Power data security transmission strategy dynamic optimization method based on gray measurement and calculation

Abstract

The invention discloses a power data security transmission strategy dynamic optimization method based on gray measurement, which comprises the following steps: S1, collecting and preprocessing multi-source heterogeneous data in a power system, and generating a time sequence data set and a structure attribute data set; s2, performing privacy protection and noise addition on the time sequence data set and the structure attribute data set by applying a differential privacy mechanism; s3, performing evaluation by adopting a gray scale measurement and calculation method, calculating a gray scale weight and generating a transmission strategy; s4, according to the gray scale weight classification data, adjusting a transmission path and a time sequence; s5, dynamically adjusting the differential privacy noise intensity and the gray scale weight according to the network bandwidth, the transmission delay and the data requirement; and S6, continuously monitoring the transmission efficiency and the privacy effect, and adjusting a transmission strategy and a protection mechanism in real time according to feedback. According to the method, the dynamic collaborative optimization of privacy protection and scheduling efficiency of the power data in the transmission process is realized by fusing a differential privacy mechanism and a gray calculation method.

Description

Technical Field

[0001] The present invention relates to the field of intelligent scheduling technology, and in particular to a method for dynamically optimizing a power data security transmission strategy based on grayscale measurement. Background Art

[0002] With the rapid development of smart grids and power information technology, the amount of data in power systems, including equipment operating status, user electricity usage behavior, and dispatch control commands, has exploded. The real-time transmission and dispatch execution of this power data typically rely on multi-level communication links, including edge gateways, dispatch centers, and cloud platforms, to coordinate tasks. However, power data transmission not only faces limited bandwidth resources and frequent link fluctuations, but also carries the risk of exposing a large amount of sensitive information. This is especially true for user-side electricity usage data, equipment fault logs, and system dispatch commands. If transmitted without effective protection, these data can easily lead to privacy leaks and scheduling anomalies. Therefore, improving the efficiency and reliability of power data transmission while ensuring data privacy has become a key research direction in current smart grid systems.

[0003] Among existing technologies, some attempts to use differential privacy mechanisms to protect power data from disturbances, adding noise to the raw data to reduce the identifiability of sensitive information. Other technologies use fixed priorities or scoring models based on static features to perform preliminary scheduling of transmission tasks. However, these approaches generally suffer from three shortcomings: First, in actual deployment, traditional differential privacy mechanisms lack specificity in noise intensity and cannot dynamically adjust according to the real-time nature of the data, transmission risk, or budget status, resulting in frequent excessive disturbances or insufficient protection. Second, transmission priority models based on static scoring are difficult to adapt to dynamic environments such as network bandwidth and latency, lacking the ability to perceive link status in real time, resulting in scheduling delays. Third, existing scheduling strategies generally ignore the diversity and structural correlation of data types, lacking refined classification and differentiated control of transmission paths and scheduling timing arrangements, affecting the overall system's response efficiency and privacy security.

[0004] Therefore, how to provide a dynamic optimization method for power data security transmission strategy based on grayscale measurement is an urgent problem that technicians in this field need to solve. Summary of the Invention

[0005] One objective of the present invention is to propose a method for dynamically optimizing power data security transmission strategies based on grayscale measurement. This method integrates differential privacy protection mechanisms, grayscale measurement models, and dynamic scheduling control technologies. It describes in detail how, during the transmission of multi-source heterogeneous power data, noise intensity and grayscale weights are dynamically adjusted based on the importance, privacy sensitivity, and real-time requirements of the data, thereby constructing an intelligent transmission strategy generation method that integrates data classification, path allocation, and time-series scheduling. This method offers the advantages of controllable privacy protection strength, precise scheduling decisions, high transmission efficiency, and strong adaptability. It can effectively improve the data transmission security and scheduling responsiveness of smart power systems in complex network environments.

[0006] According to an embodiment of the present invention, a method for dynamically optimizing a power data security transmission strategy based on grayscale measurement includes the following steps:

[0007] S1. Collect multi-source heterogeneous data in the power system and pre-process the multi-source heterogeneous data to generate standardized time series data sets and structural attribute data sets;

[0008] S2. Apply the differential privacy mechanism to protect the privacy of the time series dataset and the structural attribute dataset, and add noise to the time series dataset and the structural attribute dataset;

[0009] S3. Evaluate the time series dataset and the structural attribute dataset using a grayscale measurement method. Calculate the grayscale weight of each data transmission strategy based on the importance of the data, privacy protection requirements, and real-time requirements. Generate a transmission strategy based on the evaluation results.

[0010] S4. Classify different types of data according to grayscale weights, and adjust the data transmission path and timing according to the classification results;

[0011] S5. Dynamically adjust the noise intensity and grayscale transmission weight of the differential privacy mechanism based on real-time network bandwidth, transmission delay, and data transmission requirements to balance privacy protection and data transmission efficiency.

[0012] S6. During the data transmission process, the transmission efficiency and privacy protection effect are continuously monitored, feedback adjustments are made, and based on the feedback data, the transmission strategy and privacy protection mechanism are updated in real time to optimize the transmission delay, bandwidth utilization and privacy protection strength, thus forming a dynamic closed-loop power data transmission optimization process.

[0013] Optionally, the multi-source heterogeneous data specifically includes power equipment operation data, user power consumption data and environmental monitoring data.

[0014] Optionally, the preprocessing of multi-source heterogeneous data specifically includes data cleaning, missing value filling, standardization and format unification.

[0015] Optionally, the S2 specifically includes:

[0016] S21, for each data item x in the standardized time series data set and structural attribute data set i Perform sensitivity analysis and calculate local sensitivity LS(x i ), the local sensitivity measures the maximum output change of a data item in adjacent samples;

[0017] S22. Set the system's overall privacy budget ε total , and adopt the group allocation strategy to total Divided into ε by data type t Time Series Data Privacy Budget and ε k Structural attribute data privacy budget;

[0018] S23. Construct a dynamic Laplace mechanism and set the scale parameter of the Laplace distribution to a function form, denoted as b i (t):

[0019]

[0020] Among them, ε i (t) is the data item x i The local privacy budget at time t is, Based on the transmission risk level R t The regulation function, LS(x i ) represents the local sensitivity of the i-th data item;

[0021] S24. Generate noise terms based on mixed distribution strategy

[0022]

[0023] Among them, λ is the weighting coefficient of Laplace and Gaussian distribution, The mean is 0 and the variance is Gaussian distribution, Estimated by the data change rate, Lap(0,b i ) means taking 0 as the mean, b i is the Laplace distribution with scale parameter;

[0024] S25. Introduce a multi-scale time consistency control mechanism for time series data sets and set a short-term time window w s With the long time window w l , calculate the local time correlation coefficient respectively and the global stability correlation coefficient Introducing the weight factor ω s ,ω l :

[0025]

[0026] Among them, x t Represents the original data value at time step t in the time series data set, x′ t represents the data value after noise processing by the differential privacy mechanism at time step t, represents the mixed distribution noise value generated at time step t;

[0027] S26. For the structural attribute dataset, the self-attention mechanism is introduced to model the importance of feature fields, and each attribute vector x is defined. k The query weight matrix Q k With the key weight matrix K k , calculate the normalized importance coefficient γ through the attention function k , and integrated into the differential privacy noise formula as follows:

[0028]

[0029] Where x′ k represents the noised data item of the kth structural attribute field after differential privacy noise is added, x k Represents the kth original feature field data item in the structural attribute dataset, and softmax is a normalization operation;

[0030] S27, introduce the residual suppression mechanism and define the noise disturbance error as δ i , set an upper limit on the disturbance value |δ i |≤θ i , where θ i To control the threshold value of disturbance and suppress extreme abnormal points;

[0031] S28. Build a traceable budget update mechanism to record the privacy budget usage of each data item during each round of transmission, and dynamically update the local privacy budget based on the cumulative usage ratio and the remaining budget status at the end of the cycle;

[0032] S29. For all time series data items x t With the structure attribute data item x k Execute steps S21 to S28 respectively to obtain the privacy-processed time series dataset D′ t ={x′ t} and structural attribute dataset D′ k ={x′ k}, keep the data format consistent.

[0033] Optionally, the S3 specifically includes:

[0034] S31, construct fusion data matrix F = {x′ t ,x′ k}, the time series dataset D′ after differential privacy processing t and structural attribute dataset D′ k Align and integrate on the time axis and feature dimension to form a fusion sample;

[0035] S32, introduce a multi-index grayscale decision vector system and set three core indicators: data importance index I imp , Privacy Sensitivity Index I pri , Real-time index I tim , extracting feature quantization from data semantics, differential privacy parameters and sampling frequency respectively;

[0036] S33. Design of multivariate mapping function ψ i , each fusion data x′ i Mapping to a three-dimensional index vector:

[0037] ψ i =[I imp (x′ i ),I pri (x′ i ),I tim (x′ i )];

[0038] S34. Compare the three-dimensional indicator vector corresponding to each fused sample with a preset reference standard vector. Based on the numerical differences between the three dimensions, calculate the overall difference between the fused sample and the standard reference. The smaller the grayscale difference, the closer the fused data is to the optimal transmission target under the multi-dimensional indicator.

[0039] S35. Construct a segmented adjustable grayscale correlation coefficient function and define the correlation coefficient as:

[0040]

[0041] Among them, Δ min , Δ max is the extreme value of grayscale difference in the current sample set, ζ is the grayscale resolution coefficient, Δ i Represents the multi-index difference between the i-th fusion sample and the grayscale reference sample;

[0042] S36, introduce a weighted dynamic adjustment mechanism based on the historical communication load L i , Current privacy budget remaining rate Link reliability R i Construct influence function φ i , dynamically correct grayscale score ω i :

[0043]

[0044] Among them, θ1, θ2, θ3 are weight coefficients;

[0045] S37, introduce gray priority timing stability factor κ i , based on the Grayscale weight volatility analysis for multiple consecutive periods, calculate the sample variance:

[0046]

[0047] in, is the average grayscale weight in the sliding window, T represents the total number of time steps contained in the sliding time window used for evaluation, ω i (t) represents the gray priority score of the i-th fusion sample at time step t;

[0048] S38. According to the fluctuation range of the grayscale score of each data item in a continuous period, a stability penalty factor is introduced. When the grayscale score fluctuates more, the priority of the data will be appropriately reduced; on the contrary, the data with stable grayscale score will maintain a higher priority, and the penalty factor will be applied to the grayscale score ω. i , generate the final transmission priority weight;

[0049] S39. Sort all data items in descending order according to the final transmission priority weights and establish a transmission strategy matrix Where ID i is the data item identifier, P i is the transmission priority level, is the final transmission priority weight.

[0050] Optionally, the S4 specifically includes:

[0051] S41, extract the generated final transmission priority weight, and construct the fusion sample classification vector C = {c1, c2, ..., c n}, where c i ∈{user data, device status data, control instruction data}, indicating the data type label to which the fusion sample belongs;

[0052] S42, based on the type label c of the fusion sample i And the corresponding grayscale weight Construct a two-dimensional mapping matrix T ij , represents the priority distribution characteristics of various samples in the current scheduling cycle;

[0053] S43. Type label c based on fusion sample i and grayscale weight All fusion samples are divided into multiple priority transmission clusters G, and the samples in each cluster have similar type attributes and grayscale characteristics;

[0054] S44, for each cluster G k Assign independent data channel number R k , and generate the corresponding path set P k ,The path set is dynamically generated based on the current network topology and link bandwidth ,status;

[0055] S45. Construct a multi-factor path scheduling evaluation function for the fusion sample and introduce the path residual bandwidth Average path delay Path delay fluctuation And the scheduling sensitivity coefficient μ of the fusion sample i :

[0056]

[0057] Among them, Λ ij Represents the fusion sample x′ i Adaptation path p j The comprehensive scheduling score, μ i represents the scheduling sensitivity coefficient of the fusion sample;

[0058] S46: Perform path selection operation on each fusion sample and select the corresponding path set P. k has the maximum scheduling score Λ ij Path As the final transmission path, and build the path allocation table ID i is the unique identification number of the fusion sample, c i Indicates the type of fusion sample;

[0059] S47, according to the type of fusion sample c i , scheduling sensitivity coefficient μ i and network channel load status, calculate the transmission timing interval Δt between fusion samples i , and set the actual scheduling time Control the orderliness and real-time nature of data scheduling;

[0060] S48. Constructing a path-time scheduling mapping table Send the fused samples to the transmission channel according to the specified path and scheduling time;

[0061] S49, applying the path-time scheduling mapping table S to the data scheduling execution process, monitoring the usage status, scheduling accuracy and link performance of each path in real time during the data transmission process, and generating a transmission status log record;

[0062] S410. Collect actual transmission feedback information of the fused sample, including transmission path state changes, scheduling execution errors, and channel congestion, update the state parameters and time scheduling intervals in the path scheduling evaluation function, and complete the adaptive optimization process of the path selection and scheduling timing of the fused sample.

[0063] Optionally, the S5 specifically includes:

[0064] S51. Collect real-time network status information during the current transmission cycle, including link bandwidth utilization, data transmission delay, packet loss rate, and load level of each path channel;

[0065] S52, analyzing the transmission urgency and privacy protection level of each fusion sample in the current scheduling list, and extracting the real-time demand parameters of the sample, including the response time window, the remaining privacy budget, and the historical transmission performance;

[0066] S53. Dynamically adjust the noise intensity in the differential privacy mechanism. Based on the privacy budget consumption status of the sample and the available network bandwidth, moderately reduce the perturbation intensity of high-priority samples while ensuring privacy.

[0067] S54. Synchronously adjust the transmission weight in the grayscale measurement, dynamically reduce the grayscale priority weight of samples on high-load paths or high-latency nodes, and alleviate network congestion.

[0068] S55. Based on the sample type and its real-time requirements, a collaborative mapping relationship between privacy perturbation and weight adjustment is constructed to achieve an adaptive balance between the degree of differential privacy perturbation and the priority of grayscale transmission.

[0069] S56. Apply the adjusted noise parameters and transmission weight feedback to the next round of sample scheduling, path selection, and transmission rhythm control.

[0070] The beneficial effects of the present invention are:

[0071] This invention constructs a dynamic optimization method for secure power data transmission strategies that integrates differential privacy mechanisms with grayscale measurement models, effectively overcoming the shortcomings of existing technologies in terms of limited privacy protection, inflexible transmission scheduling, and crude path selection. Compared to traditional methods that rely on fixed noise perturbations and static priority sorting, this invention introduces local sensitivity modeling, an adaptive noise mixing mechanism, and a multi-dimensional dynamic weight adjustment strategy. This allows for dynamic adjustment of the differential privacy mechanism based on real-time network bandwidth, transmission latency, and data transmission requirements, achieving an adaptive balance between privacy protection strength and transmission efficiency.

[0072] At the same time, this invention constructs a multi-dimensional indicator evaluation system through grayscale measurement. This system comprehensively considers data importance, privacy sensitivity, and real-time scheduling, generates priority scores for fused samples, and performs fine-grained scheduling based on path status, link load, and data type. The classification results are used to allocate paths and control timing for fused samples, ensuring that high-priority sensitive data is transmitted with priority and efficiency while protecting privacy. Low-priority data is properly queued to avoid resource conflicts and path congestion.

[0073] This invention implements an end-to-end dynamic optimization process from privacy protection, scheduling decisions, to path selection, with high scenario adaptability and resource regulation capabilities. In actual power data scheduling and transmission, this invention can improve the accuracy and real-time response efficiency of transmission scheduling, significantly reduce the risk of privacy leakage and waste of network resources, and effectively support the demand for secure, efficient, and sustainable data flow management in smart grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0075] Figure 1 This is a flow chart of a method for dynamic optimization of power data security transmission strategy based on grayscale measurement proposed by the present invention;

[0076] Figure 2 This is a grayscale measurement-based dynamic weight evaluation and priority sorting flow chart of a dynamic optimization method for power data security transmission strategy proposed by the present invention. DETAILED DESCRIPTION

[0077] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0078] refer to Figure 1 and Figure 2 A method for dynamically optimizing power data security transmission strategy based on grayscale measurement includes the following steps:

[0079] S1. Collect multi-source heterogeneous data in the power system and pre-process the multi-source heterogeneous data to generate standardized time series data sets and structural attribute data sets;

[0080] S2. Apply the differential privacy mechanism to protect the privacy of the time series dataset and the structural attribute dataset, and add noise to the time series dataset and the structural attribute dataset;

[0081] S3. Evaluate the time series dataset and the structural attribute dataset using a grayscale measurement method. Calculate the grayscale weight of each data transmission strategy based on the importance of the data, privacy protection requirements, and real-time requirements. Generate a transmission strategy based on the evaluation results.

[0082] S4. Classify different types of data according to grayscale weights, and adjust the data transmission path and timing according to the classification results;

[0083] S5. Dynamically adjust the noise intensity and grayscale transmission weight of the differential privacy mechanism based on real-time network bandwidth, transmission delay, and data transmission requirements to balance privacy protection and data transmission efficiency.

[0084] S6. During the data transmission process, the transmission efficiency and privacy protection effect are continuously monitored, feedback adjustments are made, and based on the feedback data, the transmission strategy and privacy protection mechanism are updated in real time to optimize the transmission delay, bandwidth utilization and privacy protection strength, thus forming a dynamic closed-loop power data transmission optimization process.

[0085] In this embodiment, the multi-source heterogeneous data specifically includes power equipment operation data, user power consumption data and environmental monitoring data.

[0086] In this embodiment, the preprocessing of multi-source heterogeneous data specifically includes data cleaning, missing value filling, standardization and format unification.

[0087] In this embodiment, S2 specifically includes:

[0088] S21, for each data item x in the standardized time series data set and structural attribute data set i Perform sensitivity analysis and calculate local sensitivity LS(x i ), the local sensitivity measures the maximum output change of a data item in adjacent samples;

[0089] S22. Set the system's overall privacy budget ε total , and adopt the group allocation strategy to total Divided into ε by data type t Time Series Data Privacy Budget and ε k Structural attribute data privacy budget;

[0090] S23. Construct a dynamic Laplace mechanism and set the scale parameter of the Laplace distribution to a function form, denoted as b i (t):

[0091]

[0092] Among them, ε i (t) is the data item x iThe local privacy budget at time t is, Based on the transmission risk level R t The regulation function, LS(x i ) represents the local sensitivity of the i-th data item;

[0093] S24. Generate noise terms based on mixed distribution strategy

[0094]

[0095] Among them, λ is the weighting coefficient of Laplace and Gaussian distribution, The mean is 0 and the variance is Gaussian distribution, Estimated by the data change rate, Lap(0,b i ) means taking 0 as the mean, b i is the Laplace distribution with scale parameter;

[0096] S25. Introduce a multi-scale time consistency control mechanism for time series data sets and set a short-term time window w s With the long time window w l , calculate the local time correlation coefficient respectively and the global stability correlation coefficient Introducing the weight factor ω s ,ω l :

[0097]

[0098] Among them, x t Represents the original data value at time step t in the time series data set, x′ t represents the data value after noise processing by the differential privacy mechanism at time step t, represents the mixed distribution noise value generated at time step t;

[0099] S26. For the structural attribute dataset, the self-attention mechanism is introduced to model the importance of feature fields, and each attribute vector x is defined. k The query weight matrix Q k With the key weight matrix K k , calculate the normalized importance coefficient γ through the attention function k , and integrated into the differential privacy noise formula as follows:

[0100]

[0101] Where x′ k represents the noised data item of the kth structural attribute field after differential privacy noise is added, x kRepresents the kth original feature field data item in the structural attribute dataset, and softmax is a normalization operation;

[0102] S27, introduce the residual suppression mechanism and define the noise disturbance error as δ i , set an upper limit on the disturbance value |δ i |≤θ i , where θ i To control the threshold value of disturbance and suppress extreme abnormal points;

[0103] S28. Build a traceable budget update mechanism to record the privacy budget usage of each data item during each round of transmission, and dynamically update the local privacy budget based on the cumulative usage ratio and the remaining budget status at the end of the cycle;

[0104] S29. For all time series data items x t With the structure attribute data item x k Execute steps S21 to S28 respectively to obtain the privacy-processed time series dataset D′ t ={x′ t} and structural attribute dataset D′ k ={x′ k}, keep the data format consistent.

[0105] In this embodiment, S3 specifically includes:

[0106] S31, construct fusion data matrix F = {x′ t ,x′ k}, the time series dataset D′ after differential privacy processing t and structural attribute dataset D′ k Align and integrate on the time axis and feature dimension to form a fusion sample;

[0107] S32, introduce a multi-index grayscale decision vector system and set three core indicators: data importance index I imp , Privacy Sensitivity Index I pri , Real-time index I tim , extracting feature quantization from data semantics, differential privacy parameters and sampling frequency respectively;

[0108] S33. Design of multivariate mapping function ψ i , each fusion data x′ i Mapping to a three-dimensional index vector:

[0109] ψ i =[I imp (x′ i ),I pri (x′ i ),I tim(x′ i )];

[0110] S34. Compare the three-dimensional indicator vector corresponding to each fused sample with a preset reference standard vector. Based on the numerical differences between the three dimensions, calculate the overall difference between the fused sample and the standard reference. The smaller the grayscale difference, the closer the fused data is to the optimal transmission target under the multi-dimensional indicator.

[0111] S35. Construct a segmented adjustable grayscale correlation coefficient function and define the correlation coefficient as:

[0112]

[0113] Among them, Δ min , Δ max is the extreme value of grayscale difference in the current sample set, ζ is the grayscale resolution coefficient, Δ i Represents the multi-index difference between the i-th fusion sample and the grayscale reference sample;

[0114] S36, introduce a weighted dynamic adjustment mechanism based on the historical communication load L i , Current privacy budget remaining rate Link reliability R i Construct influence function φ i , dynamically correct grayscale score ω i :

[0115]

[0116] Among them, θ1, θ2, θ3 are weight coefficients;

[0117] S37, introduce gray priority timing stability factor κ i , based on the Grayscale weight volatility analysis for multiple consecutive periods, calculate the sample variance:

[0118]

[0119] in, is the average grayscale weight in the sliding window, T represents the total number of time steps contained in the sliding time window used for evaluation, ω i (t) represents the gray priority score of the i-th fusion sample at time step t;

[0120] S38. According to the fluctuation range of the grayscale score of each data item in a continuous period, a stability penalty factor is introduced. When the grayscale score fluctuates more, the priority of the data will be appropriately reduced; on the contrary, the data with stable grayscale score will maintain a higher priority, and the penalty factor will be applied to the grayscale score ω. i , generate the final transmission priority weight;

[0121] S39. Sort all data items in descending order according to the final transmission priority weights and establish a transmission strategy matrix Where ID i is the data item identifier, P i is the transmission priority level, is the final transmission priority weight.

[0122] In this embodiment, the S4 specifically includes:

[0123] S41, extract the generated final transmission priority weight, and construct the fusion sample classification vector C = {c1, c2, ..., c n}, where c i ∈{user data, device status data, control instruction data}, indicating the data type label to which the fusion sample belongs;

[0124] S42, based on the type label c of the fusion sample i And the corresponding grayscale weight Construct a two-dimensional mapping matrix T ij , represents the priority distribution characteristics of various samples in the current scheduling cycle;

[0125] S43. Type label c based on fusion sample i and grayscale weight All fusion samples are divided into multiple priority transmission clusters G, and the samples in each cluster have similar type attributes and grayscale characteristics;

[0126] S44, for each cluster G k Assign independent data channel number R k , and generate the corresponding path set P k ,The path set is dynamically generated based on the current network topology and link bandwidth ,status;

[0127] S45. Construct a multi-factor path scheduling evaluation function for the fusion sample and introduce the path residual bandwidth Average path delay Path delay fluctuation And the scheduling sensitivity coefficient μ of the fusion sample i :

[0128]

[0129] Among them, Λ ij Represents the fusion sample x i Adaptation path p j The comprehensive scheduling score, μ i represents the scheduling sensitivity coefficient of the fusion sample;

[0130] S46: Perform path selection operation on each fusion sample and select the corresponding path set P. k has the maximum scheduling score Λ ij Path As the final transmission path, and build the path allocation table ID i is the unique identification number of the fusion sample, c i Indicates the type of fusion sample;

[0131] S47, according to the type of fusion sample c i , scheduling sensitivity coefficient μ i and network channel load status, calculate the transmission timing interval Δt between fusion samples i , and set the actual scheduling time Control the orderliness and real-time nature of data scheduling;

[0132] S48. Constructing a path-time scheduling mapping table Send the fused samples to the transmission channel according to the specified path and scheduling time;

[0133] S49, applying the path-time scheduling mapping table S to the data scheduling execution process, monitoring the usage status, scheduling accuracy and link performance of each path in real time during the data transmission process, and generating a transmission status log record;

[0134] S410. Collect actual transmission feedback information of the fused sample, including transmission path state changes, scheduling execution errors, and channel congestion, update the state parameters and time scheduling intervals in the path scheduling evaluation function, and complete the adaptive optimization process of the path selection and scheduling timing of the fused sample.

[0135] In this embodiment, the S5 specifically includes:

[0136] S51. Collect real-time network status information during the current transmission cycle, including link bandwidth utilization, data transmission delay, packet loss rate, and load level of each path channel;

[0137] S52, analyzing the transmission urgency and privacy protection level of each fusion sample in the current scheduling list, and extracting the real-time demand parameters of the sample, including the response time window, the remaining privacy budget, and the historical transmission performance;

[0138] S53. Dynamically adjust the noise intensity in the differential privacy mechanism. Based on the privacy budget consumption status of the sample and the available network bandwidth, moderately reduce the perturbation intensity of high-priority samples while ensuring privacy.

[0139] S54. Synchronously adjust the transmission weight in the grayscale measurement, dynamically reduce the grayscale priority weight of samples on high-load paths or high-latency nodes, and alleviate network congestion.

[0140] S55. Based on the sample type and its real-time requirements, a collaborative mapping relationship between privacy perturbation and weight adjustment is constructed to achieve an adaptive balance between the degree of differential privacy perturbation and the priority of grayscale transmission.

[0141] S56. Apply the adjusted noise parameters and transmission weight feedback to the next round of sample scheduling, path selection, and transmission rhythm control.

[0142] Example 1:

[0143] To verify the feasibility of this invention, it was applied to a power dispatch control center. The system processes over 500 million pieces of power operation data and real-time user-side data daily, including device status samples, load regulation control instructions, and household electricity consumption data. With the increasing complexity of energy consumption structures under the dual-carbon era, the types of data are rapidly increasing, and transmission bandwidth is approaching saturation. Furthermore, a large amount of data contains sensitive information such as user behavior trajectories and distribution operation status, resulting in practical problems such as high risk of privacy leakage, large dispatch response delays, and frequent link congestion.

[0144] Previous scheduling and data transmission solutions employed fixed bandwidth allocation and static priority rules, uniformly encrypting all data. Consequently, high-priority data could not be guaranteed to be delivered in real time during peak scheduling periods, while less sensitive data consumed excessive security resources, reducing overall scheduling efficiency. In actual operation, there were numerous instances of scheduling commands not being triggered due to delayed uploads of device status data, or electricity price miscalculations due to failed user data transmissions.

[0145] To verify the effectiveness of the proposed dynamic optimization solution, the research team deployed the proposed transmission optimization algorithm, combining differential privacy with grayscale measurement, in the center's dispatch and control network system from March to May 2025, conducting A / B tests against a traditional solution. During the deployment, 1.8 million pieces of user power consumption data, device status data, and dispatch instruction data were extracted daily from 12 million data points. The data was then grouped in real time, grayscale evaluation was performed, path scheduling was performed, and privacy perturbation control was implemented.

[0146] During the experiment, the present invention periodically fine-tuned the differential privacy perturbation strength by monitoring network bandwidth and link load in real time. It also dynamically calculated grayscale weights based on the importance, sensitivity, and timeliness of the fused samples. The system then selected the optimal transmission path based on the weighted ranking and controlled the sending time, significantly reducing conflicts and delays.

[0147] Table 1 Performance comparison experimental data of the present invention and the traditional solution in power data transmission tasks

[0148]

[0149] Table 1 above clearly demonstrates the significant advantages of the dynamic optimization method for power data security transmission strategy based on grayscale measurement proposed in the present invention in practical applications, especially in terms of key performance indicators such as transmission delay control, data integrity assurance, privacy protection strength, and scheduling response efficiency, which are superior to traditional static scheduling schemes.

[0150] In terms of average transmission delay, under the traditional static scheduling scheme, the average transmission delay for user power data is as high as 220 milliseconds, while the delays for device status data and scheduling control commands are 195 milliseconds and 180 milliseconds, respectively, showing significant response lag. However, after adopting the solution of the present invention, the transmission delays for these three types of data are reduced to 97 milliseconds, 88 milliseconds, and 73 milliseconds, respectively, an average reduction of more than 55%. This greatly improves the real-time performance of data delivery, which is particularly significant for data that requires a low-latency response, such as scheduling control commands.

[0151] In terms of data integrity, traditional solutions experience packet loss rates exceeding 4% for all three types of data, with the packet loss rate for user power data reaching 5.2%. In contrast, the proposed method effectively reduces the risk of data loss through dynamic path allocation and transmission timing control, reducing the packet loss rates for each type of data to 1.1%, 0.9%, and 0.6%, respectively. This demonstrates the system's stable transmission capabilities even under high load or complex network environments.

[0152] In terms of privacy protection, this invention achieves hierarchical disturbance control for different types of data through an improved differential privacy mechanism. Traditional methods achieve an average privacy protection score ranging from 0.57 to 0.61, which is ineffective in addressing data requirements at varying privacy levels. However, under this invention, the privacy protection score for user power data reaches 0.89, for device status data 0.91, and for dispatch control commands as high as 0.94, significantly improving overall privacy strength and demonstrating that this solution can specifically enhance the privacy protection of critical data.

[0153] In terms of scheduling response accuracy, traditional solutions have a response accuracy of 81.5% to 86.0%, which is subject to scheduling task execution deviation. This invention improves the accuracy of data scheduling sorting through a dynamic grayscale weight adjustment mechanism, significantly improving the response accuracy of various tasks. The response accuracy of scheduling control commands has been increased from 86.0% to 98.1%, which has great practical value in real-world environments where the timeliness of scheduling instructions is extremely important.

[0154] Based on the above analysis, it can be seen that the solution of the present invention significantly improves the real-time, stability and scheduling execution efficiency of power data transmission while ensuring a high level of privacy protection, providing a practical, efficient and reliable solution for the secure scheduling and transmission of large-scale data in smart grids.

[0155] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A dynamic optimization method for power data security transmission strategy based on grayscale measurement, characterized in that: The steps include: S1. Collect multi-source heterogeneous data in the power system and pre-process the multi-source heterogeneous data to generate standardized time series data sets and structural attribute data sets; S2. Apply the differential privacy mechanism to protect the privacy of the time series dataset and the structural attribute dataset, and add noise to the time series dataset and the structural attribute dataset; S3. Evaluate the time series dataset and the structural attribute dataset using a grayscale measurement method. Calculate the grayscale weight of each data transmission strategy based on the importance of the data, privacy protection requirements, and real-time requirements. Generate a transmission strategy based on the evaluation results. S4. Classify different types of data according to grayscale weights, and adjust the data transmission path and timing based on the classification results; S5. Dynamically adjust the noise intensity and grayscale transmission weight of the differential privacy mechanism based on real-time network bandwidth, transmission delay, and data transmission requirements to balance privacy protection and data transmission efficiency. S6. During the data transmission process, the transmission efficiency and privacy protection effect are continuously monitored, feedback adjustments are made, and based on the feedback data, the transmission strategy and privacy protection mechanism are updated in real time to optimize the transmission delay, bandwidth utilization and privacy protection strength, thus forming a dynamic closed-loop power data transmission optimization process.

2. The method for dynamic optimization of power data security transmission strategy based on grayscale measurement according to claim 1 is characterized in that: The multi-source heterogeneous data specifically includes power equipment operation data, user power consumption data and environmental monitoring data.

3. The method for dynamic optimization of power data security transmission strategy based on grayscale measurement according to claim 1 is characterized in that: The preprocessing of multi-source heterogeneous data specifically includes data cleaning, missing value filling, standardization and format unification.

4. The method for dynamic optimization of power data security transmission strategy based on grayscale measurement according to claim 1 is characterized in that: The S2 specifically includes: S21, for each data item x in the standardized time series data set and structural attribute data set i Perform sensitivity analysis and calculate local sensitivity LS(x i ), the local sensitivity measures the maximum output change of a data item in adjacent samples; S22. Set the system's overall privacy budget ε total , and adopt the group allocation strategy to total Divided into ε by data type t Time Series Data Privacy Budget and ε k Structural attribute data privacy budget; S23. Construct a dynamic Laplace mechanism and set the scale parameter of the Laplace distribution to a function form, denoted as b i (t): Among them, ε i (t) is the data item x i The local privacy budget at time t is, Based on the transmission risk level R t The regulation function, LS(x i ) represents the local sensitivity of the i-th data item; S24. Generate noise terms based on mixed distribution strategy Among them, λ is the weighting coefficient of Laplace and Gaussian distribution, The mean is 0 and the variance is Gaussian distribution, Estimated by the data change rate, Lap(0,b i ) means taking 0 as the mean, b i is the Laplace distribution with scale parameter; S25. Introduce a multi-scale time consistency control mechanism for time series data sets and set a short-term time window w s With the long time window w l , calculate the local time correlation coefficient respectively and the global stability correlation coefficient Introducing the weight factor ω s ,ω l : Among them, x t Represents the original data value at time step t in the time series data set, x′ t represents the data value after noise processing by the differential privacy mechanism at time step t, represents the mixed distribution noise value generated at time step t; S26. For the structural attribute dataset, the self-attention mechanism is introduced to model the importance of feature fields, and each attribute vector x is defined. k The query weight matrix Q k With the key weight matrix K k , calculate the normalized importance coefficient γ through the attention function k , and integrated into the differential privacy noise formula as follows: Where x′ k represents the noised data item of the kth structural attribute field after differential privacy noise is added, x k Represents the kth original feature field data item in the structural attribute dataset, and softmax is a normalization operation; S27, introduce the residual suppression mechanism and define the noise disturbance error as δ i , set an upper limit on the disturbance value |δ i |≤θ i , where θ i To control the threshold value of disturbance and suppress extreme abnormal points; S28. Build a traceable budget update mechanism to record the privacy budget usage of each data item during each round of transmission, and dynamically update the local privacy budget based on the cumulative usage ratio and the remaining budget status at the end of the cycle; S29. For all time series data items x t With the structure attribute data item x k Execute steps S21 to S28 respectively to obtain the privacy-processed time series dataset D′ t ={x′ t } and structural attribute dataset D′ k ={x′ k }, keep the data format consistent.

5. The method for dynamic optimization of power data security transmission strategy based on grayscale measurement according to claim 1 is characterized in that: The S3 specifically includes: S31, construct fusion data matrix F = {x′ t ,x′ k }, the time series dataset D′ after differential privacy processing t and structural attribute dataset D′ k Align and integrate on the time axis and feature dimension to form a fusion sample; S32, introduce a multi-index grayscale decision vector system and set three core indicators: data importance index I imp , Privacy Sensitivity Index I pri , Real-time index I tim , extracting feature quantization from data semantics, differential privacy parameters and sampling frequency respectively; S33. Design of multivariate mapping function ψ i , each fusion data x′ i Mapping to a three-dimensional index vector: ψ i =[I imp (x′ i ),I pri (x′ i ),I tim (x′ i )]; S34. Compare the three-dimensional indicator vector corresponding to each fused sample with a preset reference standard vector. Based on the numerical differences between the three dimensions, calculate the overall difference between the fused sample and the standard reference. The smaller the grayscale difference, the closer the fused data is to the optimal transmission target under the multi-dimensional indicator. S35. Construct a segmented adjustable grayscale correlation coefficient function and define the correlation coefficient as: Among them, Δ m in, Δ m ax is the extreme value of grayscale difference in the current sample set, ζ is the grayscale resolution coefficient, Δ i Represents the multi-index difference between the i-th fusion sample and the grayscale reference sample; S36, introduce a weighted dynamic adjustment mechanism based on the historical communication load L i , Current privacy budget remaining rate Link reliability R i Construct influence function φ i , dynamically correct grayscale score ω i : Among them, θ1, θ2, θ3 are weight coefficients; S37, introduce gray priority timing stability factor κ i , based on the Grayscale weight volatility analysis for multiple consecutive periods, calculate the sample variance: in, is the average grayscale weight in the sliding window, T represents the total number of time steps contained in the sliding time window used for evaluation, ω i (t) represents the gray priority score of the i-th fusion sample at time step t; S38. According to the fluctuation range of the grayscale score of each data item in a continuous period, a stability penalty factor is introduced. When the grayscale score fluctuates more, the priority of the data will be appropriately reduced; on the contrary, the data with stable grayscale score will maintain a higher priority, and the penalty factor will be applied to the grayscale score ω. i , generate the final transmission priority weight; S39. Sort all data items in descending order according to the final transmission priority weights and establish a transmission strategy matrix Where ID i is the data item identifier, P i is the transmission priority level, is the final transmission priority weight.

6. The method for dynamic optimization of power data security transmission strategy based on grayscale measurement according to claim 1 is characterized in that: The S4 specifically includes: S41, extract the generated final transmission priority weight, and construct the fusion sample classification vector C = {c1, c2, ..., c n }, where c i ∈{user data, device status data, control instruction data}, indicating the data type label to which the fusion sample belongs; S42, based on the type label c of the fusion sample i And the corresponding grayscale weight Construct a two-dimensional mapping matrix T ij , represents the priority distribution characteristics of various samples in the current scheduling cycle; S43. Type label c based on fusion sample i and grayscale weight All fusion samples are divided into multiple priority transmission clusters G, and the samples in each cluster have similar type attributes and grayscale characteristics; S44, for each cluster G k Assign independent data channel number R k , and generate the corresponding path set P k ,The path set is dynamically generated based on the current network topology and link bandwidth ,status; S45. Construct a multi-factor path scheduling evaluation function for the fusion sample and introduce the path residual bandwidth Average path delay Path delay fluctuation And the scheduling sensitivity coefficient μ of the fusion sample i : Among them, Λ ij Represents the fusion sample x′ i Adaptation path p j The comprehensive scheduling score, μ i represents the scheduling sensitivity coefficient of the fusion sample; S46: Perform path selection operation on each fusion sample and select the corresponding path set P. k has the maximum scheduling score Λ ij Path As the final transmission path, and build the path allocation table ID i is the unique identification number of the fusion sample, c i Indicates the type of fusion sample; S47, according to the type of fusion sample c i , scheduling sensitivity coefficient μ i and network channel load status, calculate the transmission timing interval Δt between fusion samples i , and set the actual scheduling time Control the orderliness and real-time nature of data scheduling; S48. Constructing a path-time scheduling mapping table Send the fused samples to the transmission channel according to the specified path and scheduling time; S49, applying the path-time scheduling mapping table S to the data scheduling execution process, monitoring the usage status, scheduling accuracy and link performance of each path in real time during the data transmission process, and generating a transmission status log record; S410. Collect actual transmission feedback information of the fused sample, including transmission path state changes, scheduling execution errors, and channel congestion, update the state parameters and time scheduling intervals in the path scheduling evaluation function, and complete the adaptive optimization process of the path selection and scheduling timing of the fused sample.

7. The method for dynamic optimization of power data security transmission strategy based on grayscale measurement according to claim 1 is characterized in that: The S5 specifically includes: S51. Collect real-time network status information during the current transmission cycle, including link bandwidth utilization, data transmission delay, packet loss rate, and load level of each path channel; S52, analyzing the transmission urgency and privacy protection level of each fusion sample in the current scheduling list, and extracting the real-time demand parameters of the sample, including the response time window, the remaining privacy budget, and the historical transmission performance; S53. Dynamically adjust the noise intensity in the differential privacy mechanism. Based on the privacy budget consumption status of the sample and the available network bandwidth, moderately reduce the perturbation intensity of high-priority samples while ensuring privacy. S54. Synchronously adjust the transmission weight in the grayscale measurement, dynamically reduce the grayscale priority weight of samples on high-load paths or high-latency nodes, and alleviate network congestion. S55. Based on the sample type and its real-time requirements, a collaborative mapping relationship between privacy perturbation and weight adjustment is constructed to achieve an adaptive balance between the degree of differential privacy perturbation and the priority of grayscale transmission. S56. Apply the adjusted noise parameters and transmission weight feedback to the next round of sample scheduling, path selection, and transmission rhythm control.

Images