Grade difference coordinated power distribution network power data early warning method

Abstract

The invention relates to a level difference coordinated power distribution network power data early warning method, which comprises the steps of collecting power operation data of a plurality of monitoring points in a power distribution network in real time, performing cleaning and standardized preprocessing on the data, determining a current to-be-assessed early warning point according to a real-time topological structure of the power distribution network, and determining the level difference coordinated power distribution network power data according to an electrical connection relationship. Screening out upstream equipment and downstream equipment which have a level difference coordination relationship with the early warning point to form an associated equipment set, reading real-time power data of the early warning point and each equipment in the associated equipment set, calculating a difference value or a specific value between the data of the early warning point and the data of each associated equipment as a level difference coordination index, and dynamically generating an early warning threshold value suitable for the current moment and outputting an early warning signal and related diagnosis information through a preset threshold value calculation model based on the historical operation data of the early warning point, the current operation working condition and the real-time data of the associated equipment set, and starting a control measure or prompting operation and maintenance personnel to process according to a preset strategy.

Description

[0001] TECHNICAL FIELD The present application relates to the technical field of data processing, in particular to a differential coordination power data early warning method for power distribution network. BACKGROUND

[0002] With the rapid development of smart grid and power distribution network automation technology, a large number of sensing and monitoring devices are deployed in the power distribution network, generating a huge amount of power operation data including voltage, current, power, frequency, and harmonic. Real-time analysis and early warning of these data are the key to ensuring the safe, stable, and economic operation of the power distribution network. At present, the common power distribution network data early warning method is mostly based on a single threshold or fixed rule, for example, setting a fixed upper limit value for the current of a monitoring point, and triggering an alarm once it is exceeded. However, such traditional methods have significant defects. First, the power distribution network is a complex network system, and the electrical parameters between nodes and lines are mutually influenced and tightly coupled. An abnormality of a node may be caused by upstream device failure or downstream load mutation. The traditional single-point independent early warning mode cannot distinguish between local faults and cascading faults, which easily leads to false alarms or missed alarms. Second, the fixed threshold cannot adapt to the dynamic changes of the power distribution network operation mode (load tidal fluctuation, distributed power switching, network reconstruction). It may frequently misreport at the load peak, and may be slow to respond to potential risks at the load trough, lacking accuracy and adaptability in early warning. Therefore, there is an urgent need for an early warning method that can comprehensively consider the topology structure and operation characteristics of the power distribution network, realize multi-point coordination and dynamic adjustment, accurately identify real faults, suppress false alarms, and improve the accuracy and reliability of early warning.

[0003] SUMMARY In order to solve the problems existing in the prior art, the present application aims to provide a differential coordination power data early warning method for power distribution network.

[0004] The differential coordination power data early warning method for power distribution network described in the present application comprises the following steps: S101, real-time collection of power operation data of a plurality of monitoring points in the power distribution network, and cleaning and standardizing preprocessing of the data; S102, determination of a current early warning point to be evaluated according to the real-time topology structure of the power distribution network, and screening of upstream devices and downstream devices having a differential coordination relationship with the early warning point according to the electrical connection relationship, to form an associated device set; S103, reading of real-time power data of the early warning point and each device in the associated device set, and calculation of the difference or ratio between the early warning point data and each associated device data as a differential coordination index; S104, reading of real-time power data of the early warning point and each device in the associated device set, and calculation of the difference or ratio between the early warning point data and each associated device data as a differential coordination index; S105, compare the real-time data of the early warning point with the dynamic early warning threshold, and make a comprehensive logical judgment in combination with the differential coordination index, if the early warning point data is out of limit and the differential coordination index meets the preset coordination condition at the same time, an early warning signal is generated; S106, output the early warning signal and related diagnostic information, and start corresponding control measures or prompt the operation and maintenance personnel to process according to the preset strategy.

[0005] Further, in the S101 step, real-time power operation data of a plurality of monitoring points in the power distribution network is collected, and the data is cleaned and standardized pretreated, including: Through the timed collection of the data of the plurality of monitoring points in the power distribution network, an initial data set is formed, after data cleaning (mean interpolation processing of missing values) and standardized pretreatment, the data fluctuation is analyzed and the abnormal points exceeding the threshold are marked, and then combined with historical data comparison, the abnormal points conforming to the historical abnormal distribution are classified as regular fluctuations, and the remaining irregular fluctuation points are finely classified by using a support vector machine model, and finally the abnormal data classification result is output.

[0006] Further, in the S102 step, according to the real-time topological structure of the power distribution network, the current early warning point to be evaluated is determined, and according to the electrical connection relationship, the upstream equipment and downstream equipment having a differential coordination relationship with the early warning point are screened out to form an associated equipment set, including: Through topological relationship layer-by-layer association analysis, a closed loop from the early warning point to the systematic risk evaluation is realized, the differential associated equipment set of the early warning point is determined based on the topological structure, the equipment associated network is constructed, the potential risk points are marked according to the differential coordination rules combined with historical and real-time data, and the high-priority equipment is screened according to the association strength, the real-time parameter abnormality of the high-priority equipment is detected by using a support vector machine model, an abnormal fluctuation data set is generated, and the conventional and unconventional abnormalities are distinguished by comparing the historical abnormal positions; For unconventional abnormalities, the topological influence path is traced to generate a complete file revealing the associated risk of multiple equipment, each link is progressively evolved into a global risk analysis with topological and electrical relationship as the logical main line.

[0007] Further, in the S103 step, the real-time power data of the early warning point and each equipment in the associated equipment set is read, the difference or ratio between the early warning point data and each associated equipment data is calculated as a differential coordination index, including: Through multi-level data analysis, the systematic abnormal risk is identified and tracked from the real-time power parameters, the real-time power parameters of the early warning point and the related equipment are collected, and the differential index data set is generated by difference calculation and ratio analysis; The index is compared with a preset threshold value, an abnormal coordination point is marked, an electrical association path of the abnormal coordination point with a warning point is analyzed, an abnormal path list is formed, whether there is a persistent fluctuation on the path is judged by combining real-time monitoring data, fluctuation details are recorded, a support vector machine model is used to classify the trend during the fluctuation, and a potential systematic abnormality is identified; According to the classification result, the propagation direction of the abnormality between devices is traced, an abnormal propagation path record is generated, and each link uses electrical association and time sequence analysis as a chain to gradually associate, amplify, and trace the parameter abnormality, so as to realize risk tracking and positioning from point to plane.

[0008] Further, in the S104 step, based on the historical operation data of the warning point, the current operation condition, and the real-time data of the associated device set, a preset threshold value calculation model is used to dynamically generate a warning threshold value suitable for the current time, including: Through dynamic warning and association analysis, a closed loop from operation state monitoring to systematic risk tracing is realized, time-related operation change characteristics are extracted based on historical and real-time data to form an operation state set, a preset model is used to dynamically calculate the warning limit at the current time, and abnormal operation points in real-time data are marked accordingly, the association of abnormal points between devices is analyzed based on data source records to generate an abnormal association list, a support vector machine model is used to classify abnormal state changes to identify systematic abnormal risks; According to the classification result, the propagation path of the abnormality between devices is traced based on historical data to form a propagation record, and each link uses dynamic threshold value and association analysis as a chain to gradually upgrade real-time abnormal detection to the tracing of the systematic risk propagation path.

[0009] Further, in the S105 step, the real-time data of the warning point is compared with the dynamic warning threshold value, and a comprehensive logical judgment is made in combination with the differential coordination index, if the warning point data is out of limit and the differential coordination index meets the preset coordination condition at the same time, a warning signal is generated, including: Based on the comparison of real-time monitoring data and dynamic threshold value, potential abnormal points are preliminarily marked, and then verified again by the differential coordination index to exclude false positives and confirm real abnormalities, historical data is analyzed using a support vector machine model to identify abnormal trends with persistent characteristics, and associated matching is performed in combination with the operation state of surrounding devices to find systematic abnormal patterns; Finally, according to a preset rule, the abnormal points are prioritized to generate a warning trigger sequence, forming a complete closed loop from data acquisition, multi-level verification, trend analysis to warning generation, and the whole process takes time sequence data analysis and device association verification as the core, through dynamic threshold value adjustment, differential coordination judgment, and multi-dimensional association analysis, to ensure the accuracy and reliability of the warning result.

[0010] Further, in the S106 step, the early warning signal and related diagnostic information are output, and corresponding control measures are started or operation and maintenance personnel are prompted to handle according to the preset strategy, including: Through signal grading, instruction matching, execution tracking and dynamic adjustment, the abnormal state closed loop processing is realized, first, the early warning signals are integrated, the high priority signals are marked according to the intensity threshold, the control instruction sequence is generated by matching the abnormal template through the preset strategy; Real-time tracking of state feedback during execution, adjusting instruction parameters and generating operation and maintenance prompts when not expected, then associating prompts with feedback data, tracking signal change trend, and continuously updating state; Finally, the processing effect is evaluated, the processing record is generated to complete the closed loop, and each link takes signal response as the core, through the matching-execution-feedback-adjustment cycle mechanism, the effective disposal of abnormal state is ensured.

[0011] The differential coordination power distribution network power data early warning method described in the application has the advantages that by introducing the differential coordination mechanism, the electrical logical relationship between the early warning point and its upstream and downstream associated equipment is comprehensively considered, the local real fault and the electrical quantity overrun caused by the fault from other places can be effectively distinguished, thereby greatly reducing false positives and false negatives, a dynamic threshold generation strategy is adopted, the threshold can be automatically adjusted according to real-time operation data (load level, network topology), the shortcomings of poor adaptability of fixed threshold under different operation conditions are overcome, the early warning is more intelligent and more in line with the actual operation state, the generated early warning information not only contains the overrun data itself, but also contains the differential relationship with the associated equipment data, providing more rich and clear fault location and diagnosis basis for operation personnel, and improving the operation and maintenance efficiency; By discovering potential fault hidden dangers earlier and more accurately, and clearly defining the fault influence range, it is helpful for dispatchers to take targeted measures to prevent local faults from expanding into large-area power failure accidents, and significantly improves the safety and reliability of power supply of the power distribution network. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 is the flow of the differential coordination power distribution network power data early warning method described in the application Figure 1 ; Figure 2 is the flow of the differential coordination power distribution network power data early warning method described in the application Figure 2 . DETAILED DESCRIPTION

[0013] As Figures 1-2 shown, the differential coordination power distribution network power data early warning method described in the application includes: As Figures 1-2As shown, S101, real-time collection of power operation data of multiple monitoring points in the power distribution network, and cleaning and standardization preprocessing of the data.

[0014] Further, in step S101, through the multiple monitoring point devices deployed in the power distribution network, the power operation related data is collected at regular intervals and stored in the preset data warehouse to obtain an initial data set. According to the content of the initial data set, perform data cleaning operation, fill and remove for missing values and abnormal values, if the null value is detected in the data field, the mean interpolation method is used for filling, and the cleaned data set is determined. Using the cleaned data set, implement the standardization preprocessing step, convert the data of different dimensions to a unified range, if the data range of a field exceeds the preset threshold, normalize it, and determine the standardized data set. Get the standardized data set, analyze the fluctuation of each monitoring point data, mark the data points with large fluctuations, if the fluctuation amplitude of a monitoring point exceeds the preset threshold, mark it as an abnormal point, and get the marked data set. Through the marked data set, extract the distribution characteristics of the abnormal points, compare with the historical data records, if the distribution of the abnormal points is consistent with the historical abnormal distribution, classify it as a regular fluctuation, and determine the abnormal classification result. According to the abnormal classification result, filter out the abnormal point data of irregular fluctuation, use support vector machine model to further classify these data, and get the final abnormal data classification set.

[0015] Specifically, in step S101, in the process of real-time collection of power operation data of multiple monitoring points in the power distribution network and cleaning and standardization preprocessing, the following specific methods can be used to realize the automation processing of the whole process. First, assume that 100 monitoring points are deployed in the power distribution network, each monitoring point collects voltage, current and power data every minute, for example, a monitoring point collects voltage value of 220.5 volts, current value of 10.2 amperes and power value of 2249.1 watts at a certain time. These data are transmitted to the cloud server in JSON format through the Internet of Things device, and the server uses Kafka message queue to realize high-concurrency data reception to ensure the real-time of data collection. Subsequently, the data cleaning phase is entered, and the system automatically detects abnormal values through preset rules, for example, setting the normal voltage range to 200-240 volts, if the voltage value of a certain monitoring point is 190.3 volts, it is marked as abnormal and excluded, at the same time, the linear interpolation method is used to fill in the missing values, assuming that the voltage data is missing at a certain time, the interpolation result can be calculated as 220.7 volts according to the voltage values of the previous and next two times, 220.1 volts and 221.3 volts, the cleaned data is stored in the database; Then, standardization preprocessing is performed, the system uses Z-score standardization method to convert the cleaned data into a distribution with mean value of 0 and standard deviation of 1, for example, the mean value of voltage data is 220.0 volts and the standard deviation is 2.5 volts, then the standardized result of the voltage value of a certain monitoring point, 220.5 volts, is (220.5-220.0) / 2.5=0.2, ensuring that data of different dimensions are comparable; Finally, the standardized data is used for subsequent analysis, such as predicting load changes through machine learning models, assuming that the prediction result shows that the power load of a certain area will increase by 10% in the next hour, the system automatically generates alarm information and pushes it to the dispatch center, forming a complete closed-loop logic from data collection to analysis, ensuring the stability of the power distribution network.

[0016] As shown in Figures 1-2 S102, according to the real-time topology structure of the power distribution network, the current warning point to be evaluated is determined, and the upstream and downstream devices that exist in the step difference coordination relationship with the warning point are selected according to the electrical connection relationship to form the associated device set.

[0017] Further, in step S102, the current warning point information to be evaluated is obtained through the topology structure data of the power distribution network, and the upstream and downstream devices that exist in the step difference coordination with the warning point are determined in combination with the electrical connection relationship to obtain the preliminary associated device set; According to the preliminary associated device set, the connection relationship between devices is analyzed, the direct and indirect dependence paths between devices are identified using a preset topology analysis tool, and a complete device association network is determined; For the complete device association network, the historical operation records of each device are obtained, and the real-time analysis technology is combined to judge whether the operation state between devices meets the preset step difference coordination rule, if the operation state of a certain device deviates from the preset threshold, it is marked as a potential risk point, and a marked device set is obtained; From the marked device set, the electrical connection data of the potential risk point is extracted, the association strength between it and the warning point is analyzed, if the association strength exceeds the preset threshold, it is classified as a high-priority attention object, and a high-priority device list is determined; Through the high-priority device list, real-time running parameters of the relevant devices are obtained, and the support vector machine model is used for abnormal detection of the parameters to determine whether there is abnormal fluctuation. If abnormal fluctuation is detected, the fluctuation time and amplitude are recorded to obtain an abnormal fluctuation data set; According to the abnormal fluctuation data set, the topological position and electrical connection relationship of the fluctuation are analyzed, and the historical records are compared. If the fluctuation position is consistent with the historical high-frequency abnormal position, it is classified as a conventional anomaly, and the final abnormal classification result is determined. For the final abnormal classification result, the device data of the irregular abnormality is screened out, the influence path in the topological structure is traced through the preset analysis tool, and it is determined whether the influence path involves other key devices. If it involves, the corresponding associated influence record is generated to obtain a complete risk association file.

[0018] Specifically, in step S102, in the operation management of the power distribution network, the process of evaluating the early warning point and analyzing the associated device set based on the real-time topological structure can be realized by an automatic system, and the specific method is as follows. First, the system uses the digital model of the power distribution network to automatically load the topological structure data of the current power grid. Assuming that the entire power grid contains 500 nodes, the connection relationship between the nodes is analyzed by graph theory algorithm (depth-first search) to determine the early warning point to be evaluated. For example, the substation with node number N-127 is marked as an early warning point due to its high historical failure rate, and its position is in the middle section of the medium-voltage line. Subsequently, the system extracts the devices that have a step coordination relationship with N-127 according to the electrical connection relationship, calculates the electrical distance and protection range, and selects upstream devices such as the circuit breaker with number S-45 (distance N-127 is 2.5 km, protection range covers 3 upstream nodes) and downstream devices such as the load switch with number L-89 (distance N-127 is 1.8 km, protection range covers 2 downstream nodes) to form an associated device set, totaling 5 devices. Then, the system analyzes the state of the associated device set, calls the running parameters of each device in the database, for example, the opening and closing times of S-45 circuit breaker is 120 times, exceeding the normal threshold of 100 times, and it is determined as a potential risk point, while the temperature of L-89 load switch is 45.3 degrees Celsius, which is within the normal range (upper limit 50 degrees Celsius), through the weighted risk evaluation algorithm, the overall risk index of the device set is calculated, assuming that the weight of S-45 is 0.6 and the weight of L-89 is 0.4, the comprehensive risk value is 0.48 (threshold is 0.5), it is determined that there is no urgent intervention requirement, in order to form a closed loop logic, the system associates the risk evaluation result with the dispatching strategy of the distribution network, if the subsequent monitoring finds that the opening and closing times of S-45 increases to 130 times, the risk value exceeds 0.5, then the standby line switching scheme is automatically triggered to ensure the stability of the power grid, the above process is automatically executed by the system, relying on topology analysis and algorithm evaluation, forming a complete logical chain from early warning point identification to device set risk judgment.

[0019] As shown in Figures 1-2 S103, read the real-time power data of the early warning point and each device in the associated device set, calculate the difference or ratio between the early warning point data and each associated device data as the differential coordination index.

[0020] Further, in step S103, the power parameters in the real-time data are collected through the data acquisition interface of the early warning point and the associated devices to obtain a preliminary power parameter set; According to the preliminary power parameter set, the difference calculation and ratio analysis are performed on the power parameters between the early warning point and each associated device to determine the differential index data set; Using a preset threshold range, each index in the differential index data set is compared, if a certain index exceeds the preset threshold range, it is marked as an abnormal coordination point to obtain a marked index set; From the marked index set, the related power parameters of the abnormal coordination point are extracted, and the electrical association path between the abnormal coordination point and the early warning point is analyzed in combination with the coordination relationship data in the device set to determine the abnormal association path list; Obtain each path data in the abnormal association path list, and combine the power monitoring records in the real-time data to determine whether there is a persistent fluctuation on the path, if a persistent fluctuation is detected, record the fluctuation time period and the related device identifier to obtain a fluctuation record data set; Through the fluctuation record data set, the power parameter change trend of each associated device in the fluctuation time period is analyzed, and the support vector machine model is used for classification processing to determine whether there is a potential systematic abnormality to obtain the classified trend analysis result; According to the classified trend analysis result, the related equipment identifier of the potential systematic abnormality is extracted, the propagation direction of the abnormality in the equipment set is traced by combining the electrical correlation path data, and an abnormal propagation path record is obtained.

[0021] Specifically, in step S103, in the intelligent monitoring of the power distribution network, the system reads the real-time power data of the warning point and the associated equipment set through automation technology, and calculates the differential coordination index to evaluate the operation matching between devices. Assuming that the warning point is power distribution transformer No. N-215, its rated voltage is 10 kV, the real-time monitoring voltage value is 10.2 kV, and the current value is 150 A; The system first extracts the data of the associated equipment set from the database. The upstream device is circuit breaker No. B-33, the real-time voltage is 10.3 kV, and the current is 148 A. The downstream device is fuse No. F-67, the real-time voltage is 9.9 kV, and the current is 152 A; Then, the system uses the difference calculation method to analyze the voltage difference and current difference between the warning point and each device, and obtains the voltage difference between N-215 and B-33 as 0.1 kV, the current difference as 2 A, the voltage difference between N-215 and F-67 as 0.3 kV, and the current difference as 2 A; Subsequently, the system further calculates the ratio index, taking the ratio of the voltage difference to the rated voltage as the coordination parameter. The ratio of N-215 to B-33 is 0.01, and the ratio of N-215 to F-67 is 0.03. The coordination threshold is set to 0.02, and the analysis result shows that the coordination with F-67 is low; To deepen the analysis, the system introduces the power factor comparison. The power factor of N-215 is 0.92, the power factor of F-67 is 0.88, and the difference is 0.04. Combined with the voltage ratio, the coordination is determined to be optimized. The system automatically stores the analysis result in the log and compares it with the historical data. If the difference continues to expand, the parameter calibration logic is triggered to generate optimization suggestions to ensure the operation matching between devices, forming a complete closed loop from data collection to coordination evaluation.

[0022] As shown in Figures 1-2 S104, based on the historical operation data of the warning point, the current operating condition, and the real-time data of the associated equipment set, a pre-set threshold calculation model is used to dynamically generate a warning threshold applicable to the current time.

[0023] Further, in step S104, the historical records and real-time monitoring operating state data are obtained through the data acquisition interface of the warning point and the associated equipment, and an initial operating data set is obtained; According to the initial operating data set, the running state of the warning point and the associated equipment is extracted, and the time-related running change characteristics are extracted in combination with the time identifier of the current time to determine a running state set; The running state set is combined with a preset threshold adjustment model to dynamically calculate each feature in the running state set, and a warning limit set suitable for the current time is obtained; For the warning limit set, in combination with the monitoring data monitored in real time, if the running value of a certain data exceeds the corresponding limit in the warning limit set, it is marked as an abnormal running point, and an abnormal marking data set is obtained; Through the abnormal marking data set, the related running state and associated equipment information of the abnormal running point are obtained, the correlation of the abnormal running point between devices is analyzed in combination with the record of the data source, and an abnormal correlation list is determined; According to the abnormal correlation list, a support vector machine model is used to classify and process the running state changes of the abnormal running point, to determine whether there is a systematic abnormal risk, and a classification analysis result is obtained; Through the classification analysis result, the related equipment identifier of the systematic abnormal risk is extracted, and the running state data in the historical record is combined to trace the propagation path of the abnormality between devices, and an abnormal propagation record is obtained.

[0024] Specifically, in step S104, in the intelligent monitoring system of the power distribution network, for the dynamic warning threshold generation of the warning point and its associated device set, the system first automatically calls the historical operation database, extracts the operation data of the power distribution transformer numbered T-108 in the past 30 days, finds that the average load rate is 75%, the highest load peak is 90%, and in combination with the current operation condition, the load rate is 82% is collected in real time; Subsequently, the system synchronously acquires the data of the associated device set, the upstream device is the switch cabinet numbered S-12, the real-time load rate is 80%, and the downstream device is the line terminal numbered L-45, the load rate is 85%; Then, the system calculates the historical load rate and the current working condition by a preset threshold calculation model, adopts the formula: dynamic threshold = historical average load rate x 0.6 + current load rate x 0.4, and calculates that the preliminary threshold of T-108 is 78.2%; In order to further optimize the threshold, the system introduces the influence factor of the load rate of the associated equipment, sets the influence weight of the upstream device to 0.3 and the downstream device to 0.2, calculates the comprehensive influence value: (80% x 0.3 + 85% x 0.2) = 41%, and finally adjusts the dynamic threshold to 78.2% + 41% x 0.5 = 80.25%; At the same time, the system analyzes the standard deviation of the load rate fluctuation in the historical data, which is 3.5%, sets the safety margin as 1.5 times of the standard deviation, i.e. 5.25%, and thus the final early warning threshold is set as 85.5%. If the current load rate exceeds this value, the system automatically records the anomaly and associates it with the ambient temperature data. It is found that the temperature is 35 degrees Celsius, which is beyond the normal range, and it is inferred that high temperature may cause load anomaly. The complete logical chain from extracting historical data to threshold dynamic adjustment to anomaly reason speculation is formed, ensuring the adaptability of the threshold.

[0025] As shown in Figures 1-2 S105, the real-time data of the early warning point is compared with the dynamic early warning threshold, and a comprehensive logical judgment is made in combination with the differential coordination index. If the early warning point data is out of limit and the differential coordination index simultaneously meets the preset coordination condition, an early warning signal is generated.

[0026] Further, in step S105, real-time data of the early warning point is obtained, monitoring data is continuously updated through a data acquisition interface, and each group of data is time-stamped to obtain a real-time monitoring data set with time identification; According to the real-time monitoring data set, the data is compared with the dynamic threshold generated by the preset model item by item. If a certain monitoring data exceeds the corresponding dynamic threshold range, it is marked as a potential abnormal point to obtain an abnormal marking set; For the abnormal marking set, the calculation result of the differential index is combined for secondary verification. If the differential index of the potential abnormal point simultaneously meets the preset coordination condition, the point is confirmed as an abnormal state, and an abnormal confirmation list is determined; Through the abnormal confirmation list, the historical operation records of the related early warning point are extracted, a support vector machine model is used to classify the distribution law of the abnormal state, and it is judged whether the anomaly has a persistent characteristic to obtain a classification judgment result; According to the classification judgment result, the information of the abnormal point with persistent characteristic is obtained, the comparison and analysis of real-time data and historical records are combined to determine the change trend of the abnormal point, and a trend analysis record is obtained; For the trend analysis record, the running state of the abnormal point and the surrounding equipment is matched through a data correlation tool. If the change trend is consistent with the running state of other equipment, it is marked as an associated anomaly to obtain an associated anomaly list; Using the associated anomaly list, the priority of the abnormal point is sorted according to the preset condition, and the high-priority abnormal point is filtered out through the sorting result to determine the final early warning signal triggering sequence.

[0027] Specifically, in step S105, in the intelligent monitoring system of the power distribution network, the comparison of real-time data of a certain early warning point with the dynamic early warning threshold and the comprehensive logical judgment, the system first automatically collects the operation data of the power distribution transformer numbered T-205 at the current time, and obtains the real-time voltage value of 10.8 kV. The dynamic early warning threshold generated by the previous calculation is 10.5 kV. The system compares and finds that the real-time voltage value exceeds the threshold by 0.3 kV, triggering the preliminary out-of-limit condition; Subsequently, the system calls the differential coordination index calculated in the step, assuming that the index value is 0.85, and the preset coordination condition is that the index value needs to be greater than 0.8. The system judges that 0.85 is greater than 0.8, which meets the coordination condition; Next, the system enters the comprehensive logical judgment link, combines the out-of-limit condition and the coordination index, and uses a logical algorithm: if the out-of-limit value ratio (i.e. 0.3 / 10.5=2.86%) is greater than 2% and the coordination index is greater than 0.8, the abnormality is confirmed to be established, and a warning signal is generated; The system further associates the voltage data of the upstream and downstream equipment of the early warning point, finds that the voltage of the upstream switch cabinet numbered S-18 is 10.9 kV, and the voltage of the downstream line terminal numbered L-32 is 10.7 kV. The average value of the upstream and downstream voltages is 10.8 kV, which is consistent with the real-time value of T-205. It is inferred that the abnormality may be caused by the overall voltage fluctuation of the system. A complete logical chain is formed from data acquisition, threshold comparison to coordination judgment and abnormality confirmation. The system automatically records the warning signal to the database and marks the abnormal time point as the current time, which is convenient for subsequent analysis and tracing, and ensures the comprehensiveness and accuracy of monitoring.

[0028] As shown in Figures 1-2 S106, output the warning signal and related diagnostic information, and start the corresponding control measures or prompt the operation and maintenance personnel to process according to the preset strategy.

[0029] Further, in step S106, the warning signal and related diagnostic information are obtained, the signal and information are classified and arranged through a data integration tool, and if the signal strength exceeds the preset threshold range, it is marked as a high-priority signal to obtain a classified signal list; For the classified signal list, the high-priority signal is matched and analyzed by using a preset strategy. If the matching mechanism identifies that the signal is consistent with the pre-stored abnormal state template, a corresponding control instruction is generated, and the instruction distribution sequence is determined; Through the instruction distribution sequence, the execution process of the control instruction is triggered, and the execution process is tracked in real time by using a state monitoring module to obtain feedback data of the execution state; According to the feedback data of the execution state, the response processing result of the abnormal state is analyzed, and if the response processing does not reach the preset standard, the parameter configuration of the control instruction is adjusted to determine the adjusted instruction scheme According to the adjusted instruction scheme, the corresponding operation and maintenance prompt is generated by the information generation tool, the prompt content is associated with the state feedback by using the data transmission interface, and the integrated prompt information is obtained; According to the integrated prompt information, the subsequent change of the early warning signal is tracked by the signal output module, if the change trend is still associated with the abnormal state, the state feedback data is updated, and the subsequent processing direction is determined; According to the updated state feedback data, the effect of response processing is evaluated by using the strategy triggering mechanism, the final processing record is generated by the data comparison tool, and the closed-loop result of the abnormal state is determined.

[0030] Specifically, in step S106, in the intelligent monitoring system of the power distribution network, for the output and subsequent processing of the early warning signal, the system first automatically generates the early warning signal for the power distribution equipment numbered T-312, the signal content includes the abnormal type of current overload, the real-time current value is 250 amperes, and the preset safety threshold is 220 amperes, the overload amplitude is 30 amperes, the proportion is about 13.64%, the overload severity index is calculated by algorithm as 0.75 (the calculation formula is overload proportion multiplied by weight coefficient 0.05), and the index is compared with the preset severity threshold 0.6, confirming that the abnormal level is secondary; Subsequently, the system automatically extracts relevant diagnostic information, calls the historical data analysis module, finds that the average current value of T-312 device in the past 24 hours is 210 amperes, and the fluctuation range is between 200 and 230 amperes, combined with the current value 250 amperes, it is inferred that there may be device aging or load surge; To further verify, the system associates adjacent device data, queries the current value of downstream terminal number L-45 as 240 amperes, the current value of upstream switch cabinet number S-27 as 255 amperes, and the regional average current value as 248.33 amperes, close to the real-time value of T-312, and infers that the abnormality may be related to uneven regional load distribution; Then, the system starts control measures according to the pre-set strategy, automatically adjusts the load distribution ratio of the branch where T-312 is located, calculates the adjustment coefficient as 1.01 through intelligent algorithm (load balancing coefficient = current load / regional average load), triggers load transfer instruction, and transfers part of the load to standby line L-46, the target reduces T-312 current to below 215 amperes; At the same time, the system generates a detailed diagnostic report containing abnormal time, current data, associated device state and adjustment record, automatically uploads to the cloud database, and pushes the processing suggestion to the operation and maintenance platform, the suggestion content includes checking T-312 device running environment and load access condition, forming a complete closed-loop logic from early warning output, diagnostic analysis to control execution.

[0031] The above description is only the preferred embodiment of one or more embodiments of the specification, and is not used to limit one or more embodiments of the specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of the specification should be included in the protection range of one or more embodiments of the specification.

Claims

1. A method for early warning of power data in a distribution network with differential coordination, characterized in that, include: S101. Real-time acquisition of power operation data from multiple monitoring points in the distribution network, and cleaning and standardization preprocessing of the data; S102. Based on the real-time topology of the distribution network, determine the current early warning point to be evaluated, and based on the electrical connection relationship, screen out the upstream and downstream equipment that have a graded coordination relationship with the early warning point to form a set of associated equipment. S103. Read the real-time power data of the warning point and each device in the associated device set, and calculate the difference or ratio between the data of the warning point and the data of each associated device as a differential coordination index. S104. Read the real-time power data of the warning point and each device in the associated device set, and calculate the difference or ratio between the data of the warning point and the data of each associated device as a differential coordination index. S105. Compare the real-time data of the warning point with the dynamic warning threshold, and make a comprehensive logical judgment in combination with the differential coordination index. If the data of the warning point exceeds the limit and the differential coordination index meets the preset coordination conditions, then generate a warning signal. S106. Output the warning signal and related diagnostic information, and initiate corresponding control measures or prompt maintenance personnel to handle the situation according to the preset strategy.

2. The method for early warning of power data in a distribution network based on differential coordination according to claim 1, characterized in that, Data cleaning and standardization preprocessing include: handling missing values ​​using mean interpolation and converting the data to a uniform dimension using Z-score standardization.

3. The method for early warning of power data in a distribution network based on differential coordination according to claim 1, characterized in that, The steps for screening related device sets include: identifying electrical connection paths between devices based on topology analysis tools, and marking potential risk points in conjunction with historical operating data.

4. The method for early warning of power data in a distribution network based on differential coordination according to claim 1, characterized in that, The differential coordination index includes at least one of voltage difference, current difference, ratio index, or power factor difference.

5. The method for early warning of power data in a distribution network with differential coordination according to claim 1, characterized in that, The generation of dynamic early warning thresholds includes: weighted calculation based on historical operating data and real-time data, and adjustment by introducing the influence factors of related equipment.

6. The method for early warning of power data in a distribution network with differential coordination according to claim 1, characterized in that, The comprehensive logical judgment includes: if the proportion of out-of-limit values ​​is greater than the preset proportion and the differential coordination indicator meets the coordination conditions, then an anomaly is confirmed and an early warning signal is generated.

7. The method for early warning of power data in a distribution network based on differential coordination according to claim 1, characterized in that, The steps for outputting early warning signals and diagnostic information include: matching preset strategies based on the severity index of the anomaly, generating control commands or maintenance prompts, and recording the processing to form a closed loop.

8. The method for early warning of power data in a distribution network with differential coordination according to claim 1, characterized in that, Use the support vector machine model to classify and analyze trends in outlier data.

Images