Adaptive Dynamic Network Security Strategy Intelligent Control Method

Abstract

This invention relates to the field of network security technology, specifically disclosing an adaptive dynamic network security strategy intelligent control method. By real-time monitoring of the qubit error rate and communication channel signal strength in network traffic, abnormal feature values ​​of the communication link and channel disturbance feature values ​​are extracted respectively. Specifically, Bayesian inference is used to achieve high-sensitivity identification of quantum noise interference, and Haar wavelet transform is used to improve the robustness of external interference detection. The two types of features are fused into a comprehensive security risk feature vector, which is input into a gradient boosting tree model for training. With the goal of minimizing the prediction scoring error, a network security score is output, and threat levels are classified accordingly. The system automatically adjusts the protection strategy according to the security level and forms a closed-loop optimization control through continuous feedback, improving the adaptability and proactive defense capabilities of the network system. This solves the problems of lagging security response and inaccurate strategy adjustment in existing technologies under complex environments.

Description

Technical Field

[0001] This invention relates to the field of network security technology, and more specifically to an adaptive dynamic network security strategy intelligent control method. Background Technology

[0002] As quantum communication and traditional networks become increasingly integrated, the challenges to network security are becoming more complex. Existing network security protection systems are primarily designed for traditional communication environments, typically relying on static or semi-dynamic security policy configurations, lacking effective awareness of changes in the state of quantum communication links. In practical applications, network attacks are becoming increasingly covert and diverse, especially weak noise interference and sudden external disturbances targeting quantum channels, which are difficult to identify promptly using conventional threshold judgment methods. Furthermore, when facing multi-source heterogeneous security risks, existing systems often employ isolated analysis to address different types of threat characteristics, lacking a unified risk fusion assessment mechanism. This results in delayed security responses and untimely policy adjustments, failing to meet the high-security, highly adaptive requirements of modern network operations.

[0003] Current technologies lack the ability to collaboratively perceive weak quantum noise interference and sudden channel disturbances in a network environment that integrates quantum communication characteristics, and to build an intelligent security control mechanism with closed-loop feedback capabilities based on this. Traditional network security systems generally neglect the role of micro-indicators such as qubit error rate in overall risk assessment, and also fail to fully consider the attack intent information implied by channel signal strength fluctuations. This invention, by introducing mathematical tools such as Bayesian inference and Haar wavelet transform, combined with the intelligent learning capabilities of the gradient boosting tree model, achieves for the first time fully automated closed-loop control from low-level physical feature extraction to high-level policy regulation. This significantly improves the network system's proactive defense and dynamic adaptation capabilities in complex threat environments, filling the current technological gap in intelligent security control in quantum-classical hybrid networks. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive dynamic network security strategy intelligent control method to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] An adaptive dynamic network security strategy intelligent control method includes the following steps:

[0007] S1: During network operation, monitor the quantum bit error rate and communication channel signal strength in network traffic data in real time;

[0008] S2: Analyze the error rate of qubits and calculate the abnormal characteristic value of the communication link based on its fluctuation level to assess whether there is quantum noise interference in network transmission;

[0009] S3: Analyze the signal strength of the communication channel and calculate the channel disturbance characteristic value based on its stability changes to determine whether there is external interference;

[0010] S4: Construct a comprehensive security risk feature vector by combining communication link anomaly feature values ​​and channel disturbance feature values, input it into the intelligent control model for fusion analysis, and identify the current network security threat level;

[0011] S5: Dynamically adjust network security policies based on the identified security threat levels to achieve closed-loop optimization control of network security protection and improve the overall security and adaptability of the system.

[0012] As a further aspect of the present invention: the assessment of whether quantum noise interference exists in network transmission specifically includes:

[0013] During network operation, the qubit error rate in network traffic data is monitored in real time. The qubit error rate is analyzed, and the abnormal characteristic value of the communication link is calculated based on its fluctuation. It is then determined whether the abnormal characteristic value of the communication link is greater than or equal to a preset threshold. If it is, there is quantum noise interference in the network transmission; otherwise, there is no quantum noise interference in the network transmission.

[0014] As a further aspect of the present invention: the process for obtaining the abnormal feature value of the communication link is as follows:

[0015] The link is set to have quantum noise interference events. Initialize prior probabilities , representing the initial probability that the communication link is affected by quantum noise in the initial state of network operation;

[0016] Real-time acquisition of qubit error rates from network traffic data forms an observation dataset. ;

[0017] Calculate the interference in the link separately. and no interference Data observed under conditions likelihood function and ;

[0018] Calculate the posterior probability using Bayes' theorem;

[0019] posterior probability The nonlinear transformation maps the abnormal feature values ​​of the communication link.

[0020] As a further aspect of the present invention: the determination of whether external interference exists specifically includes:

[0021] During network operation, the signal strength of the communication channel in the network traffic data is monitored in real time. The signal strength of the communication channel is analyzed, and the channel disturbance characteristic value is calculated based on its stability changes. It is then determined whether the channel disturbance characteristic value is greater than or equal to a preset threshold. If it is, there is external interference; otherwise, there is no external interference.

[0022] As a further aspect of the present invention: the process of obtaining the channel disturbance feature value is as follows:

[0023] During network operation, a data sequence of communication channel signal strength is collected, and the data sequence of communication channel signal strength is subjected to a first-level Haar wavelet transform using the Haar wavelet transform algorithm to obtain the detail coefficients corresponding to each layer.

[0024] Calculate the energy of each level of detail coefficients;

[0025] Based on the energy distribution of detail coefficients at each layer, the energy ratio of detail coefficients between adjacent layers is calculated. The energy ratios of detail coefficients between all adjacent layers are summed to obtain the channel disturbance characteristic value.

[0026] As a further aspect of the present invention: the step of constructing a comprehensive security risk feature vector from communication link anomaly feature values ​​and channel disturbance feature values, and inputting it into an intelligent control model for fusion analysis, specifically includes:

[0027] During network operation, abnormal communication link feature values ​​and channel disturbance feature values ​​are acquired and constructed into a comprehensive security risk feature vector, which is used as input to the intelligent control model to minimize the error between the predicted network security score and the actual network security score. This vector serves as the training target for the intelligent control model, which is then trained. Based on the trained intelligent control model, a network security score is output. The intelligent control model is a gradient boosting number model.

[0028] As a further aspect of the present invention: the process for obtaining the abnormal feature value of the communication link is as follows:

[0029] A smart regulation model based on the gradient boosting tree algorithm is constructed. The comprehensive security risk feature vector is used as the model input, and the output is the predicted network security score. The actual network security scores in historical data are collected as labels, and the training set and test set are divided by cross-validation. The intelligent regulation model is iteratively trained using the training set. The optimization objective is to minimize the mean square error between the predicted network security score and the actual network security score. The model performance is evaluated using the test set to prevent overfitting.

[0030] As a further aspect of the present invention: the identification of the current network security threat level specifically includes:

[0031] Determine whether the current network security score is greater than or equal to a preset first threshold. If yes, it is recorded as a high-risk threat level. If no, determine whether the current network security score is less than or equal to a preset second threshold. If yes, it is recorded as a low-risk threat level. If no, it is recorded as a medium-risk threat level.

[0032] As a further aspect of the present invention: the step of dynamically adjusting the network security strategy based on the identified security threat level to achieve closed-loop optimization control of network security protection specifically includes:

[0033] For network states identified as high-risk threats, the system will initiate enhanced security defenses, including increasing the frequency of quantum key distribution, enabling multi-layered data encryption, and implementing stricter access control policies. For medium-risk threats, the system will take moderately enhanced defenses, including strengthening monitoring and auditing functions and regularly updating firewall rules. For low-risk threats, the system will maintain the current basic security protection configuration and continue to monitor network traffic data.

[0034] The beneficial effects of this invention are:

[0035] (1) This invention constructs a network security risk perception system with high mathematical rigor and engineering feasibility by introducing a communication link anomaly feature extraction mechanism based on Bayesian inference and a channel disturbance feature analysis method based on Haar wavelet transform. Specifically, at the quantum communication link level, a Bayesian inference framework is used to dynamically model the error rate of the real-time acquired qubits. Combining the prior interference probability and the likelihood function of the observed data, the posterior probability of the link being interfered with by quantum noise is calculated. This probability is then transformed into communication link anomaly feature values ​​through a nonlinear mapping in the form of natural logarithm, thereby achieving high-sensitivity identification of weak noise interference. At the classical communication channel level, Haar wavelet transform is used to decompose the communication channel signal strength sequence into multiple scales, extract the energy distribution features of detail coefficients, and construct channel disturbance feature values ​​by summing the energy ratios of adjacent levels, effectively capturing short-term sudden external interference events. Compared to traditional risk identification methods that rely on fixed thresholds or single statistical features, this invention not only overcomes the problem of misjudgment caused by environmental changes, but also improves the robustness and adaptability of feature extraction by combining multi-scale mathematical tools with statistical inference. This significantly enhances the system's ability to identify potential security threats and its response efficiency in complex network environments, providing high-quality input for subsequent intelligent control models and laying the theoretical and data foundation for the implementation of network adaptive security strategies.

[0036] (2) This invention employs a gradient boosting tree-based machine learning model to construct an intelligent control system. It fully leverages the advantages of this model in nonlinear feature fusion and high-dimensional data modeling, effectively integrating multi-dimensional security indicators such as communication link anomaly features from quantum communication links and channel disturbance features from classical channel levels to construct a comprehensive security risk feature vector. This vector serves as the model input, enabling in-depth perception and quantitative assessment of network security status. During model training, the system introduces an optimization strategy aimed at minimizing the mean square error between the predicted network security score and the actual score. A cross-validation mechanism is used to divide the training and testing sets, ensuring the model possesses good generalization ability and anti-overfitting performance. Iterative optimization algorithms continuously adjust model parameters, enabling the intelligent control model to accurately capture the evolution trend of network risks, thereby achieving high-precision prediction of network security scores. Based on this, the system further compares the output network security score with a preset threshold to automatically identify the current network security threat level (high risk, medium risk, or low risk), and dynamically matches corresponding network security protection strategies, including multi-layered defense measures such as key update frequency adjustment, encryption strength enhancement, and access control policy changes. Meanwhile, by continuously collecting new network operation data and feeding it back to the intelligent control model, the system forms a closed-loop optimization mechanism. This enables the entire security strategy system to respond in real time, adapt and continuously optimize when facing complex and ever-changing network attack patterns and environmental disturbances. It significantly improves the network system's proactive defense level, operational stability and intelligent management capabilities, providing solid support for building a highly secure, resilient and scalable next-generation network protection system. Attached Figure Description

[0037] The invention will now be further described with reference to the accompanying drawings.

[0038] Figure 1 This is a flowchart of the adaptive dynamic network security strategy intelligent control method of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Please see Figure 1 As shown, this invention is an adaptive dynamic network security strategy intelligent control method, comprising the following steps:

[0041] S1: During network operation, monitor the quantum bit error rate and communication channel signal strength in network traffic data in real time;

[0042] S2: Analyze the error rate of qubits and calculate the abnormal characteristic value of the communication link based on its fluctuation level to assess whether there is quantum noise interference in network transmission;

[0043] S3: Analyze the signal strength of the communication channel and calculate the channel disturbance characteristic value based on its stability changes to determine whether there is external interference;

[0044] S4: Construct a comprehensive security risk feature vector by combining communication link anomaly feature values ​​and channel disturbance feature values, input it into the intelligent control model for fusion analysis, and identify the current network security threat level;

[0045] S5: Dynamically adjust network security policies based on the identified security threat levels to achieve closed-loop optimization control of network security protection and improve the overall security and adaptability of the system.

[0046] In S1, during network operation, the qubit error rate and communication channel signal strength in network traffic data are monitored in real time, specifically including:

[0047] During network operation, the first step in real-time monitoring of the qubit error rate and communication channel signal strength in network traffic data is to deploy a specially designed data acquisition module. This module includes a high-precision quantum state measurement device and a channel state information (CSI) acquisition device. The former is used to accurately record the error rate of qubits transmitted through the network at each node, while the latter is responsible for acquiring the signal strength information of the communication channel in real time. To ensure the accuracy and reliability of the data, the quantum state measurement device employs advanced error correction algorithms to filter out the influence of environmental noise on the qubits, while the CSI acquisition device utilizes multi-antenna technology to enhance signal acquisition capabilities, combined with filters to eliminate interference signals.

[0048] Next, the collected raw data is transmitted to the central processing unit for preprocessing to meet the needs of subsequent analysis. The preprocessing process first involves data cleaning to remove outliers caused by hardware failures or transient interference. Then, time synchronization technology is used to calibrate data from different sources, ensuring the consistency of the qubit error rate with the communication channel signal strength on the time axis. Finally, all preprocessed data is formatted into a unified standard format and stored in a database for use in subsequent steps, including feature extraction, analysis, and training of intelligent control models. This process ensures the quality of the collected data, providing a solid foundation for accurately assessing network security.

[0049] In S2, the error rate of qubits is analyzed, and anomaly characteristic values ​​of the communication link are calculated based on their fluctuation level. This is used to assess whether quantum noise interference exists in network transmission, specifically including:

[0050] During network operation, the qubit error rate in network traffic data is monitored in real time. The qubit error rate is analyzed, and the abnormal characteristic value of the communication link is calculated based on its fluctuation. It is then determined whether the abnormal characteristic value of the communication link is greater than or equal to a preset threshold. If it is, there is quantum noise interference in the network transmission; otherwise, there is no quantum noise interference in the network transmission.

[0051] The process for obtaining the abnormal feature values ​​of the communication link is as follows:

[0052] The link is set to have quantum noise interference events. Initialize prior probabilities , representing the initial probability that the communication link is affected by quantum noise in the initial state of network operation;

[0053] Real-time acquisition of qubit error rates from network traffic data forms an observation dataset. ;

[0054] Calculate the interference in the link separately. and no interference Data observed under conditions likelihood function and ;

[0055] The posterior probability is calculated using Bayes' theorem, and the expression is as follows: ;

[0056] in, This represents the prior probability that the link is not disturbed. Indicating in the observation data The probability of quantum noise interference in the downlink. and This indicates the presence of interference. and no interference Data observed under conditions The likelihood function;

[0057] posterior probability The nonlinear transformation maps the data to communication link anomaly characteristic values, and the calculation expression is as follows: ;

[0058] in, Indicates abnormal characteristic values ​​of the communication link. This represents the natural logarithm function (with base e).

[0059] It should be noted that intelligent identification and quantitative assessment of quantum noise interference are achieved through real-time monitoring and statistical analysis of the qubit error rate. This method first sets the prior probability of link interference events and constructs an observation model based on real-time collected qubit error rate data. It then uses Bayes' theorem to calculate the posterior probability of quantum noise interference in the current network state, and finally maps it to anomaly characteristic values ​​of the communication link through a nonlinear transformation in the form of the natural logarithm, effectively improving the sensitivity and accuracy of judgment for weak noise interference. Compared with traditional threshold detection methods, this method has stronger adaptability and theoretical rigor, enabling it to dynamically adapt to the complex and ever-changing quantum communication environment. It improves the system's ability to perceive abnormal states in the quantum network and enhances its security early warning level, providing highly reliable input for subsequent comprehensive risk assessment and intelligent control.

[0060] In S3, the signal strength of the communication channel is analyzed, and the channel disturbance characteristic value is calculated based on its stability changes to determine whether external interference exists. Specifically, this includes:

[0061] During network operation, the signal strength of the communication channel in the network traffic data is monitored in real time. The signal strength of the communication channel is analyzed, and the channel disturbance characteristic value is calculated based on its stability changes. It is then determined whether the channel disturbance characteristic value is greater than or equal to a preset threshold. If it is, there is external interference; otherwise, there is no external interference.

[0062] The process for obtaining the channel disturbance feature value is as follows:

[0063] During network operation, a data sequence of communication channel signal strength is collected, and the data sequence of communication channel signal strength is subjected to a first-level Haar wavelet transform using the Haar wavelet transform algorithm to obtain the detail coefficients corresponding to each layer.

[0064] The energy for each level of detail coefficients is calculated using the following expression: ;

[0065] In the formula, in the formula, Indicates the first The energy of the detail coefficients in layer decomposition. Indicates the first Layer decomposition A detailed coefficient value, This represents the value of the detail factor. The Haar wavelet transform is represented by the first... The total number of detail coefficients after layer decomposition;

[0066] Based on the energy distribution of detail coefficients at each layer, the energy ratio of detail coefficients between adjacent layers is calculated. The energy ratios of detail coefficients between all adjacent layers are summed to obtain the channel disturbance characteristic value.

[0067] It should be noted that this technique achieves high-sensitivity detection of external interference by mathematically modeling the stability changes of communication channel signal strength. This method utilizes Haar wavelet transform to decompose the signal strength data sequence into multiple scales, extracting the energy distribution characteristics of detail coefficients at each level. Channel disturbance characteristic values ​​are constructed by calculating the sum of the energy ratios of detail coefficients at adjacent levels, thus effectively characterizing the dynamic fluctuations of the channel state. Compared to traditional time-domain or frequency-domain analysis methods, this method can more accurately capture short-term, sudden signal disturbances, improving the accuracy and real-time performance of external interference identification. It also exhibits good robustness and adaptability, providing reliable data support for subsequent network security risk assessments.

[0068] In S4, communication link anomaly characteristics and channel disturbance characteristics are combined to construct a comprehensive security risk feature vector, which is then input into the intelligent control model for fusion analysis to identify the current network security threat level, specifically including:

[0069] During network operation, abnormal communication link feature values ​​and channel disturbance feature values ​​are acquired and constructed into a comprehensive security risk feature vector, which is used as input to the intelligent control model to minimize the error between the predicted network security score and the actual network security score. This vector serves as the training target for the intelligent control model, which is then trained. Based on the trained intelligent control model, a network security score is output. The intelligent control model is a gradient boosting number model.

[0070] The process for obtaining the abnormal feature values ​​of the communication link is as follows:

[0071] A smart regulation model based on the gradient boosting tree algorithm is constructed. The comprehensive security risk feature vector is used as the model input, and the output is the predicted network security score. The actual network security scores in historical data are collected as labels, and the training set and test set are divided by cross-validation. The intelligent regulation model is iteratively trained using the training set. The optimization objective is to minimize the error loss function between the predicted network security score and the actual network security score. The model performance is evaluated using the test set to prevent overfitting.

[0072] Determine whether the current network security score is greater than or equal to a preset first threshold. If yes, it is recorded as a high-risk threat level. If no, determine whether the current network security score is less than or equal to a preset second threshold. If yes, it is recorded as a low-risk threat level. If no, it is recorded as a medium-risk threat level.

[0073] It should be noted that by fusing abnormal communication link feature values ​​with channel disturbance feature values ​​to construct a comprehensive security risk feature vector, and building an intelligent control model based on the gradient boosting tree algorithm, accurate prediction and dynamic evaluation of network security scores are achieved. This method fully considers the impact of quantum noise interference and external environmental disturbances at the feature extraction level, enhancing the model's ability to perceive network security vulnerabilities. The introduction of cross-validation mechanisms and error minimization optimization objectives during model training effectively improves the model's generalization performance and anti-overfitting ability. By setting multi-level scoring thresholds, dynamic identification of network threat levels is achieved, providing a scientific basis for the adaptive adjustment of subsequent security strategies, thereby significantly enhancing the proactive defense capabilities and intelligence level of the network system.

[0074] In S5, network security policies are dynamically adjusted based on the identified security threat levels, achieving closed-loop optimization control of network security protection and improving the overall security and adaptability of the system. Specifically, this includes:

[0075] For network states identified as high-risk threats, the system will initiate enhanced security defenses, such as increasing the frequency of quantum key distribution, enabling multi-layered data encryption, and implementing stricter access control policies. For medium-risk threats, the system will take moderately enhanced defenses, such as strengthening monitoring and auditing functions and regularly updating firewall rules. For low-risk threats, the system will maintain the current basic security protection configuration and continue to monitor network traffic data.

[0076] Secondly, after implementing security policy adjustments, the system will continuously collect qubit error rates and communication channel signal strength from new network traffic data, and evaluate changes in network security scores in real time using a pre-trained intelligent control model. If the new security score is found to be inconsistent with expectations or the network environment changes again, the system can quickly respond and reassess the current security threat level, readjusting the network security policy to form a dynamic feedback loop. This ensures that network security protection measures always match the current network threat level, thereby achieving closed-loop optimization control of network security protection.

[0077] The working principle of this invention is as follows: This invention aims to enhance the perception capability of quantum communication link anomalies and channel disturbances in complex network environments, and to achieve real-time assessment of network security status and dynamic optimization of security strategies. Firstly, during network operation, a specially designed data acquisition module is deployed, including a high-precision quantum state measurement device and a channel state information capture device, used to monitor the qubit error rate and communication channel signal strength in network traffic data in real time. The acquired raw data, after cleaning, time synchronization, and format standardization, provides high-quality input for subsequent feature extraction. In the feature extraction stage, the system models the qubit error rate based on a Bayesian inference mechanism, calculates the posterior probability of quantum noise interference in the link by combining prior probability and observation likelihood function, and maps it to communication link anomaly feature values ​​through a nonlinear transformation in the form of natural logarithm, thereby achieving high-sensitivity identification of weak quantum noise. Simultaneously, the Haar wavelet transform algorithm is used to perform multi-scale decomposition of the communication channel signal strength sequence, extracting the energy distribution characteristics of detail coefficients at each level, and constructing channel disturbance feature values ​​by summing the energy ratios of adjacent levels to determine whether external interference exists. Subsequently, the two types of feature values ​​are used to construct a comprehensive security risk feature vector, which serves as the input to the gradient boosting tree model. The model is trained using historical network security scoring data, with the objective function being to minimize the error between the predicted and actual scores. This process iteratively optimizes the model parameters and validates them on a test set to prevent overfitting. After training, the model can output the current network security score in real time and classify the network status into high-risk, medium-risk, or low-risk threat levels based on preset first and second thresholds. Finally, in S5, the system automatically matches corresponding security policy configurations based on the identified security threat level, such as enhanced key distribution, multi-layered encryption, or access control enhancement measures. Simultaneously, the system continuously collects new data and feeds it back to the intelligent control model, forming a closed-loop feedback mechanism to ensure that security protection strategies can dynamically evolve with the network threat situation, ultimately achieving a comprehensive improvement in the security, stability, and adaptability of the network system. This method integrates the characteristics of quantum communication with traditional communication channel state analysis, constructing a highly intelligent network security control system with closed-loop response capabilities, demonstrating promising engineering application prospects and widespread application value.

[0078] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. An adaptive dynamic network security strategy intelligent control method, characterized in that, Includes the following steps: S1: During network operation, monitor the quantum bit error rate and communication channel signal strength in network traffic data in real time; S2: Analyze the error rate of qubits and calculate the abnormal characteristic value of the communication link based on its fluctuation level to assess whether there is quantum noise interference in network transmission; The process for obtaining the abnormal feature values ​​of the communication link is as follows: The link is set to have quantum noise interference events. Initialize prior probabilities , representing the initial probability that the communication link is affected by quantum noise in the initial state of network operation; Real-time acquisition of qubit error rates from network traffic data forms an observation dataset. ; Calculate the interference in the link separately. and no interference Data observed under conditions likelihood function and ; Calculate the posterior probability using Bayes' theorem; posterior probability The abnormal characteristic value of the communication link is mapped through a nonlinear transformation. S3: Analyze the signal strength of the communication channel and calculate the channel disturbance characteristic value based on its stability changes to determine whether there is external interference; The process for obtaining the channel disturbance feature value is as follows: During network operation, a data sequence of communication channel signal strength is collected, and the data sequence of communication channel signal strength is subjected to a first-level Haar wavelet transform using the Haar wavelet transform algorithm to obtain the detail coefficients corresponding to each layer. Calculate the energy of each level of detail coefficients; Based on the energy distribution of detail coefficients at each layer, the energy ratio of detail coefficients between adjacent layers is calculated, and the energy ratios of detail coefficients between all adjacent layers are summed to obtain the channel disturbance characteristic value. S4: Construct a comprehensive security risk feature vector by combining communication link anomaly feature values ​​and channel disturbance feature values, input it into the intelligent control model for fusion analysis, and identify the current network security threat level; S5: Dynamically adjust network security policies based on the identified security threat levels to achieve closed-loop optimization control of network security protection and improve the overall security and adaptability of the system.

2. The adaptive dynamic network security strategy intelligent control method according to claim 1, characterized in that, The assessment of whether quantum noise interference exists in network transmission specifically includes: During network operation, the qubit error rate in network traffic data is monitored in real time. The qubit error rate is analyzed, and the abnormal characteristic value of the communication link is calculated based on its fluctuation. It is then determined whether the abnormal characteristic value of the communication link is greater than or equal to a preset threshold. If it is, there is quantum noise interference in the network transmission; otherwise, there is no quantum noise interference in the network transmission.

3. The adaptive dynamic network security strategy intelligent control method according to claim 1, characterized in that, The determination of whether external interference exists specifically includes: During network operation, the signal strength of the communication channel in the network traffic data is monitored in real time. The signal strength of the communication channel is analyzed, and the channel disturbance characteristic value is calculated based on its stability changes. It is then determined whether the channel disturbance characteristic value is greater than or equal to a preset threshold. If it is, there is external interference; otherwise, there is no external interference.

4. The adaptive dynamic network security strategy intelligent control method according to claim 1, characterized in that, The process of constructing a comprehensive security risk feature vector from communication link anomaly feature values ​​and channel disturbance feature values, and inputting it into the intelligent control model for fusion analysis, specifically includes: During network operation, abnormal communication link feature values ​​and channel disturbance feature values ​​are acquired and constructed into a comprehensive security risk feature vector, which is used as input to the intelligent control model to minimize the error between the predicted network security score and the actual network security score. This vector serves as the training target for the intelligent control model, which is then trained. Based on the trained intelligent control model, a network security score is output. The intelligent control model is a gradient boosting number model.

5. The adaptive dynamic network security strategy intelligent control method according to claim 4, characterized in that, The training of the intelligent regulation model specifically includes: A smart regulation model based on the gradient boosting tree algorithm is constructed. The comprehensive security risk feature vector is used as the model input, and the output is the predicted network security score. The actual network security scores in historical data are collected as labels, and the training set and test set are divided by cross-validation. The intelligent regulation model is iteratively trained using the training set. The optimization objective is to minimize the mean square error between the predicted network security score and the actual network security score. The model performance is evaluated using the test set to prevent overfitting.

6. The adaptive dynamic network security strategy intelligent control method according to claim 1, characterized in that, The identification of the current network security threat level specifically includes: Determine whether the current network security score is greater than or equal to a preset first threshold. If yes, it is recorded as a high-risk threat level. If no, determine whether the current network security score is less than or equal to a preset second threshold. If yes, it is recorded as a low-risk threat level. If no, it is recorded as a medium-risk threat level.

7. The adaptive dynamic network security strategy intelligent control method according to claim 1, characterized in that, The method of dynamically adjusting network security strategies based on the identified security threat levels to achieve closed-loop optimization control of network security protection specifically includes: For network states identified as high-risk threats, the system will initiate enhanced security defenses, including increasing the frequency of quantum key distribution, enabling multi-layered data encryption, and implementing stricter access control policies. For medium-risk threats, the system will take moderately enhanced defenses, including strengthening monitoring and auditing functions and regularly updating firewall rules. For low-risk threats, the system will maintain the current basic security protection configuration and continue to monitor network traffic data.

Images