A network security data privacy protection management method and system

Abstract

A network security data privacy protection management method and system. The input data stream is divided into multiple flow units and a three-dimensional logical coordinate is allocated in the discrete space grid to initialize the privacy flow field; the sensitivity entropy, threat pressure and resource density of each flow unit are calculated in real time; through a continuous differentiable mapping function, the protection strength at the current time is mapped, and the privacy flow field is dynamically updated; further, the nonlinear protection behavior is triggered to adjust the protection strength and switch to the corresponding data protection strategy; the partial differential equation is constructed and the protection strength is updated in real time; based on the three-dimensional space, the detection is carried out and the privacy vortex suppression loop is activated, so as to suppress the fluctuation of the protection strength in the privacy vortex area; based on the updated protection strength, the protected data stream is output; and a visual interface is provided to receive the interactive operation of the user to trigger the re-equilibrium of the privacy flow field. The application improves the security and reliability of data privacy protection.

Description

Technical Field

[0001] This invention belongs to the field of network security technology, specifically relating to a network security data privacy protection and management method and system. Background Technology

[0002] Network security data privacy protection management is a key technical area for preventing the leakage of sensitive information during the data collection, transmission, storage, processing, and sharing processes of information systems. With the widespread application of cloud computing, edge computing, and big data analytics, data streams are characterized by high dynamism, multiple sources, and cross-domain collaboration, placing demands on a balance between real-time performance, scalability, and availability for privacy protection.

[0003] Patent application CN118378300A achieves privacy control by pre-setting a fixed protection strategy before data publication or querying. This method first classifies the data set to identify sensitive attributes; then, it performs a uniform transformation on the sensitive attributes according to predefined rules or parameters, such as adding noise of a fixed amplitude, applying encryption with a fixed key, or performing attribute generalization; finally, the transformed data is output to downstream modules. This method typically relies on a static configuration table or a single threshold judgment in its implementation, ensuring that the protection strength remains consistent throughout the data lifecycle and does not adjust with changes in state or external environment.

[0004] However, existing technologies cannot dynamically adjust the protection strength based on the real-time status of data streams and resources, resulting in insufficient protection in high-threat scenarios or excessive sacrifice of availability in high-frequency legitimate access scenarios. Summary of the Invention

[0005] To address the issue that existing technologies cannot update protection strength in real time based on data flow, this invention provides a network security data privacy protection management method and system.

[0006] The first aspect of this invention discloses a network security data privacy protection management method, which adopts the following technical solution: The input data stream is divided into multiple rheological units according to time, source, and type; and a three-dimensional logical coordinate is assigned to each rheological unit in a discrete spatial grid to initialize the privacy rheological field; each grid unit in the privacy rheological field corresponds to a protection strength value; The sensitivity entropy, threat pressure, and resource density of each rheological unit are calculated in real time; the sensitivity entropy, threat pressure, and resource density are mapped to the protection strength at the current moment through a continuously differentiable mapping function, and the privacy rheological field is updated. Based on the rate of change of protection strength and the spatial gradient of threat pressure, nonlinear protection behavior is triggered and the protection strength is adjusted; the corresponding data protection strategy is switched according to the adjusted protection strength; the nonlinear protection behavior includes shear thickening and shear thinning. Based on the adjusted protection strength, the instantaneous rate of change of the protection strength is calculated and the strength field distribution is updated; the instantaneous rate of change of the protection strength is a partial differential equation consisting of a diffusion term, a threat pressure coupling term, a resource density coupling term, and a vortex suppression term; Calculate the curl of the gradient vector of the scalar field consisting of the ratio of threat pressure to resource density in three-dimensional space; detect privacy vortex regions based on the curl modulus and activate privacy vortex suppression loops; suppress fluctuations in protection strength within privacy vortex regions based on suppression mechanisms; Based on the updated protection strength, a corresponding protection strategy is selected for each rheostat unit from a preset mapping table, and the protection strategy is applied to the data within the rheostat unit to output a protected data stream.

[0007] Furthermore, the process of dividing the input data stream into multiple rheological units based on time, source, and type, and assigning a three-dimensional logical coordinate to each rheological unit in a virtual space grid to initialize the privacy rheological field includes: Based on a preset time window and a logical topology defined by data source identifiers and data type labels, the data stream is segmented to generate multiple rheostat units containing a preset minimum number of data packets. Each rheological unit is assigned a unique logical coordinate in a three-dimensional discrete spatial grid based on a hash function; Each rheological unit is embedded into the grid cell corresponding to its logical coordinate in the discrete spatial grid with an initial protection strength to complete the initialization of the privacy rheological field; The three-dimensional discrete space grid has a side length of The initial protection strength of the cubic mesh is 0. When the rheological element is embedded in the mesh, the initial protection strength of the corresponding mesh element is updated to the weighted value of the total number of rheological element fields and the average length of the corresponding field.

[0008] Furthermore, the real-time calculation of the sensitivity entropy, threat pressure, and resource density of each rheological unit includes: The sensitivity entropy is calculated by: calculating the empirical probability of each attribute field within the sliding window corresponding to the rheological unit; calculating the Shannon entropy of each attribute field based on the empirical probability; and then normalizing the weighted average of the Shannon entropies of all attribute fields to obtain the sensitivity entropy. The empirical probability is the ratio of the frequency of occurrence of the corresponding value of the attribute field within the sliding window to the capacity of the sliding window.

[0009] Furthermore, the threat pressure is calculated by weighting the query frequency per unit time, the abnormal access detection score, and the external threat intelligence matching degree. Among them, the counter at time intervals Total number of query requests and time intervals for internal rheostat units The ratio is used to obtain the query frequency per unit time; The access pattern vector of the rheological unit is input into the isolated forest model, and the output value of the model is normalized to obtain the abnormal access detection score; the access pattern vector consists of the source IP, time interval and operation type. The external threat intelligence database is queried using the source identifier and type label of the rheological unit; from all matching entries, the maximum confidence value is selected and normalized to obtain the external threat intelligence matching degree.

[0010] Furthermore, the resource density is calculated by calculating the ratio of available CPU cores to total CPU cores, and the ratio of memory resources to total memory resources; the two ratios are then geometrically averaged to obtain the resource density.

[0011] Furthermore, the sensitivity entropy, threat pressure, and resource density are mapped to the protection strength at the current moment, including one of the following two methods: First, a control point network is predefined in a three-dimensional space; the positions of sensitivity, threat pressure and resource density in the predefined three-dimensional grid are determined, and the eight vertices corresponding to the positions are found; based on the protection strength values ​​of the eight vertices, the protection strength value corresponding to the current triple is calculated using trilinear interpolation and B-spline convolution mathematical methods. Secondly, select in advance in a three-dimensional space One center point; based on radial basis function network, using sensitivity entropy, threat pressure and resource density as input vectors, and calculating and... The protection strength value corresponding to the current ternary is calculated by combining the distance of each center point with the weight of each center point.

[0012] Further, triggering the nonlinear protection behavior and adjusting the protection strength includes: Based on time point The last time cycle Internally, calculate threat pressure The rate of change; The rate of change of threat pressure is calculated along the X, Y, and Z axes in three-dimensional space to obtain the spatial gradient of threat pressure. If the rate of change is positive and the spatial gradient exceeds a preset first pressure threshold, then shear thickening behavior is triggered. When shear thickening behavior is triggered, the current protection strength is multiplied by an amplification factor greater than 1, and the system switches to the first protection strategy.

[0013] Furthermore, the triggering of nonlinear protection behavior also includes: If the rate of change of threat pressure is less than the preset second pressure threshold, and the current query frequency per unit time is higher than the historical average query frequency, and the current rheological unit is not in the privacy vortex zone, then shear thinning behavior is triggered, the current protection strength is divided by a reduction factor greater than 1, and the second protection strategy is switched. When the shear thickening behavior and the shear thinning behavior are triggered simultaneously, the shear thickening behavior is executed first.

[0014] Furthermore, the step of switching to the corresponding data protection strategy based on the adjusted protection strength includes: A continuous mapping table is pre-established, which defines multiple protection strength intervals arranged in ascending order, and each interval corresponds to a protection strategy; Based on the current protection strength value, the corresponding protection strategy is determined by querying the continuous mapping table. The protection strategy includes: local perturbation, partial homomorphic encryption, fully homomorphic encryption, and differential privacy noise.

[0015] Furthermore, the partial differential equations described above are constructed by: The finite volume method is used to solve the differential partial equations in discrete space. The differential partial equations include diffusion terms, threat-pressure coupling terms, resource density coupling terms, and vortex suppression terms. The diffusion term is calculated from the diffusion term of the Laplace of the protection strength; The threat pressure coupling term is obtained by multiplying the threat pressure by the threat pressure coupling coefficient; The resource density coupling term is obtained by multiplying the resource density by the resource density coupling coefficient; The vortex suppression term is obtained by multiplying the difference between the protection strength and its mean by the vortex activation function by the vortex suppression coefficient. At each time step, the sum of the diffusion term, threat pressure positive coupling term, resource density negative coupling term, and vortex suppression term for each grid cell is calculated to obtain the rate of change of the current protection strength; Based on the rate of change of the current protection strength, the protection strength of all grid cells in the privacy rheological field is updated in real time using an explicit time format; The value of the vortex activation function is determined by the result of the privacy vortex detection.

[0016] Further, the curl of the gradient vector in three-dimensional space is calculated; privacy vortex region detection based on curl modulus includes: In a scalar field consisting of the ratio of threat pressure to resource density, the curl of the gradient vector is calculated at the center of each grid cell; the curl consists of three curl components in the X, Y, and Z axes. The curl modulus is calculated based on the three curl components. If the curl modulus exceeds the preset first curl threshold, the grid cell is marked as a privacy vortex region.

[0017] Furthermore, the step of selecting a corresponding protection strategy for each rheological unit from a preset mapping table includes: Establish a continuous mapping table from protection strength values ​​to protection strategies; Based on the current protection strength value of the rheostat unit, a query is performed in the mapping table to match the corresponding protection strategy; Iterate through each attribute field within the rheological unit, and for each attribute field, call the corresponding data transformation algorithm according to its type, sensitivity level, and current protection strength to generate the transformed field value; All transformed field values ​​are encapsulated into a protected data stream.

[0018] Furthermore, it also includes: The protection intensity field, its gradient field, and privacy vortex region are visualized; and an interface is provided to receive user interaction based on the visualization results; the interaction is interpreted as a modification of the boundary conditions or local conditions of the protection intensity field to trigger the rebalancing of the privacy rheological field.

[0019] Furthermore, the visualization of the protection intensity field, its gradient field, and the privacy vortex region includes: From the three-dimensional protective intensity field, a two-dimensional plane with fixed coordinates is selected for slicing; The protection strength value of each grid cell on the slice is mapped to color to generate a two-dimensional heat map; The gradient of the protection intensity field is calculated using central difference and six-neighborhood, and streamlines of the gradient direction and intensity are plotted on the two-dimensional thermogram. On the two-dimensional heatmap, the grid cells containing the privacy vortex region are dynamically highlighted.

[0020] Furthermore, the interactive operation is parsed as a modification of the boundary conditions or local conditions of the protection intensity field, including: It receives the user's drag operation on the visualized streamline and parses the operation into a modification instruction for the gradient boundary conditions of the protection intensity field; It receives user clicks at specific locations on the visualization interface and parses the clicks into commands to inject damping enhancement at the corresponding grid cells. According to the modification instruction and the injection instruction, the partial differential equation is adjusted to trigger the rebalancing of the protection intensity field.

[0021] The second aspect of this invention discloses a network security data privacy protection management system, which implements the data privacy protection management method as described in the first aspect of this invention. The system includes: Rheological unit construction module: Divides the input data stream into multiple rheological units according to time, source and type; and assigns a three-dimensional logical coordinate in a discrete spatial grid to each rheological unit to initialize the privacy rheological field; each grid unit in the privacy rheological field corresponds to a protection strength value; 3D parametric modeling module: calculates the sensitivity entropy, threat pressure, and resource density of each rheological unit in real time; maps the sensitivity entropy, threat pressure, and resource density to the protection strength at the current moment through a continuously differentiable mapping function, and updates the privacy rheological field; Nonlinear protection trigger module: Based on the rate of change of protection strength and the spatial gradient of threat pressure, triggers nonlinear protection behavior and adjusts the protection strength; switches to the corresponding data protection strategy according to the adjusted protection strength; the nonlinear protection behavior includes shear thickening and shear thinning; Partial differential equation construction module: Based on the adjusted protection strength, calculate the instantaneous rate of change of the protection strength and update the strength field distribution; the instantaneous rate of change of the protection strength is a partial differential equation composed of diffusion term, threat pressure coupling term, resource density coupling term and vortex suppression term; Vortex Detection and Suppression Module: Calculates the curl of the gradient vector of the scalar field composed of the ratio of threat pressure to resource density in three-dimensional space; detects privacy vortex regions based on the curl modulus and activates the privacy vortex suppression loop; suppresses fluctuations in protection strength within privacy vortex regions based on the suppression mechanism; Policy execution module: Based on the updated protection strength, it selects the corresponding protection policy for each rheological unit from the preset mapping table, and executes the protection policy on the data within the rheological unit to output the protected data stream.

[0022] The beneficial effects of this invention are that, compared with the prior art, 1. Real-time modeling of privacy rheological fields and non-Newtonian adaptive behavior mechanism: By calculating the sensitivity entropy, threat pressure and resource density of each rheological unit in real time and mapping them to continuously differentiable protection strength, combined with shear thickening (automatically amplifying the strength and switching to fully homomorphic encryption when there is a high threat) and shear thinning (reducing the strength and turning it into a light perturbation when there is a high frequency of legitimate queries), the protection strength is dynamically and adaptively adjusted according to the state, which solves the problem of insufficient protection or excessive sacrifice of availability caused by fixed protection strength in the existing technology.

[0023] 2. Field-driven evolution and vortex suppression loop based on partial differential equations: The global evolution of the protection intensity field is driven by partial differential equations including diffusion terms, positive coupling terms of threat pressure, negative coupling terms of resource density, and vortex suppression terms. In conjunction with real-time calculation of the curl of the threat pressure to resource density ratio to detect privacy vortices, and forced convergence to the neighborhood average value through a virtual damping source, the instability of the protection strategy caused by local field oscillations is prevented, thus solving the technical defects of the lack of global coordination and stability control in the existing technology.

[0024] 3. Visualized projection and closed-loop control of artificial field intervention: The three-dimensional protection intensity field slices are projected into heat maps and streamline maps, and the vortex region is dynamically highlighted. It supports the analysis of gradient boundary conditions by dragging the streamline or injecting it as local damping enhancement and triggering global rebalancing by clicking. This enables managers to visualize the field state and make precise artificial intervention, solving the technical problems of the protection process being unexplainable and inflexible in existing technologies. Attached Figure Description

[0025] Figure 1 This is an overall flowchart of an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this invention are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0027] As an embodiment of the present invention, a specific implementation method for network security data privacy protection and management is disclosed, and the execution flow of the method embodiment is as follows: Figure 1 .

[0028] S1: As one implementation method, the data flow entering the information system is divided into several flow units according to its source, type, and time window, and logical coordinates are assigned to each flow unit; specifically including: S1.1: The data stream is segmented using a fixed time window and logical topology to obtain rheodynamic units; Specifically, the received input data stream consists of multiple data packets, each containing a source identifier, type label, and timestamp field. The fixed time window length is... The logical topology is defined by the device number or network subnet address. Fragmentation operations are performed within the input buffer, following these rules: Regarding time Data packets within, based on time With fixed time window length Ratio calculation time slice index ; Grouping by source identifier and type label, and aggregating packets with the same combination into candidate units; When the number of data packets in the candidate unit reaches the preset minimum threshold At that time, rheological units are generated. And record the start time. End time ; Among them, rheological unit The data structure includes a field list, metadata, and a statistical summary. The field list stores the raw attribute value sequence; the metadata includes source identifiers, type labels, etc. , The statistical summary contains frequency counters for each field value, which are used for subsequent sensitivity entropy calculation.

[0029] The lifecycle of a rheostat unit is controlled by a time window. Current time satisfy When the cell is removed from the grid, memory is freed, and its statistical summary is archived to the history buffer for use in subsequent threat stress calculations as historical mean statistics.

[0030] S1.2: A hash function based on source, type, and timestamp generates three-dimensional logical coordinates for each rheological unit, and embeds the three-dimensional logical coordinates into a discrete spatial grid to initialize the privacy rheological field.

[0031] Furthermore, the hash function H uses the MurmurHash3 algorithm, with the input being a concatenated string. The output is a 128-bit unsigned integer h. Logical coordinates Calculated using the following piecewise modulo mapping: ; ; ; in, , , Discrete space grids in , , Number of elements in the direction, total mesh size is mod means modulo.

[0032] In a further embodiment, a discrete spatial grid It is a uniform cubic mesh with a side length of . Unit volume During initialization, the protection strength of all mesh cells. Set to 0. Rheological unit. After embedding, its corresponding mesh cell The value is updated to the initial value. The calculation formula is: ; in, for Total number of fields in the middle The average length of the field. , This is a preset constant.

[0033] In a further implementation, a rheological unit index table is maintained, with logical coordinates as the keys. The value is The memory pointer supports constant-time insertion and querying. When a new rheological unit is generated, if the target coordinates are already occupied, a merge operation is performed: the field list of the new unit is appended to the existing unit, and the statistical summary and time range are updated. If the total number of data packets after merging exceeds... If so, then splitting is performed: the data packet is divided according to the midpoint of time, generating two new units, which are mapped to the original coordinates and the nearest free adjacent coordinates respectively.

[0034] The mesh uses periodic boundary conditions: a modulo operation is performed when coordinates exceed the range. , , ; The fragmentation and mapping processes are executed in parallel. Upon arrival, data packets enter the fragmentation queue. When the queue reaches its capacity or the time window ends, batch hash calculation and coordinate embedding are triggered. After embedding, an update notification is sent to the privacy-preserving rheological field 3D parametric modeling module, containing a list of affected coordinates and corresponding rheological unit pointers.

[0035] The privacy rheological field is a three-dimensional, discrete, continuously evolving field used to mathematically and physically simulate the entire real-time state.

[0036] S2: As one implementation method, the sensitivity entropy, threat pressure, and resource density are calculated in real time for each rheological unit, and the protection strength in the privacy rheological field is constructed through a predefined continuously differentiable mapping function. The specific steps are as follows: S2.1: Calculate the sensitivity entropy based on the empirical probability distribution of each attribute field within the sliding window; Specifically, for each rheological unit Maintain a sliding window of fixed length. Window capacity is A historical data packet. The window is implemented using a first-in, first-out queue; the oldest record is removed when a new data packet arrives. Sensitivity entropy. The calculation is for Each attribute field implement: Statistics Window inner field The distribution of values, denoted by different values The frequency of occurrence is ; Calculate empirical probability ; Applying Shannon's entropy formula: ,in, For the number of times, The sequence number for each distinct value. For fields The number of unique possible values; right The entropy of all fields is taken as a weighted average, with weights... Determined by the predefined sensitivity level of the field; sensitivity entropy With weight The calculation formula is: , ; Among them, attribute fields =1, 2, 3, ..., K; the sum of all weights equals 1, which is a standard normalization process that ensures the calculated weights are equal to 1. It is a standardized value between 0 and 1, which facilitates comparison and integration with other indicators (such as threat pressure).

[0037] S2.2: Calculate threat pressure by weighted fusion of query frequency per unit time, abnormal access detection score, and external threat intelligence matching degree; Specifically, threats and pressure The calculation fuses the three signals. Frequency is queried per unit time. By using a counter at time intervals Internal accumulation pairs The number of query requests, divided by Obtained. Anomaly access detection score. Output from the built-in isolated forest model, input is The access pattern vector (source IP, time interval, operation type) is normalized to [0,1]. External threat intelligence matching degree. Matching using STIX format intelligence database The source identifier and type label are used, and the highest confidence level is taken when a match is found, normalized to [0,1]. The weighted fusion calculation formula for the three signals is as follows: ; in, , , These are the weights for query frequency per unit time, abnormal access detection score, and external threat intelligence matching degree, respectively. .

[0038] Fusion value After being compressed to the [0,1] range by the sigmoid function, the calculation formula is as follows: ;in , To adjust the parameters.

[0039] S2.3: Calculate resource density by the ratio of available computing resources and memory resources to total resources; resource density The calculation is based on the node resource monitoring interface. It periodically reads the number of available CPU cores. Available memory (bytes), and total number of cores Total memory , The calculation formula is: ; A geometric mean is used to balance the impact of the two resources.

[0040] S2.4: Construct a continuously differentiable mapping function using three-variable spline interpolation or radial basis functions to map sensitivity entropy, threat pressure, and resource density into protection strength.

[0041] A continuously differentiable mapping function f will ( , , Mapping to protection strength Two implementation paths are provided, and one can be selected at runtime via a configuration flag.

[0042] Furthermore, path one employs three-variable third-order spline interpolation. A control point grid is pre-constructed in space, with a grid resolution of [resolution missing]. Each control point stores The value is used to form the tensor product B-spline basis function. During real-time mapping: determine ( , , The voxel in question; Calculate the local coordinates of 8 corner points → perform trilinear interpolation to obtain the base values ​​→ apply B-spline convolution to obtain the final coordinates. .

[0043] Furthermore, path two employs a radial basis function network. M center points are pre-defined. Weight of each center With polynomial coefficients. The mapping formula is: ; in, It is a cubic radial basis. For linear polynomial compensation, The center point and weights were determined through offline least squares fitting.

[0044] Output of mapping function f Directly write to the corresponding mesh cell of the privacy rheological field After the calculation is complete, a parameter update packet containing coordinates is sent to the non-Newtonian privacy behavior triggering module and the partial differential evolution engine. With triples ( , , Update packages are pushed out in an event-driven manner to ensure that downstream modules use consistent parameters at the same time step.

[0045] S3: As one implementation method, based on the protection strength, its rate of change, and the gradient of threat pressure, non-Newtonian privacy behavior is triggered, including shear thickening and shear thinning, to adjust the protection strength and switch the corresponding protection strategy; specifically including: S3.1: When the threat pressure rises rapidly and its gradient modulus exceeds the preset first pressure threshold, shear thickening is triggered, the protection strength is amplified by a coefficient greater than 1, and the system switches to a high-strength protection strategy (first protection strategy). Specifically, at each time step For each rheological unit Maintenance and protection strength current value Compared to the previous time step value Calculate the rate of change: ; Simultaneously, threat pressure is calculated on a discrete grid. Central difference gradient The module length is: ; Using six adjacent grid cells Difference the values.

[0046] The shear thickening trigger condition consists of two parts that must be satisfied simultaneously: First, the rate of increase in threat pressure: Greater than 0; Second, gradient magnitude Exceeding the preset first threshold When the conditions are met, an amplification operation is performed, and the new protection strength calculation formula is: ; in: This is the magnification factor. ; This is a preset positive coefficient. After magnification, Immediately cover the original The unit is then marked as entering a high-strength protection state.

[0047] The high-strength protection strategy corresponds to the protection strength value output by the continuously differentiable mapping function. When the data stream exceeds a preset first intensity threshold, the highest level of privacy protection measures are implemented. The specific content of these measures is proportional to the current rheological unit sensitivity level.

[0048] S3.2: When the threat pressure is stable and the query frequency is higher than the historical average, trigger shear thinning, reduce the protection strength by a coefficient greater than 1 and switch to a lightweight protection strategy (second protection strategy). The shear thinning trigger condition consists of three parts that must be met simultaneously: First, the absolute value of the threat pressure change is less than a preset second pressure threshold; second, the query frequency per unit time. Exceeding the historical average from the past The calculation involves an exponentially weighted moving average over several time windows; third, the area is not currently in a privacy vortex zone. When these conditions are met, a reduction operation is performed, and the new protection strength is calculated using the following formula: ; in, It is a constant coefficient greater than 1. The larger the value, the greater the reduction in protection strength.

[0049] The privacy vortex region refers to the critical area where the protection intensity field experiences localized and severe oscillations due to multi-source interference. Within this region, the spatial curl of the protection intensity is significantly non-zero; this curl is detected to locate and suppress such oscillations.

[0050] The protection strength field specifically refers to the protection strength on each grid cell in the privacy rheological field. The spatial distribution of values. It is a scalar field.

[0051] S3.3: Establish a continuous mapping table from protection strength to protection strategy in advance to achieve seamless connection between protection strength and specific protection measures.

[0052] The continuous mapping table is stored in memory as a piecewise linear function, containing multiple interval endpoints. With the corresponding protection strategy identifier The table structure is an ordered array, with endpoints ordered by... Sort in ascending order. The query process is as follows: Given Perform a binary search to locate the interval. Output strategy The protection strategies include plaintext transmission, local perturbation, partial homomorphic encryption, fully homomorphic encryption, and strong differential privacy noise addition. Each strategy is associated with a specific execution function pointer.

[0053] After triggering non-Newtonian behavior, the updated version will be... With strategy identifier Write to the shared event queue and push to the protection policy execution module. Within the same time step, if both shear thickening and shear thinning conditions are met, shear thickening is executed first, and the shear thinning flag is suppressed until the vortex suppression loop or the next time step is re-evaluated. Historical average The update is performed at the end of each time window, using a smoothing coefficient. Based on the current actual frequency and the historical frequency, the formula for calculating the updated historical frequency is as follows: ; in, This refers to the current actual frequency, that is, the number of queries or frequency actually monitored within the current time window; Historical frequency refers to historical data saved from the previous calculation cycle.

[0054] In one embodiment of the present invention, the high-strength protection strategy corresponds to when the protection strength value... Exceeding the preset first intensity threshold (e.g.) When the protection strength value is 0.8), the highest security level protection method is determined by querying the continuous mapping table, including but not limited to fully homomorphic encryption or strong differential privacy noise addition. The lightweight protection strategy corresponds to when the protection strength value... Below the preset second intensity threshold (e.g.) When 0.3), the lowest computational overhead protection means is determined by querying the continuous mapping table, including but not limited to plaintext transmission or local perturbation by adding extremely small-scale Laplace noise.

[0055] S4: As one implementation of the example, the real-time evolution of the privacy rheological field is driven by a partial differential equation that includes diffusion terms, threat pressure coupling terms, resource density coupling terms, and vortex suppression terms.

[0056] Specifically, the partial differential equations are discretized spatially and temporally using the finite volume method. At each time step, the sum of the diffusion term, the positive coupling term of threat pressure, the negative coupling term of resource density, and the vortex suppression term is calculated to update the protection intensity field distribution.

[0057] In discrete three-dimensional mesh Solving the protection strength field The evolution is governed by the following equations: ; in, The diffusion coefficient is a constant. , , These are the threat-pressure coupling coefficient, resource-density coupling coefficient, and vortex suppression coefficient, respectively. As a threat of pressure, For resource density, For vortex activation function, This represents the average protection strength of the current grid cell's 26 neighboring cells. The value of the vortex activation function, representing the curl, is determined by the results of privacy vortex detection.

[0058] The finite volume method divides the mesh into control volumes, with each control volume storing data at its center. Value. When spatially discrete, for the diffusion term Traverse the six faces of the control volume and calculate adjacent units. The net flux is obtained by summing the dot product of the difference and the surface normal vector, and then dividing by the control volume. The diffusion contribution is subsequently obtained. Threat pressure positive coupling term. Take the current grid cell directly Value multiplied by Resource density negative coupling term Take the current Value multiplied by The vortex suppression term is output in the privacy vortex detection module. Activated when =1, calculation and Difference, multiplied by Later added.

[0059] Time discretization employs an explicit Euler scheme. Each time step... At the beginning, back up the current scene. Move to a temporary buffer. Then, iterate through all grid cells in parallel, calculating the sum of the four terms to obtain the local rate of change. The updated formula is: ; in, For the first The predicted protection intensity field at each time step For the first The protection intensity field at each time step; After the update is complete, check the Courant-Friedrichs-Lewy condition to ensure... To maintain numerical stability, where, This is the spatial step size, i.e., the grid resolution. It is automatically halved when the condition is not met. And recalculate the current time step.

[0060] The evolution engine interacts with upstream modules via a double-buffered queue. The privacy-preserving rheological field 3D parametric modeling module pushes... and During updates, the engine writes the new value to the field copy, which only takes effect at time step boundaries. This is not triggered by non-Newtonian privacy behavior. Local adjustments are injected as source terms: direct addition or subtraction at the corresponding coordinates. Record it in the event log.

[0061] vortex suppression term The calculation uses three nested loops to traverse the 26-neighborhood and accumulates the results. Divide the value by the neighborhood size. Provided in real time by the privacy vortex detection module, the engine only reads the flag bits.

[0062] After evolution is complete, The entire field is marked as available, notifying the protection strategy execution module and the visualization module. Boundary conditions are of the Neumann type: the diffusion flux at the mesh edge faces is set to 0. The engine supports dynamic coefficient adjustment, which can be modified via an interface through the manual intervention module. , , , The changes will take effect in the next time step.

[0063] The entire solution process is executed in a multi-threaded environment, with the mesh divided into sub-blocks. Each thread is responsible for calculating the diffusion and coupling terms of one sub-block. Synchronization barriers ensure that all threads enter the global field copying phase after completing local updates.

[0064] S5: As one implementation method, the curl of the ratio of threat pressure to resource density in the privacy rheological field is calculated, privacy vortices are detected, and a privacy vortex suppression loop is activated when the curl exceeds a preset threshold. The protection strength is forced to converge to the local average value through a damping term. The specific steps are as follows: S5.1: Calculate the curl component of the vector field formed by the ratio of threat pressure to resource density in the grid cell and take its modulus; Specifically, at each time step Internally, read current threat pressures from the privacy rheological field. and resource density Constructing a scalar field To avoid division by zero, Below the preset lower limit Time setting Scalar field gradient vector curl It is calculated at the center of each grid cell, using a finite difference approximation: Calculate the x-component of the curl vector : ; Calculate the y-direction component of the curl vector : ; Calculate the z-direction component of the curl vector : ; After obtaining the three components, the curl modulus is: Maintaining the global curl field ,and Storage of the same size in the field; where, if A large value indicates that there is violent rotation or instability in the region; if A value close to 0 indicates that the region is smooth and irrotational.

[0065] S5.2: When the curl modulus exceeds the preset first curl threshold, it is marked as a privacy vortex region and the privacy vortex suppression loop is activated; The detection process traverses all grid cells in parallel. When a certain cell... If a region is identified as a privacy vortex region, its coordinates are recorded in the vortex list. A preset first curl threshold is set. The vortex center is defined as the point with the maximum local curl modulus, determined through a 3×3×3 neighborhood search. During the search, the curl modulus of 27 elements is compared. The largest value is taken as the center coordinate C.

[0066] The 3×3×3 neighborhood is centered on the candidate grid cell currently marked as the privacy vortex region, including the candidate cell itself and its 26 surrounding neighboring grid cells, for a total of 27 cells. By comparing the curl moduli of these 27 cells, the grid cell with the largest curl moduli is determined as the vortex center C of this local vortex region. Subsequently, starting from this center C, a complete privacy vortex cluster is formed through connectivity search. The "vortex center" is the grid cell with the largest curl moduli within the 3×3×3 neighborhood.

[0067] S5.3: Insert a virtual damping source at the center of the vortex, and force the protection strength to be pulled to the neighborhood average value through the damping term in the partial differential equation, while temporarily disabling the shear thinning mechanism in the vortex region.

[0068] Furthermore, after activating the privacy vortex suppression loop, three operations are performed. First, settings are configured in the partial differential evolution engine. =1, triggering the damping term Recalculate at the vortex center C, using an average of a 5×5×5 neighborhood (125 elements) centered at C. Value. Second, inserting a virtual damping source into the mesh element at the vortex center C results in a local increase in the vortex suppression coefficient, i.e. ;in, This is the fundamental vortex suppression coefficient in the partial differential equation. The preset magnification factor ( >1). Enhanced coefficients It only applies to the vortex center C and its neighborhood. Third, a suppression command is sent to the non-Newtonian privacy behavior module to temporarily disable shear thinning triggering in all cells within the vortex region until... Down to The following, among which This is a preset second curl threshold.

[0069] The vortex region is dynamically expanded: starting from center C, a breadth-first search is performed to include all connected regions. The units form connected vortex clusters. Cluster boundary units are marked as transition zones, where shear thickening can still be triggered. Vortex event records include center coordinates, maximum... Cluster size, start and end times are recorded and written to the audit log.

[0070] In a further implementation, before the end of each time step, each cluster in the vortex list is recalculated. If the largest within the cluster Then remove the mark and restore Return to the original value, disable shear thinning, and remove from the list. The damping effect of the suppression loop decays naturally through the evolution of the partial differential equation. to No additional cleaning steps are required.

[0071] Curl calculation and detection are performed in a separate thread pool, interacting with the partial differential evolution engine through a locked free-circular buffer. The detection thread outputs a vortex event packet, containing the center coordinates, The value and disable flag are read and applied by the evolutionary engine in the next time step. Non-Newtonian behavior modules subscribe to the same event package and update the local trigger mask in real time.

[0072] S6: As one implementation method, a corresponding protection strategy is selected from a preset continuous mapping table based on the real-time updated protection strength, and the protected data stream is output after data transformation is performed on the rheostat unit; the specific steps are as follows: S6.1: Based on the current protection strength value, query the continuous mapping table to determine specific protection measures such as disturbance, encryption, or suppression; Specifically, at the end of each time step, the updated protection strength is received from the partial differential evolution engine. and their corresponding coordinates The continuous mapping table is stored as an ordered key-value structure in memory, where the key is the endpoint of the protection strength interval. The value is the policy executor identifier. The query operation is performed on each rheostat unit. Execute: with the current Given the input, use binary search to locate the minimum value that makes... , obtain Policy executor identifier A function pointer that points to one of the following five implementation types: Plaintext transmission: Directly copy the original field value to the output buffer; Local perturbation: Add Laplace noise to numerical fields, with the noise scale adjusted by... The sub-range of the mapping is determined; this is performed on categorical fields. -Anonymous generalization, Value follows Increase; Partial homomorphic encryption: The Paillier scheme is used to encrypt numeric fields, and the key is managed globally; Fully homomorphic encryption: All fields are encrypted using the BFV scheme, supporting addition and multiplication operations; Strongly Differential Privacy Noise: Injecting globally sensitive normalized Gaussian noise into the query results for privacy budgeting. Based on protection strength Reverse derivation.

[0073] S6.2: Execute the selected protection strategy on each field of the data in the rheostat unit, generate the protected data stream, and output it to the downstream module.

[0074] The execution process iterates by field. A list of fields. For each field Check its type label and call The corresponding sub-function. The sub-function takes primitive values ​​as input. Field sensitivity level current The output is the transformed value. The transformed values ​​are written to the output data packet in the original order, and metadata is appended to mark the transformation type and parameter summary.

[0075] Data transformations are performed in parallel at the rheological unit granularity. A thread pool is maintained, with each thread bound to one... Threads obtain from the input queue pointers and After identification and transformation, the protected data packet is pushed into the output queue. The output queue adopts a double-buffering mechanism: one buffer is used for writing, and the other buffer is used for reading by downstream modules. The switching is controlled by a semaphore.

[0076] Protected data streams are encapsulated in a uniform format: the header contains a rheological unit identifier, a timestamp, and a transformation summary; the payload is a sequence of transformed fields. The summary field records the policy identifier, noise scale, or encryption scheme version used, facilitating downstream auditing. Output to downstream modules is achieved through zero-copy memory mapping, reducing serialization overhead.

[0077] Verification before execution Validity: When If the value exceeds the defined range of the mapping table, the upper bound is taken as the maximum strategy, and the lower bound is taken as the plaintext. After execution, the data is released. The original memory is used to retain the transformation log in a circular buffer. The log contains coordinates x, y, and y. value, Identification and execution duration. Logs are archived regularly for self-inspection and compliance reporting.

[0078] S7: As one implementation method, the privacy rheological field is projected as a two-dimensional heat map and streamline diagram, and the privacy vortex region is dynamically highlighted. Artificial field intervention is supported by dragging streamlines or clicking to inject damping sources. The specific steps are as follows: S7.1: The dimensional protection intensity field is sliced ​​on a fixed coordinate plane to generate a color-coded two-dimensional heat map; Specifically, within each visualization refresh cycle, the current protection strength field is obtained from the partial differential evolution engine. A complete snapshot. The fixed coordinate plane defaults to... The slicing operation can be used to select intermediate layers, or switch to any integer layer in the x or y direction via interface controls. The slicing operation extracts the values ​​of all mesh cells in that layer. The values ​​are used to form a two-dimensional array.

[0079] Heatmap generation uses linear color mapping: The minimum value corresponds to dark blue. The maximum value corresponds to bright red, and intermediate values ​​are interpolated using HSV color space. The rendering engine uses WebGL shaders to compute the color value of each pixel in parallel on the GPU. Pixel resolution matches grid resolution, and edge units undergo bilinear upsampling for smooth display.

[0080] S7.2: Calculate the protection strength gradient field and draw streamline diagrams, overlaying and displaying the gradient directions; The protection intensity gradient field is a vector field representing the protection intensity. The rate of change in space, each component representing Rate of change along the corresponding coordinate direction. Streamline diagram based on the protection intensity gradient field. Calculations are performed. The gradient is approximated using the central difference method in each grid cell, involving six adjacent cells. Streamlines originate from uniformly distributed seed points and are pursued using a fourth-order Runge-Kutta integral, with the step size adaptively controlled between 0.5Δx and 2Δx. The pursuit terminates when the streamline exits the slice boundary or enters a low-gradient region. Each streamline is drawn with a semi-transparent white arrow, and the arrow density varies with the gradient. Increase.

[0081] S7.3: Render dynamic red vortex icons for regions where the curl exceeds the threshold; The privacy vortex region dynamic highlighting obtains the current vortex cluster list from the vortex detection module. For each cluster, its center coordinates and connectivity are projected onto the slice plane. When the cluster center is located in the current slice, a rotating red vortex icon is drawn at the corresponding pixel. If the cluster spans multiple layers, a semi-transparent projection outline is drawn at the slice boundary.

[0082] S7.4: Provides an interactive interface that interprets user dragging streamline operations as gradient boundary conditions or click operations as local damping enhancements, triggering global rebalancing of the privacy rheological field.

[0083] In this embodiment, the following process is executed cyclically during runtime: First, the input data stream is segmented into rheological units according to a fixed time window and logical topology, mapped to three-dimensional grid coordinates using a hash function, and the privacy rheological field is initialized. Next, the sensitivity entropy, threat pressure, and resource density of each rheological unit are calculated in real time, and the protection strength is generated using a continuously differentiable mapping function. Then, non-Newtonian behavior is triggered based on the rate of change of protection strength and the threat pressure gradient, prioritizing shear thickening and suppressing shear thinning within the vortex region. Simultaneously, the global evolution of the field is driven by solving partial differential equations containing diffusion, coupling, and damping terms using the finite volume method. During the evolution process, the curl of the threat pressure to resource density ratio is calculated in real time, privacy vortices are detected, and suppression loops are activated, using virtual damping... The protection strength of the source and the accelerated damping term is forced to converge to the neighborhood average. After the evolution is completed, the strategy is selected by querying the continuous mapping table based on the updated protection strength. Plaintext transmission, perturbation, homomorphic encryption or strong noise addition are performed on each rheological unit field to generate a protected data stream output to the downstream. Finally, the field slice is projected into a heat map and streamline map, the vortex region is dynamically highlighted, and the manually dragged streamline is interpreted as a gradient boundary condition or the click injection is a local damping enhancement through the interactive interface, triggering the global rebalancing of the field. Non-Newtonian behavior, vortex suppression and artificial intervention are coordinated through priority queue and state mask to ensure that shear thickening dominates in the vortex region, artificial injection is given priority, and the damping term continues to converge to the stable anchor point until the curl drops below the threshold and normal evolution is restored.

[0084] Example 2: Application in a Medical Data Sharing Platform The medical data sharing platform receives patient record streams from hospital terminals, containing diagnostic text, test values, and image metadata. The platform segments the data stream into rheological units based on hospital ID and examination type, with a time window of 5 minutes. Sensitivity entropy calculation prioritizes the distribution of unique values ​​in diagnostic fields, with a weight increased to 0.6. Threat pressure analysis integrates external attack intelligence with hospital IP addresses, with a weight increased to 0.4. Resource density monitoring tracks the CPU and memory usage of platform edge nodes.

[0085] The protection strength mapping function employs trivariate spline interpolation, and the control point grid resolution is doubled to 64×64×64 to ensure fine-grained control. In non-Newtonian behavior, the shear thickening threshold is reduced by 30%, and abnormal access is immediately switched to fully homomorphic encryption. Shear thinning is triggered only when doctors query frequently and there are no vortices, transforming into local perturbations to preserve diagnostic keywords.

[0086] The partial differential evolution engine reduces the diffusion coefficient by half, enhancing the protection strength and smoothing the space to prevent excessive differences in protection between adjacent patient records. The vortex detection range is expanded to a 7×7×7 neighborhood, and the suppression loop acceleration factor is increased by 5 times, quickly calming field oscillations caused by sudden batch queries.

[0087] The visualization interface is projected onto the diagnostic department's topology plane, and the heatmap color mapping is adjusted to a medical blue-green gradient. Flow lines display the flow direction of patient records. Manual intervention allows department heads to drag flow lines to adjust protection gradients or click on high-risk records to inject damping, with real-time rebalancing ensuring compliant output.

[0088] When the protection strategy is implemented, the diagnostic text is subjected to strong differential privacy noise, the test values ​​are encrypted using partial homomorphic encryption, and the image metadata only generalizes the hospital identifier. The protected data stream is output to the research terminal, preserving statistical availability while meeting privacy regulations.

[0089] As an embodiment of the present invention, a network security data privacy protection and management system is disclosed, employing the specific implementation method described above for the network security data privacy protection and management method. The system includes: Rheological unit construction module: Divides the input data stream into multiple rheological units according to time, source and type; and assigns a three-dimensional logical coordinate in a discrete spatial grid to each rheological unit to initialize the privacy rheological field; each grid unit in the privacy rheological field corresponds to a protection strength value; 3D parametric modeling module: calculates the sensitivity entropy, threat pressure, and resource density of each rheological unit in real time; maps the sensitivity entropy, threat pressure, and resource density to the protection strength at the current moment through a continuously differentiable mapping function, and updates the privacy rheological field; Nonlinear protection trigger module: Based on the rate of change of protection strength and the spatial gradient of threat pressure, triggers nonlinear protection behavior and adjusts the protection strength; switches to the corresponding data protection strategy according to the adjusted protection strength; the nonlinear protection behavior includes shear thickening and shear thinning; Partial differential equation construction module: Based on the adjusted protection strength, calculate the instantaneous rate of change of the protection strength and update the strength field distribution; the instantaneous rate of change of the protection strength is a partial differential equation composed of diffusion term, threat pressure coupling term, resource density coupling term and vortex suppression term; Vortex Detection and Suppression Module: Calculates the curl of the gradient vector of the scalar field composed of the ratio of threat pressure to resource density in three-dimensional space; detects privacy vortex regions based on the curl modulus and activates the privacy vortex suppression loop; suppresses fluctuations in protection strength within privacy vortex regions based on the suppression mechanism; Policy execution module: Based on the updated protection strength, it selects the corresponding protection policy for each rheological unit from the preset mapping table, and executes the protection policy on the data within the rheological unit to output the protected data stream.

[0090] As an embodiment of the present invention, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it employs the specific implementation method described above for the secure data privacy protection management method.

[0091] As an embodiment of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, adopts the specific implementation method described above for the secure data privacy protection management method.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for network security data privacy protection and management, characterized in that, include: The input data stream is divided into multiple rheological units according to time, source, and type; Each rheological unit is assigned a three-dimensional logical coordinate in a discrete spatial grid to initialize the privacy rheological field; each grid unit in the privacy rheological field corresponds to a protection strength value; The sensitivity entropy, threat pressure, and resource density of each rheological unit are calculated in real time; the sensitivity entropy, threat pressure, and resource density are mapped to the protection strength at the current moment through a continuously differentiable mapping function, and the privacy rheological field is updated. Based on the rate of change of protection strength and the spatial gradient of threat pressure, nonlinear protection behavior is triggered and the protection strength is adjusted. The corresponding data protection strategy is switched according to the adjusted protection strength; the nonlinear protection behavior includes shear thickening and shear thinning. Based on the adjusted protection strength, the instantaneous rate of change of the protection strength is calculated and the strength field distribution is updated; the instantaneous rate of change of the protection strength is a partial differential equation consisting of a diffusion term, a threat pressure coupling term, a resource density coupling term, and a vortex suppression term; Calculate the curl of the gradient vector of the scalar field consisting of the ratio of threat pressure to resource density in three-dimensional space; detect privacy vortex regions based on the curl modulus and activate privacy vortex suppression loops; The fluctuation of protection strength within the privacy vortex region is suppressed based on the suppression mechanism; Based on the updated protection strength, a corresponding protection strategy is selected for each rheostat unit from a preset mapping table, and the protection strategy is applied to the data within the rheostat unit to output a protected data stream.

2. The network security data privacy protection management method according to claim 1, characterized in that, The input data stream is divided into multiple rheological units based on time, source, and type; Each rheological unit is assigned a three-dimensional logical coordinate in the virtual space grid to initialize the privacy rheological field, including: Based on a preset time window and a logical topology defined by data source identifiers and data type labels, the data stream is segmented to generate multiple rheostat units containing a preset minimum number of data packets. Each rheological unit is assigned a unique logical coordinate in a three-dimensional discrete spatial grid based on a hash function; Each rheological unit is embedded into the grid cell corresponding to its logical coordinate in the discrete spatial grid with an initial protection strength to complete the initialization of the privacy rheological field; The three-dimensional discrete space grid has a side length of The initial protection strength of the cubic mesh is 0. When the rheological element is embedded in the mesh, the initial protection strength of the corresponding mesh element is updated to the weighted value of the total number of rheological element fields and the average length of the corresponding field.

3. The network security data privacy protection management method according to claim 1, characterized in that, The real-time calculation of the sensitivity entropy, threat pressure, and resource density of each rheological unit includes: The sensitivity entropy is calculated by: calculating the empirical probability of each attribute field within the sliding window corresponding to the rheological unit; calculating the Shannon entropy of each attribute field based on the empirical probability; and then normalizing the weighted average of the Shannon entropies of all attribute fields to obtain the sensitivity entropy. The empirical probability is the ratio of the frequency of occurrence of the corresponding value of the attribute field within the sliding window to the capacity of the sliding window.

4. The network security data privacy protection management method according to claim 3, characterized in that, The threat pressure is calculated by weighting the query frequency per unit time, the abnormal access detection score, and the external threat intelligence matching degree. Among them, the counter at time intervals Total number of query requests and time intervals for internal rheostat units The ratio is used to obtain the query frequency per unit time; The access pattern vector of the rheological unit is input into the isolated forest model, and the output value of the model is normalized to obtain the abnormal access detection score; the access pattern vector consists of the source IP, time interval and operation type. The external threat intelligence database is queried using the source identifier and type label of the rheological unit; from all matching entries, the maximum confidence value is selected and normalized to obtain the external threat intelligence matching degree.

5. A network security data privacy protection management method according to claim 3, characterized in that, The resource density is calculated by calculating the ratio of available CPU cores to total CPU cores, and the ratio of memory resources to total memory resources; then, the two ratios are geometrically averaged to obtain the resource density.

6. The network security data privacy protection management method according to claim 1, characterized in that, The sensitivity entropy, threat pressure, and resource density are mapped to the protection strength at the current moment, including one of the following two methods: First, a control point network is predefined in a three-dimensional space; the positions of sensitivity, threat pressure and resource density in the predefined three-dimensional grid are determined, and the eight vertices corresponding to the positions are found; based on the protection strength values ​​of the eight vertices, the protection strength value corresponding to the current triple is calculated using trilinear interpolation and B-spline convolution mathematical methods. Secondly, select in advance in a three-dimensional space One center point; based on radial basis function network, using sensitivity entropy, threat pressure and resource density as input vectors, and calculating and... The protection strength value corresponding to the current ternary is calculated by combining the distance of each center point with the weight of each center point.

7. The network security data privacy protection management method according to claim 1, characterized in that, The triggering of nonlinear protection behavior and adjustment of protection strength includes: Based on time point The last time cycle Internally, calculate threat pressure The rate of change; The rate of change of threat pressure is calculated along the X, Y, and Z axes in three-dimensional space to obtain the spatial gradient of threat pressure. If the rate of change is positive and the spatial gradient exceeds a preset first pressure threshold, then shear thickening behavior is triggered. When shear thickening behavior is triggered, the current protection strength is multiplied by an amplification factor greater than 1, and the system switches to the first protection strategy.

8. A network security data privacy protection management method according to claim 7, characterized in that, The triggering of nonlinear protection behavior also includes: If the rate of change of threat pressure is less than the preset second pressure threshold, and the current query frequency per unit time is higher than the historical average query frequency, and the current rheological unit is not in the privacy vortex zone, then shear thinning behavior is triggered, the current protection strength is divided by a reduction factor greater than 1, and the second protection strategy is switched. When the shear thickening behavior and the shear thinning behavior are triggered simultaneously, the shear thickening behavior is executed first.

9. A network security data privacy protection management method according to claim 1, characterized in that, The step of switching to the corresponding data protection strategy based on the adjusted protection strength includes: A continuous mapping table is pre-established, which defines multiple protection strength intervals arranged in ascending order, and each interval corresponds to a protection strategy; Based on the current protection strength value, the corresponding protection strategy is determined by querying the continuous mapping table. The protection strategy includes: local perturbation, partial homomorphic encryption, fully homomorphic encryption, and differential privacy noise.

10. A network security data privacy protection management method according to claim 1, characterized in that, The partial differential equations described above include: The finite volume method is used to solve the differential partial equations in discrete space. The differential partial equations include diffusion terms, threat-pressure coupling terms, resource density coupling terms, and vortex suppression terms. The diffusion term is calculated from the diffusion term of the Laplace of the protection strength; The threat pressure coupling term is obtained by multiplying the threat pressure by the threat pressure coupling coefficient; The resource density coupling term is obtained by multiplying the resource density by the resource density coupling coefficient; The vortex suppression term is obtained by multiplying the difference between the protection strength and its mean by the vortex activation function by the vortex suppression coefficient. At each time step, the sum of the diffusion term, threat pressure positive coupling term, resource density negative coupling term, and vortex suppression term for each grid cell is calculated to obtain the rate of change of the current protection strength; Based on the rate of change of the current protection strength, the protection strength of all grid cells in the privacy rheological field is updated in real time using an explicit time format; The value of the vortex activation function is determined by the result of the privacy vortex detection.

11. A network security data privacy protection management method according to claim 1, characterized in that, Calculate the curl of the gradient vector in three-dimensional space; Curl-based modulus-length detection of privacy-preserving vortex regions includes: In a scalar field consisting of the ratio of threat pressure to resource density, the curl of the gradient vector is calculated at the center of each grid cell. The curl is composed of three curl components in the X, Y, and Z axis directions; The curl modulus is calculated based on the three curl components. If the curl modulus exceeds the preset first curl threshold, the grid cell is marked as a privacy vortex region.

12. The network security data privacy protection management method according to claim 1, characterized in that, The step of selecting a corresponding protection strategy for each rheological unit from a preset mapping table includes: Establish a continuous mapping table from protection strength values ​​to protection strategies; Based on the current protection strength value of the rheostat unit, a query is performed in the mapping table to match the corresponding protection strategy; Iterate through each attribute field within the rheological unit, and for each attribute field, call the corresponding data transformation algorithm according to its type, sensitivity level, and current protection strength to generate the transformed field value; All transformed field values ​​are encapsulated into a protected data stream.

13. A network security data privacy protection management method according to claim 1, characterized in that, Also includes: Visualize the protection intensity field, its gradient field, and the privacy vortex region; It also provides an interface to receive user interaction based on the visual results; The interactive operation is interpreted as a modification of the boundary or local conditions of the protection intensity field to trigger the rebalancing of the privacy rheological field.

14. A network security data privacy protection management method according to claim 13, characterized in that, The visualization of the protection intensity field, its gradient field, and the privacy vortex region includes: From the three-dimensional protective intensity field, a two-dimensional plane with fixed coordinates is selected for slicing; The protection strength value of each grid cell on the slice is mapped to color to generate a two-dimensional heat map; The gradient of the protection intensity field is calculated using central difference and six-neighborhood, and streamlines of the gradient direction and intensity are plotted on the two-dimensional thermogram. On the two-dimensional heatmap, the grid cells containing the privacy vortex region are dynamically highlighted.

15. A network security data privacy protection management method according to claim 13, characterized in that, The interactive operation is parsed as a modification of the boundary conditions or local conditions of the protection intensity field, including: It receives the user's drag operation on the visualized streamline and parses the operation into a modification instruction for the gradient boundary conditions of the protection intensity field; It receives user clicks at specific locations on the visualization interface and parses the clicks into commands to inject damping enhancement at the corresponding grid cells. According to the modification instruction and the injection instruction, the partial differential equation is adjusted to trigger the rebalancing of the protection intensity field.

16. A network security data privacy protection management system, performing the secure data privacy protection management as described in any one of claims 1-15, characterized in that, The system includes: Rheological unit construction module: Divides the input data stream into multiple rheological units according to time, source and type; and assigns a three-dimensional logical coordinate in a discrete spatial grid to each rheological unit to initialize the privacy rheological field; each grid unit in the privacy rheological field corresponds to a protection strength value; 3D parametric modeling module: calculates the sensitivity entropy, threat pressure, and resource density of each rheological unit in real time; maps the sensitivity entropy, threat pressure, and resource density to the protection strength at the current moment through a continuously differentiable mapping function, and updates the privacy rheological field; Nonlinear protection trigger module: Based on the rate of change of protection strength and the spatial gradient of threat pressure, triggers nonlinear protection behavior and adjusts the protection strength; switches to the corresponding data protection strategy according to the adjusted protection strength; the nonlinear protection behavior includes shear thickening and shear thinning; Partial differential equation construction module: Based on the adjusted protection strength, calculate the instantaneous rate of change of the protection strength and update the strength field distribution; the instantaneous rate of change of the protection strength is a partial differential equation composed of diffusion term, threat pressure coupling term, resource density coupling term and vortex suppression term; Vortex Detection and Suppression Module: Calculates the curl of the gradient vector of the scalar field composed of the ratio of threat pressure to resource density in three-dimensional space; detects privacy vortex regions based on the curl modulus and activates the privacy vortex suppression loop; suppresses fluctuations in protection strength within privacy vortex regions based on the suppression mechanism; Policy execution module: Based on the updated protection strength, it selects the corresponding protection policy for each rheological unit from the preset mapping table, and executes the protection policy on the data within the rheological unit to output the protected data stream.

17. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements the secure data privacy protection management method according to any one of claims 1-15.

18. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the secure data privacy protection management method according to any one of claims 1-15.

Images