Grid-based power distribution network intelligent monitoring and management system
By constructing a grid-based intelligent monitoring system for distribution networks, dynamically adjusting the parameter acquisition frequency, and building a parameter correlation matrix and fault propagation model, the problems of resource waste and fault diagnosis misjudgment in existing distribution network monitoring systems are solved, and efficient fault handling and resource optimization are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU HANGGANG CLOUD COMPUTING DATA CENTER CO LTD
- Filing Date
- 2025-11-20
- Publication Date
- 2026-07-14
AI Technical Summary
Existing power distribution network monitoring systems struggle to dynamically adjust data acquisition strategies under abnormal scenarios, leading to wasted computing power or missed key data collection. Fault diagnosis struggles to distinguish between occasional fluctuations and system anomalies, the scope of fault propagation is unclear, and there is a lack of scientific prioritization of handling, resulting in chaotic and inefficient scheduling of operation and maintenance resources.
The grid-based intelligent monitoring system for distribution networks dynamically adjusts the parameter acquisition frequency, constructs a parameter correlation matrix, calculates the probability of coordinated anomalies and the probability of fault propagation, quantifies the priority of handling, and forms a standardized dispatch list by building a grid anomaly parameter extraction module, a fault diagnosis module, and a fault propagation identification module.
It achieves precise adaptation of parameter acquisition, improves the accuracy and efficiency of fault diagnosis, accurately defines the scope of fault propagation, optimizes the scheduling of operation and maintenance resources, and shortens the fault handling cycle.
Smart Images
Figure CN121507735B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent monitoring and management technology of power distribution networks, and specifically relates to a grid-based intelligent monitoring and management system for power distribution networks. Background Technology
[0002] Existing power distribution network monitoring systems mostly use fixed-frequency, full-domain parameter acquisition. This makes it difficult to dynamically adjust acquisition strategies under abnormal scenarios, easily leading to wasted computing power or missed key data. Fault diagnosis often relies on single-parameter judgment, making it difficult to distinguish between occasional fluctuations and system anomalies, and exhibiting poor adaptability to new types of faults. Furthermore, the lack of clear scope definition for fault propagation and the absence of scientifically quantified handling priorities result in chaotic scheduling of operation and maintenance resources and low efficiency in fault handling. Specifically, the following technical issues are identified:
[0003] The existing power distribution network parameter acquisition mode is fixed, which cannot dynamically adapt the acquisition frequency according to the degree of parameter anomaly, and it is difficult to reliably distinguish between occasional parameter fluctuations and substantial anomalies, affecting the accuracy of subsequent fault analysis.
[0004] Existing power distribution network fault diagnosis technologies are unable to establish correlations between parameters to accurately determine systemic anomalies, nor can they efficiently match fault types, especially lacking adaptability to unverified novel or compound faults.
[0005] Existing distribution network fault propagation analysis is difficult to accurately define the scope of fault impact and calculate propagation probability, and lacks a scientific method for quantifying handling priorities, resulting in inefficient scheduling of operation and maintenance resources and prolonged fault handling cycles. To address this, we propose a grid-based intelligent monitoring and management system for distribution networks. Summary of the Invention
[0006] The purpose of this invention is to provide a grid-based intelligent monitoring and management system for power distribution networks to solve the problems mentioned in the background art.
[0007] To achieve the above objectives, the present invention provides the following technical solution: a grid-based intelligent monitoring and management system for power distribution networks, comprising:
[0008] Mesh anomaly parameter extraction module: Collects multi-dimensional parameters of the mesh state of basic mesh cells, analyzes the parameter anomaly index of the mesh state parameters, adjusts the parameter collection frequency, marks potential anomaly parameters, extracts the anomaly feature values of potential anomaly parameters, calculates the feature influence values, marks and confirms anomaly parameters, and constructs a set of anomaly feature influence values.
[0009] Mesh fault diagnosis module: Constructs a dimension correlation matrix of mesh state parameters, uses confirmed abnormal parameters as trigger sources to carry out correlation verification, calculates the probability of collaborative anomalies, identifies abnormal mesh cells, analyzes the causal matching degree with fault types, and matches fault types;
[0010] Grid fault propagation identification module: Constructs a grid topology association matrix, generates anomaly-adjacent grid combinations, matches fault propagation characteristic parameters, calculates fault propagation probability by combining geographical propagation distance and distance attenuation benchmark values, filters fault propagation units, quantifies and sorts the priority values for handling, and packages the information to be pushed to the operation and maintenance end for scheduling and handling.
[0011] Preferably, the specific process for analyzing the parameter anomaly index of the grid state parameters is as follows:
[0012] Obtain the pre-divided basic grid units from the existing power information system of the distribution network area to be monitored; for each basic grid unit, collect the corresponding multi-dimensional parameters of grid status;
[0013] For each grid state parameter, the time series data of the parameter within the preset security assessment window is obtained, and the dispersion of the series is calculated using the standard deviation to obtain the parameter fluctuation dispersion.
[0014] Obtain the safe threshold range of the grid state parameter, and calculate the cumulative time that the parameter exceeds the range within the window as the duration of parameter exceeding the limit;
[0015] Based on time series data, the parameter change rate of adjacent acquisition time units is calculated and the average value is taken to obtain the parameter time series change rate;
[0016] After normalizing and dimensionlessly processing the above parameter fluctuation dispersion, parameter over-limit duration, and parameter time series change rate, the parameter anomaly index is obtained by weighting them according to a preset weight coefficient that sums to one.
[0017] Preferably, the specific process of adjusting the parameter acquisition frequency and marking potentially abnormal parameters is as follows:
[0018] If the parameter anomaly index is greater than or equal to the corresponding preset threshold, the acquisition update frequency adapted to the parameter anomaly index being greater than or equal to the corresponding preset threshold is calculated by combining the initial acquisition frequency of the grid status parameters with the preset correction coefficient.
[0019] If the parameter anomaly index is less than the corresponding preset threshold, the acquisition update frequency adapted to the parameter anomaly index being less than the corresponding preset threshold is calculated by combining the initial acquisition frequency and the preset correction coefficient.
[0020] The acquisition and update frequency determined within the current preset security assessment window will be used as the initial acquisition frequency for the grid status parameter in the next preset security assessment window.
[0021] Grid state parameters whose parameter anomaly index is greater than or equal to the corresponding preset threshold are marked as potential anomaly parameters.
[0022] Preferably, the specific process of calculating the feature influence value, marking and confirming abnormal parameters, and constructing the abnormal feature influence value set is as follows:
[0023] For each potential abnormal parameter in the multi-dimensional abnormal parameter dataset, its normal threshold range is obtained, and parameter values that exceed the range within the preset security assessment window are selected as abnormal feature values, and the original abnormal feature set is formed.
[0024] The security assessment window is divided into several sub-intervals. The closer the sub-interval is to the current time, the greater its weight. The maximum value of the abnormal feature value in each sub-interval is taken as the abnormal statistical value. The abnormal time weighted value is calculated in combination with the weight.
[0025] The parameters of the original abnormal feature set are fitted by time series linear regression, and the contribution value of the abnormal trend is calculated by the absolute value of the slope of the fitted line and the maximum abnormal feature value.
[0026] Calculate the anomaly concentration by combining the total number of samples and the number of occurrences of outlier characteristic values within the statistical window;
[0027] A comprehensive analysis of the above-mentioned abnormal time weighted values, abnormal trend contribution values, and abnormal concentration is conducted to obtain characteristic impact values;
[0028] If the feature's influence value is greater than or equal to the corresponding preset threshold, it is marked as a confirmed abnormal parameter;
[0029] The characteristic impact values of each confirmed abnormal parameter during the evaluation period are compiled to form the abnormal characteristic impact value set of the basic grid unit.
[0030] Preferably, the specific process of constructing a correlation matrix of grid state parameters to conduct correlation verification based on the confirmed abnormal parameters is as follows:
[0031] Obtain all grid state parameters within the basic grid cell, and combine all grid state parameters in pairs to obtain several grid state parameter pairs;
[0032] For each grid state parameter pair, the number of times the two anomalies are coordinated based on historical fault data is counted. The correlation degree of the grid state parameter pair is obtained by the ratio of the number of coordinated anomalies to the total number of faults.
[0033] Combine all grid state parameter pairs and their corresponding correlation degrees to construct a dimensional correlation matrix;
[0034] If the current basic grid cell detects a confirmed abnormal parameter, then each confirmed abnormal parameter is used as an independent trigger source to trigger a correlation check. The correlation check needs to count the number of grid state parameters that simultaneously meet the conditions of "being a confirmed abnormal parameter" and "the correlation degree of the parameter pair formed with the confirmed abnormal parameter that triggered the check is ≥ the corresponding preset threshold", and this number is recorded as the number of collaborative abnormal parameters.
[0035] Preferably, the specific process for calculating the cooperative anomaly probability and determining the anomalous grid cell is as follows:
[0036] The total number of strongly correlated parameters is recorded as the total number of grid state parameters whose correlation with the confirmed abnormal parameters is greater than or equal to the corresponding preset threshold.
[0037] The probability of collaborative anomalies is obtained by the ratio of the number of collaborative anomaly parameters to the total number of strongly correlated parameters.
[0038] If the probability of a collaborative anomaly is less than the corresponding preset threshold, it is determined to be an occasional anomaly, and there is no systematic anomaly in the current basic grid cell.
[0039] If the probability of collaborative anomaly is greater than or equal to a preset threshold, the current basic grid cell is determined to be abnormal and marked as an abnormal grid cell.
[0040] Preferably, the causal matching degree between the fault type and the fault type is analyzed. The specific process of matching the fault type is as follows:
[0041] Construct a fault-feature cause-effect graph. For each fault type, define the necessary associated combination of confirmed abnormal parameters, which is denoted as the fault feature combination set.
[0042] For each abnormal grid cell, extract all confirmed abnormal parameters to form a set of grid abnormal parameters;
[0043] Substitute the set of grid anomaly parameters into the fault-feature causal graph, perform causal matching with the fault feature combination set corresponding to each fault type, and calculate the causal matching degree.
[0044] The fault type with the highest causal matching degree that is greater than or equal to the corresponding preset matching threshold is selected as the identification result and recorded as the grid anomaly matching fault.
[0045] If the causal match degree with all fault types is less than the preset matching threshold, it is determined that there is no clear matching fault type, and the fallback handling mechanism is triggered.
[0046] Preferably, the specific process for calculating the fault propagation probability and screening fault propagation units is as follows:
[0047] A grid topology correlation matrix is constructed based on the geographical boundaries and electrical connections of the distribution network area to be monitored. For any two basic grid cells, if they are adjacent, the matrix element value is 1, otherwise it is 0.
[0048] For each anomalous mesh cell, obtain all basic mesh cells with an element value of 1 from the mesh topology correlation matrix, and combine them with the anomalous mesh cell to form several anomalous-adjacent mesh combinations;
[0049] Construct a fault type-fault propagation characteristic parameter mapping table, in which each fault type corresponds to a specific and quantifiable fault propagation characteristic parameter;
[0050] Based on the matching fault corresponding to the current abnormal mesh cell, the corresponding fault propagation characteristic parameters are matched;
[0051] For each pair of anomaly-adjacent grid combinations, the Euclidean distance between the center coordinates of the two basic grid cells is obtained to obtain the geographic transmission distance;
[0052] To match faults, multiple geographic conduction distance intervals and corresponding preset distance attenuation benchmark values are set, and the corresponding distance attenuation benchmark value is output based on the geographic conduction distance of the current anomaly-adjacent grid combination.
[0053] The fault propagation probability is obtained by comprehensively analyzing the fault propagation characteristic parameters of the current anomaly-adjacent grid combination, the geographical propagation distance, and the distance attenuation benchmark value.
[0054] If the probability of fault propagation is greater than or equal to the corresponding preset threshold, then the adjacent basic grid cells in the current anomaly-adjacent grid combination will be marked as fault propagation cells.
[0055] Preferably, the specific process of quantifying and prioritizing the handling of cases, packaging the information, and pushing it to the operations and maintenance end for scheduling and handling is as follows:
[0056] Organize all fault propagation cells caused by the current abnormal mesh cell to obtain a fault propagation list;
[0057] For each fault propagation unit, a comprehensive analysis is conducted based on the corresponding fault hazard propagation degree, grid comprehensive value degree, and fault propagation probability to obtain the priority value for handling.
[0058] By sorting each fault propagation unit in the fault propagation list in descending order of its handling priority value, a fault propagation unit handling priority sequence list is obtained.
[0059] The abnormal information of the abnormal grid unit and the corresponding fault propagation unit handling priority sequence list are packaged and sent to the operation and maintenance management center and the on-site operation and maintenance terminal to perform operation and maintenance scheduling processing of the abnormal grid unit.
[0060] Compared with the prior art, the beneficial effects of the present invention are:
[0061] (1) The grid-based intelligent monitoring and management system for distribution networks quantifies the degree of anomalies by constructing a parameter anomaly index model, with parameter fluctuation dispersion, duration of exceeding limits, and time-series change rate as the core dimensions. It dynamically adjusts the collection frequency—intensifying collection to capture key data when the anomaly risk is high, and maintaining a regular frequency to save computing power when the risk is low, thus avoiding resource waste or data omission caused by fixed collection. At the same time, by combining the anomaly time weighting value, trend contribution value, and concentration multi-dimensional calculation characteristic influence value, it can effectively distinguish between accidental parameter fluctuations and substantive anomalies, providing an accurate dataset of confirmed abnormal parameters for subsequent fault diagnosis, and improving the accuracy of analysis from the source.
[0062] (2) The grid-based intelligent monitoring and management system for distribution networks constructs a grid state parameter dimension correlation matrix based on historical fault data, accurately determines systemic anomalies through collaborative anomaly probability, and solves the problem of misjudgment caused by single parameter judgment; it realizes rapid matching of anomaly parameter set and fault type based on fault-feature causal graph, which greatly shortens the fault location time; for new or compound faults not included in the data, it continuously expands the fault identification range through a fallback mechanism, which significantly improves the system's adaptability to complex fault scenarios.
[0063] (3) The grid-based intelligent monitoring and management system for distribution networks constructs a grid association matrix based on geographical and electrical topology relationships to delineate the potential impact range of fault propagation and avoid efficiency losses caused by indiscriminate investigation. It accurately calculates the fault propagation probability and screens out real propagation units by combining the inherent attributes of equipment, dynamic status, risk accumulation and other differentiated parameter values and geographical distance attenuation characteristics. At the same time, it quantifies the handling priority by comprehensively considering the fault hazard propagation degree, grid comprehensive value degree and propagation probability, and forms a standardized scheduling list to be pushed to the operation and maintenance end, so as to realize the priority handling of "high hazard, high value and high urgency" units, coordinate resource scheduling and on-site execution, and significantly shorten the fault handling cycle. Attached Figure Description
[0064] Figure 1 This is a flowchart of the present invention. Detailed Implementation
[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0066] Example 1;
[0067] Please see Figure 1The present invention provides a grid-based intelligent monitoring and management system for distribution networks, including: a grid anomaly parameter extraction module, a grid fault diagnosis module, and a grid fault propagation identification module;
[0068] The grid anomaly parameter extraction module collects multi-dimensional parameters of the grid state of basic grid cells, analyzes the parameter anomaly index of the grid state parameters, adjusts the parameter collection frequency, marks potential anomaly parameters, extracts the anomaly feature values of potential anomaly parameters, calculates the feature influence values, marks and confirms anomaly parameters, and constructs a set of anomaly feature influence values. The specific process is as follows:
[0069] From the existing power information system of the distribution network area to be monitored, obtain the divided basic grid units. Each basic grid unit is an independent management and control unit with a clear geographical boundary, containing core power facilities such as distribution lines, distribution transformers, and switching equipment in a specific section, and corresponding to a unique power supply user group. Its electrical topology and geographical boundary information are all derived from the standardized data interface of the above system.
[0070] For each basic grid cell, collect its corresponding multi-dimensional grid state parameters, including:
[0071] Electrical parameters: measured values of main line voltage, measured values of main line current, measured values of total active power of distribution transformers, and measured values of line loss within the grid.
[0072] Non-electrical parameters: measured values of distribution transformer winding temperature, measured values of ice thickness on overhead lines, measured values of humidity in the installation environment of switchgear, and measured values of vibration amplitude of distribution transformers;
[0073] For each grid state parameter, time series data of that grid state parameter within a preset security assessment window are obtained, and the dispersion of the series is calculated using the standard deviation formula to obtain the parameter fluctuation dispersion LS.
[0074] Obtain the safety threshold range of the grid state parameter, and count the cumulative time that the grid state parameter exceeds the corresponding safety threshold range within the preset safety assessment window. Record the cumulative time as the parameter over-limit duration LT.
[0075] Based on the time series data of grid state parameters, the rate of change of grid state parameters in adjacent acquisition time units is calculated, and the average value is taken to obtain the time series change rate CB of the parameters.
[0076] After normalizing and removing the dimensionless processing of the parameter fluctuation dispersion LS, parameter over-limit duration LT, and parameter time series change rate CB corresponding to the grid state parameters within the current preset safety assessment window, the parameter anomaly index R is obtained using the formula: R=LS×a1+LT×a2+CB×a3, where a1, a2, and a3 are preset weight coefficients, and a1+a2+a3=1.
[0077] The preset parameter is the abnormal index threshold R0. If R ≥ R0, the formula is used: The acquisition update frequency is obtained. ,in, The initial sampling frequency for the grid state parameters. This is the preset correction factor;
[0078] If R < R0, use the formula: The acquisition update frequency is obtained. ;
[0079] Obtain the acquisition and update frequency of the grid status parameter within the current preset security assessment window, and use this acquisition frequency as the initial acquisition frequency of the grid status parameter in the next preset security assessment window;
[0080] Grid state parameters whose parameter anomaly index is greater than or equal to the corresponding preset threshold are marked as potential anomaly parameters;
[0081] The real-time collected data of each potential abnormal parameter within the current preset security assessment window are integrated to form a multi-dimensional abnormal parameter dataset of the basic grid unit.
[0082] For each potential anomalous parameter in the multidimensional anomalous parameter dataset, obtain its corresponding normal threshold range. ;in, This represents the lower limit of the normal threshold range for potentially abnormal parameters. This represents the upper limit of the normal threshold range for potential abnormal parameters;
[0083] Filter within the preset security assessment window that meet the requirements The parameter values are denoted as abnormal feature values and organized into the original abnormal feature set. Where i is the potential anomaly parameter label, and j is the time-series sampling point number. Let be the real-time value of the i-th type of potential anomaly parameter at the j-th sampling point;
[0084] The preset security assessment window is divided into n sub-intervals, and the weight of the k-th sub-interval is... ( In this context, the closer a sub-region is to the current time, the greater its corresponding weight, with the aim of highlighting the impact of recent anomalies.
[0085] Based on the abnormal feature set, the maximum value of the abnormal feature value in each sub-interval is obtained and recorded as the abnormal statistical value. By analyzing the outlier statistics corresponding to each sub-interval Weights of the corresponding sub-intervals Substitute into the formula:
[0086] We obtain the abnormal time weighted value W;
[0087] in, This represents the outlier statistics for the k-th sub-interval;
[0088] The parameter values in the original anomaly feature set are fitted with a straight line based on linear regression over time. The formula for the straight line fitting is: Where t is the time variable, a is the slope of the line, and b is the intercept. These are fitted values for abnormal times;
[0089] Obtain the absolute value of the slope |a| of the straight line fitting formula and denote it as the trend rate;
[0090] Using the formula: The contribution value of abnormal trends is obtained. ;
[0091] in, It represents the maximum value of the abnormal feature value in the original abnormal feature set;
[0092] Count the total number of samples N within the preset security assessment window, and simultaneously count the number of occurrences M of abnormal feature values, using the formula: The abnormal concentration C is obtained;
[0093] By weighting the potential anomaly parameters within a preset security assessment window by the anomaly time weighting value W and the anomaly trend contribution value... And after normalizing and dimensionlessly processing the abnormal concentration C,
[0094] Using the formula: , obtain the characteristic influence value ;
[0095] in, ;
[0096] A preset threshold for the influence value of a feature is set. If the influence value of a potential abnormal parameter is greater than the corresponding preset threshold, the potential abnormal parameter is determined to be a confirmed abnormal parameter.
[0097] The characteristic influence values of each confirmed abnormal parameter of the basic grid cell during the evaluation period are organized to obtain the abnormal characteristic influence value set of the basic grid cell.
[0098] It should be noted that:
[0099] Achieving precise adaptation and resource optimization in parameter acquisition: This module simultaneously covers multi-dimensional data acquisition of both electrical and non-electrical parameters, comprehensively capturing the operating status of grid cells; it dynamically adjusts the acquisition frequency through parameter anomaly index, encrypting acquisition when the anomaly risk is high to ensure data integrity, and maintaining a normal frequency when the risk is low to save computing power, avoiding resource waste caused by blind acquisition, and balancing monitoring accuracy and operating efficiency.
[0100] Ensuring the comprehensiveness and accuracy of anomaly identification: This module quantifies the impact of potential anomalies from multiple dimensions by filtering anomaly feature values, calculating time-weighted values (highlighting the impact of recent anomalies), anomaly trend contribution values (capturing patterns of change), and anomaly concentration (reflecting the density of anomalies). It effectively distinguishes between random fluctuations and substantive anomalies, ensures the reliability of confirmed anomaly parameters, provides accurate targets for subsequent fault diagnosis, and reduces the risk of misjudgment and missed judgment.
[0101] Building a standardized data foundation to support end-to-end collaboration: The abnormal feature impact value set ultimately formed by this module integrates key information for confirming abnormal parameters. It provides unified and standardized data input for the correlation verification and causal matching of the grid fault diagnosis module, as well as the transmission probability calculation and handling priority ranking of the fault transmission identification module. This ensures the smooth connection and efficient operation of all links in the entire monitoring system and improves the systematicness and pertinence of power distribution network anomaly handling.
[0102] The grid fault diagnosis module constructs a dimensional correlation matrix of grid state parameters, uses confirmed abnormal parameters as trigger sources to conduct correlation verification, calculates the probability of collaborative anomalies, identifies abnormal grid cells based on the collaborative anomaly probability, analyzes the causal matching degree with fault types, and matches fault types. The specific process is as follows:
[0103] Obtain all grid state parameters within the basic grid cell, and combine all grid state parameters in pairs to obtain several grid state parameter pairs;
[0104] For each grid state parameter pair, the number of times the two anomalies are coordinated is counted based on historical fault data, and the correlation degree of the grid state parameter pair is obtained by taking the number of coordinated anomalies as a percentage of the total number of faults.
[0105] Combine all grid state parameter pairs and their corresponding correlation degrees to construct a dimensional correlation matrix;
[0106] If an abnormal parameter is detected in the current basic grid cell, a correlation check will be triggered once for each abnormal parameter as an independent trigger source. The specific process of the correlation check is as follows:
[0107] The number of grid state parameters that simultaneously satisfy the following two conditions is denoted as the number of cooperative anomaly parameters (CS):
[0108] The grid status parameters are used to confirm the anomaly.
[0109] The correlation between the grid state parameter pair, consisting of the grid state parameter and the confirmed abnormal parameter of the trigger verification, is greater than or equal to the corresponding preset threshold.
[0110] Count all grid state parameter pairs that are associated with confirmed abnormal parameters and whose correlation is greater than or equal to the corresponding preset threshold grid state parameter, and record them as the total number of strongly correlated parameters K;
[0111] Using the formula: P=Cs / K, the probability of collaborative anomaly P can be obtained;
[0112] Preset collaborative anomaly probability threshold ,like If the anomaly is not found, it is considered an accidental anomaly; there is no systematic anomaly in the current basic grid cell.
[0113] like If the current basic grid cell is found to be abnormal, it will be marked as an abnormal grid cell.
[0114] Construct a fault-feature cause-effect graph. For each fault type in the library, define the necessary associated combination of confirmed abnormal parameters, denoted as the fault feature combination set SF.
[0115] For each abnormal mesh cell, extract all confirmed abnormal parameters to form a mesh abnormal parameter set S;
[0116] Substitute the grid anomaly parameter set S into the fault-feature causal graph and perform causal matching with the fault feature combination set corresponding to each fault type. Calculate the causal matching degree Q using the following formula:
[0117]
[0118] in, The number of elements in the set;
[0119] The fault type with the highest causal matching degree and a causal matching degree greater than or equal to the corresponding preset matching threshold is selected as the identification result and recorded as a grid anomaly matching fault.
[0120] If the causal matching degree with all fault types is less than the preset matching threshold, it is determined that there is no clear matching fault type, triggering the fallback handling mechanism, namely: generating an abnormal parameter combination early warning work order and pushing it to the operation and maintenance backend (marking all confirmed abnormal parameters and corresponding characteristic impact values); initiating the manual review process, combining equipment operation logs and on-site environmental data to determine whether it is a new fault or a compound fault; if it is a new fault type, updating the fault-feature causal graph and incorporating its fault feature combination set into the graph.
[0121] It should be noted that:
[0122] To accurately distinguish between systematic and random anomalies and reduce the risk of misjudgment: First, construct a correlation matrix of grid state parameters based on historical fault data to clarify the correlation strength between parameters; then, conduct correlation verification using confirmed abnormal parameters as trigger sources, and obtain the probability of coordinated anomalies by calculating the ratio of "the number of coordinated abnormal parameters to the total number of strongly correlated parameters"—only when the probability of coordinated anomalies reaches a threshold is it determined that there is a systematic anomaly in the grid cell, rather than a random anomaly caused by a single parameter fluctuation, thus avoiding misjudging isolated parameter anomalies as system faults and improving the reliability of anomaly judgment;
[0123] Accurate matching of fault types is achieved by relying on causal graphs, reducing troubleshooting costs: The constructed fault-feature causal graph pre-sets "inevitably associated confirmed abnormal parameter combinations" for each fault type. During diagnosis, by calculating the matching degree between the set of abnormal parameters in the grid and the set of fault feature combinations, the fault type with the highest matching degree can be directly located. There is no need for maintenance personnel to check potential fault points one by one, which greatly shortens the fault location time and reduces the complexity and cost of manual troubleshooting.
[0124] By employing a fallback mechanism and graph iteration, the system's adaptability to complex faults is enhanced: For new or compound faults not yet included in the database, the module triggers a fallback mechanism of "early warning work order push + manual review" to ensure effective response even when there is no clear matching fault type; at the same time, feature combinations of new fault types are incorporated into the causal graph, enabling dynamic iteration of the graph, allowing the system to continuously expand the fault identification range as operational experience accumulates, and adapt to the complex and ever-changing fault scenarios of the distribution network in the long term.
[0125] The grid fault propagation identification module constructs a grid topology association matrix, generates anomaly-adjacent grid combinations, matches fault propagation characteristic parameters, calculates fault propagation probability based on geographical propagation distance and distance attenuation benchmarks, filters fault propagation units, quantifies and prioritizes handling, packages the information, and pushes it to the operations and maintenance end for scheduling and handling. The specific process is as follows:
[0126] Based on the geographical boundaries and electrical connections of the distribution network area to be monitored, a grid topology correlation matrix is constructed. For any two basic grid cells in the grid topology correlation matrix, if the two basic grid cells are adjacent grid cells, the matrix element value is assigned = 1, otherwise the matrix element value is assigned = 0.
[0127] For each abnormal grid cell, obtain all the basic grid cells with matrix element values of 1 from the grid topology correlation matrix, and combine them with the current abnormal grid cell to obtain several abnormal-adjacent grid combinations;
[0128] Construct a fault type-fault propagation characteristic parameter mapping table. For each fault type, there is a clear and quantifiable fault propagation characteristic parameter in the table. This parameter is a core physical or environmental parameter that is directly related to the corresponding fault propagation law. All of them are derived from existing data systems of the distribution network (such as equipment management system, meteorological monitoring system, load monitoring system, etc.) and are existing data that can be directly retrieved.
[0129] By substituting the matching fault corresponding to the current abnormal mesh cell into the fault type-fault propagation characteristic parameter mapping table, the corresponding fault propagation characteristic parameter is matched.
[0130] Furthermore, the values for the fault propagation characteristic parameters are determined as follows:
[0131] If the parameter is an inherent property of the equipment (such as the impedance of the connecting line): the value is a fixed design value of the parameter (not a statistical value, derived from the equipment ledger, and does not fluctuate over time).
[0132] If the parameter is dynamically changing state-type data (such as environmental similarity, current carrying capacity margin): the value is the average value within the preset safety assessment window;
[0133] If the parameter is risk accumulation data (such as parameters related to the equipment's service life): the value is the current real-time statistical value (reflecting the cumulative status up to the assessment time);
[0134] By substituting the fault type corresponding to the current abnormal mesh cell into the fault type-fault propagation characteristic parameter mapping table, the corresponding fault propagation characteristic parameter is matched.
[0135] For each pair of anomalies-adjacent grid combinations, the Euclidean distance between the center coordinates of the two basic grid cells is obtained, and the geographic transmission distance DL is obtained.
[0136] For the matching faults corresponding to the current abnormal grid cells, several geographic transmission distance intervals are set, and each geographic transmission distance interval is preset to correspond to a distance attenuation benchmark value.
[0137] Match the geographic transmission distance of the current anomaly-adjacent grid combination with all geographic transmission distance intervals of the corresponding fault type, and output the corresponding distance attenuation benchmark value GS.
[0138] After normalizing and dimensionlessly processing the fault propagation characteristic parameters GT, geographic propagation distance DL, and distance attenuation baseline value GS of the current anomaly-adjacent grid combination, the following formula is used: The fault propagation probability GP is obtained, where γ is the preset fault influence coefficient.
[0139] A preset fault propagation probability threshold is set. If the fault propagation probability of the current anomaly-adjacent grid combination is greater than or equal to the corresponding preset threshold, then the adjacent basic grid cells in the current anomaly-adjacent grid combination are marked as fault propagation cells.
[0140] Organize all fault propagation cells caused by the current abnormal mesh cell to obtain a fault propagation list;
[0141] For each fault propagation unit, the formula is used: The priority value for disposal is YXZ;
[0142] Where r1, r2, and r3 are preset weight coefficients, and r1 + r2 + r3 = 1;
[0143] GD is the fault hazard propagation degree: it is obtained by weighting the inherent hazard level of the fault with the fault propagation probability of the abnormal grid cell;
[0144] WZ stands for Grid Comprehensive Value: It is obtained by weighting the user level and the grid load density normalized value with a fixed weight; the user level is a directly obtained parameter derived from the power supply guarantee priority classification of the user file system; the grid load density normalized value is a calculation parameter obtained by the ratio of the grid's real-time maximum load to the distribution network grid's maximum design load;
[0145] GP represents the fault propagation probability between fault propagation cells and abnormal mesh cells;
[0146] When calculating the priority value, all three parameters were normalized and dimensionless.
[0147] By sorting each fault propagation unit in the fault propagation list in descending order of its handling priority value, a fault propagation unit handling priority sequence list is obtained.
[0148] By packaging and sending the abnormal information (identifier, fault type, confirmed abnormal parameters and characteristic impact values) of abnormal grid units and the corresponding fault propagation unit handling priority sequence list to the operation and maintenance management center and the on-site operation and maintenance terminal, the abnormal grid units are scheduled for operation and maintenance. The operation and maintenance management center can use this information to coordinate and schedule operation and maintenance resources and issue remote control commands, while the on-site operation and maintenance terminal can use this information to carry out precise handling of the corresponding fault propagation units and provide feedback on the execution results.
[0149] It should be noted that:
[0150] Accurately delineate the fault propagation range based on the topology matrix to avoid ineffective monitoring: Construct a grid topology association matrix based on geographical boundaries and electrical connection relationships. Directly distinguish between adjacent and non-adjacent grids through matrix element values. Then generate "abnormal-adjacent grid combinations" for abnormal grid cells. Focus only on the cells directly associated with abnormal grids to analyze the propagation risk, rather than indiscriminately checking all grids. This greatly reduces the propagation analysis range and reduces ineffective computing power consumption and monitoring costs.
[0151] Multi-dimensional adaptation of conduction characteristic parameters improves the accuracy of conduction probability calculation: Differentiated value methods are designed for different types of fault conduction characteristic parameters (inherent equipment attributes, dynamic state type, and risk accumulation type): fixed design values are used for inherent equipment attributes to ensure stability, window average values are used for dynamic data to reflect real-time status, and real-time statistical values are used for risk accumulation data to reflect the cumulative effect; combined with the distance attenuation characteristics of geographical conduction distance, the conduction probability is calculated after multi-dimensional data normalization, so that the probability results are more in line with the actual fault conduction law, providing an accurate basis for screening real conduction units;
[0152] Quantify disposal priorities to achieve optimal scheduling of operation and maintenance resources: When calculating disposal priority values, three core factors are comprehensively considered: "fault severity (fault severity propagation degree), grid importance (grid comprehensive value degree), and propagation urgency (fault propagation probability)". The difference in units is eliminated through normalization. Finally, a list is formed by prioritizing the grid units. Operation and maintenance personnel can directly prioritize the disposal of grid units with high hazard, high value and rapid propagation, avoid resource misallocation (such as prioritizing the disposal of low-value grids and delaying the disposal of high-value grids) and improve fault management efficiency.
[0153] Full-chain information synchronization supports efficient closed-loop handling: Key information of abnormal grids (identifier, fault type, confirmed abnormal parameters) and handling priority list are packaged and pushed to the operation and maintenance control center and field terminals. The control center can coordinate and dispatch personnel, materials and other resources accordingly, and the field terminals can directly obtain accurate handling objects and background information, realizing seamless information connection between back-end coordination and on-site execution, forming a closed-loop management from transmission and identification to handling implementation, and shortening the fault handling cycle.
[0154] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A grid-based intelligent monitoring and management system for power distribution networks, characterized in that: include: Mesh anomaly parameter extraction module: Collects multi-dimensional parameters of the mesh state of basic mesh cells, analyzes the parameter anomaly index of the mesh state parameters, adjusts the parameter collection frequency, marks potential anomaly parameters, extracts the anomaly feature values of potential anomaly parameters, calculates the feature influence values, marks and confirms anomaly parameters, and constructs a set of anomaly feature influence values. A basic grid unit is an independent management and control unit with a clear geographical boundary, containing a specific section of power distribution lines, power distribution transformers, and switching equipment, and corresponding to a unique group of power supply users; When calculating the feature impact value, the three dimensions of abnormal time weighting value, abnormal trend contribution value and abnormal concentration are integrated, and the result is obtained by normalization and weighting according to preset weights. The abnormal time weighting value is calculated by statistically analyzing the maximum value of the sub-interval with high weight in the most recent time. Mesh fault diagnosis module: Constructs a dimension correlation matrix of mesh state parameters, uses confirmed abnormal parameters as trigger sources to carry out correlation verification, calculates the probability of collaborative anomalies, identifies abnormal mesh cells, analyzes the causal matching degree with fault types, and matches fault types; If the causal matching degree with all fault types is less than the preset matching threshold, the fallback mechanism is triggered. After manual review and confirmation of the new fault type, the fault-feature causal graph is dynamically updated. Grid Fault Propagation Identification Module: Constructs a grid topology association matrix, generates anomaly-adjacent grid combinations, matches fault propagation characteristic parameters, calculates fault propagation probability by combining geographical propagation distance and distance attenuation benchmark values, filters fault propagation units, quantifies and sorts the priority values for handling, and packages the information to push to the operation and maintenance end for scheduling and handling. When matching fault propagation characteristic parameters, differentiated value selection methods are adopted for three types of parameters: inherent equipment attributes, dynamic state type, and risk accumulation type. The probability of fault propagation is calculated using an exponential decay model based on geographical distance; The priority value for handling is obtained by integrating three dimensions: fault hazard propagation degree, grid comprehensive value degree, and fault propagation probability, and then weighting them according to preset weights after normalization. The overall value of the grid system is incorporated into the priority factors for user power supply security. The specific process for calculating the fault propagation probability and screening fault propagation units is as follows: A grid topology correlation matrix is constructed based on the geographical boundaries and electrical connections of the distribution network area to be monitored. For any two basic grid cells, if they are adjacent, the matrix element value is 1, otherwise it is 0. For each anomalous mesh cell, obtain all basic mesh cells with an element value of 1 from the mesh topology correlation matrix, and combine them with the anomalous mesh cell to form several anomalous-adjacent mesh combinations; Construct a fault type-fault propagation characteristic parameter mapping table, in which each fault type corresponds to a specific and quantifiable fault propagation characteristic parameter; Based on the matching fault corresponding to the current abnormal mesh cell, the corresponding fault propagation characteristic parameters are matched; For each pair of anomaly-adjacent grid combinations, the Euclidean distance between the center coordinates of the two basic grid cells is obtained to obtain the geographic transmission distance; To match faults, multiple geographic conduction distance intervals and corresponding preset distance attenuation benchmark values are set, and the corresponding distance attenuation benchmark value is output based on the geographic conduction distance of the current anomaly-adjacent grid combination. The fault propagation probability is obtained by comprehensively analyzing the fault propagation characteristic parameters of the current anomaly-adjacent grid combination, the geographical propagation distance, and the distance attenuation benchmark value. If the probability of fault propagation is greater than or equal to the corresponding preset threshold, then the adjacent basic grid cells in the current anomaly-adjacent grid combination will be marked as fault propagation cells.
2. The grid-based intelligent monitoring and management system for power distribution networks according to claim 1, characterized in that: The specific process for analyzing the parametric anomaly index of grid state parameters is as follows: Obtain the pre-divided basic grid units from the existing power information system of the distribution network area to be monitored; for each basic grid unit, collect the corresponding multi-dimensional parameters of grid status; For each grid state parameter, the time series data of the parameter within the preset security assessment window is obtained, and the dispersion of the series is calculated using the standard deviation to obtain the parameter fluctuation dispersion. Obtain the safe threshold range of the grid state parameter, and calculate the cumulative time that the parameter exceeds the range within the window as the duration of parameter exceeding the limit; Based on time series data, the parameter change rate of adjacent acquisition time units is calculated and the average value is taken to obtain the parameter time series change rate; After normalizing and dimensionlessly processing the above parameter fluctuation dispersion, parameter over-limit duration, and parameter time series change rate, the parameter anomaly index is obtained by weighting them according to a preset weight coefficient that sums to one.
3. The grid-based intelligent monitoring and management system for power distribution networks according to claim 2, characterized in that: The specific process of adjusting the parameter acquisition frequency and marking potentially abnormal parameters is as follows: If the parameter anomaly index is greater than or equal to the corresponding preset threshold, the acquisition update frequency adapted to the parameter anomaly index being greater than or equal to the corresponding preset threshold is calculated by combining the initial acquisition frequency of the grid status parameters with the preset correction coefficient. If the parameter anomaly index is less than the corresponding preset threshold, the acquisition update frequency adapted to the parameter anomaly index being less than the corresponding preset threshold is calculated by combining the initial acquisition frequency and the preset correction coefficient. The acquisition and update frequency determined within the current preset security assessment window will be used as the initial acquisition frequency for the grid status parameter in the next preset security assessment window. Grid state parameters whose parameter anomaly index is greater than or equal to the corresponding preset threshold are marked as potential anomaly parameters.
4. The grid-based intelligent monitoring and management system for power distribution networks according to claim 3, characterized in that: The specific process of calculating the influence values of features, identifying and confirming anomalous parameters, and constructing a set of anomalous feature influence values is as follows: For each potential abnormal parameter in the multi-dimensional abnormal parameter dataset, its normal threshold range is obtained, and parameter values that exceed the range within the preset security assessment window are selected as abnormal feature values, and the original abnormal feature set is formed. The security assessment window is divided into several sub-intervals. The closer the sub-interval is to the current time, the greater its weight. The maximum value of the abnormal feature value in each sub-interval is taken as the abnormal statistical value. The abnormal time weighted value is calculated in combination with the weight. The parameters of the original abnormal feature set are fitted by time series linear regression, and the contribution value of the abnormal trend is calculated by the absolute value of the slope of the fitted line and the maximum abnormal feature value. Calculate the anomaly concentration by combining the total number of samples and the number of occurrences of outlier characteristic values within the statistical window; A comprehensive analysis of the above-mentioned abnormal time weighted values, abnormal trend contribution values, and abnormal concentration is conducted to obtain characteristic impact values; If the feature's influence value is greater than or equal to the corresponding preset threshold, it is marked as a confirmed abnormal parameter; The characteristic impact values of each confirmed abnormal parameter during the evaluation period are compiled to form the abnormal characteristic impact value set of the basic grid unit.
5. The grid-based intelligent monitoring and management system for power distribution networks according to claim 4, characterized in that: The specific process of constructing the correlation matrix of grid state parameters and conducting correlation verification based on the confirmed abnormal parameters is as follows: Obtain all grid state parameters within the basic grid cell, and combine all grid state parameters in pairs to obtain several grid state parameter pairs; For each grid state parameter pair, the number of times the two anomalies are coordinated based on historical fault data is counted. The correlation degree of the grid state parameter pair is obtained by the ratio of the number of coordinated anomalies to the total number of faults. Combine all grid state parameter pairs and their corresponding correlation degrees to construct a dimensional correlation matrix; If the current basic grid cell detects a confirmed abnormal parameter, then each confirmed abnormal parameter is used as an independent trigger source to trigger a correlation check. The correlation check needs to count the number of grid state parameters that simultaneously satisfy the correlation degree of the parameter pair consisting of the confirmed abnormal parameter and the confirmed abnormal parameter that triggers the check ≥ the corresponding preset threshold, and record it as the number of collaborative abnormal parameters.
6. The grid-based intelligent monitoring and management system for power distribution networks according to claim 5, characterized in that: The specific process for calculating the cooperative anomaly probability and determining the anomalous mesh cell is as follows: The total number of strongly correlated parameters is recorded as the total number of grid state parameters whose correlation with the confirmed abnormal parameters is greater than or equal to the corresponding preset threshold. The probability of collaborative anomalies is obtained by the ratio of the number of collaborative anomaly parameters to the total number of strongly correlated parameters. If the probability of a collaborative anomaly is less than the corresponding preset threshold, it is determined to be an occasional anomaly, and there is no systematic anomaly in the current basic grid cell. If the probability of collaborative anomaly is greater than or equal to the corresponding preset threshold, the current basic grid cell is determined to be abnormal and marked as an abnormal grid cell.
7. The grid-based intelligent monitoring and management system for power distribution networks according to claim 6, characterized in that: The analysis of causal matching degree between fault types and fault types, and the specific process of matching fault types are as follows: Construct a fault-feature cause-effect graph. For each fault type, define the necessary associated combination of confirmed abnormal parameters, which is denoted as the fault feature combination set. For each abnormal grid cell, extract all confirmed abnormal parameters to form a set of grid abnormal parameters; Substitute the set of grid anomaly parameters into the fault-feature causal graph, perform causal matching with the fault feature combination set corresponding to each fault type, and calculate the causal matching degree. The fault type with the highest causal matching degree that is greater than or equal to the corresponding preset matching threshold is selected as the identification result and recorded as the grid anomaly matching fault. If the causal match degree with all fault types is less than the preset matching threshold, it is determined that there is no clear matching fault type, and the fallback handling mechanism is triggered.
8. The grid-based intelligent monitoring and management system for power distribution networks according to claim 7, characterized in that: The specific process of quantifying and prioritizing disposal, packaging the information, and pushing it to the operations and maintenance end for scheduling and disposal is as follows: Organize all fault propagation cells caused by the current abnormal mesh cell to obtain a fault propagation list; For each fault propagation unit, a comprehensive analysis is conducted based on the corresponding fault hazard propagation degree, grid comprehensive value degree, and fault propagation probability to obtain the priority value for handling. Sort each fault propagation unit in the fault propagation list in descending order of its handling priority value to obtain a fault propagation unit handling priority sequence list. The abnormal information of the abnormal grid unit and the corresponding fault propagation unit handling priority sequence list are packaged and sent to the operation and maintenance control center and the on-site operation and maintenance terminal for operation and maintenance scheduling of the abnormal grid unit.