Data center monitoring system and method based on artificial intelligence
Through the artificial intelligence monitoring system, the network attacks and risks of the data middle platform are identified and predicted, and the security situation prediction model is built, which solves the security threats of the data middle platform in the collection, transmission, analysis and permission management stages, and improves data security and accuracy .
Patent Information
- Application Number
- CN202411132053.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-08-19
AI Technical Summary
The data middle platform is vulnerable to security threats such as network attacks and data leakage and tampering during the data collection, transmission, analysis and permission management stages, resulting in data leakage and security risks. It is difficult to effectively prevent data abuse and leakage from different business departments. .
Adopt artificial intelligence-based data middle-end monitoring system to identify and predict network attacks and data risks through risk identification, behavior sequence construction, prediction model and interpretation algorithm, calculate risk contribution and membership, and build security using Bayesian neural network Situation prediction model, predicts the security situation of the data middle platform.
It improves the accuracy of security situation prediction of the data middle platform, reduces the risks of data leakage and tampering, and ensures the security and reliability of data.
Smart Images

Figure CN118656836B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data management and artificial intelligence technology, and specifically to an artificial intelligence-based data center monitoring system and method. Background Art
[0002] With the rapid development of informatization and the widespread application of digital technology, the development of various fields involves the management of massive data. In the context of technical needs, the data middle platform came into being. As a centralized management and service platform for internal enterprise data, the data middle platform collects data in real time through core big data management and big data analysis technologies; governs and stores data according to unified data standards; retrieves data resources through data engines, empowers key business scenarios in the industry, and becomes the mainstay of digital operations for enterprises.
[0003] As the scale of internal enterprise data continues to expand, data security issues have become more prominent. The data center needs to centrally manage and securely control internal enterprise data. Otherwise, once the data center service is tampered with or other security incidents occur, it will bring security risks to the entire data center, and may lead to data leakage and security issues. Data security is difficult to ensure. First, the risks that may be encountered during data collection will lead to the inability to obtain valuable data during operational analysis or even data leakage. Second, the data center is vulnerable to various network attacks at various stages of processing, and will face various data leakage and tampering problems. Third, different business departments have different data access and usage permissions. How to prevent data leakage and abuse while using it efficiently has become crucial.
[0004] Therefore, it becomes crucial to predict the security situation of the data center to prevent problems such as data leakage and tampering. Summary of the invention
[0005] The purpose of the present invention is to provide an artificial intelligence-based data middle platform monitoring system and method. By considering various network attacks and manual operation risks or hardware equipment risks at each stage of data middle platform processing, the security situation of the data middle platform is predicted, thereby preventing various problems such as data leakage and tampering to a certain extent.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0007] The data middle station monitoring method based on artificial intelligence includes the following steps:
[0008] Identify risks in the data collection stage, data transmission stage, data analysis stage, and data rights management stage;
[0009] Clean the identified risk data, classify the risk behaviors corresponding to the cleaned data, and construct a risk behavior sequence;
[0010] A risk behavior prediction model is constructed based on historical risk sequences and their corresponding risk indexes, and the currently acquired risk behavior sequence is input into the risk behavior prediction model to predict risk behavior;
[0011] Based on the historical risk behavior data, the SHAP interpretation algorithm is used to calculate the contribution of the risk data to the risk behavior change, and based on the contribution, the overall membership of the risk behavior prediction and the overall average risk level are calculated;
[0012] Based on the overall membership of historical risk behavior data and its corresponding risk behavior prediction, the overall average risk level and the security situation of the data center, a security situation prediction model is constructed using a Bayesian neural network;
[0013] Based on the predicted risk behavior data, the security situation prediction model is used to predict the security situation of the data center.
[0014] According to the above technical solution, the risk identification step includes:
[0015] Identify network attacks during the data collection, data transmission, data analysis, and data rights management stages; network attacks include but are not limited to man-in-the-middle attacks, denial of service attacks, phishing attacks, zero-day attacks, supply chain attacks, tampering attacks, forgery attacks, eavesdropping attacks, weak password and reused password attacks, and vulnerability exploitation attacks;
[0016] Calculate the data connection risks in the data collection stage; the data connection risks refer to the various forms from direct connection to DDI, network loading, FTP, etc., which need to deal with various ERP management software on the market and have requirements for technology updates. Or traditional data collection companies may focus on data processing and neglect technology development, resulting in the current technical capabilities being out of touch with reality and unable to provide more suitable services;
[0017] Calculate the data transmission risks existing in the data transmission stage; the data transmission risks include insecure transmission operation, insecure mobile media, insecure transmission equipment and network, unencrypted or insufficiently encrypted data, insecure transmission protocol, etc.;
[0018] Calculate the data quality risks that exist during the data analysis phase;
[0019] Calculate the data permission risks existing in the data permission management stage; the data permission risks include permission abuse risk, permission leakage risk, permission configuration error risk, permission audit deficiency risk, and shared account risk.
[0020] According to the above technical solution, the data connection risk : ;
[0021] The data transmission risk is calculated by multiplication method or matrix method according to the threat, vulnerability, asset value and loss degree;
[0022] Data quality risks : ;
[0023] The data permission risks described: ;
[0024] In the formula, Indicates data connection risk. represents the risk calculation function, Indicates the possibility of security incidents occurring during the data collection phase, Represents the loss after a security incident occurs during the data collection phase. Indicates data quality risk, Indicates the proportion of data error rate. Represents the error rate of the data, Indicates the proportion of missing data. represents the missing rate of data, Indicates the proportion of inconsistencies between data from different data sources or at different time points. Indicates the inconsistency ratio between data from different data sources or at different time points. Indicates the proportion of the timeliness difference of data. Indicates that the data is out of date. Indicates data permission risk, Indicates the possibility of security threats occurring during the data rights management phase. Indicates the impact of a possible security threat.
[0025] According to the above technical solution, network attack behaviors, data connection risks, data transmission risks, data quality risks and data permission risks that may lead to the same risk behavior are classified into the same category, thereby constructing the risk behavior sequence;
[0026] Among them, data connection risk, data transmission risk, data quality risk, data permission risk and data corresponding to the same network attack behavior exist in at least one risk behavior sequence. The risk behavior is not limited to data leakage, data tampering, denial of service, etc.;
[0027] According to the above technical solution, the risk behavior prediction model is constructed using a multi-category neural network based on risk data, risk sequences corresponding to risk behaviors and their corresponding risk indexes. The specific steps include:
[0028] Based on the risk data, extracting risk sequence feature data corresponding to the risk behavior from the risk data and establishing a feature set of the multi-category neural network, and then normalizing the feature set;
[0029] The normalized feature set is trained in batches to establish a risk behavior prediction model.
[0030] According to the above technical solution, the security situation prediction step of the data center includes:
[0031] Determine the risk level corresponding to the risk behavior based on the risk behavior prediction value;
[0032] The SHAP interpretation algorithm is used to obtain the contribution of the risk data identified in each spatial grid point to the change of risk behavior, and the contribution of different risk behaviors at the same time is divided;
[0033] Calculate the overall membership of risk behavior prediction based on the divided contribution areas of risk behavior, and calculate the average risk level of the overall risk behavior prediction;
[0034] The risk behavior prediction value, overall membership degree and overall average risk level are input into the security situation prediction model to predict the security situation of the data center.
[0035] According to the above technical solution, the contribution degree is less than or equal to the threshold The risk behavior is divided into one category, and the contribution level is marked as 1; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 2; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 3; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 4; the contribution level is greater than the threshold The risk behaviors are classified into one category and marked with a contribution level of 5.
[0036] The SHAP interpretation algorithm obtains the contribution of the risk data identified in each spatial grid point to the change in risk behavior. Since one set of risk data corresponds to multiple sets of risk behaviors, the contribution can be used to explore the correlation of each risk behavior. However, due to the large number of risk behaviors, the workload of exploring the correlation of each risk behavior is too large. In order to reduce the workload, the risk behaviors are divided according to the contribution, and the correlation of the overall risk behavior is explored according to the divided risk areas, that is, the overall membership of the risk behavior is calculated.
[0037] At the same time, since different risk behaviors may cause the same security situation, in order to distinguish them to a certain extent, the overall average risk level of the risk behavior is used as a restriction condition.
[0038] According to the above technical solution, the overall membership of the risk behavior prediction :
[0039] ;
[0040] In the formula, represents the overall membership of risk behavior prediction, Indicates The contribution level of each risk behavior classification area mark, Indicates The contribution level of each risk behavior classification area mark, Indicates The number of risk behaviors contained in each risk behavior classification area, Indicates The number of risk behaviors contained in each risk behavior classification area.
[0041] Also included is an embodiment of a data middle station monitoring system based on artificial intelligence, including:
[0042] The risk identification module identifies risks in the data collection stage, data transmission stage, data analysis stage, and data authority management stage;
[0043] The risk behavior sequence construction module cleans the identified risk data, classifies the risk behaviors corresponding to the cleaned data, and constructs a risk behavior sequence;
[0044] The risk behavior prediction module builds a risk behavior prediction model based on the historical risk sequence and its corresponding risk index, and inputs the currently acquired risk behavior sequence into the risk behavior prediction model to predict the risk behavior;
[0045] The security situation prediction model construction module calculates the contribution of risk data to the change of risk behavior based on historical risk behavior data using the SHAP interpretation algorithm, and calculates the overall membership of risk behavior prediction and the overall average risk level based on the contribution; constructs a security situation prediction model based on the overall membership of historical risk behavior data and its corresponding risk behavior prediction, the overall average risk level and the security situation of the data center using a Bayesian neural network;
[0046] The security situation prediction module uses the security situation prediction model to predict the security situation of the data center based on the predicted risk behavior data.
[0047] According to the above technical solution, the specific execution steps of the security situation prediction module include:
[0048] Determine the risk level corresponding to the risk behavior based on the risk behavior prediction value;
[0049] The SHAP interpretation algorithm is used to obtain the contribution of the risk data identified in each spatial grid point to the change of risk behavior, and the contribution of different risk behaviors at the same time is divided;
[0050] Calculate the overall membership of risk behavior prediction based on the divided contribution areas of risk behavior, and calculate the average risk level of the overall risk behavior prediction;
[0051] The risk behavior prediction value, overall membership degree and overall average risk level are input into the security situation prediction model to predict the security situation of the data center.
[0052] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention predicts risk behaviors by acquiring various risk data existing in the middle station data during operation, and calculates the overall membership of risk behaviors by calculating the contribution of risk data to risk behaviors. The essence of its membership is to explore the correlation between various risk behaviors, because the correlation between different risk behaviors will also lead to different security situations, and then use the security situation prediction model to predict the security situation of the data middle station based on the predicted risk behavior, membership, and average risk level. The present invention takes into account the correlation between various risk behaviors, and uses the average risk level of risk behaviors as a constraint condition to predict the security situation of the data middle station, thereby improving the accuracy of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0053] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0054] Figure 1 It is a flowchart of the steps of the data middle station monitoring method based on artificial intelligence of the present invention;
[0055] Figure 2 It is a flowchart of the security situation prediction steps of the data center. DETAILED DESCRIPTION
[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0057] Embodiment 1: A data center monitoring method based on artificial intelligence, comprising the following steps:
[0058] S1. Identify risks in the data collection stage, data transmission stage, data analysis stage and data authority management stage. Specific implementation:
[0059] Identify network attacks during the data collection, data transmission, data analysis, and data rights management stages, including man-in-the-middle attacks, denial of service attacks, phishing attacks, zero-day attacks, supply chain attacks, tampering attacks, forgery attacks, eavesdropping attacks, weak password and reused password attacks, and vulnerability exploitation attacks;
[0060] Calculate the data connection risks in the data collection stage, data connection risks : ;
[0061] Calculate the data transmission risk in the data transmission stage. The data transmission risk is calculated based on the threat, vulnerability, asset value and loss degree using the multiplication method or matrix method.
[0062] Calculate the data quality risks in the data analysis phase, data quality risks : ;
[0063] Calculate the data permission risks that exist in the data permission management stage. Data permission risks: ;
[0064] In the formula, Indicates data connection risk. represents the risk calculation function, Indicates the possibility of security incidents occurring during the data collection phase, Represents the loss after a security incident occurs during the data collection phase. Indicates data quality risk, Indicates the proportion of data error rate. Represents the error rate of the data, Indicates the proportion of missing data. represents the missing rate of data, Indicates the proportion of inconsistencies between data from different data sources or at different time points. Indicates the inconsistency ratio between data from different data sources or at different time points. Indicates the proportion of the timeliness difference of data. Indicates that the data is out of date. Indicates data permission risk, Indicates the possibility of security threats occurring during the data rights management phase. Indicates the impact of a possible security threat.
[0065] S2. Clean the identified risk data, and classify the risk behaviors (data leakage, data tampering, denial of service) corresponding to the cleaned data to construct a risk behavior sequence; specifically, classify the network attack behaviors, data connection risks, data transmission risks, data quality risks, and data permission risks that may lead to the same risk behavior into the same category, and then construct the risk behavior sequence;
[0066] Among them, data connection risk, data transmission risk, data quality risk, data permission risk and data corresponding to the same network attack behavior exist in at least one risk behavior sequence. Risk behaviors include but are not limited to data leakage, data tampering, and denial of service;
[0067] Among them, the risk behavior sequence corresponding to the data leakage risk behavior includes but is not limited to data connection risk, data transmission risk, data permission risk, eavesdropping attack, weak password and reused password attack, vulnerability exploitation attack, and zero-day attack data;
[0068] The risk behavior sequence corresponding to the data tampering risk behavior includes but is not limited to data connection risk, data transmission risk, data permission risk, data quality risk, man-in-the-middle attack, tampering attack, forgery attack, weak password and reused password attack, vulnerability exploitation attack, and phishing attack data;
[0069] The risk behavior sequence corresponding to the denial of service risk behavior includes but is not limited to data corresponding to denial of service attacks, weak password and reused password attacks, data permission risks, and vulnerability exploitation attacks.
[0070] S3. Construct a risk behavior prediction model based on historical risk sequences and their corresponding risk indexes, and input the currently acquired risk behavior sequence into the risk behavior prediction model to predict risk behavior. Specifically, the following steps are performed: based on the risk data, extract the risk sequence feature data corresponding to the risk behavior from the risk data and establish a feature set of the multi-category neural network, and then normalize the feature set; batch train the normalized feature set to establish a risk behavior prediction model, and use the risk behavior prediction model to predict risk behavior.
[0071] S4. Based on the historical risk behavior data, the contribution of the risk data to the risk behavior change is calculated using the SHAP interpretation algorithm, and based on the contribution, the overall membership of the risk behavior prediction and the overall average risk level are calculated;
[0072] Risk behaviors such as data leakage, data tampering, and denial of service have different proportions in the total risk behaviors, and the overall security situation of the corresponding data platform is also different. Therefore, the training set and validation set are divided according to the risk behavior, the overall membership degree of risk behavior prediction, the overall average risk level, and the corresponding overall security situation of the data platform. Based on the training set and validation set, a security situation prediction model is constructed using the Bayesian neural network.
[0073] S5. Based on the predicted risk behavior data, the security situation prediction model is used to predict the security situation of the data center. The specific steps include:
[0074] S501. Determine a risk level corresponding to the risk behavior based on the risk behavior prediction value.
[0075] S502: Use the SHAP interpretation algorithm to obtain the contribution of the risk data identified in each spatial grid point to the change of risk behavior, and divide the contribution of different risk behaviors at the same time. Specifically, the contribution is less than or equal to the threshold value. The risk behavior is divided into one category, and the contribution level is marked as 1; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 2; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 3; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 4; the contribution level is greater than the threshold The risk behaviors are classified into one category and marked with a contribution level of 5.
[0076] S503: Calculate the overall membership of risk behavior prediction based on the divided risk behavior contribution areas, and calculate the overall average risk level of risk behavior prediction; the overall membership of risk behavior prediction :
[0077] ;
[0078] In the formula, represents the overall membership of risk behavior prediction, Indicates The contribution level of each risk behavior classification area mark, Indicates The contribution level of each risk behavior classification area mark, Indicates The number of risk behaviors contained in each risk behavior classification area, Indicates The number of risk behaviors contained in each risk behavior classification area.
[0079] S504. Input the risk behavior prediction value, the overall membership degree and the overall average risk level into the security situation prediction model to predict the security situation of the data center.
[0080] Embodiment 2: A data middle platform monitoring system based on artificial intelligence, comprising:
[0081] The risk identification module identifies risks in the data collection stage, data transmission stage, data analysis stage, and data authority management stage;
[0082] The risk behavior sequence construction module cleans the identified risk data, classifies the risk behaviors corresponding to the cleaned data, and constructs a risk behavior sequence;
[0083] The risk behavior prediction module builds a risk behavior prediction model based on the historical risk sequence and its corresponding risk index, and inputs the currently acquired risk behavior sequence into the risk behavior prediction model to predict the risk behavior;
[0084] The security situation prediction model construction module calculates the contribution of risk data to the change of risk behavior based on historical risk behavior data using the SHAP interpretation algorithm, and calculates the overall membership of risk behavior prediction and the overall average risk level based on the contribution; constructs a security situation prediction model based on the overall membership of historical risk behavior data and its corresponding risk behavior prediction, the overall average risk level and the security situation of the data center using a Bayesian neural network;
[0085] The security situation prediction module uses the security situation prediction model to predict the security situation of the data center based on the predicted risk behavior data. The specific execution steps include:
[0086] Determine the risk level corresponding to the risk behavior based on the risk behavior prediction value;
[0087] The SHAP interpretation algorithm is used to obtain the contribution of the risk data identified in each spatial grid point to the change of risk behavior, and the contribution of different risk behaviors at the same time is divided;
[0088] Calculate the overall membership of risk behavior prediction based on the divided contribution areas of risk behavior, and calculate the average risk level of the overall risk behavior prediction;
[0089] The risk behavior prediction value, overall membership degree and overall average risk level are input into the security situation prediction model to predict the security situation of the data center.
[0090] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0091] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. The data center monitoring method based on artificial intelligence is characterized by: The steps include: Identify risks in the data collection stage, data transmission stage, data analysis stage, and data rights management stage; Clean the identified risk data, classify the risk behaviors corresponding to the cleaned data, and construct a risk behavior sequence; A risk behavior prediction model is constructed based on historical risk sequences and their corresponding risk indexes, and the currently acquired risk behavior sequence is input into the risk behavior prediction model to predict risk behavior; Based on the historical risk behavior data, the SHAP interpretation algorithm is used to calculate the contribution of the risk data to the risk behavior change, and based on the contribution, the overall membership of the risk behavior prediction and the overall average risk level are calculated; Based on the overall membership of historical risk behavior data and its corresponding risk behavior prediction, the overall average risk level and the security situation of the data center, a security situation prediction model is constructed using a Bayesian neural network; The security situation prediction model is used to predict the security situation of the data center based on the predicted risk behavior data. The security situation prediction steps of the data center include: Determine the risk level corresponding to the risk behavior based on the risk behavior prediction value; The SHAP interpretation algorithm is used to obtain the contribution of the risk data identified in each spatial grid point to the change of risk behavior, and the contribution of different risk behaviors at the same time is divided; The overall membership of the risk behavior prediction is calculated based on the contribution area of the divided risk behavior, and the average risk level of the risk behavior prediction as a whole is calculated; the overall membership of the risk behavior prediction : ; In the formula, represents the overall membership of risk behavior prediction, Indicates The contribution level of each risk behavior classification area mark, Indicates The contribution level of each risk behavior classification area mark, Indicates The number of risk behaviors contained in each risk behavior classification area, Indicates The number of risk behaviors contained in each risk behavior classification area; The risk behavior prediction value, overall membership degree and overall average risk level are input into the security situation prediction model to predict the security situation of the data center.
2. The data center monitoring method based on artificial intelligence according to claim 1 is characterized in that: The risk identification steps include: Identify cyber attacks during the data collection, data transmission, data analysis, and data rights management stages; Identify cyber attacks during the data collection, data transmission, data analysis, and data rights management stages; Calculate the data connection risks during the data collection phase; Calculate the data transmission risks during the data transmission phase; Calculate the data quality risks that exist during the data analysis phase; Calculate the data permission risks that exist during the data permission management phase.
3. The data center monitoring method based on artificial intelligence according to claim 2 is characterized in that: The risks of data connection : ; The data transmission risk is calculated by multiplication method or matrix method according to the threat, vulnerability, asset value and loss degree; Data quality risks : ; The data permission risks described: ; In the formula, Indicates data connection risk. represents the risk calculation function, Indicates the possibility of security incidents occurring during the data collection phase, Represents the loss after a security incident occurs during the data collection phase. Indicates data quality risk, Indicates the proportion of data error rate. Represents the error rate of the data, Indicates the proportion of missing data. represents the missing rate of data, Indicates the proportion of inconsistencies between data from different data sources or at different time points. Indicates the inconsistency ratio between data from different data sources or at different time points. Indicates the proportion of the timeliness difference of data. Indicates that the data is out of date. Indicates data permission risk, Indicates the possibility of security threats occurring during the data rights management phase. Indicates the impact of a possible security threat.
4. The data center monitoring method based on artificial intelligence according to claim 1 is characterized in that: Classify network attack behaviors, data connection risks, data transmission risks, data quality risks, and data permission risks that may lead to the same risk behavior into the same category, and then construct the risk behavior sequence; Among them, data connection risks, data transmission risks, data quality risks, data authority risks, and data corresponding to the same type of network attack behavior exist in at least one risk behavior sequence.
5. The data center monitoring method based on artificial intelligence according to claim 1 is characterized in that: The risk behavior prediction model is constructed using a multi-category neural network based on risk data, risk sequences corresponding to risk behaviors and their corresponding risk indexes.
6. The data center monitoring method based on artificial intelligence according to claim 1 is characterized in that: The contribution is less than or equal to the threshold The risk behavior is divided into one category, and the contribution level is marked as 1; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 2; the contribution level is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 3; The contribution is greater than the threshold and less than or equal to the threshold The risk behaviors are classified into one category, and the contribution level is marked as 4; The contribution is greater than the threshold The risk behaviors are classified into one category and marked with a contribution level of 5.
7. The data center monitoring system based on artificial intelligence is characterized by: include: The risk identification module identifies risks in the data collection stage, data transmission stage, data analysis stage, and data authority management stage; The risk behavior sequence construction module cleans the identified risk data, classifies the risk behaviors corresponding to the cleaned data, and constructs a risk behavior sequence; The risk behavior prediction module builds a risk behavior prediction model based on the historical risk sequence and its corresponding risk index, and inputs the currently acquired risk behavior sequence into the risk behavior prediction model to predict the risk behavior; The security situation prediction model building module calculates the contribution of risk data to the change of risk behavior based on historical risk behavior data using the SHAP interpretation algorithm, and calculates the overall membership of risk behavior prediction and the overall risk average level based on the contribution; Based on the overall membership of historical risk behavior data and its corresponding risk behavior prediction, the overall average risk level and the security situation of the data center, a security situation prediction model is constructed using a Bayesian neural network; The security situation prediction module uses the security situation prediction model to predict the security situation of the data center based on the predicted risk behavior data; The specific execution steps of the security situation prediction module include: Determine the risk level corresponding to the risk behavior based on the risk behavior prediction value; The SHAP interpretation algorithm is used to obtain the contribution of the risk data identified in each spatial grid point to the change of risk behavior, and the contribution of different risk behaviors at the same time is divided; The overall membership of the risk behavior prediction is calculated based on the contribution area of the divided risk behavior, and the average risk level of the risk behavior prediction as a whole is calculated; the overall membership of the risk behavior prediction : ; In the formula, represents the overall membership of risk behavior prediction, Indicates The contribution level of each risk behavior classification area mark, Indicates The contribution level of each risk behavior classification area mark, Indicates The number of risk behaviors contained in each risk behavior classification area, Indicates The number of risk behaviors contained in each risk behavior classification area; The risk behavior prediction value, overall membership degree and overall average risk level are input into the security situation prediction model to predict the security situation of the data center.
