer data anomalies; The intelligent storage layer uses the relational database MYSQL for primary data storage and MongoDB for storing unstructured or semi-structured data. The interface layer provides standardized API interfaces to support the integration of multiple systems and ensures that all requests are authenticated and authorized through a zero-trust approach, minimizing the trust radius.

2. According to claim 1, a power personalized master data management system combined with a multi-tenant management mode is characterized in that: The tenant management layer includes a tenant registration and allocation module and a tenant authentication module, as follows: The tenant registration and allocation module is used for tenant account registration, including tenant basic information entry, business scenario selection, power service scope and storage resource allocation, and allocates tenant unique identification for data isolation and identification in master data and system management; The tenant authentication module, when called by a user or system, authenticates and authorizes the tenant request according to the tenant identification and authentication credentials, ensuring that only verified tenants can access the specified data.

3. According to claim 1, a power personalized master data management system combined with a multi-tenant management mode is characterized in that: The field standardization adopts NPL, which processes unstructured or semi-structured fields and converts them into a standardized data format that complies with unified specifications. The details are as follows First, use regular expressions to clean up redundant symbols, numbers, or meaningless strings, and based on domain rules, correct the format of address or device model data; Then use the NLP model to extract the key entity values ​​in the field and use the classification probability distribution to estimate The optimal criteria for extracting entities; ； in, For a given input text x field belongs to entity category e k The probability of; h is the feature vector generated by the NLP model; W l ,b l are the weight and bias of the NLP model respectively; N is the total number of entity categories; l is the category index; W k 、b k are the weight and bias of the kth class respectively, and T is the transpose; Finally, the best standardization result is determined by converting the field text into a vector representation and calculating the similarity between the field and the records in the standard library.

4. The personalized power master data management system combined with a multi-tenant management mode according to claim 3 is characterized in that: The duplicate record matching is to find and merge the duplicate records in the master data, as follows: Use NLP technology to vectorize text fields and generate semantic representations; normalize numeric fields to a unified range; For multiple fields of two records i and j, calculate their similarity, and comprehensively weight the multiple fields to obtain the overall similarity of the two records. : ; in, , Represents the gth field of record i and j respectively; w g is the weight of field g; m is the number of fields; Based on the similarity matrix, density clustering is used to identify clusters of duplicate records. The most comprehensive records are retained in each cluster, and other redundant records are deleted.

5. The electric power personalized master data management system combined with a multi-tenant management mode according to claim 1, characterized in that: The master data intelligent modeling module includes a tenant behavior data collection submodule, a preference analysis and recommendation submodule, and a dynamic data model generation submodule; the tenant behavior data collection submodule records the frequency of tenants' access to fields and model configuration preference data; The preference analysis and recommendation submodule implements collaborative filtering or matrix decomposition to generate personalized field recommendations; The dynamic data model generation submodule uses the field recommendation results to dynamically create a data model and apply it to the main data storage.

6. The personalized power master data management system combined with a multi-tenant management mode according to claim 5, characterized in that: The preference analysis and recommendation submodule implements collaborative filtering or matrix decomposition to generate personalized field recommendations, as follows: Calculate the similarity between fields, find the set of fields that match the tenant's preferences, and calculate the similarity between the two fields using cosine similarity: ； in: For tenants k For Field access behavior; For tenants k For Field access behavior; For Field and Fields similarity; n is the number of tenants; For the fields; Recommend fields similar to the fields that tenants use most frequently: ； in, For tenants k The priority score of the recommended field f; A collection of fields that tenants have previously accessed frequently; For problems with sparse behavior matrices, matrix decomposition techniques are used to predict tenant preferences: Decompose the tenant behavior matrix R into a latent vector matrix: ； Among them: U is the mapping matrix between tenants and potential features; S is a diagonal matrix representing feature importance; V is the mapping matrix between fields and potential features; Combine the mapping matrix between tenants and latent features with the latent feature vector of the field to predict the tenant's preference score for a field. : Score based on predicted preference , generates the top several most likely to be used field sets for the tenant as part of the personalized data model.

7. The electric power personalized master data management system combined with a multi-tenant management mode according to claim 1, characterized in that: The anomaly detection module extracts long-term trends and periodicity through time series decomposition, and detects obvious anomalies through residual errors. At the same time, combined with the time series prediction model, it further monitors anomalies that are not captured by decomposition through the residuals between the predicted value and the actual value.

8. The electric power personalized master data management system combined with a multi-tenant management mode according to claim 1, characterized in that: The intelligent storage layer shards the database by tenant or business module according to the business volume and tenant scale to improve read and write performance and horizontal expansion capabilities; each tenant has an independent schema and a master data shared table structure to ensure isolation between data; a tenant_id field is added to each data table to filter data according to the tenant.

9. The personalized power master data management system combined with a multi-tenant management mode according to claim 8, characterized in that: The tenant_id field is added to each data table, and data is filtered according to the tenant, as follows: In all data tables that need to distinguish tenants, add a tenant_id field to identify the tenant to which each piece of data belongs; and use tenant_id as part of the composite primary key; each time a data operation is performed, add the tenant_id condition to the SQL query based on the current tenant context; this ensures that each tenant can only access and operate the data that belongs to him / her, and holds the context information of tenant_id in the application layer session, which is usually stored after the user is authenticated successfully and used in subsequent data operations; In database operations, tenant_id is automatically used as a condition for data filtering. Database query is a filtering process with the formula: Q filtered =Q base ∩{x∣x.tenant_id=current_tenant_id} Among them, Q base Represents the original database query result, Q filtered is the result set after applying the tenant_id filter condition; x is the input data, x.tenant_id is the tenant_id and current_tenant_id of x is the identifier of the current tenant; The composite primary key of the data is composed of tenant_id and other fields to ensure that the data of each tenant is unique and complete. For each x, there is: in, Represents the joint primary key space of the entire data set; is the user id of x; Indicates user id; Create a view to display data specifically for a specific tenant. Embed the WHERE tenant_id = :tenant_id condition in the view, eliminating the need to repeat the condition in every query.

10. The electric power personalized master data management system combined with a multi-tenant management mode according to claim 1, characterized in that: The zero-trust approach ensures that all requests are authenticated and authorized, minimizing the trust radius, as follows: Based on the dynamic authorization mechanism, it uses real-time context to decide whether to grant access rights, integrates identity and access management services, and implements fine-grained role permission control according to predefined policies; Fine-grained network segmentation, limiting east-west traffic within each segment, using network virtualization technology to authorize network connections on data paths and prevent unauthorized access; Set the default policy to deny any unauthenticated traffic, ensure that requests comply with security policies before they are allowed to pass, and that every request must provide authentication credentials, regardless of whether the request comes from the external or internal network.

Images