Power data intelligent transmission system and method based on multi-source fusion perception

Abstract

The application discloses a kind of power data intelligent transmission system and method based on multi-source fusion perception, it is related to power system monitoring technical field, including the real-time information entropy of analysis key monitoring data flow in sliding time window and the spatial distribution entropy between similar monitoring points;Below steady-state entropy threshold, execute long-period data aggregation and key point sampling based on entropy weight, generate steady-state feature abstract and upload;Exceed transient state entropy threshold or entropy change rate exceeds entropy change rate threshold, execute high-frequency data capture, and according to entropy change trend, send resource pre-scheduling request to main station system;Steady-state feature abstract deviates from expected value and corresponding entropy value is below model deviation entropy threshold, generate accurate data acquisition instruction and issue corresponding edge perception node, make it start high-precision perception to specified monitoring target and report data;When target corresponding entropy value recovers to below steady-state entropy threshold, control edge perception node to restore default transmission strategy, and send resource release request to main station system.

Description

Technical Field

[0001] This invention belongs to the field of power system monitoring technology, specifically relating to an intelligent power data transmission system and method based on multi-source fusion sensing. Background Technology

[0002] With the continuous development of smart grids, the scale of power systems is expanding rapidly, and the deployment of various sensors, smart terminals, and monitoring equipment is growing exponentially. At the same time, the types of sensors are becoming increasingly diversified, covering multiple dimensions such as voltage, current, temperature, vibration, and partial discharge, forming massive, heterogeneous, and high-dimensional real-time monitoring data streams. This data forms the foundation for power system status perception, fault early warning, and operation optimization, and its efficient and reliable transmission is a key link in realizing the intelligent operation of the power grid.

[0003] However, in actual operation, how to efficiently transmit this data faces significant technical challenges. On the one hand, when the power grid is in a steady state, monitoring data often exhibits highly stable and repetitive characteristics. If the traditional fixed-period reporting mode is adopted, a large amount of redundant data will continuously occupy limited communication bandwidth, resulting in a serious waste of resources. On the other hand, when the power grid experiences a fault or enters a transient process, the data changes drastically. At this time, the system is required to capture critical details at a higher frequency. However, if the network is congested due to normal data usage, or if the sampling frequency of sensing nodes is insufficient, critical data can easily be lost or delayed, hindering fault assessment and handling. Therefore, how to effectively reduce the transmission overhead of redundant data while ensuring reliable transmission of critical data has become a core technical problem that urgently needs to be solved in this field.

[0004] To address the aforementioned issues, existing technologies have primarily explored data compression, periodic adjustment, and threshold triggering. A common approach is to employ a fixed-period data reporting strategy. Under this strategy, edge sensing nodes periodically upload collected monitoring data to the main station system at preset time intervals. The advantage of this method is its simplicity and clear logic, but its disadvantage lies in its inability to adapt to the dynamic changes in the data itself. During steady-state operation of the power grid, a large amount of repetitive and stable data still consumes transmission resources, while during transient changes, a fixed sampling frequency is insufficient to guarantee complete capture of rapidly changing processes.

[0005] Another common approach is to introduce a simple threshold triggering mechanism. Specifically, the system sets one or more numerical thresholds for the monitored data. When the monitored value exceeds the threshold range, a high-frequency data report or alarm is triggered; when the monitored value is within the threshold range, the reporting frequency is reduced or only statistical values ​​are reported. This method reduces the amount of steady-state data transmitted to some extent, but its judgment relies solely on the magnitude of a single data point, lacking a comprehensive assessment of the overall data change trend and uncertainty. For example, when data drifts slowly but never exceeds the threshold, the system may fail to detect its abnormal trend, leading to missed detections of potential faults.

[0006] Furthermore, some existing technologies attempt to reduce the amount of data transmitted through data compression or filtering algorithms. For example, differential coding and wavelet transform are used to compress the original data before transmission, or Kalman filtering and other algorithms are used to smooth the data. While these methods can effectively reduce the amount of data transmitted, they are often indiscriminate processing and fail to differentiate the information value of the data. During compression, subtle fluctuation characteristics that are important for fault diagnosis may be lost. At the same time, the above-mentioned existing solutions generally lack the ability to intelligently interact with upper-level applications such as digital twin models, making it difficult to form a closed-loop optimization from perception to decision-making, resulting in a disconnect between the adjustment of transmission strategies and the actual operating state of the power grid.

[0007] In summary, while existing technologies have alleviated data transmission pressure to some extent, they still have the following fundamental shortcomings: First, there is a lack of dynamic assessment capabilities for the value of data information. Existing methods mostly use fixed periods or simple numerical thresholds as the basis for adjusting transmission strategies, failing to quantify the uncertainty and information content of data streams in the time and space dimensions from an information theory perspective. This leads to the continuous transmission of a large amount of data with high information redundancy during steady-state operation, resulting in inefficient use of communication resources; while during transient changes, the sampling strategy may fail to respond in time, leading to the loss of critical data with high information value.

[0008] Second, there is a disconnect between transmission decisions and the actual operating status of the power grid. Data transmission in existing solutions is typically isolated, failing to organically link data acquisition from edge sensing nodes with status verification and resource scheduling in the main station system. In particular, when monitoring data exhibits slow numerical drift while the data sequence itself remains stable, existing threshold triggering mechanisms struggle to effectively identify this, thus missing the optimal window for early warning.

[0009] Third, there is a lack of resource guarantee mechanisms for high-value data streams. When transient anomalies occur in the power grid, it is often necessary to transmit a large amount of high-frequency collected data in a short period of time. However, existing technologies fail to pre-schedule network resources based on data change trends before the anomaly occurs, which may lead to network congestion affecting real-time data transmission at critical moments.

[0010] Therefore, it is necessary to design an intelligent power data transmission system and method based on multi-source fusion sensing to solve the above problems. Summary of the Invention

[0011] The purpose of this invention is to provide a power data intelligent transmission system and method based on multi-source fusion sensing, which aims to fundamentally solve the contradiction between low data transmission value density and weak critical event protection capability, and improve the efficiency, reliability and intelligence level of power system data transmission.

[0012] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for intelligent transmission of power data based on multi-source fusion sensing includes the following steps: S1, based on the results of multi-source fusion sensing, calculate the real-time information entropy of key monitoring data streams within a sliding time window, and calculate the spatial distribution entropy among similar monitoring points; the results of multi-source fusion sensing refer to the structured monitoring data streams obtained after spatiotemporal alignment and normalization of raw data collected by at least two types of sensors; the key monitoring data streams refer to the monitoring data sequences reflecting the status of core equipment selected according to preset rules. S2, based on the real-time information entropy value and the trend of information entropy value change, when the real-time information entropy value is lower than the steady-state entropy threshold, long-cycle data aggregation and key point sampling based on entropy weight are performed to generate a steady-state feature summary and upload it; when the real-time information entropy value exceeds the transient entropy threshold or the real-time information entropy change rate exceeds the entropy change rate threshold, high-frequency data capture is performed, and a resource pre-scheduling request is sent to the main station system according to the entropy change trend. S3: Based on the pre-built digital twin model, obtain the expected value, compare the received steady-state feature summary with the expected value of the digital twin model, and if the steady-state feature summary deviates from the expected value and the corresponding real-time information entropy value is lower than the model deviation entropy threshold, then generate a precise data acquisition instruction. S4 sends the precise data acquisition command to the corresponding edge sensing node, enabling the edge sensing node to start high-precision sensing for the monitoring target specified by the command and report the generated sensing data. S5, when the real-time information entropy value corresponding to the monitored target recovers to below the steady-state entropy threshold, the control edge sensing node restores the default transmission strategy and sends a resource release request to the main station system.

[0013] Furthermore, the entropy weight refers to the quantized weight value assigned to each data point within a sliding time window or long-period data aggregation window, based on the probability density of the value interval of each data point and its contribution to the overall information entropy of the window. The entropy weight is used to identify key data points with high information value in a steady-state data stream. The calculation method for the entropy weight is as follows: Suppose the range of values ​​in the data sequence within the window is divided into n intervals, and the probability density of the i-th interval is... The information entropy contribution value corresponding to this interval is For any data point x falling into the i-th interval, its entropy weight Defined as: ; That is, the entropy contribution value corresponding to this interval.

[0014] Preferably, step S1 includes: S110, based on the results of multi-source fusion sensing, for a selected single monitoring data sequence, within a preset sliding time window, calculates the probability distribution characteristics of the data sequence, and determines the real-time information entropy value within this window according to the information entropy formula. The real-time information entropy value is used to characterize the uncertainty and information content of the data sequence within this time period. The formula is: ; Where H is the real-time information entropy value; n is the total number of data value intervals; This represents the probability density of the range of values ​​for the i-th data point; To sum the calculation results for all n data value intervals; The single monitoring data sequence is a continuous data sequence of the same type and the same monitoring dimension extracted from the key monitoring data stream. The extraction process must maintain the temporal continuity and integrity of the data. The probability distribution characteristics are obtained by statistically analyzing the value distribution of the data sequence within a sliding time window, specifically including the interval division of data values ​​and the proportion of data samples in each interval. S120, within a time window or the same aggregation period, based on fused sensing data from multiple monitoring points of the same category at different spatial locations, calculate the discrete distribution characteristics of the data in the spatial dimension; in a preferred embodiment, the discrete distribution characteristics in the spatial dimension include dividing the monitoring area into spatial grids of fixed size, statistically analyzing the number of monitoring points and the data value distribution within each grid; and determining the spatial distribution entropy value based on the information entropy formula. ; in, The value represents the spatial distribution entropy; m represents the total number of spatial grids in the monitoring area. This represents the probability of the j-th spatial grid, which is the ratio of the number of similar monitoring points in this grid to the total number of all similar monitoring points. This represents the summation of the calculation results for all m spatial grids; the spatial distribution entropy value is used to characterize the degree of spatial consistency of the regional state reflected by the multiple monitoring points.

[0015] Preferably, step S2 includes: S210, If the real-time information entropy value is lower than the steady-state entropy threshold, then perform long-cycle data aggregation, which includes extending the time window length of data aggregation; S220, during the long-cycle data aggregation process, based on the distribution of real-time information entropy values ​​within the time window, key data points with entropy values ​​higher than the local key entropy threshold are identified and extracted; a steady-state feature summary is constructed from the key data points, and the steady-state feature summary is uploaded; S230, if the real-time information entropy value exceeds the transient entropy threshold, or the real-time information entropy change rate exceeds the entropy change rate threshold, then it is determined that the high-frequency data capture condition is met; the real-time information entropy change rate is calculated by the ratio of the difference between the real-time information entropy values ​​within two adjacent sliding time windows to the time interval, and the formula is: ; Where r represents the rate of change of real-time information entropy. This represents the real-time information entropy value of the k-th sliding time window. This represents the real-time information entropy value of the (k-1)th sliding time window. This represents the time interval between two adjacent sliding time windows; the determination of meeting the high-frequency data capture conditions adopts OR logic. If the real-time information entropy value exceeds the standard or the entropy change rate exceeds the standard, and at least one condition is met, high-frequency data capture is triggered to ensure a rapid response to abnormal data changes; the setting of the transient entropy threshold and the entropy change rate threshold needs to be combined with the characteristics of transient faults in the power system to ensure that transient anomalies can be accurately identified while avoiding false triggering. S240, when the high-frequency data capture condition is met, perform high-frequency data capture, which includes shortening the sampling interval or pausing data aggregation; S250: Based on the trend of change of the real-time information entropy value before it exceeds the transient entropy threshold, predict the data traffic in the future time period, and send a resource pre-scheduling request containing the predicted data traffic to the main station system.

[0016] Furthermore, the "local key entropy threshold" refers to the screening threshold determined based on the statistical distribution of the entropy weights of all data points within a long-term data aggregation window, used to distinguish between key data points and non-key data points.

[0017] Furthermore, the trend of change is obtained through linear fitting or slope calculation.

[0018] Preferably, step S3 includes: S310, based on the steady-state feature summary, obtain the digital twin model of the monitoring target corresponding to the steady-state feature summary, and extract the expected value or expected value range of the digital twin model under the current operating conditions; S320, compare the key data features in the steady-state feature summary with the expected value or expected value range, and determine whether the key data features deviate from the expected value or exceed the expected value range; S330, when a key data feature is determined to deviate from the expected value or exceed the expected value range, query the real-time information entropy value of the monitoring target within the time window for generating the steady-state feature summary; S340, if the real-time information entropy value obtained is lower than the model deviation entropy threshold, it is determined that the deviation of the steady-state feature summary belongs to the latent deviation in the steady state, and a precise data acquisition instruction containing the monitoring target identifier and high-precision perception requirements is generated.

[0019] Preferably, step S4 includes: S410 parses the precise data acquisition command, obtains the monitoring target identifier and high-precision perception requirements specified in the command; and sends the precise data acquisition command to the edge perception node responsible for the monitoring target based on the monitoring target identifier. S420 After receiving the instruction, the edge sensing node configures the sensing parameters for the monitored target according to the high-precision sensing requirements. The configuration of sensing parameters includes, but is not limited to, increasing the sampling frequency, switching the sensing mode, or enabling backup sensing resources. In a preferred embodiment, the edge sensing node is configured with multiple sensors or sensors with different precision modes, and the configuration of sensing parameters includes switching to a higher precision sensor or enabling a backup sensing unit. S430, the edge sensing node initiates high-precision data sensing for the monitored target based on the configured sensing parameters; S440: The edge sensing node encapsulates the data generated during the high-precision sensing process into sensing data messages and reports them to the main station.

[0020] Preferably, step S5 includes: S510, continuously monitor the high-precision sensing data stream corresponding to the monitoring target, and calculate its real-time information entropy value within the sliding time window; S520: When the real-time information entropy value recovers to below the steady-state entropy threshold and the duration reaches the preset value, it is determined that the state of the monitored target has recovered to a steady state. The preset value is the preset steady-state determination duration, which is preset based on the device response time or historical steady-state recovery time statistics. S530, based on the judgment result, sends a policy reset command to the edge sensing node. The policy reset command is used to instruct the edge sensing node to stop the current high-precision sensing mode and restore the mode of dynamically selecting the data transmission strategy based on the real-time information entropy value. S540 sends a resource release request to the master station system. The resource release request is used to instruct the master station system to reclaim communication resources that have been pre-scheduled or reserved for the monitoring target.

[0021] Preferably, a power data intelligent transmission system based on multi-source fusion sensing is provided for executing the power data intelligent transmission method based on multi-source fusion sensing. The system includes: The entropy assessment module is used to calculate the real-time information entropy of key monitoring data streams within a sliding time window based on the results of multi-source fusion sensing, and to calculate the spatial distribution entropy among similar monitoring points; based on the real-time information entropy value and its changing trend, it assesses the real-time value and urgency of data transmission. The transmission decision module is used to dynamically select and execute the data transmission strategy based on the output of the entropy evaluation module, and to initiate resource pre-scheduling when the resource pre-scheduling conditions are met. In the preferred scheme, resource pre-scheduling is initiated to the main station system when the resource pre-scheduling conditions are met, including when a high-value data flow is predicted. The instruction generation module compares the received steady-state feature summary with the expected value of the digital twin model, and generates accurate data acquisition instructions when a latent deviation in the steady state is identified.

[0022] Preferably, the entropy evaluation module includes: The real-time entropy unit is used to calculate the probability distribution characteristics of a selected single monitoring data sequence within a sliding time window of a preset length, based on the results of multi-source fusion sensing, and to determine the real-time information entropy value within the window according to the information entropy formula. The spatial entropy unit is used to calculate the discrete distribution characteristics of the data in the spatial dimension based on fused sensing data from multiple monitoring points of the same category in different spatial locations within a time window or the same aggregation period, and to determine the spatial distribution entropy value according to the information entropy formula.

[0023] Preferably, the transmission decision module includes: The summary generation unit is used to perform long-cycle data aggregation when the real-time information entropy value is lower than the steady-state entropy threshold, and to identify key data points based on entropy weights to form a steady-state feature summary. The pre-scheduling unit is used to perform high-frequency data capture and send a resource pre-scheduling request to the main station system according to the entropy trend when the real-time information entropy value exceeds the transient entropy threshold or its rate of change exceeds the entropy change rate threshold.

[0024] Preferably, the instruction generation module includes: The deviation determination unit is used to compare the key data features in the steady-state feature summary with the expected values ​​of the digital twin model and query the corresponding real-time information entropy value. The precision instruction unit is used to generate precise data acquisition instructions that include the monitoring target identifier and high-precision perception requirements when key data features deviate from the expected value and the real-time information entropy value is lower than the model deviation entropy threshold.

[0025] Furthermore, it also includes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the intelligent power data transmission method based on multi-source fusion sensing.

[0026] Furthermore, it also includes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the described intelligent power data transmission method based on multi-source fusion sensing.

[0027] The beneficial effects of the intelligent power data transmission system and method based on multi-source fusion sensing provided by this invention are as follows: 1. This invention introduces information entropy as a core metric to automatically identify the steady and transient periods of data flow; it dynamically matches network bandwidth usage with the actual information value of the data, thereby improving bandwidth utilization. 2. This invention predicts bandwidth demand in advance and performs resource pre-scheduling by analyzing entropy change trends, thus avoiding network congestion and delays during critical data transmission; and improves effective information density and decision response speed by performing secondary verification of steady-state summaries through a digital twin model. 3. This invention deeply integrates the data value perception, dynamic aggregation, and network pre-scheduling of edge sensing nodes with the model verification and command issuance on the master station side, making the power grid operation safer, smarter, and more stable. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating the steps of the intelligent power data transmission method based on multi-source fusion sensing of the present invention. Figure 2 This is a schematic diagram of the intelligent power data transmission system based on multi-source fusion sensing according to the present invention; Figure 3 This is a schematic diagram of the computer device structure in an embodiment of the present invention. Detailed Implementation

[0029] Example 1: like Figure 1 As shown, a method for intelligent transmission of power data based on multi-source fusion sensing includes the following steps: S1, based on the results of multi-source fusion sensing, calculate the real-time information entropy of key monitoring data streams within a sliding time window, and calculate the spatial distribution entropy among similar monitoring points; the results of multi-source fusion sensing refer to the structured monitoring data streams obtained after spatiotemporal alignment and normalization of raw data collected by at least two types of sensors; the key monitoring data streams refer to the monitoring data sequences reflecting the status of core equipment selected according to preset rules. S2, based on the real-time information entropy value and the trend of information entropy value change, when the real-time information entropy value is lower than the steady-state entropy threshold, long-cycle data aggregation and key point sampling based on entropy weight are performed to generate a steady-state feature summary and upload it; when the real-time information entropy value exceeds the transient entropy threshold or the real-time information entropy change rate exceeds the entropy change rate threshold, high-frequency data capture is performed, and a resource pre-scheduling request is sent to the main station system according to the entropy change trend. S3: Based on the pre-built digital twin model, obtain the expected value, compare the received steady-state feature summary with the expected value of the digital twin model, and if the steady-state feature summary deviates from the expected value and the corresponding real-time information entropy value is lower than the model deviation entropy threshold, then generate a precise data acquisition instruction. S4 sends the precise data acquisition command to the corresponding edge sensing node, enabling the edge sensing node to start high-precision sensing for the monitoring target specified by the command and report the generated sensing data. S5, when the real-time information entropy value corresponding to the monitored target recovers to below the steady-state entropy threshold, the control edge sensing node restores the default transmission strategy and sends a resource release request to the main station system.

[0030] Furthermore, the entropy weight refers to the quantized weight value assigned to each data point within a sliding time window or long-period data aggregation window, based on the probability density of the value interval of each data point and its contribution to the overall information entropy of the window. The entropy weight is used to identify key data points with high information value in a steady-state data stream. The calculation method for the entropy weight is as follows: Suppose the range of values ​​in the data sequence within the window is divided into n intervals, and the probability density of the i-th interval is... The information entropy contribution value corresponding to this interval is For any data point x falling into the i-th interval, its entropy weight Defined as: ; That is, the entropy contribution value corresponding to this interval.

[0031] Furthermore, in this embodiment, the multi-source fusion sensing result refers to: Within the power system monitoring area, raw data is collected by multiple sensing devices, including at least voltage sensors, current sensors, and temperature sensors. The raw data is then processed by edge sensing nodes or data aggregation units through timestamp alignment, spatial coordinate association, dimensional normalization, and outlier removal, resulting in a structured monitoring data stream with a unified data format and spatiotemporal labels.

[0032] Preferably, step S1 includes: S110, based on the results of multi-source fusion sensing, for a selected single monitoring data sequence, within a preset sliding time window, calculates the probability distribution characteristics of the data sequence, and determines the real-time information entropy value within this window according to the information entropy formula. The real-time information entropy value is used to characterize the uncertainty and information content of the data sequence within this time period. The formula is: ; Where H is the real-time information entropy value; n is the total number of data value intervals; This represents the probability density of the range of values ​​for the i-th data point; To sum the calculation results for all n data value intervals; The single monitoring data sequence is a continuous data sequence of the same type and the same monitoring dimension extracted from the key monitoring data stream. The extraction process must maintain the temporal continuity and integrity of the data. The probability distribution characteristics are obtained by statistically analyzing the value distribution of the data sequence within a sliding time window, specifically including the interval division of data values ​​and the proportion of data samples in each interval. S120, within a time window or the same aggregation period, based on fused sensing data from multiple monitoring points of the same category at different spatial locations, calculate the discrete distribution characteristics of the data in the spatial dimension; in a preferred embodiment, the discrete distribution characteristics in the spatial dimension include dividing the monitoring area into spatial grids of fixed size, statistically analyzing the number of monitoring points and the data value distribution within each grid; and determining the spatial distribution entropy value based on the information entropy formula. ; in, The value represents the spatial distribution entropy; m represents the total number of spatial grids in the monitoring area. This represents the probability of the j-th spatial grid, which is the ratio of the number of similar monitoring points in this grid to the total number of all similar monitoring points. This represents the summation of the calculation results for all m spatial grids; the spatial distribution entropy value is used to characterize the degree of spatial consistency of the regional state reflected by the multiple monitoring points.

[0033] Preferably, step S2 includes: S210, If the real-time information entropy value is lower than the steady-state entropy threshold, then perform long-cycle data aggregation, which includes extending the time window length of data aggregation; S220, during the long-cycle data aggregation process, based on the distribution of real-time information entropy values ​​within the time window, key data points with entropy values ​​higher than the local key entropy threshold are identified and extracted; a steady-state feature summary is constructed from the key data points, and the steady-state feature summary is uploaded; S230, if the real-time information entropy value exceeds the transient entropy threshold, or the real-time information entropy change rate exceeds the entropy change rate threshold, then it is determined that the high-frequency data capture condition is met; the real-time information entropy change rate is calculated by the ratio of the difference between the real-time information entropy values ​​within two adjacent sliding time windows to the time interval, and the formula is: ; Where r represents the rate of change of real-time information entropy. This represents the real-time information entropy value of the k-th sliding time window. This represents the real-time information entropy value of the (k-1)th sliding time window. This represents the time interval between two adjacent sliding time windows; the determination of meeting the high-frequency data capture conditions adopts OR logic. If the real-time information entropy value exceeds the standard or the entropy change rate exceeds the standard, and at least one condition is met, high-frequency data capture is triggered to ensure a rapid response to abnormal data changes; the setting of the transient entropy threshold and the entropy change rate threshold needs to be combined with the characteristics of transient faults in the power system to ensure that transient anomalies can be accurately identified while avoiding false triggering. S240, when the high-frequency data capture condition is met, perform high-frequency data capture, which includes shortening the sampling interval or pausing data aggregation; S250: Based on the trend of change of the real-time information entropy value before it exceeds the transient entropy threshold, predict the data traffic in the future time period, and send a resource pre-scheduling request containing the predicted data traffic to the main station system.

[0034] Furthermore, the "local key entropy threshold" refers to the screening threshold determined based on the statistical distribution of the entropy weights of all data points within a long-term data aggregation window, used to distinguish between key data points and non-key data points.

[0035] Furthermore, the trend of change is obtained through linear fitting or slope calculation.

[0036] Preferably, step S3 includes: S310, based on the steady-state feature summary, obtain the digital twin model of the monitoring target corresponding to the steady-state feature summary, and extract the expected value or expected value range of the digital twin model under the current operating conditions; S320, compare the key data features in the steady-state feature summary with the expected value or expected value range, and determine whether the key data features deviate from the expected value or exceed the expected value range; S330, when a key data feature is determined to deviate from the expected value or exceed the expected value range, query the real-time information entropy value of the monitoring target within the time window for generating the steady-state feature summary; S340, if the real-time information entropy value obtained is lower than the model deviation entropy threshold, it is determined that the deviation of the steady-state feature summary belongs to the latent deviation in the steady state, and a precise data acquisition instruction containing the monitoring target identifier and high-precision perception requirements is generated.

[0037] Preferably, step S4 includes: S410 parses the precise data acquisition command, obtains the monitoring target identifier and high-precision perception requirements specified in the command; and sends the precise data acquisition command to the edge perception node responsible for the monitoring target based on the monitoring target identifier. S420 After receiving the instruction, the edge sensing node configures the sensing parameters for the monitored target according to the high-precision sensing requirements. The configuration of sensing parameters includes, but is not limited to, increasing the sampling frequency, switching the sensing mode, or enabling backup sensing resources. In a preferred embodiment, the edge sensing node is configured with multiple sensors or sensors with different precision modes, and the configuration of sensing parameters includes switching to a higher precision sensor or enabling a backup sensing unit. S430, the edge sensing node initiates high-precision data sensing for the monitored target based on the configured sensing parameters; S440: The edge sensing node encapsulates the data generated during the high-precision sensing process into sensing data messages and reports them to the main station.

[0038] Preferably, step S5 includes: S510, continuously monitor the high-precision sensing data stream corresponding to the monitoring target, and calculate its real-time information entropy value within the sliding time window; S520: When the real-time information entropy value recovers to below the steady-state entropy threshold and the duration reaches the preset value, it is determined that the state of the monitored target has recovered to a steady state. The preset value is the preset steady-state determination duration, which is preset based on the device response time or historical steady-state recovery time statistics. S530, based on the judgment result, sends a policy reset command to the edge sensing node. The policy reset command is used to instruct the edge sensing node to stop the current high-precision sensing mode and restore the mode of dynamically selecting the data transmission strategy based on the real-time information entropy value. S540 sends a resource release request to the master station system. The resource release request is used to instruct the master station system to reclaim communication resources that have been pre-scheduled or reserved for the monitoring target.

[0039] like Figure 2 As shown, a power data intelligent transmission system based on multi-source fusion sensing is used to execute the power data intelligent transmission method based on multi-source fusion sensing. The system includes: The entropy assessment module is used to calculate the real-time information entropy of key monitoring data streams within a sliding time window based on the results of multi-source fusion sensing, and to calculate the spatial distribution entropy among similar monitoring points; based on the real-time information entropy value and its changing trend, it assesses the real-time value and urgency of data transmission. The transmission decision module is used to dynamically select and execute the data transmission strategy based on the output of the entropy evaluation module, and to initiate resource pre-scheduling when the resource pre-scheduling conditions are met. In the preferred scheme, resource pre-scheduling is initiated to the main station system when the resource pre-scheduling conditions are met, including when a high-value data flow is predicted. The instruction generation module compares the received steady-state feature summary with the expected value of the digital twin model, and generates accurate data acquisition instructions when a latent deviation in the steady state is identified.

[0040] Preferably, the entropy evaluation module includes: The real-time entropy unit is used to calculate the probability distribution characteristics of a selected single monitoring data sequence based on the results of multi-source fusion sensing, within a sliding time window of a preset length, and determine the real-time information entropy value within the window according to the information entropy formula. The spatial entropy unit is used to calculate the discrete distribution characteristics of the data in the spatial dimension based on fused sensing data from multiple monitoring points of the same category at different spatial locations within a time window or the same aggregation period, and to determine the spatial distribution entropy value according to the information entropy formula.

[0041] Preferably, the transmission decision module includes: The summary generation unit is used to perform long-cycle data aggregation when the real-time information entropy value is lower than the steady-state entropy threshold, and to identify key data points based on entropy weights to form a steady-state feature summary. The pre-scheduling unit is used to perform high-frequency data capture and send a resource pre-scheduling request to the main station system according to the entropy trend when the real-time information entropy value exceeds the transient entropy threshold or its rate of change exceeds the entropy change rate threshold.

[0042] Preferably, the instruction generation module includes: The deviation determination unit is used to compare the key data features in the steady-state feature summary with the expected values ​​of the digital twin model and query the corresponding real-time information entropy value. The precision instruction unit is used to generate precise data acquisition instructions that include the monitoring target identifier and high-precision perception requirements when key data features deviate from the expected value and the real-time information entropy value is lower than the model deviation entropy threshold.

[0043] Example 2: This embodiment provides a technical solution: a power data intelligent transmission method based on multi-source fusion sensing, the method comprising: S100. Based on the results of multi-source fusion sensing, calculate the real-time information entropy of key monitoring data streams within the sliding time window, and calculate the spatial distribution entropy among similar monitoring points. Specifically, step S100 includes: S110. Based on the results of multi-source fusion sensing, for a selected single monitoring data sequence, which is a continuous data sequence of the same type and the same monitoring dimension extracted from the key monitoring data stream, the extraction process must maintain the temporal continuity and integrity of the data; within a preset sliding time window, the probability distribution characteristics of the data sequence are calculated. The probability distribution characteristics are obtained by statistically analyzing the value distribution of the data sequence within the sliding time window, specifically including the interval division of data values ​​and the proportion of data samples in each interval; the real-time information entropy value within the window is determined according to the information entropy formula. The information entropy formula adopts the Shannon entropy calculation model, and the formula is: ; Where H is the real-time information entropy value; n is the total number of data value intervals; This represents the probability density of the range of values ​​for the i-th data point; To sum the calculation results for all n data value intervals; Real-time information entropy is used to characterize the uncertainty and information content of a data sequence within a given time period. For example, within a time window t1 to t10, i.e., 10 minutes, the temperature sequence reported by temperature measurement point 1 is [25.1, 25.0, 25.2, 24.9, 25.1, 25.0, 25.3, 24.8, 25.2, 25.1]. The temperature range is divided into intervals, such as [24.5, 24.7), [24.7, 24.9), ..., [25.3, 25.5). The number of data points in each interval is counted, and the probability is calculated. Substituting into the entropy formula, the real-time information entropy value of the window is calculated. Since the real-time information entropy value is less than the steady-state entropy threshold, the system initially determines that the data stream is in a stable state. S120. Within a time window or the same aggregation period, based on fused sensing data from multiple monitoring points of the same category at different spatial locations, calculate the discrete distribution characteristics of this data in the spatial dimension. The discrete distribution characteristics in the spatial dimension include dividing the monitoring area into spatial grids of fixed size and statistically analyzing the number of monitoring points and the data value distribution within each grid. The spatial distribution entropy is calculated using the Shannon entropy calculation model, with the following formula: ; in, The value represents the spatial distribution entropy; m represents the total number of spatial grids in the monitoring area. This represents the probability of the j-th spatial grid, which is the ratio of the number of similar monitoring points in this grid to the total number of all similar monitoring points. This represents the summation of the calculation results for all m spatial grids; the spatial distribution entropy value is used to characterize the degree of spatial consistency of the regional state reflected by the multiple monitoring points.

[0044] For example: at the same time t10, the temperature values ​​of the 5 temperature measurement points are [25.1, 25.5, 24.8, 25.9, 25.2]; the line area is divided into 3 grids, and the distribution of temperature points in each grid is statistically analyzed; assuming the distribution probability is q = [0.4, 0.4, 0.2]; substituting into the spatial entropy formula, the result is... ; S200. Based on the real-time information entropy value and its changing trend, when the real-time information entropy value is lower than the steady-state entropy threshold, long-cycle data aggregation and key point sampling based on entropy weight are performed. Entropy weight refers to the weight assigned to a data point based on its information entropy contribution within a time window; the higher the entropy value, the greater the weight, used to identify key points with high information content in steady-state data. A steady-state feature summary is generated and uploaded. When the real-time information entropy value exceeds the transient entropy threshold or the real-time information entropy change rate exceeds the entropy change rate threshold, high-frequency data capture is performed, and a resource pre-scheduling request is sent to the main station system based on the entropy change trend. The changing trend of the real-time information entropy value is obtained by fitting and analyzing the entropy value calculation results within multiple consecutive sliding time windows, used to characterize key points. The system monitors the fluctuation patterns and changing trends of key data streams; steady-state entropy thresholds, transient entropy thresholds, and entropy change rate thresholds are pre-set based on historical power system operating data and safe operation standards, and calibrated in conjunction with simulation analysis results; key point sampling based on entropy weight refers to quantifying the entropy weight of a data point by calculating its contribution to the overall information entropy within a sliding time window, specifically the product of the probability density of the data interval where the data point is located and its information content; in a long-period aggregation window, data points with entropy weights higher than the weights corresponding to the local key entropy thresholds are selected as key data points constituting the steady-state feature summary; resource pre-scheduling requests are used to apply to the master station system for communication resources adapted to subsequent high-frequency data transmission, ensuring the transmission stability after high-frequency data acquisition. Specifically, step S200 includes: S210. If the real-time information entropy value is lower than the steady-state entropy threshold, then long-cycle data aggregation is performed. Long-cycle data aggregation includes extending the time window length of data aggregation. The extension of the time window length of long-cycle data aggregation is determined based on the difference between the steady-state entropy threshold and the real-time information entropy value. The larger the difference, the larger the extension, ensuring that in steady-state scenarios with low data uncertainty, resource consumption is reduced by reducing the frequency of data transmission. For example: Since the real-time information entropy value is less than the steady-state entropy threshold, the steady-state processing branch is entered; long-cycle aggregation is executed; the difference between the steady-state entropy threshold and the real-time information entropy value is 0.2, which is small, so the aggregation window is extended to 30 minutes; this is only an example and is not a limitation. S220. During long-cycle data aggregation, based on the distribution of real-time information entropy values ​​within a time window, key data points with entropy values ​​higher than the local key entropy threshold are identified and extracted; a steady-state feature summary is constructed from the key data points and uploaded; the local key entropy threshold is determined based on the distribution characteristics of real-time information entropy values ​​within the long-cycle data aggregation time window, specifically the combined calculation result of the mean and standard deviation of the entropy value distribution within the time window, used to distinguish between key data points and non-key data points; For example: within a 30-minute aggregation window, calculate the real-time entropy value sequence for each minute; if the mean is... Standard deviation If the local key entropy threshold is 0.4, then identify the original temperature data points corresponding to the entropy values ​​higher than 0.4, such as the 5th, 15th, and 25th minutes, and use them as key data points; construct a steady-state feature summary {timestamp: [t5, t15, t25], temperature value: 25.2, 24.9, 25.5]} from these points and upload it to the main site; S230. If the real-time information entropy value exceeds the transient entropy threshold, or the real-time information entropy change rate exceeds the entropy change rate threshold, then it is determined that the high-frequency data capture condition is met. The real-time information entropy change rate is calculated by the ratio of the difference between the real-time information entropy values ​​within two adjacent sliding time windows to the time interval, as shown in the formula: Where r represents the rate of change of real-time information entropy, H k H represents the real-time information entropy value of the k-th sliding time window. k-1 This represents the real-time information entropy value of the (k-1)th sliding time window. The time interval between two adjacent sliding time windows is represented. The judgment of meeting the high-frequency data capture conditions adopts OR logic, that is, as long as the real-time information entropy value exceeds the standard or the entropy change rate exceeds the standard, high-frequency data capture is triggered to achieve a rapid response to abnormal data changes. The setting of transient entropy threshold and entropy change rate threshold is combined with the characteristics of transient faults in the power system to achieve accurate capture of transient data changes and avoid false triggering. S240. When the conditions for high-frequency data capture are met, high-frequency data capture is performed. High-frequency data capture includes shortening the sampling interval or pausing data aggregation. The extent to which the sampling interval is shortened is determined based on the magnitude by which the real-time information entropy value exceeds the transient entropy threshold or the magnitude by which the entropy change rate exceeds the entropy change rate threshold. The greater the magnitude of the exceedance, the shorter the sampling interval, ensuring accurate capture of transient data changes. Pausing data aggregation means stopping the execution of long-cycle data aggregation operations and using the method of real-time acquisition of raw data to avoid the loss of transient data information caused by the aggregation process. For example, shortening the temperature sampling interval from 1 minute to 10 seconds, pausing aggregation, and directly reporting the raw data. S250. Based on the trend of real-time information entropy value before it exceeds the transient entropy threshold, predict the data traffic in the future time period and send a resource pre-scheduling request containing the predicted data traffic to the main station system. The duration of the future time period is determined based on the fitting result of the entropy change trend, which is the time required for the predicted data traffic to stabilize or reach the network capacity limit. The data traffic prediction adopts a trend fitting algorithm, using the entropy value data and corresponding data traffic of multiple time windows before the real-time information entropy value exceeds the transient entropy threshold as input, to establish a mapping relationship between entropy value and data traffic, and obtain the predicted data traffic. For example, analyzing the upward trend of entropy value before it exceeds 2.0, predicting that the data traffic will increase by 5 times in the next 2 minutes, and sending a pre-scheduling request to the main station system. This is only an example and is not a limitation.

[0045] S300. Compare the received steady-state feature summary with the expected value of the digital twin model. If the steady-state feature summary deviates from the expected value and the corresponding real-time information entropy value is lower than the model deviation entropy threshold, then generate a precise data acquisition instruction. Specifically, step S300 includes: S310. Based on the steady-state feature summary, obtain the digital twin model of the monitoring target corresponding to the steady-state feature summary, and extract the expected value or expected value range of the digital twin model under the current operating conditions. The correspondence between the steady-state feature summary and the monitoring target is established through the monitoring target identifier carried in the summary. After receiving the steady-state feature summary, the master station retrieves the corresponding monitoring target digital twin model from the digital twin model library based on the identifier. The current operating conditions are determined by the power system operating parameters loaded in real time by the digital twin model. The expected value or expected value range is calculated by the digital twin model based on the current operating conditions, combined with historical steady-state operating data, equipment technical parameters and system safety operation standards, and is used to characterize the data value standard of the monitoring target under normal operating conditions. S320. Compare the key data features in the steady-state feature summary with the expected values ​​or expected value ranges to determine whether the key data features deviate from the expected values ​​or exceed the expected value range. The comparison employs a dimensional comparison approach. For each key data feature in the steady-state feature summary, the deviation from the expected value of the digital twin model is calculated for each monitoring dimension, or it is determined whether the deviation falls within the expected value range. The deviation is calculated using the absolute value of the difference between the key data feature value and the expected value, and its mathematical expression is as follows: ;in, This represents the deviation, and x represents the key data feature value. This represents the expected value of the digital twin model; when the deviation exceeds a preset deviation threshold, it is determined to deviate from the expected value; for the expected value range... ,in, This is the lower limit of the expected value. The upper limit of the expected value is the value of the key data feature. When this occurs, it is determined to be outside the expected range; S330. When a key data feature is determined to deviate from or exceed the expected value range, query the real-time information entropy value of the monitoring target within the time window for generating the steady-state feature summary. The query of the real-time information entropy value is achieved through the data acquisition timestamp and monitoring target identifier in the steady-state feature summary. The main station retrieves the real-time information entropy value within the corresponding time period from the historical entropy value database based on the timestamp and identifier. During the query process, it is necessary to ensure that the retrieved real-time information entropy value is completely matched with the generation time window of the steady-state feature summary to avoid judgment errors caused by inconsistent time dimensions. S340. If the real-time information entropy value found is lower than the model deviation entropy threshold, the deviation of the steady-state feature summary is determined to be a latent deviation in a steady state, and a precise data acquisition instruction containing the monitoring target identifier and high-precision perception requirements is generated. The model deviation entropy threshold is a comprehensive preset based on the information entropy statistical distribution of the monitoring data stream under historical normal operation data and the prediction error tolerance range of the digital twin model under the corresponding steady-state operating conditions. It is used to distinguish between latent deviation and explicit deviation. When the real-time information entropy value is lower than this threshold, it indicates that although the monitoring target data deviates from the expectation, the data stream itself is still in a steady state, and is therefore determined to be a latent deviation. For example: The main station receives a steady-state feature summary {temperature value: [25.2, 24.9, 25.5]}; based on the identified line AX phase, it calls its digital twin model; the model calculates the expected range of conductor temperature under the current operating condition as [24.5, 25.5]; the key data features in the summary are compared with the expected range [24.5, 25.5]; 25.5 is on the boundary; The query records the corresponding historical real-time entropy values ​​within the 30-minute window that generated the summary. For example, if the query results are [0.3, 0.35, 0.28, ...], they are all less than the model deviation entropy threshold. Since 25.5 reaches the expected upper limit and the corresponding entropy value is always lower than 0.8, it is judged as a latent deviation in a stable state. Generate a precise data acquisition instruction: {Target: Line AX phase; Requirement: Temperature; Accuracy: ±0.1°C; Frequency: 1 time / second; Duration: 5 minutes}. This is only an example and no restrictions are imposed.

[0046] S400: Send the precise data acquisition command to the corresponding edge sensing node, so that the edge sensing node can start high-precision sensing for the monitoring target specified by the command and report the generated sensing data. Specifically, step S400 includes: S410. Parse the precise data acquisition command to obtain the monitoring target identifier and high-precision sensing requirements specified in the command; based on the monitoring target identifier, send the precise data acquisition command to the edge sensing node responsible for the monitoring target; the command parsing process includes decrypting the command, format parsing, and content verification to ensure the integrity and security of the command; the edge sensing node responsible for the monitoring target is determined through the power system edge sensing node topology mapping relationship. This mapping relationship is established based on the geographical location of the monitoring target, the communication link affiliation, and the coverage of the edge sensing node to ensure that the command can be accurately sent to the corresponding edge sensing node; After receiving the command, the S420 edge sensing node configures the sensing parameters for the monitored target according to the high-precision sensing requirements. Configuring these parameters includes increasing the sampling frequency, switching to a higher-precision sensor, or activating a backup sensing unit. The configured parameters are combined with the performance of the sensing equipment mounted on the edge sensing node to ensure that the configured parameters are within the range supported by the device hardware. Increasing the sampling frequency is adjusted based on the frequency standard in the high-precision sensing requirements to ensure the timeliness of data acquisition. Switching to a higher-precision sensor requires a self-test of the sensor to confirm that its performance meets the high-precision sensing requirements before the switch is completed. Activating the backup sensing unit ensures the normal execution of high-precision sensing when the main sensing unit malfunctions or lacks sufficient accuracy. S430 and edge sensing nodes initiate high-precision data sensing of the monitored target based on the configured sensing parameters. Before the high-precision data sensing is initiated, the edge sensing nodes perform a comprehensive self-check of the operating status of the sensing equipment, the connection status of the communication module, and the power supply status. The sensing operation is initiated only after all modules are confirmed to be normal. During the sensing process, the operating status of the equipment is monitored in real time. If equipment failure or abnormal data acquisition occurs, the fault alarm mechanism is immediately triggered, and an attempt is made to switch to backup equipment or adjust the parameters. The S440 edge sensing node encapsulates the data generated during the high-precision sensing process into sensing data packets and reports them to the main station. The encapsulation of the sensing data packets must follow a preset format standard. The encapsulated content includes the monitoring target identifier, data acquisition timestamp, sensing data value, data accuracy level, device operating status identifier, and data verification code. The data verification code is calculated using a preset encryption algorithm and is used by the main station to verify the integrity and authenticity of the data after receiving it. The reporting process adopts a priority transmission mechanism to ensure that the transmission priority of high-precision sensing data is higher than that of ordinary steady-state data, thus ensuring the timeliness of data reporting. For example: the instruction is sent to the edge sensing node responsible for the line; the node parses the instruction, adjusts the sampling frequency of the corresponding temperature sensor from 1 time / minute to 1 time / second, and switches to a higher precision level; it starts 5 seconds of high-precision temperature measurement to obtain a set of high-density data [25.52, 25.49, 25.51, ...]; it encapsulates it into a message and uploads it to the main station with high priority. This is just an example and is not a limitation.

[0047] S500: When the real-time information entropy value corresponding to the monitored target recovers to below the steady-state entropy threshold, the control edge sensing node restores the default transmission strategy and sends a resource release request to the main station system. Specifically, step S500 includes: S510. Continuously monitor the high-precision sensing data stream corresponding to the target and calculate its real-time information entropy value within the sliding time window; the frequency of continuous monitoring is consistent with the sampling frequency of high-precision sensing to ensure real-time tracking of the high-precision sensing data stream. S520. When the real-time information entropy value recovers to below the steady-state entropy threshold and the duration reaches the preset stable judgment duration, it is determined that the state of the monitored target has recovered to a steady state. S530. Based on the judgment result, a policy reset command is sent to the edge sensing node. The policy reset command is used to instruct the edge sensing node to stop the current high-precision sensing mode and restore the mode of dynamically selecting the data transmission strategy based on the real-time information entropy value. The policy reset command includes the monitoring target identifier, the reset time node, and the baseline parameters of the restored transmission strategy. After receiving the command, the edge sensing node first stops the operation of the high-precision sensing device, and then configures the default transmission strategy according to the baseline parameters. The default transmission strategy is the transmission mode of dynamically switching between long-cycle data aggregation and high-frequency data capture based on the real-time information entropy value, ensuring that the restored transmission strategy is consistent with the initial transmission logic. S540. Send a resource release request to the master station system. The resource release request is used to instruct the master station system to reclaim communication resources that have been pre-scheduled or reserved for the monitoring target. The resource release request includes the monitoring target identifier, the type of pre-scheduled resource, the resource allocation duration, and the release time node. After receiving the request, the master station system checks the pre-scheduled communication resources. After confirming that the resources are not occupied by other high-priority tasks, it performs the resource reclamation operation and sends the resource release completion information back to the master station to ensure the efficient recycling of communication resources. For example: The main station continuously monitors newly reported high-precision temperature flow and calculates its real-time entropy value. Assuming that after processing, the temperature returns to a stable state, the calculated entropy value drops to 0.4 and remains below the steady-state entropy threshold for 3 minutes; the system determines that the state has returned to a steady state and sends a policy reset command to the edge sensing nodes; the edge sensing nodes restore the temperature sensor sampling frequency to 1 time / minute; and send a resource release request to the main station system to reclaim the network bandwidth temporarily reserved for this high-precision data transmission.

[0048] This invention provides another technical solution: a power data intelligent transmission system based on multi-source fusion sensing, which includes: an entropy evaluation module, a transmission decision module, and an instruction generation module. The entropy assessment module is used to calculate the real-time information entropy of key monitoring data streams within a sliding time window based on the results of multi-source fusion sensing, and to calculate the spatial distribution entropy among similar monitoring points; based on the real-time information entropy value and its changing trend, it assesses the real-time value and urgency of data transmission. The transmission decision module is used to dynamically select and execute data transmission strategies based on the output of the entropy evaluation module, and to initiate resource pre-scheduling to the main station system when a high-value data stream is predicted. The instruction generation module compares the received steady-state feature summary with the expected value of the digital twin model, and generates accurate data acquisition instructions when a latent deviation in the steady state is identified.

[0049] The entropy evaluation module includes a real-time entropy unit and a spatial entropy unit. The real-time entropy unit is used to calculate the probability distribution characteristics of a selected single monitoring data sequence within a preset sliding time window based on the results of multi-source fusion sensing, and to determine the real-time information entropy value within the window according to the information entropy formula. The spatial entropy unit is used to calculate the discrete distribution characteristics of the fused sensing data from multiple monitoring points of the same category in different spatial locations within the time window or the same aggregation period, and to determine the spatial distribution entropy value according to the information entropy formula.

[0050] The transmission decision module includes a summary generation unit and a pre-scheduling unit. The summary generation unit is used to perform long-cycle data aggregation when the real-time information entropy value is lower than the steady-state entropy threshold, and to identify key data points based on entropy weight to form a steady-state feature summary. The pre-scheduling unit is used to perform high-frequency data capture when the real-time information entropy value exceeds the transient entropy threshold or its rate of change exceeds the entropy change rate threshold, and to send a resource pre-scheduling request to the main station system according to the entropy change trend.

[0051] The instruction generation module includes a deviation determination unit and a precise instruction unit. The deviation determination unit is used to compare the key data features in the steady-state feature summary with the expected values ​​of the digital twin model and query the corresponding real-time information entropy value. The precise instruction unit is used to generate precise data acquisition instructions containing the monitoring target identifier and high-precision perception requirements when the key data features deviate from the expected values ​​and the real-time information entropy value is lower than the model deviation entropy threshold.

[0052] Example 3: like Figure 3 As shown, embodiments of the present invention also provide a computer device. Figure 3 Taking a single processor 10 as an example, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other using different buses and can be mounted on a common motherboard or otherwise installed as needed. The processor can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices, such as display devices coupled to the interface. In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations, for example, as a server array, a group of blade servers, or a multiprocessor system.

[0053] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0054] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0055] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0056] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0057] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0058] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

Claims

1. A method for intelligent transmission of power data based on multi-source fusion sensing, characterized in that, Includes the following steps: S1, based on the results of multi-source fusion sensing, calculate the real-time information entropy of key monitoring data streams within a sliding time window, and calculate the spatial distribution entropy among similar monitoring points; the results of multi-source fusion sensing refer to the structured monitoring data streams obtained after spatiotemporal alignment and normalization of raw data collected by at least two types of sensors. Key monitoring data streams refer to the monitoring data sequences that reflect the status of core equipment, selected according to preset rules. S2, based on the real-time information entropy value and the trend of information entropy value change, when the real-time information entropy value is lower than the steady-state entropy threshold, long-cycle data aggregation and key point sampling based on entropy weight are performed to generate a steady-state feature summary and upload it; When the real-time information entropy value exceeds the transient entropy threshold or the real-time information entropy change rate exceeds the entropy change rate threshold, high-frequency data capture is performed, and a resource pre-scheduling request is sent to the main station system according to the entropy change trend. S3: Based on the pre-built digital twin model, obtain the expected value, compare the received steady-state feature summary with the expected value of the digital twin model, and if the steady-state feature summary deviates from the expected value and the corresponding real-time information entropy value is lower than the model deviation entropy threshold, then generate a precise data acquisition instruction. S4 sends the precise data acquisition command to the corresponding edge sensing node, enabling the edge sensing node to start high-precision sensing for the monitoring target specified by the command and report the generated sensing data. S5, when the real-time information entropy value corresponding to the monitored target recovers to below the steady-state entropy threshold, the control edge sensing node restores the default transmission strategy and sends a resource release request to the main station system.

2. The intelligent power data transmission method based on multi-source fusion sensing according to claim 1, characterized in that, Step S1 includes: S110, based on the results of multi-source fusion sensing, for a selected single monitoring data sequence, within a preset sliding time window, calculates the probability distribution characteristics of the data sequence, and determines the real-time information entropy value within this window according to the information entropy formula. The real-time information entropy value is used to characterize the uncertainty and information content of the data sequence within this time period. The formula is: ； Where H is the real-time information entropy value; n is the total number of data value intervals; This represents the probability density of the range of values ​​for the i-th data point; To sum the calculation results for all n data value intervals; S120, within a time window or the same aggregation period, based on fused sensing data from multiple monitoring points of the same category at different spatial locations, calculate the discrete distribution characteristics of this data set in the spatial dimension, and determine the spatial distribution entropy value according to the information entropy formula: ； in, The value represents the spatial distribution entropy; m represents the total number of spatial grids in the monitoring area. This represents the probability of the j-th spatial grid, which is the ratio of the number of similar monitoring points in this grid to the total number of all similar monitoring points. This represents the summation of the calculation results for all m spatial grids; the spatial distribution entropy value is used to characterize the degree of spatial consistency of the regional state reflected by the multiple monitoring points.

3. The intelligent power data transmission method based on multi-source fusion sensing according to claim 1, characterized in that, Step S2 includes: S210, If the real-time information entropy value is lower than the steady-state entropy threshold, then perform long-cycle data aggregation, which includes extending the time window length of data aggregation; S220, during the long-cycle data aggregation process, based on the distribution of real-time information entropy values ​​within the time window, key data points with entropy values ​​higher than the local key entropy threshold are identified and extracted; a steady-state feature summary is constructed from the key data points, and the steady-state feature summary is uploaded; S230, if the real-time information entropy value exceeds the transient entropy threshold, or the real-time information entropy change rate exceeds the entropy change rate threshold, then it is determined that the high-frequency data capture condition is met; the real-time information entropy change rate is calculated by the ratio of the difference between the real-time information entropy values ​​within two adjacent sliding time windows to the time interval, and the formula is: ； Where r represents the rate of change of real-time information entropy. This represents the real-time information entropy value of the k-th sliding time window. This represents the real-time information entropy value of the (k-1)th sliding time window. This represents the time interval between two adjacent sliding time windows; the determination of whether the high-frequency data capture condition is met uses OR logic. If the real-time information entropy value exceeds the standard or the entropy change rate exceeds the standard, and at least one condition is met, high-frequency data capture is triggered. S240, when the high-frequency data capture condition is met, perform high-frequency data capture, which includes shortening the sampling interval or pausing data aggregation; S250: Based on the trend of change of the real-time information entropy value before it exceeds the transient entropy threshold, predict the data traffic in the future time period, and send a resource pre-scheduling request containing the predicted data traffic to the main station system.

4. The intelligent power data transmission method based on multi-source fusion sensing according to claim 1, characterized in that, Step S3 includes: S310, based on the steady-state feature summary, obtain the digital twin model of the monitoring target corresponding to the steady-state feature summary, and extract the expected value or expected value range of the digital twin model under the current operating conditions; S320, compare the key data features in the steady-state feature summary with the expected value or expected value range, and determine whether the key data features deviate from the expected value or exceed the expected value range; S330, when a key data feature is determined to deviate from the expected value or exceed the expected value range, query the real-time information entropy value of the monitoring target within the time window for generating the steady-state feature summary; S340, if the real-time information entropy value obtained is lower than the model deviation entropy threshold, it is determined that the deviation of the steady-state feature summary belongs to the latent deviation in the steady state, and a precise data acquisition instruction containing the monitoring target identifier and high-precision perception requirements is generated.

5. The intelligent power data transmission method based on multi-source fusion sensing according to claim 1, characterized in that, Step S4 includes: S410 parses the precise data acquisition command, obtains the monitoring target identifier and high-precision perception requirements specified in the command; and sends the precise data acquisition command to the edge perception node responsible for the monitoring target based on the monitoring target identifier. After receiving the instruction, the edge sensing node in S420 configures the sensing parameters for the monitored target according to the high-precision sensing requirements. The configuration of sensing parameters includes, but is not limited to, increasing the sampling frequency, switching the sensing mode, or enabling backup sensing resources. S430, the edge sensing node initiates high-precision data sensing for the monitored target based on the configured sensing parameters; S440: The edge sensing node encapsulates the data generated during the high-precision sensing process into sensing data messages and reports them to the main station.

6. The intelligent power data transmission method based on multi-source fusion sensing according to claim 1, characterized in that, Step S5 includes: S510, continuously monitor the high-precision sensing data stream corresponding to the monitoring target, and calculate its real-time information entropy value within the sliding time window; S520: When the real-time information entropy value recovers to below the steady-state entropy threshold and the duration reaches the preset value, it is determined that the state of the monitored target has recovered to a steady state. The preset value is the preset stable determination duration. S530, based on the judgment result, sends a policy reset command to the edge sensing node. The policy reset command is used to instruct the edge sensing node to stop the current high-precision sensing mode and restore the mode of dynamically selecting the data transmission strategy based on the real-time information entropy value. S540 sends a resource release request to the master station system. The resource release request is used to instruct the master station system to reclaim communication resources that have been pre-scheduled or reserved for the monitoring target.

7. A power data intelligent transmission system based on multi-source fusion sensing, used to execute the power data intelligent transmission method based on multi-source fusion sensing as described in any one of claims 1-6, characterized in that, The system includes: The entropy assessment module is used to calculate the real-time information entropy of key monitoring data streams within a sliding time window based on the results of multi-source fusion sensing, and to calculate the spatial distribution entropy among similar monitoring points; based on the real-time information entropy value and its changing trend, it assesses the real-time value and urgency of data transmission. The transmission decision module is used to dynamically select and execute the data transmission strategy based on the output of the entropy evaluation module, and to initiate resource pre-scheduling when the resource pre-scheduling conditions are met. The instruction generation module compares the received steady-state feature summary with the expected value of the digital twin model, and generates accurate data acquisition instructions when a latent deviation in the steady state is identified.

8. The intelligent power data transmission system based on multi-source fusion sensing according to claim 7, characterized in that, The entropy evaluation module includes: The real-time entropy unit is used to calculate the probability distribution characteristics of a selected single monitoring data sequence based on the results of multi-source fusion sensing, within a sliding time window of a preset length, and determine the real-time information entropy value within the window according to the information entropy formula. The spatial entropy unit is used to calculate the discrete distribution characteristics of the data in the spatial dimension based on fused sensing data from multiple monitoring points of the same category at different spatial locations within a time window or the same aggregation period, and to determine the spatial distribution entropy value according to the information entropy formula.

9. The intelligent power data transmission system based on multi-source fusion sensing according to claim 7, characterized in that, The transmission decision module includes: The summary generation unit is used to perform long-cycle data aggregation when the real-time information entropy value is lower than the steady-state entropy threshold, and to identify key data points based on entropy weights to form a steady-state feature summary. The pre-scheduling unit is used to perform high-frequency data capture and send a resource pre-scheduling request to the main station system according to the entropy trend when the real-time information entropy value exceeds the transient entropy threshold or its rate of change exceeds the entropy change rate threshold.

10. The intelligent power data transmission system based on multi-source fusion sensing according to claim 7, characterized in that, The instruction generation module includes: The deviation determination unit is used to compare the key data features in the steady-state feature summary with the expected values ​​of the digital twin model and query the corresponding real-time information entropy value. The precision instruction unit is used to generate precise data acquisition instructions that include the monitoring target identifier and high-precision perception requirements when key data features deviate from the expected value and the real-time information entropy value is lower than the model deviation entropy threshold.

Images