A Management Method for IDC Operations and Maintenance Knowledge Graph

Abstract

This invention discloses a management method for an IDC (Internet Data Center) operation and maintenance knowledge graph, relating to the field of data center operation and maintenance engineering technology. It obtains data center operation and maintenance management features by collecting and processing data center operation and maintenance distribution parameters, operation and maintenance graph construction parameters, and operation and maintenance management correlation coefficients. Using these features and corresponding operation and maintenance management methods, a multimodal operation and maintenance management model is constructed. The method receives data center operation and maintenance management features to be processed, and the constructed model processes these features to generate real-time operation and maintenance management methods. This invention employs a feature-corrected knowledge graph method and sets a post-operation and maintenance improvement ratio parameter. It uses a graph network model constructed with operation and maintenance knowledge keywords and knowledge graph parameters to obtain operation and maintenance management methods. This achieves unique data processing for data center operation and maintenance, making operation and maintenance management more automated, helping to save operation and maintenance resources, reduce operation and maintenance risks, and enhance the rationality of operation and maintenance management.

Description

Technical Field

[0001] This invention belongs to the field of data center operation and maintenance engineering technology, and in particular relates to a management method for an IDC operation and maintenance knowledge graph. Background Technology

[0002] In today's highly information-driven era, the stable operation of data centers (IDCs) has become a critical guarantee for business continuity. However, traditional operation and maintenance methods are proving inadequate in the face of increasingly complex IT environments, mainly due to issues such as knowledge fragmentation, low diagnostic efficiency, and difficulty in transferring experience. In recent years, with the development of artificial intelligence technology, knowledge graphs, as an efficient knowledge representation and reasoning tool, are fundamentally changing traditional operation and maintenance models.

[0003] Some literature mentions methods and systems for similarity assessment of patient medical records based on knowledge graphs. Furthermore, in pharmaceutical projects, knowledge graphs are used to integrate statistical models with deep learning techniques for named entity recognition.

[0004] Existing technologies employ deep learning methods to extract information from databases, using convolutional neural networks multiple times to mine and extract important knowledge from projects. They extend the concept of knowledge graphs from specific domains to professional domains, constructing knowledge graph networks under the current situation of relatively limited entity types, thereby improving the accuracy of evaluation.

[0005] However, the above methods are not applicable to data center operations and maintenance (O&M), and the applied technologies are extremely complex. Directly applying knowledge graphs to obtain O&M methods for data centers has significant shortcomings, and obtaining more detailed O&M methods for data centers is even more difficult. A method for classifying and obtaining knowledge graphs for data center O&M methods has not yet been invented. To obtain more accurate and efficient O&M methods for data center O&M management scenarios, the challenges facing current technologies include: how to process specific relevant parameters of the data center to obtain data center O&M management characteristics; how to use these characteristics and corresponding O&M management methods to construct a multimodal O&M management model; how to adaptively improve the model based on the scenario; and how to obtain O&M management methods using a graph network model constructed with O&M knowledge keywords and knowledge graph parameters. This would enable unique data processing for data center O&M, making O&M management more automated, helping to save O&M resources, reduce O&M risks, and enhance the rationality of O&M management. Summary of the Invention

[0006] To address the aforementioned technical issues, this invention proposes a management method for IDC (Internet Data Center) operation and maintenance knowledge graphs.

[0007] In a first aspect of the present invention, a method for managing an IDC (Internet Data Center) operations and maintenance knowledge graph is provided, the method comprising the following steps:

[0008] Obtain the operation and maintenance distribution parameters of the data center, and use knowledge graph analysis methods to calculate the operation and maintenance graph construction parameters. Set the improvement ratio parameters before and after data center operation and maintenance to obtain the operation and maintenance management correlation coefficients, and collect the corresponding operation and maintenance management methods.

[0009] The operation and maintenance distribution parameters, the operation and maintenance map construction parameters, and the operation and maintenance management correlation coefficients are processed to form data center operation and maintenance management characteristics;

[0010] A multimodal operation and maintenance management model is constructed using the data center operation and maintenance management characteristics and the operation and maintenance management method. The multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph.

[0011] The multimodal operation and maintenance management model is used to process the characteristics of the data center to be operated and maintained, and a real-time operation and maintenance management method is generated for operation and maintenance. The data center is then operated and maintained according to the real-time operation and maintenance management method.

[0012] Furthermore, the operation and maintenance distribution parameters include operation and maintenance security level coefficients, operation and maintenance data center levels, and operation and maintenance status data.

[0013] Furthermore, before calculating the operation and maintenance graph construction parameters using the knowledge graph analysis method, a relationship graph between operation and maintenance entities and fault types is constructed. A knowledge graph network is constructed based on the top R glyphs of the operation and maintenance entities and the top T glyphs of the fault types. The knowledge graph network is constructed by processing the top R glyphs of the operation and maintenance entities and the top T glyphs of the fault types using natural language processing technology, and then representing them in the form of the degree of graph network nodes and edges.

[0014] Furthermore, the operation and maintenance graph construction parameters are calculated based on the knowledge graph network relationship calculation formula, specifically using the feature values ​​of operation and maintenance entity characters, the feature values ​​of fault type characters, and the degree of knowledge graph network edges.

[0015] Furthermore, the operation and maintenance management correlation coefficient is calculated using the improvement ratio parameter set before and after data center operation and maintenance:

[0016]

[0017] In the formula, The operation and maintenance management correlation coefficient is... and These represent the dimensionless values ​​of data center disk utilization after maintenance and the dimensionless values ​​of data center disk utilization before maintenance, respectively. and These represent the dimensionless values ​​of the total data center energy consumption set after maintenance and the dimensionless values ​​of the total data center energy consumption before maintenance, respectively. To facilitate data processing and operation and maintenance management of the model, relevant coefficients were adjusted.

[0018] Furthermore, the data center operation and maintenance management characteristics are obtained by horizontally linking and splicing the operation and maintenance distribution parameters, the operation and maintenance map construction parameters, and the operation and maintenance management correlation coefficients.

[0019] Furthermore, the multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph, and its activation function calculation formula is shown below:

[0020]

[0021] X represents the activation function value, and X represents the input data center operation and maintenance management characteristics. Parameters are used to construct the operation and maintenance map corresponding to the input data center operation and maintenance management characteristics. The sum of the parameters used to construct the operation and maintenance graph for training the model. The number of training sets used for model training.

[0022] A management system for an IDC (Internet Data Center) operations and maintenance knowledge graph is also provided. The system includes a data center operations and maintenance distribution parameter acquisition module, a knowledge graph network construction module, an operations and maintenance graph construction parameter generation module, an operations and maintenance management correlation coefficient acquisition module, a multimodal operations and maintenance management model construction module, and an operations and maintenance management method generation module.

[0023] The data center operation and maintenance distribution parameter acquisition module acquires the operation and maintenance distribution parameters of the data center.

[0024] The knowledge graph network construction module constructs a knowledge graph network based on the glyphs of operation and maintenance entities and the glyphs of fault types.

[0025] The operation and maintenance graph construction parameter generation module calculates the operation and maintenance graph construction parameters using knowledge graph analysis methods.

[0026] The operation and maintenance management correlation coefficient acquisition module: processes the improvement ratio parameters set before and after data center operation and maintenance to obtain the operation and maintenance management correlation coefficient;

[0027] The multimodal operation and maintenance management model construction module processes the operation and maintenance distribution parameters, the operation and maintenance graph construction parameters, and the operation and maintenance management correlation coefficients to form data center operation and maintenance management features, collects corresponding operation and maintenance management methods, and constructs a multimodal operation and maintenance management model using the data center operation and maintenance management features and the operation and maintenance management methods. The multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph.

[0028] The operation and maintenance management method generation module: uses the multimodal operation and maintenance management model to process the characteristics of the data center to be operated and maintained, generates a real-time operation and maintenance management method for operation and maintenance, and performs operation and maintenance management on the data center according to the real-time operation and maintenance management method.

[0029] Furthermore, the operation and maintenance distribution parameters include operation and maintenance security level coefficients, operation and maintenance data center levels, and operation and maintenance status data.

[0030] Furthermore, the multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph, and its activation function calculation formula is shown below:

[0031]

[0032] X represents the activation function value, and X represents the input data center operation and maintenance management characteristics. Parameters are used to construct the operation and maintenance map corresponding to the input data center operation and maintenance management characteristics. The sum of the parameters used to construct the operation and maintenance graph for training the model. The number of training sets used for model training.

[0033] This invention discloses a management method for an IDC (Internet Data Center) operation and maintenance (IDC) knowledge graph. It collects and processes data center operation and maintenance distribution parameters, operation and maintenance graph construction parameters, and operation and maintenance management correlation coefficients to obtain data center operation and maintenance management features. These features, along with corresponding operation and maintenance management methods, are used to construct a multimodal operation and maintenance management model. The model is based on a graph neural network model with ReLU activation function-based scenario adaptation improvements. It receives data center operation and maintenance features to be processed, processes these features, and generates real-time operation and maintenance management methods. This invention employs a feature-corrected knowledge graph method and data center operation and maintenance improvement ratio parameters. It uses a graph network model constructed with operation and maintenance knowledge keywords and knowledge graph parameters to obtain operation and maintenance management methods. This achieves unique data processing for data center operation and maintenance, making operation and maintenance management more automated, helping to save operation and maintenance resources, reduce operation and maintenance risks, and enhance the rationality of operation and maintenance management. Attached Figure Description

[0034] Figure 1 This is a flowchart of a management method for an IDC (Internet Data Center) operation and maintenance knowledge graph according to the present invention;

[0035] Figure 2 This is a schematic diagram of the management system structure for an IDC operation and maintenance knowledge graph according to the present invention;

[0036] Figure 3 This is a schematic diagram of the knowledge graph network construction in this invention;

[0037] Figure 4 This is a schematic diagram of the neural network principle in this invention;

[0038] Figure 5 This is a schematic diagram of the electronic device structure for implementing the method of the present invention in an embodiment of the present invention. Detailed Implementation

[0039] The invention will now be further described in conjunction with the accompanying drawings and specific embodiments. The graph neural network model used in this invention is an improvement on the neural network model in the application scenario of IDC operation and maintenance management.

[0040] Graph neural networks (GNNs) are neural networks that operate directly on graph-structured data to process feature information of nodes, edges, or the entire graph. Their core idea is to update the representation of the current node by aggregating the feature information of neighboring nodes, thereby capturing the dependencies and topological features between nodes in the graph. Therefore, the operation and maintenance management of the networked knowledge graph presented in this invention is highly suitable for the scenario-based application of GNN models.

[0041] like Figure 2 As shown, the management method of the IDC operation and maintenance knowledge graph of the present invention belongs to the field of data center operation and maintenance management engineering technology research and development technology, and belongs to other engineering technology research and development industries other than marine engineering equipment, new materials, biotechnology, new energy, new energy vehicles, energy conservation, and environmental protection. Therefore, it belongs to new technology and innovation and entrepreneurship services.

[0042] First embodiment of the present invention:

[0043] In a first aspect of the present invention, a method for managing an IDC (Internet Data Center) operations and maintenance knowledge graph is provided, the method comprising the following steps:

[0044] Obtain the operation and maintenance distribution parameters of the data center, and use knowledge graph analysis methods to calculate the operation and maintenance graph construction parameters. Set the improvement ratio parameters before and after data center operation and maintenance to obtain the operation and maintenance management correlation coefficients, and collect the corresponding operation and maintenance management methods.

[0045] The operation and maintenance distribution parameters, the operation and maintenance map construction parameters, and the operation and maintenance management correlation coefficients are processed to form data center operation and maintenance management characteristics;

[0046] A multimodal operation and maintenance management model is constructed using the data center operation and maintenance management characteristics and the operation and maintenance management method. The multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph.

[0047] The multimodal operation and maintenance management model is used to process the characteristics of the data center to be operated and maintained, and a real-time operation and maintenance management method is generated for operation and maintenance. The data center is then operated and maintained according to the real-time operation and maintenance management method.

[0048] Furthermore, the operation and maintenance distribution parameters include the operation and maintenance security level coefficient, the operation and maintenance data center level, and the operation and maintenance status data, and their calculation formula is as follows:

[0049]

[0050] In the formula, Here are the operation and maintenance distribution parameters, where i represents the i-th fault log and m represents the total number of fault logs (m). For the operation and maintenance security level coefficient, This represents the number of fault entries in the i-th fault log. Its value is derived from the total number of fault entries in the fault log. The operation and maintenance security level coefficient is divided into three values: 1, 2, and 5, corresponding to no fault entries, less than or equal to 2 fault entries, and greater than 2 fault entries, respectively. Let represent the operational data center level corresponding to the i-th fault log. This level is determined by taking values ​​based on the number of standard racks in the data center. The values ​​are 2, 4, 6, and 8 for each of the following standard rack counts: less than 1000, greater than or equal to 1000 but less than or equal to 3000, greater than 3000 but less than or equal to 5000, and greater than 5000. These values ​​represent the optimal range of operational distribution parameters for model training. This is a creative representation of the data parameters used to train the model. Here, j represents the j-th operational operation of the data center, and n represents the total number of operational operations. This represents the actual time consumed during the j-th maintenance operation. This represents the theoretical time consumed in the j-th maintenance operation. The ratio of maintenance time consumed can objectively reflect the actual maintenance situation of the data center. Combining the maintenance security level coefficient, the maintenance data center level, and the maintenance situation data can provide a good feedback on the acquisition of maintenance management methods. Since the maintenance management method is affected by the required security level and the specific level of the data center, the applicability of most maintenance management methods depends on the maintenance data center level and the maintenance situation. Therefore, this invention fully considers the impact of this type of data on the rationality of maintenance management methods and optimizes the calculation of maintenance distribution parameters suitable for subsequent graph neural network model classification to classify maintenance management methods.

[0051] Furthermore, before calculating the operation and maintenance graph construction parameters using knowledge graph analysis methods, a relationship graph between operation and maintenance entities and fault types is constructed. A knowledge graph network is built based on the top R most frequent glyphs of the operation and maintenance entities and the top T most frequent glyphs of the fault types. The knowledge graph network is constructed by processing the top R most frequent glyphs of the operation and maintenance entities and the top T most frequent glyphs of the fault types using natural language processing techniques, and then representing them in the form of the degree of graph network nodes and edges.

[0052] Some simplified forms of knowledge graph network construction can be found in [link to relevant documentation]. Figure 3As shown, the operation and maintenance entity glyphs and fault type glyphs are used as nodes, and the node values ​​are the glyph feature values ​​transformed from the operation and maintenance entity glyphs and fault type glyphs. The association between glyphs is constructed by connecting the edges and the weight values ​​of the edges.

[0053] Furthermore, the parameters for constructing the operation and maintenance graph are calculated based on the knowledge graph network relationship calculation formula, specifically using the feature values ​​of the operation and maintenance entity glyphs, the feature values ​​of the fault type glyphs, and the degree of the knowledge graph network edges:

[0054]

[0055] In the formula Parameters are constructed for the operation and maintenance map. The number of the top R types of glyphs most frequently used in operation and maintenance entity glyphs. The number of the top T types of glyphs related to the number of glyphs of the fault type. Let a represent the most frequently used operation and maintenance entity glyph, and b represent the b-th glyph related to the number of fault type glyphs. This represents the glyph feature value of the a-th glyph that is used the most in operations and maintenance. This represents the glyph feature value of the b-th glyph related to the number of glyphs for the fault type. This represents the baseline value for calculating the correlation degree. Since some operation and maintenance entity glyphs may simultaneously form knowledge graph associations with multiple fault type glyphs, the average correlation degree between each glyph is adjusted using the maximum degree and the degree value in the knowledge graph relationship. This represents the maximum number of edges in the knowledge graph network. This represents the maximum weight of the edge. The minimum weight of the edge. Since the more fault type-related glyphs appear in operations and maintenance, the more difficult and detailed the operations and maintenance management becomes, the operations and maintenance graph construction parameters are obtained by constructing knowledge graph network relationships using operations and maintenance entity glyphs and fault type glyphs. Specifically, the similarity calculation is performed using the corresponding relevance function based on the glyph feature values ​​to obtain the operations and maintenance graph construction parameters, thereby correcting the use of operations and maintenance management methods.

[0056] Furthermore, the operation and maintenance management correlation coefficient is calculated using the improvement ratio parameter set before and after data center operation and maintenance:

[0057]

[0058] In the formula, The operation and maintenance management correlation coefficient refers to the fact that, in this invention, changes in relevant parameters of the data center before and after operation and maintenance can, to a certain extent, reflect changes in the operation and maintenance management method. and These represent the dimensionless values ​​of data center disk utilization after maintenance and the dimensionless values ​​of data center disk utilization before maintenance, respectively. and These represent the dimensionless values ​​of the total data center energy consumption set after maintenance and the dimensionless values ​​of the total data center energy consumption before maintenance, respectively. To facilitate data processing by the model, the operation and maintenance management correlation coefficient correction parameter is set to a value range of 2-10, ensuring that the operation and maintenance management correlation coefficient used for model training meets the model input requirements.

[0059] Furthermore, the data center operation and maintenance management characteristics are obtained by horizontally linking and splicing the operation and maintenance distribution parameters, the operation and maintenance map construction parameters, and the operation and maintenance management correlation coefficients.

[0060] Furthermore, the multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph, and its activation function calculation formula is shown below:

[0061]

[0062] X represents the activation function value, and X represents the input data center operation and maintenance management characteristics. Parameters are used to construct the operation and maintenance map corresponding to the input data center operation and maintenance management characteristics. The sum of the parameters used to construct the operation and maintenance graph for training the model. The number of training sets used for model training, based on the knowledge graph analysis in this invention, allows the activation function to introduce nonlinear transformations into the neural network, enabling it to fit arbitrarily complex functions, limiting neuron outputs, and adapting to scenario requirements. The knowledge graph network in this invention has a significant impact on the construction of the neural network. By utilizing the operation and maintenance graph construction parameters, scene-adaptive gradient propagation adjustment of the graph neural network can be performed, resulting in optimized model training performance. To adapt to changes using variations of the basic activation function, comparing the average operation and maintenance graph construction parameters of the training set with the operation and maintenance graph construction parameters corresponding to the input data center operation and maintenance management characteristics and taking the corresponding logarithmic function value can greatly alleviate the gradient vanishing problem, making it possible to train the graph neural network. The stable propagation of gradients ensures that each layer of the network can effectively learn, providing strong support for model construction. Figure 4 The diagram shown illustrates the general principle of a neural network, demonstrating the crucial role of activation functions.

[0063] The graph neural network used in this invention is an improved form of the conventional graph neural network. The conventional graph neural network generally uses the ReLU activation function. However, the conventional ReLU activation function obviously suffers from the vanishing gradient problem in the classification and recognition of operation and maintenance management methods in knowledge graph-based scenarios. It cannot obtain a suitable activation function value, resulting in a chaotic output of the IDC operation and maintenance management method. Through the innovative variant of the ReLU activation function in this invention, the gradient vanishing problem of the model in the output process of the operation and maintenance management method can be alleviated, making it possible to train the graph neural network model.

[0064] The second embodiment of the present invention:

[0065] A management system for an IDC (Internet Data Center) operations and maintenance knowledge graph is also provided. The system includes a data center operations and maintenance distribution parameter acquisition module, a knowledge graph network construction module, an operations and maintenance graph construction parameter generation module, an operations and maintenance management correlation coefficient acquisition module, a multimodal operations and maintenance management model construction module, and an operations and maintenance management method generation module.

[0066] The data center operation and maintenance distribution parameter acquisition module acquires the operation and maintenance distribution parameters of the data center.

[0067] The knowledge graph network construction module constructs a knowledge graph network based on the glyphs of operation and maintenance entities and the glyphs of fault types.

[0068] The operation and maintenance graph construction parameter generation module calculates the operation and maintenance graph construction parameters using knowledge graph analysis methods.

[0069] The operation and maintenance management correlation coefficient acquisition module: processes the improvement ratio parameters set before and after data center operation and maintenance to obtain the operation and maintenance management correlation coefficient;

[0070] The multimodal operation and maintenance management model construction module processes the operation and maintenance distribution parameters, the operation and maintenance graph construction parameters, and the operation and maintenance management correlation coefficients to form data center operation and maintenance management features, collects corresponding operation and maintenance management methods, and constructs a multimodal operation and maintenance management model using the data center operation and maintenance management features and the operation and maintenance management methods. The multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph.

[0071] The operation and maintenance management method generation module: uses the multimodal operation and maintenance management model to process the characteristics of the data center to be operated and maintained, generates a real-time operation and maintenance management method for operation and maintenance, and performs operation and maintenance management on the data center according to the real-time operation and maintenance management method.

[0072] Furthermore, the operation and maintenance distribution parameters include the operation and maintenance security level coefficient, the operation and maintenance data center level, and the operation and maintenance status data, and their calculation formula is as follows:

[0073]

[0074] In the formula, Here are the operation and maintenance distribution parameters, where i represents the i-th fault log and m represents the total number of fault logs (m). For the operation and maintenance security level coefficient, This represents the number of fault entries in the i-th fault log. Its value is derived from the total number of fault entries in the fault log. The operation and maintenance security level coefficient is divided into three values: 1, 2, and 5, corresponding to no fault entries, less than or equal to 2 fault entries, and greater than 2 fault entries, respectively. Let represent the operational data center level corresponding to the i-th fault log. This level is determined by taking values ​​based on the number of standard racks in the data center. The values ​​are 2, 4, 6, and 8 for each of the following standard rack counts: less than 1000, greater than or equal to 1000 but less than or equal to 3000, greater than 3000 but less than or equal to 5000, and greater than 5000. These values ​​represent the optimal range of operational distribution parameters for model training. This is a creative representation of the data parameters used to train the model. Here, j represents the j-th operational operation of the data center, and n represents the total number of operational operations. This represents the actual time consumed during the j-th maintenance operation. This represents the theoretical time consumed in the j-th maintenance operation. The ratio of maintenance time consumed can objectively reflect the actual maintenance situation of the data center. Combining the maintenance security level coefficient, the maintenance data center level, and the maintenance situation data can provide a good feedback on the acquisition of maintenance management methods. Since the maintenance management method is affected by the required security level and the specific level of the data center to be maintained, the applicability of most maintenance management methods depends on the maintenance data center level and the maintenance situation. Therefore, this invention fully considers the impact of this type of data on the rationality of maintenance management methods and optimizes the calculation of maintenance distribution parameters suitable for subsequent graph neural network model classification to classify maintenance management methods.

[0075] Furthermore, the multimodal operation and maintenance management model adopts a graph network model based on an improved operation and maintenance graph, and its activation function calculation formula is shown below:

[0076]

[0077] X represents the activation function value, and X represents the input data center operation and maintenance management characteristics. Parameters are used to construct the operation and maintenance map corresponding to the input data center operation and maintenance management characteristics. The sum of the parameters used to construct the operation and maintenance graph for training the model. The number of training sets used for model training, based on the knowledge graph analysis described in this invention, allows activation functions to introduce nonlinear transformations into neural networks, enabling them to fit arbitrarily complex functions and restrict neuron outputs to adapt to scenario requirements. The knowledge graph network in this invention has a significant impact on neural network construction. By utilizing the operation and maintenance graph construction parameters, scene-adaptive gradient propagation adjustment of the graph neural network can be performed, resulting in optimized model training performance. To adapt to changes using variations of the basic activation function, comparing the average operation and maintenance graph construction parameters of the training set with the operation and maintenance graph construction parameters corresponding to the input data center operation and maintenance management characteristics and taking the corresponding logarithmic function value can greatly alleviate the gradient vanishing problem, making graph neural network training possible. The stable propagation of gradients ensures that each layer of the network can effectively learn, providing strong support for model construction.

[0078] This invention discloses a management method for an IDC (Internet Data Center) operation and maintenance (IDC) knowledge graph. It collects and processes data center operation and maintenance distribution parameters, operation and maintenance graph construction parameters, and operation and maintenance management correlation coefficients to obtain data center operation and maintenance management features. These features, along with corresponding operation and maintenance management methods, are used to construct a multimodal operation and maintenance management model. The model is based on a graph neural network model with ReLU activation function-based scenario adaptation improvements. It receives data center operation and maintenance features to be processed, processes these features, and generates real-time operation and maintenance management methods. This invention employs a feature-corrected knowledge graph method and data center operation and maintenance improvement ratio parameters. It uses a graph network model constructed with operation and maintenance knowledge keywords and knowledge graph parameters to obtain operation and maintenance management methods. This achieves unique data processing for data center operation and maintenance, making operation and maintenance management more automated, helping to save operation and maintenance resources, reduce operation and maintenance risks, and enhance the rationality of operation and maintenance management.

[0079] like Figure 5 The diagram shows the necessary computer-related equipment for implementing the IDC operation and maintenance knowledge graph management method of the present invention.

[0080] The combination of multiple embodiments of the present invention can achieve all the above effects, but it is not required that each embodiment of the present invention achieve all the above advantages and effects, because each embodiment of the present invention can constitute a separate technical solution and make one or more contributions to the prior art.

[0081] For any module structures not specifically defined in this invention, the existing technical specifications shall prevail. The existing technical specifications mentioned in the foregoing background and specific embodiments sections are considered part of this invention and are used to understand the meaning of certain technical features or parameters. The scope of protection of this invention is determined by the actual contents of the claims.

Claims

1. A method for managing an IDC operation and maintenance knowledge graph, characterized in that, The method comprises the following steps: Obtain the operation and maintenance distribution parameters of the data center, calculate the operation and maintenance graph construction parameters using a knowledge graph analysis method, process the operation and maintenance management correlation coefficients according to the set improvement proportion parameters before and after the operation and maintenance of the data center, and collect the corresponding operation and maintenance management methods; Process the operation and maintenance distribution parameters, the operation and maintenance graph construction parameters, and the operation and maintenance management correlation coefficients to form the data center operation and maintenance management features; Use the data center operation and maintenance management features and the operation and maintenance management methods to construct a multi-modal operation and maintenance management model, which is an improved graph network model based on the operation and maintenance graph; Use the multi-modal operation and maintenance management model to process the data center operation and maintenance management features, generate real-time operation and maintenance management methods for operation and maintenance, and manage the data center according to the real-time operation and maintenance management methods; The operation and maintenance distribution parameters include operation and maintenance safety level coefficients, operation and maintenance data center levels, and operation and maintenance situation data, and the calculation formula is: ， In the formula, is the operation and maintenance distribution parameter, i represents the i th fault log, m represents the total number of fault logs, n represents the total number of operation and maintenance times, is the level of the operation and maintenance data center corresponding to the i th fault log, represents the number of faults of the i th fault log, represents the actual time consumed by the j th operation and maintenance, represents the theoretical time consumed by the j th operation and maintenance; The operation and maintenance graph construction parameters are calculated according to the knowledge graph network relationship calculation formula, and are specifically calculated using the feature values of operation and maintenance entity glyphs, the feature values of fault type glyphs, and the degrees of knowledge graph network edges; ， In the formula is the operation and maintenance graph construction parameter, is the number of the first R types of character shapes most frequently used by the operation and maintenance entity, is the number of the first T types of character shapes related to the number of fault type character shapes, represents the a-th most frequently used operation and maintenance entity character shape, and b represents the b-th character shape related to the number of fault type character shapes, represents the character feature value of the a-th most frequently used operation and maintenance entity character shape, represents the character feature value of the b-th character shape related to the number of fault type character shapes, represents the correlation degree calculation reference value, is the number of the largest edge in the knowledge graph network, is the maximum weight of the edge, is the minimum weight of the edge; The operation and maintenance management correlation coefficients are calculated using the set improvement proportion parameters before and after the operation and maintenance of the data center: ， In the formula, is the operation and maintenance management correlation coefficient, and respectively represent the de-dimensioned value of the data center disk usage rate set after operation and maintenance, and the de-dimensioned value of the data center disk usage rate before operation and maintenance, and respectively represent the de-dimensioned value of the total energy consumption of the data center set after operation and maintenance, and the de-dimensioned value of the total energy consumption of the data center before operation and maintenance, is an operation and maintenance management correlation coefficient correction parameter for facilitating data processing of the model.

2. The IDC operation and maintenance knowledge graph management method of claim 1, characterized in that: Before the operation and maintenance graph construction parameters are calculated using the knowledge graph analysis method, a relationship graph of operation and maintenance entities and fault types is constructed, a knowledge graph network is constructed according to the top R glyphs of operation and maintenance entities and the top T glyphs of fault types that appear most frequently, and the top R glyphs of operation and maintenance entities and the top T glyphs of fault types that appear most frequently are represented in the form of graph network nodes and edge degrees after being processed using natural language processing technology.

3. The IDC operation and maintenance knowledge graph management method of claim 1, characterized in that: The data center operation and maintenance management features are obtained by transversely linking and splicing the operation and maintenance distribution parameters, the operation and maintenance graph construction parameters, and the operation and maintenance management correlation coefficients.

4. The IDC operation and maintenance knowledge graph management method of claim 3, characterized in that: The multi-modal operation and maintenance management model is an improved graph network model based on the operation and maintenance graph, and the activation function calculation formula is as follows: ， Xi is the input data center operation and maintenance feature, Xi is the input data center operation and maintenance feature corresponding to the operation and maintenance graph construction parameter, Xi is the sum of the operation and maintenance graph construction parameters of the M training set data used for training the model, Xi is the number of training sets used for model training.

5. An IDC operation and maintenance knowledge graph management system, comprising a data center operation and maintenance distribution parameter acquisition module, a knowledge graph network construction module, an operation and maintenance graph construction parameter generation module, an operation and maintenance management correlation coefficient acquisition module, a multi-modal operation and maintenance management model construction module, and an operation and maintenance management method generation module, characterized in that: The data center operation and maintenance distribution parameter acquisition module acquires the operation and maintenance distribution parameters of the data center; The knowledge graph network construction module constructs a knowledge graph network according to the operation and maintenance entity glyphs and the fault type glyphs; The operation and maintenance graph construction parameter generation module calculates the operation and maintenance graph construction parameters using a knowledge graph analysis method; The multi-modal operation and maintenance management model construction module uses the data center operation and maintenance management features and the operation and maintenance management methods to construct a multi-modal operation and maintenance management model, which is an improved graph network model based on the operation and maintenance graph; and The operation and maintenance management method generation module generates real-time operation and maintenance management methods for operation and maintenance. The operation and maintenance management correlation coefficient acquisition module: for the data center before and after operation and maintenance, the set improvement ratio parameter processing obtains the operation and maintenance management correlation coefficient; The multi-modal operation and maintenance management model construction module: processing the operation and maintenance distribution parameter, the operation and maintenance graph construction parameter and the operation and maintenance management correlation coefficient forms the data center operation and maintenance management feature, and collects the corresponding operation and maintenance management method, uses the data center operation and maintenance management feature and the operation and maintenance management method to construct the multi-modal operation and maintenance management model, the multi-modal operation and maintenance management model adopts the graph network model based on the operation and maintenance graph improvement; The operation and maintenance management method generation module: using the multi-modal operation and maintenance management model to process the data center to be operated and maintained management feature, generating the real-time operation and maintenance management method for operation and maintenance, according to the real-time operation and maintenance management method, the data center is operated and maintained management; The operation and maintenance distribution parameter includes operation and maintenance security level coefficient, operation and maintenance data center level and operation and maintenance condition data, its calculation formula is: ， In the formula, is the operation and maintenance distribution parameter, i represents the i th fault log, m represents the total number of fault logs, n represents the total number of operation and maintenance times, is the level of the operation and maintenance data center corresponding to the i th fault log, represents the number of faults of the i th fault log, represents the actual time consumed by the j th operation and maintenance, represents the theoretical time consumed by the j th operation and maintenance; The operation and maintenance graph construction parameter is calculated according to the knowledge graph network relationship calculation formula, which is calculated according to the characteristic value of the operation and maintenance entity character, the characteristic value of the fault type character and the degree of the knowledge graph network edge: ， In the formula is the operation and maintenance graph construction parameter, is the number of the first R types of character shapes most frequently used by the operation and maintenance entity, is the number of the first T types of character shapes related to the number of fault type character shapes, represents the a-th most frequently used operation and maintenance entity character shape, and b represents the b-th character shape related to the number of fault type character shapes, represents the character feature value of the a-th most frequently used operation and maintenance entity character shape, represents the character feature value of the b-th character shape related to the number of fault type character shapes, represents the correlation degree calculation reference value, is the number of the largest edge in the knowledge graph network, is the maximum weight of the edge, is the minimum weight of the edge; The operation and maintenance management correlation coefficient is calculated and obtained by using the set improvement ratio parameter before and after the operation and maintenance of the data center: ， In the formula, is the operation and maintenance management correlation coefficient, and respectively represent the de-dimensioned value of the data center disk usage rate set after operation and maintenance and the de-dimensioned value of the data center disk usage rate before operation and maintenance, and respectively represent the de-dimensioned value of the total energy consumption of the data center set after operation and maintenance and the de-dimensioned value of the total energy consumption of the data center before operation and maintenance, is an operation and maintenance management correlation coefficient correction parameter for facilitating data processing of the model.

6. The IDC operation and maintenance knowledge graph management system of claim 5, characterized in that: The multi-modal operation and maintenance management model adopts the graph network model based on the operation and maintenance graph improvement, and the activation function calculation formula is as follows: ， Xi is the input data center operation and maintenance feature, Xi is the input data center operation and maintenance feature corresponding to the operation and maintenance graph construction parameter, Xi is the input data center operation and maintenance feature corresponding to the operation and maintenance graph construction parameter, Xi is the input data center operation and maintenance feature corresponding to the operation and maintenance graph construction parameter,

Images