A method, apparatus, equipment, medium, and program product for monitoring and detecting anomalies in power data flow behavior.
By preprocessing multi-source data from the power system and evaluating it using AI models, and by comparing the evaluation results within the detection cycle, an optimized and upgraded early warning signal is generated. This solves the problems of insufficient accuracy and intelligence in power data flow monitoring, and enables the power system to achieve self-sensing feedback and reliable operation.
Patent Information
- Application Number
- CN202611300520.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-26
- Publication Date
- 2026-11-17
AI Technical Summary
Existing power data flow monitoring methods have shortcomings in data integration and analysis capabilities, accuracy of anomaly identification, and level of intelligence, making it difficult to effectively ensure the safe and stable operation of the power system.
By collecting multi-source data from the power system, the data is preprocessed and then input into a pre-trained anomaly detection and evaluation model for risk assessment. The assessment results within the detection period are compared with the actual diagnostic results to generate early warning signals for system optimization and upgrade. AI models such as convolutional neural networks and recurrent neural networks are used for comprehensive reasoning and analysis.
It improves the accuracy and comprehensiveness of risk assessment of power data flow behavior, realizes the self-awareness and proactive feedback of system performance, reduces the burden of operation and maintenance personnel, and ensures the long-term reliability and assessment quality of the system.
Smart Images

Figure CN123116006A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of big data technology, and in particular to a method, apparatus, equipment, medium, and program product for monitoring and detecting anomalies in power data flow behavior. Background Technology
[0002] As a critical national infrastructure, the safe and stable operation of the power system is directly related to the normal order of the national economy and social life. With the in-depth advancement of smart grid construction and the continuous improvement of power information technology, the power system generates massive amounts of high-value data during operation, including dispatch operation data, load forecasting data, electricity consumption information collection data, power grid topology data, and equipment status monitoring data. These data come from a wide range of sources, are diverse in format, are updated frequently, and carry important business implications and security attributes.
[0003] In the current power system operation and management system, the aforementioned data does not exist in isolation but continuously flows between multiple business platforms, information systems, and functional departments. For example, the SCADA system needs to upload real-time collected grid operation data to the dispatch master station for decision analysis; the metering automation system reports massive amounts of user electricity consumption data to the data center for load forecasting and electricity billing; the energy management system and the power trading platform need to share some load and transaction data to support market operations; in addition, the big data analysis platform also needs to integrate and call relevant data from multiple business systems for power load forecasting model training and grid situational awareness analysis. The above data flow behavior runs through all aspects of the power system—generation, transmission, transformation, distribution, and consumption—and constitutes an important part of the full lifecycle management of power data.
[0004] Power data faces multiple risks during cross-system and cross-departmental transfers. On the one hand, data is vulnerable to tampering, leakage, or unauthorized access during transmission and sharing; on the other hand, abnormal data transfer behavior may reflect system malfunctions, network attacks, or internal operational errors. Therefore, real-time monitoring of power data transfer behavior and timely detection of anomalies are crucial for ensuring the safe and stable operation of the power system and preventing data security risks.
[0005] In related technologies, the power industry generally employs relatively traditional methods for data flow monitoring. Common technical approaches include: setting fixed monitoring indicators for a single data source and using threshold judgments to identify whether data exceeds the normal range; or matching and verifying data flow behavior based on preset rule templates, triggering alarms when data characteristics meet specific rule conditions. In addition, some systems use simple statistical analysis methods, such as calculating the mean, variance, and other statistical quantities of the data and comparing them with historical baselines to determine whether the current data deviates from its target range.
[0006] However, current anomaly detection methods have the following technical problems:
[0007] Existing monitoring methods have shortcomings in terms of data integration and analysis capabilities, accuracy of anomaly identification, and level of intelligence, and need to be optimized. Summary of the Invention
[0008] Therefore, it is necessary to provide a method, device, computer equipment, computer-readable storage medium, and computer program product for monitoring and detecting anomalies in power data flow behavior that can improve the accuracy of diagnosis and the early prediction rate, in response to the above-mentioned technical problems.
[0009] Firstly, this application provides a method for monitoring and detecting anomalies in power data flow behavior. The method includes:
[0010] Collect multi-source power data from the power system, including operation time, user role, proportion of sensitive fields, and SQL length.
[0011] The multi-source power data is preprocessed to obtain standardized power data. The preprocessing includes data cleaning, data organization, and data standardization.
[0012] The standardized power data is input into a pre-trained anomaly detection and evaluation model. Based on the anomaly detection and evaluation model, inference calculations are performed to output the risk assessment results of the multi-source power data.
[0013] Multiple risk assessment results within a preset detection period are obtained. Based on the actual diagnostic results corresponding to each risk assessment result, the accuracy of each risk assessment result is determined, and a system optimization and upgrade early warning signal is generated based on the accuracy determination result.
[0014] In one embodiment, the step of acquiring multiple risk assessment results within a preset detection period, determining the accuracy of each risk assessment result based on the actual diagnostic results corresponding to each risk assessment result, and generating a system optimization and upgrade early warning signal based on the accuracy determination result includes:
[0015] Each risk assessment result is evaluated to determine whether it matches the actual diagnosis result. Any inconsistent risk assessment results are marked as risk discrepancies.
[0016] The number of risk anomaly assessment results within the detection period is obtained, and the ratio of the number of risk anomaly assessment results to the total number of risk assessment results within the detection period is calculated to obtain the risk anomaly assessment judgment value.
[0017] The risk assessment judgment value is compared with the preset risk assessment judgment threshold. When the risk assessment judgment value exceeds the preset risk assessment judgment threshold, the optimization and upgrade early warning signal is generated.
[0018] In one embodiment, comparing the risk anomaly assessment value with a preset risk anomaly assessment threshold, and generating the optimization and upgrade early warning signal when the risk anomaly assessment value exceeds the preset risk anomaly assessment threshold, includes:
[0019] The analysis duration corresponding to each of the risk assessment results is obtained, and the analysis duration is the interval between the output time of the risk assessment result and the collection time of the corresponding data.
[0020] Each analysis duration is compared with a preset analysis duration threshold, and the analysis duration exceeding the preset analysis duration threshold is marked as an abnormal analysis duration;
[0021] Obtain the number of outlier durations within the detection cycle, and calculate the ratio of the number of outlier durations to the total number of risk assessment results within the detection cycle to obtain the outlier count value;
[0022] Based on the risk assessment judgment value and the anomaly analysis value, a system optimization and upgrade early warning signal is generated.
[0023] In one embodiment, before generating the system optimization and upgrade early warning signal based on the accuracy judgment result, the method further includes:
[0024] Several sets of pre-prepared test data packets are obtained, and multiple power multi-source data risk assessments are performed on each set of test data packets to obtain multiple risk assessment results corresponding to each set of test data packets.
[0025] For each group of test data packets, determine whether there are differences between the multiple risk assessment results corresponding to that group. If there are differences, mark the group of test data packets as unstable test packets.
[0026] Obtain the number of unstable test packets and calculate the ratio of the number of unstable test packets to the total number of test data packets to obtain the abnormal test packet count.
[0027] The abnormal measurement percentage is compared with a preset abnormal measurement percentage threshold. When the abnormal measurement percentage exceeds the preset abnormal measurement percentage threshold, an evaluation unstable signal is generated.
[0028] In one embodiment, determining whether there are differences among the multiple risk assessment results corresponding to each group of test data packets, and marking the group of test data packets as unstable test packets if differences exist, includes:
[0029] Obtain the analysis duration corresponding to each of the multiple risk assessments for the test data packets in this group, wherein the analysis duration is the interval between the output time of the risk assessment result and the input time of the corresponding test data packet;
[0030] Calculate the variance of all analysis durations corresponding to the test data packets in the group to obtain the analysis time bias value. Compare the analysis time bias value with a preset analysis time bias threshold. When the analysis time bias value exceeds the preset analysis time bias threshold, mark the test data packets in the group as unstable test packets.
[0031] In one embodiment, the anomaly detection and evaluation model is any one of convolutional neural network, recurrent neural network, fully connected neural network, random forest or support vector machine; the risk assessment result includes at least one of risk score, risk level or predicted anomaly rate, and the risk level includes low risk, medium risk and high risk.
[0032] Secondly, this application also provides a device for monitoring and detecting anomalies in power data flow behavior. The device includes:
[0033] The data acquisition module is used to collect multi-source power data from the power system. The multi-source power data includes operation time, user role, proportion of sensitive fields, and SQL length.
[0034] The data processing module is used to preprocess the multi-source power data to obtain standardized power data. The preprocessing includes data cleaning, data organization, and data standardization.
[0035] The model evaluation module is used to input the standardized power data into a pre-trained anomaly detection and evaluation model, perform inference calculations based on the anomaly detection and evaluation model, and output the risk assessment results of the multi-source power data.
[0036] The early warning optimization module is used to acquire multiple risk assessment results within a preset detection period, determine the accuracy of each risk assessment result based on the actual diagnostic results corresponding to each risk assessment result, and generate a system optimization and upgrade early warning signal based on the accuracy judgment result.
[0037] In one embodiment, the early warning optimization module is further configured to:
[0038] Each risk assessment result is evaluated to determine whether it matches the actual diagnosis result. Any inconsistent risk assessment results are marked as risk discrepancies.
[0039] The number of risk anomaly assessment results within the detection period is obtained, and the ratio of the number of risk anomaly assessment results to the total number of risk assessment results within the detection period is calculated to obtain the risk anomaly assessment judgment value.
[0040] The risk assessment judgment value is compared with the preset risk assessment judgment threshold. When the risk assessment judgment value exceeds the preset risk assessment judgment threshold, the optimization and upgrade early warning signal is generated.
[0041] In one embodiment, the early warning optimization module is further configured to:
[0042] The analysis duration corresponding to each of the risk assessment results is obtained, and the analysis duration is the interval between the output time of the risk assessment result and the collection time of the corresponding data.
[0043] Each analysis duration is compared with a preset analysis duration threshold, and the analysis duration exceeding the preset analysis duration threshold is marked as an abnormal analysis duration;
[0044] Obtain the number of outlier durations within the detection cycle, and calculate the ratio of the number of outlier durations to the total number of risk assessment results within the detection cycle to obtain the outlier count value;
[0045] Based on the risk assessment judgment value and the anomaly analysis value, a system optimization and upgrade early warning signal is generated.
[0046] In one embodiment, the early warning optimization module is further configured to:
[0047] Several sets of pre-prepared test data packets are obtained, and multiple power multi-source data risk assessments are performed on each set of test data packets to obtain multiple risk assessment results corresponding to each set of test data packets.
[0048] For each group of test data packets, determine whether there are differences between the multiple risk assessment results corresponding to that group. If there are differences, mark the group of test data packets as unstable test packets.
[0049] Obtain the number of unstable test packets and calculate the ratio of the number of unstable test packets to the total number of test data packets to obtain the abnormal test packet count.
[0050] The abnormal measurement percentage is compared with a preset abnormal measurement percentage threshold. When the abnormal measurement percentage exceeds the preset abnormal measurement percentage threshold, an evaluation unstable signal is generated.
[0051] In one embodiment, the early warning optimization module is further configured to:
[0052] Obtain the analysis duration corresponding to each of the multiple risk assessments for the test data packets in this group, wherein the analysis duration is the interval between the output time of the risk assessment result and the input time of the corresponding test data packet;
[0053] Calculate the variance of all analysis durations corresponding to the test data packets in the group to obtain the analysis time bias value. Compare the analysis time bias value with a preset analysis time bias threshold. When the analysis time bias value exceeds the preset analysis time bias threshold, mark the test data packets in the group as unstable test packets.
[0054] In one embodiment, the anomaly detection and evaluation model is any one of convolutional neural network, recurrent neural network, fully connected neural network, random forest or support vector machine; the risk assessment result includes at least one of risk score, risk level or predicted anomaly rate, and the risk level includes low risk, medium risk and high risk.
[0055] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of a method for monitoring and detecting anomalies in power data flow behavior as described in any embodiment of the first aspect.
[0056] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of a method for monitoring and detecting anomalies in power data flow behavior as described in any embodiment of the first aspect.
[0057] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of a method for monitoring and detecting anomalies in power data flow behavior as described in any embodiment of the first aspect.
[0058] The above-described method, apparatus, computer equipment, storage medium, and computer program product for monitoring and detecting anomalies in power data flow, derived from the technical features in the embodiments, can achieve the following beneficial effects to address the technical problems in the background art:
[0059] This application provides a method for monitoring and detecting anomalies in power data flow. In implementation, by collecting multi-source data generated during power system operation and sequentially performing data preprocessing, AI model inference and evaluation, and self-verification of the evaluation results' accuracy, automated monitoring and anomaly detection of power data flow are achieved. Compared to existing monitoring methods that rely on a single data source and fixed threshold judgments, this solution integrates multi-dimensional information such as operation time, user roles, sensitive field ratios, and SQL length, utilizing a pre-trained AI model for comprehensive inference and analysis, effectively improving the accuracy and comprehensiveness of risk assessment. Simultaneously, by comparing and analyzing the evaluation results with actual diagnostic results within a preset detection period, the system's own evaluation accuracy can be quantitatively measured, and an optimization and upgrade warning signal can be proactively generated when accuracy is insufficient, realizing a self-awareness and proactive feedback mechanism for system performance. Therefore, this solution not only improves the intelligence level and diagnostic accuracy of power data anomaly detection but also ensures the long-term reliability and evaluation quality of the system through a built-in self-verification mechanism, reducing the manual verification burden on maintenance personnel and providing strong technical support for the safe flow and efficient management of power data. Attached Figure Description
[0060] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0061] Figure 1 This is an application environment diagram of a method for monitoring and detecting anomalies in power data flow behavior in one embodiment;
[0062] Figure 2 This is a schematic diagram of the first process of a method for monitoring and detecting anomalies in power data flow behavior in one embodiment;
[0063] Figure 3 This is a schematic diagram of the second process of a method for monitoring and detecting anomalies in power data flow behavior in another embodiment;
[0064] Figure 4 This is a schematic diagram of the third process of a method for monitoring and detecting anomalies in power data flow behavior in another embodiment;
[0065] Figure 5 This is a schematic diagram of the fourth process of a method for monitoring and detecting anomalies in power data flow behavior in another embodiment;
[0066] Figure 6This is a schematic diagram of the fifth process of a method for monitoring and detecting anomalies in power data flow behavior in another embodiment;
[0067] Figure 7 This is a structural block diagram of a power data flow monitoring and anomaly detection device in one embodiment;
[0068] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0069] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0070] This application provides a method for monitoring and detecting anomalies in power data flow behavior, which can be applied to, for example... Figure 1 The application environment shown can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services.
[0071] In one embodiment, such as Figure 2 As shown, a method for monitoring and detecting anomalies in power data flow behavior is provided, and this method is applied to... Figure 1 Taking the server in the example, the following steps are included:
[0072] Step 202: Collect multi-source power data from the power system, including operation time, user role, proportion of sensitive fields, and SQL length.
[0073] Step 204: Preprocess the multi-source power data to obtain standardized power data. The preprocessing includes data cleaning, data organization, and data standardization.
[0074] Step 206: Input the standardized power data into the pre-trained anomaly detection and evaluation model, perform inference calculations based on the anomaly detection and evaluation model, and output the risk assessment results of the multi-source power data.
[0075] Step 208: Obtain multiple risk assessment results within a preset detection period, determine the accuracy of each risk assessment result based on the actual diagnostic results corresponding to each risk assessment result, and generate a system optimization and upgrade early warning signal based on the accuracy judgment result.
[0076] In the above-mentioned method for monitoring and detecting anomalies in power data flow, reasonable derivation based on the technical features in the embodiments achieves the beneficial effect of solving the technical problems raised in the background art:
[0077] This application provides a method for monitoring and detecting anomalies in power data flow. In implementation, by collecting multi-source data generated during power system operation and sequentially performing data preprocessing, AI model inference and evaluation, and self-verification of the evaluation results' accuracy, automated monitoring and anomaly detection of power data flow are achieved. Compared to existing monitoring methods that rely on a single data source and fixed threshold judgments, this solution integrates multi-dimensional information such as operation time, user roles, sensitive field ratios, and SQL length, utilizing a pre-trained AI model for comprehensive inference and analysis, effectively improving the accuracy and comprehensiveness of risk assessment. Simultaneously, by comparing and analyzing the evaluation results with actual diagnostic results within a preset detection period, the system's own evaluation accuracy can be quantitatively measured, and an optimization and upgrade warning signal can be proactively generated when accuracy is insufficient, realizing a self-awareness and proactive feedback mechanism for system performance. Therefore, this solution not only improves the intelligence level and diagnostic accuracy of power data anomaly detection but also ensures the long-term reliability and evaluation quality of the system through a built-in self-verification mechanism, reducing the manual verification burden on maintenance personnel and providing strong technical support for the safe flow and efficient management of power data.
[0078] In one embodiment, it can be as follows Figure 3 As shown, step 208 includes:
[0079] Step 302: Determine whether each risk assessment result matches the actual diagnosis result, and mark the inconsistent risk assessment results as risk discrepancies.
[0080] Step 304: Obtain the number of risk anomaly assessment results within the detection period, and calculate the ratio of the number of risk anomaly assessment results to the total number of risk assessment results within the detection period to obtain the risk anomaly assessment judgment value.
[0081] Step 306: Compare the risk anomaly assessment value with the preset risk anomaly assessment threshold. When the risk anomaly assessment value exceeds the preset risk anomaly assessment threshold, generate the optimization and upgrade early warning signal.
[0082] In this embodiment, by comparing each risk assessment result with the actual diagnostic result within the detection period, discrepancies are marked as risk anomalies, and their proportion is calculated as the risk anomaly judgment value. This judgment value is then compared with a preset threshold to determine whether to generate an optimization and upgrade warning signal. This achieves quantitative evaluation and automated feedback of the system's assessment accuracy, objectively reflecting the system's assessment quality level within the detection period. It also proactively triggers warnings when accuracy declines, providing a clear basis for system optimization and upgrades, and effectively ensuring the long-term reliability and accuracy of power data risk assessment.
[0083] In one embodiment, it can be as follows Figure 4 As shown, step 306 includes:
[0084] Step 402: Obtain the analysis duration corresponding to each of the risk assessment results, wherein the analysis duration is the interval between the output time of the risk assessment result and the collection time of the corresponding data.
[0085] Step 404: Compare each of the analysis durations with a preset analysis duration threshold, and mark the analysis durations that exceed the preset analysis duration threshold as heterogeneous analysis durations.
[0086] Step 406: Obtain the number of anomaly durations within the detection cycle, and calculate the ratio of the number of anomaly durations to the total number of risk assessment results within the detection cycle to obtain the anomaly count value.
[0087] Step 408: Generate a system optimization and upgrade early warning signal based on the risk anomaly assessment value and the anomaly analysis value.
[0088] For example, the server can obtain the number of risk anomaly assessment results within the detection period and calculate the risk anomaly assessment judgment value by comparing it with the total number of power multi-source data risk assessment results. The larger the risk anomaly assessment judgment value, the worse the accuracy of the system's risk assessment results for power multi-source data within the detection period. The risk anomaly assessment judgment value is compared with a preset risk anomaly assessment judgment threshold. If the risk anomaly assessment judgment value exceeds the preset risk anomaly assessment judgment threshold, it indicates that the accuracy of the system's risk assessment results for power multi-source data within the detection period is poor, and an optimization and upgrade early warning signal is generated.
[0089] Furthermore, if the risk anomaly assessment value does not exceed the preset risk anomaly assessment threshold, the analysis time corresponding to the risk assessment result of the corresponding power multi-source data is collected (i.e., the interval between the output time of the risk assessment result and the data collection time). The analysis time is then compared with the preset analysis time threshold. If the analysis time exceeds the preset analysis time threshold, it indicates that the processing and analysis efficiency of the corresponding power multi-source data risk assessment is low, and the corresponding analysis time is marked as the anomaly analysis time. The number of anomaly analysis times within the detection period is obtained, and the ratio is calculated with the total number of power multi-source data risk assessment results to obtain the anomaly analysis status value. The average of all analysis times within the detection period is then calculated to obtain the analysis effectiveness value.
[0090] For example, the server can use the formula WP=ry1*WF+ry2*WK+ry3*WN / (ry1+ry2) to numerically calculate the risk anomaly assessment value WF, the anomaly status value WK, and the analysis effectiveness value WN to obtain the optimization and upgrade warning value WP. Here, ry1, ry2, and ry3 are preset proportional coefficients, where ry1>ry2>ry3>0. Furthermore, the larger the value of the optimization and upgrade warning value WP, the worse the overall system evaluation performance within the detection period. The above formula is a dimensionless numerical calculation, derived from software simulation using collected large amounts of data to obtain the most recent real-world data. The preset parameters in the formula are set by those skilled in the art according to actual conditions. The optimization and upgrade warning value WP is compared with a preset optimization and upgrade warning threshold. If the optimization and upgrade warning value WP exceeds the preset threshold, it indicates that the overall system evaluation performance within the detection period is poor, and an optimization and upgrade warning signal is generated.
[0091] In this embodiment, based on the accuracy assessment, analysis time is further introduced as a dimension for evaluating system efficiency. By obtaining the analysis time corresponding to each risk assessment result and comparing it with a preset threshold, overdue results are marked as abnormal analysis time, and their proportion is calculated to obtain the abnormal analysis status value. The risk abnormal assessment judgment value and the abnormal analysis status value are combined to determine whether to generate an optimization and upgrade warning signal. Thus, a dual-dimensional quantitative evaluation of system assessment accuracy and analysis efficiency is achieved, which can more comprehensively reflect the overall performance level of the system. When the system deteriorates in either the accuracy or efficiency dimension, an early warning can be triggered, providing a more sufficient decision-making basis for system optimization and upgrade.
[0092] In one embodiment, it can be as follows Figure 5 As shown, before step 208, the procedure further includes:
[0093] Step 502: Obtain several sets of pre-prepared test data packets, and perform multiple power multi-source data risk assessments on each set of test data packets to obtain multiple risk assessment results corresponding to each set of test data packets.
[0094] Step 504: For each group of test data packets, determine whether there are differences between the multiple risk assessment results corresponding to that group. If there are differences, mark the group of test data packets as unstable test packets.
[0095] Step 506: Obtain the number of unstable test packets and calculate the ratio of the number of unstable test packets to the total number of test data packets to obtain the abnormal test packet count.
[0096] Step 508: Compare the abnormal measurement percentage with a preset abnormal measurement percentage threshold. When the abnormal measurement percentage exceeds the preset abnormal measurement percentage threshold, generate an evaluation unstable signal.
[0097] For example, when the management terminal receives an unstable assessment signal, it issues a corresponding warning, which can accurately reflect the stability and efficiency of the system assessment results and promptly remind system operation and maintenance personnel to check and optimize the system as needed, thereby ensuring the accuracy and stability of subsequent risk assessments of multi-source power data. The specific test and analysis process of the verification and analysis module may include: acquiring several sets of data packets prepared in advance for system assessment testing, preferably no less than fifteen sets of data packets; sending the corresponding data packets to the data acquisition module, and performing several risk assessments of multi-source power data based on the corresponding data packets, preferably no less than five times for each data packet; if there are differences in the results of several risk assessments for the corresponding data packets, indicating that the test results for the corresponding data packets are unstable, then assigning a judgment symbol RP-1 to the corresponding data packet.
[0098] For example, if there is no difference in the results of several risk assessments for the corresponding data packet, it indicates that the test results for the corresponding data packet are relatively stable. Then, the variance of several analysis times for the corresponding data packet is calculated to obtain the analysis time bias value. The analysis time bias value is compared with the preset analysis time bias threshold. If the analysis time bias value exceeds the preset analysis time bias threshold, it indicates that the test efficiency stability for the corresponding data packet is relatively poor. Then, the judgment symbol RP-1 is assigned to the corresponding data packet.
[0099] The server can obtain the number of data packets assigned the judgment symbol RP-1 and mark them as outlier detection values. The ratio of the outlier detection value to the number of data packets is used to calculate the outlier percentage. The larger the outlier percentage value, the worse the overall stability of the system evaluation. The outlier percentage value is compared with a preset outlier percentage threshold. If the outlier percentage value exceeds the preset outlier percentage threshold, it indicates that the overall stability of the system evaluation is poor, and an evaluation instability signal is generated.
[0100] In this embodiment, a system evaluation stability verification step is added before generating the optimization and upgrade early warning signal. Multiple risk assessments are performed using several sets of test data packets to determine if there are discrepancies between the assessment results. Test packets with discrepancies are marked as unstable test packets, and their proportion is calculated to obtain the outlier percentage. When the percentage exceeds a threshold, an evaluation instability signal is generated. This achieves quantitative detection of the consistency of system evaluation results, effectively identifying evaluation instability issues caused by model fluctuations or system anomalies. It provides objective quantitative evidence for system stability assessment, helping to ensure the consistency and reliability of power data risk assessment results.
[0101] In one embodiment, it can be as follows Figure 6 As shown, step 504 includes:
[0102] Step 602: Obtain the analysis duration corresponding to each of the multiple risk assessments for the test data packets in this group. The analysis duration is the interval between the output time of the risk assessment result and the input time of the corresponding test data packet.
[0103] Step 604: Calculate the variance of all analysis durations corresponding to the test data packets in this group to obtain the analysis time bias value. Compare the analysis time bias value with a preset analysis time bias threshold. When the analysis time bias value exceeds the preset analysis time bias threshold, mark the test data packets in this group as unstable test packets.
[0104] In this embodiment, based on the consistency of evaluation results, the stability of analysis time is further introduced as a dimension for system stability evaluation. By obtaining the analysis time corresponding to multiple risk assessments of each group of test data packets and calculating their variance, the analysis time bias is obtained. When the analysis time bias exceeds a preset threshold, the corresponding test data packet is marked as an unstable test packet. This achieves quantitative detection of the stability of system evaluation efficiency, effectively identifying inconsistencies in analysis time caused by system resource fluctuations or processing anomalies. It provides a more comprehensive quantitative basis for system stability evaluation and helps ensure the efficiency consistency of power data risk assessment in long-term operation.
[0105] In one embodiment, the anomaly detection and evaluation model is any one of convolutional neural network, recurrent neural network, fully connected neural network, random forest or support vector machine; the risk assessment result includes at least one of risk score, risk level or predicted anomaly rate, and the risk level includes low risk, medium risk and high risk.
[0106] In this embodiment, the abnormal data detection and evaluation model can be any one of convolutional neural networks, recurrent neural networks, fully connected neural networks, random forests, or support vector machines. The risk assessment result includes at least one of risk score, risk level, or predicted anomaly rate, with risk levels further subdivided into low risk, medium risk, and high risk. This provides multiple flexible implementation methods for AI model selection and result output, adapting to the deployment needs of different power application scenarios. Simultaneously, the standardized risk level classification enhances the intuitiveness and interpretability of the assessment results, facilitating quick understanding and response by operation and maintenance personnel.
[0107] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0108] Based on the same inventive concept, this application also provides a device for monitoring and detecting anomalies in power data flow behavior to implement the aforementioned method for monitoring and detecting anomalies in power data flow behavior. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the device for monitoring and detecting anomalies in power data flow behavior provided below can be found in the limitations of the method for monitoring and detecting anomalies in power data flow behavior described above, and will not be repeated here.
[0109] In one embodiment, such as Figure 7 As shown, a device for monitoring and detecting anomalies in power data flow behavior is provided, comprising: a data acquisition module, a data processing module, a model evaluation module, and an early warning and optimization module, wherein:
[0110] The data acquisition module is used to collect multi-source power data from the power system. The multi-source power data includes operation time, user role, proportion of sensitive fields, and SQL length.
[0111] The data processing module is used to preprocess the multi-source power data to obtain standardized power data. The preprocessing includes data cleaning, data organization, and data standardization.
[0112] The model evaluation module is used to input the standardized power data into a pre-trained anomaly detection and evaluation model, perform inference calculations based on the anomaly detection and evaluation model, and output the risk assessment results of the multi-source power data.
[0113] The early warning optimization module is used to acquire multiple risk assessment results within a preset detection period, determine the accuracy of each risk assessment result based on the actual diagnostic results corresponding to each risk assessment result, and generate a system optimization and upgrade early warning signal based on the accuracy judgment result.
[0114] In one embodiment, the early warning optimization module is further configured to:
[0115] Each risk assessment result is evaluated to determine whether it matches the actual diagnosis result. Any inconsistent risk assessment results are marked as risk discrepancies.
[0116] The number of risk anomaly assessment results within the detection period is obtained, and the ratio of the number of risk anomaly assessment results to the total number of risk assessment results within the detection period is calculated to obtain the risk anomaly assessment judgment value.
[0117] The risk assessment judgment value is compared with the preset risk assessment judgment threshold. When the risk assessment judgment value exceeds the preset risk assessment judgment threshold, the optimization and upgrade early warning signal is generated.
[0118] In one embodiment, the early warning optimization module is further configured to:
[0119] The analysis duration corresponding to each of the risk assessment results is obtained, and the analysis duration is the interval between the output time of the risk assessment result and the collection time of the corresponding data.
[0120] Each analysis duration is compared with a preset analysis duration threshold, and the analysis duration exceeding the preset analysis duration threshold is marked as an abnormal analysis duration;
[0121] Obtain the number of outlier durations within the detection cycle, and calculate the ratio of the number of outlier durations to the total number of risk assessment results within the detection cycle to obtain the outlier count value;
[0122] Based on the risk assessment judgment value and the anomaly analysis value, a system optimization and upgrade early warning signal is generated.
[0123] In one embodiment, the early warning optimization module is further configured to:
[0124] Several sets of pre-prepared test data packets are obtained, and multiple power multi-source data risk assessments are performed on each set of test data packets to obtain multiple risk assessment results corresponding to each set of test data packets.
[0125] For each group of test data packets, determine whether there are differences between the multiple risk assessment results corresponding to that group. If there are differences, mark the group of test data packets as unstable test packets.
[0126] Obtain the number of unstable test packets and calculate the ratio of the number of unstable test packets to the total number of test data packets to obtain the abnormal test packet count.
[0127] The abnormal measurement percentage is compared with a preset abnormal measurement percentage threshold. When the abnormal measurement percentage exceeds the preset abnormal measurement percentage threshold, an evaluation unstable signal is generated.
[0128] In one embodiment, the early warning optimization module is further configured to:
[0129] Obtain the analysis duration corresponding to each of the multiple risk assessments for the test data packets in this group, wherein the analysis duration is the interval between the output time of the risk assessment result and the input time of the corresponding test data packet;
[0130] Calculate the variance of all analysis durations corresponding to the test data packets in the group to obtain the analysis time bias value. Compare the analysis time bias value with a preset analysis time bias threshold. When the analysis time bias value exceeds the preset analysis time bias threshold, mark the test data packets in the group as unstable test packets.
[0131] In one embodiment, the anomaly detection and evaluation model is any one of convolutional neural network, recurrent neural network, fully connected neural network, random forest or support vector machine; the risk assessment result includes at least one of risk score, risk level or predicted anomaly rate, and the risk level includes low risk, medium risk and high risk.
[0132] The modules in the aforementioned power data flow monitoring and anomaly detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.
[0133] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a method for monitoring and detecting anomalies in power data flow behavior.
[0134] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0135] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.
[0136] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.
[0137] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.
[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.
[0139] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0140] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0141] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for monitoring and detecting anomalies in power data flow behavior, characterized in that, The method includes: Collect multi-source power data from the power system, including operation time, user role, proportion of sensitive fields, and SQL length. The multi-source power data is preprocessed to obtain standardized power data. The preprocessing includes data cleaning, data organization, and data standardization. The standardized power data is input into a pre-trained anomaly detection and evaluation model. Based on the anomaly detection and evaluation model, inference calculations are performed to output the risk assessment results of the multi-source power data. Multiple risk assessment results within a preset detection period are obtained. Based on the actual diagnostic results corresponding to each risk assessment result, the accuracy of each risk assessment result is determined, and a system optimization and upgrade early warning signal is generated based on the accuracy determination result.
2. The method according to claim 1, characterized in that, The process of acquiring multiple risk assessment results within a preset detection period, determining the accuracy of each risk assessment result based on the actual diagnostic results corresponding to each risk assessment result, and generating a system optimization and upgrade early warning signal based on the accuracy determination results includes: Each risk assessment result is evaluated to determine whether it matches the actual diagnosis result. Any inconsistent risk assessment results are marked as risk discrepancies. The number of risk anomaly assessment results within the detection period is obtained, and the ratio of the number of risk anomaly assessment results to the total number of risk assessment results within the detection period is calculated to obtain the risk anomaly assessment judgment value. The risk assessment judgment value is compared with the preset risk assessment judgment threshold. When the risk assessment judgment value exceeds the preset risk assessment judgment threshold, the optimization and upgrade early warning signal is generated.
3. The method according to claim 2, characterized in that, The step of comparing the risk anomaly assessment value with a preset risk anomaly assessment threshold, and generating the optimization and upgrade early warning signal when the risk anomaly assessment value exceeds the preset risk anomaly assessment threshold, includes: The analysis duration corresponding to each of the risk assessment results is obtained, and the analysis duration is the interval between the output time of the risk assessment result and the collection time of the corresponding data. Each analysis duration is compared with a preset analysis duration threshold, and the analysis duration exceeding the preset analysis duration threshold is marked as an outlier analysis duration; Obtain the number of outlier durations within the detection cycle, and calculate the ratio of the number of outlier durations to the total number of risk assessment results within the detection cycle to obtain the outlier count value; Based on the risk assessment judgment value and the anomaly analysis value, a system optimization and upgrade early warning signal is generated.
4. The method according to claim 1, characterized in that, Before generating the system optimization and upgrade early warning signal based on the accuracy judgment result, the following steps are also included: Several sets of pre-prepared test data packets are obtained, and multiple power multi-source data risk assessments are performed on each set of test data packets to obtain multiple risk assessment results corresponding to each set of test data packets. For each group of test data packets, determine whether there are differences between the multiple risk assessment results corresponding to that group. If there are differences, mark the group of test data packets as unstable test packets. Obtain the number of unstable test packets and calculate the ratio of the number of unstable test packets to the total number of test data packets to obtain the abnormal test packet count. The abnormal measurement percentage is compared with a preset abnormal measurement percentage threshold. When the abnormal measurement percentage exceeds the preset abnormal measurement percentage threshold, an evaluation unstable signal is generated.
5. The method according to claim 4, characterized in that, For each group of test data packets, determining whether there are differences among the multiple risk assessment results corresponding to that group, and marking that group of test data packets as unstable test packets if differences exist, includes: Obtain the analysis duration corresponding to each of the multiple risk assessments for the test data packets in this group, wherein the analysis duration is the interval between the output time of the risk assessment result and the input time of the corresponding test data packet; Calculate the variance of all analysis durations corresponding to the test data packets in the group to obtain the analysis time bias value. Compare the analysis time bias value with a preset analysis time bias threshold. When the analysis time bias value exceeds the preset analysis time bias threshold, mark the test data packets in the group as unstable test packets.
6. The method according to claim 1, characterized in that, The abnormal data detection and evaluation model is any one of convolutional neural network, recurrent neural network, fully connected neural network, random forest or support vector machine; the risk assessment result includes at least one of risk score, risk level or predicted anomaly rate, and the risk level includes low risk, medium risk and high risk.
7. A device for monitoring and detecting anomalies in power data flow behavior, characterized in that, The device includes: The data acquisition module is used to collect multi-source power data from the power system. The multi-source power data includes operation time, user role, proportion of sensitive fields, and SQL length. The data processing module is used to preprocess the multi-source power data to obtain standardized power data. The preprocessing includes data cleaning, data organization, and data standardization. The model evaluation module is used to input the standardized power data into a pre-trained anomaly detection and evaluation model, perform inference calculations based on the anomaly detection and evaluation model, and output the risk assessment results of the multi-source power data. The early warning optimization module is used to acquire multiple risk assessment results within a preset detection period, determine the accuracy of each risk assessment result based on the actual diagnostic results corresponding to each risk assessment result, and generate a system optimization and upgrade early warning signal based on the accuracy judgment result.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
