Fault analysis method and system for intelligent power distribution network

Abstract

The invention discloses a fault analysis method and system for an intelligent power distribution network, and relates to the technical field of power distribution network fault diagnosis, and the key points of the technical scheme comprise the following steps: determining a target region which needs fault analysis and comprises a target detection node in the intelligent power distribution network, collecting network topology data, environment parameters and historical fault data of the target area in the historical period; extracting a target detection node as well as a first adjacent node and a second adjacent node which are distributed adjacent to the target detection node from the target area; according to the network topology data, the environmental parameters and the historical fault data, extracting a regional association factor influencing the fault diffusion range of the first adjacent node and the second adjacent node; and predicting the operation condition of the target detection node in the current period according to the operation fluctuation condition of the target detection node in different historical periods to obtain a current state prediction value. The method has the effect of accurately analyzing the operation state and fault probability of the target detection node and the adjacent nodes.

Description

Technical Field

[0001] This invention relates to the field of power distribution network fault diagnosis technology, and in particular to a fault analysis method and system for smart power distribution networks. Background Technology

[0002] As the intelligence level of power grids continues to improve, the scale of smart distribution networks is expanding, their topologies are becoming increasingly complex, and the number of nodes is growing exponentially. Furthermore, the integration of various new types of equipment and distributed energy sources makes network operation more volatile. Simultaneously, environmental factors are having a more significant impact on distribution networks; frequent extreme weather events and surrounding electromagnetic interference increase the uncertainty of equipment failures. Under these circumstances, traditional fault analysis methods, due to their limited data collection methods and lack of in-depth historical data mining and dynamic correlation analysis, struggle to accurately predict the occurrence and spread of faults.

[0003] Most existing fault analysis technologies focus only on the operating parameters of individual devices, failing to fully consider the combined impact of network topology changes and environmental disturbances, thus unable to comprehensively reflect the actual operating status of smart distribution networks. For example, when dealing with fault associations between adjacent nodes, traditional methods lack combined analysis of historical fault propagation patterns and current network conditions, leading to inaccurate fault prediction probabilities. Furthermore, in determining the fault impact range, they struggle to dynamically adapt to real-time changes in network topology and the environment, making it impossible to accurately define the boundaries of the fault impact domain. In addition, for assessing the risk of cascading faults, traditional technologies lack quantitative assessment models and scientific risk indicators, making it difficult to provide early warnings of fault chain reactions, thus preventing maintenance personnel from taking timely and effective measures. Summary of the Invention

[0004] This invention provides a fault analysis method and system for smart power distribution networks, solving the technical problem of how to improve the accuracy of fault prediction in smart power distribution networks and ensure the safe and stable operation of power distribution networks.

[0005] The first aspect of this invention provides a fault analysis method for smart power distribution networks, comprising:

[0006] Acquire network topology data, environmental parameters and historical fault data of the target area of ​​the smart power distribution network within a historical period, and extract the target detection node and the adjacent first and second adjacent nodes from the target area;

[0007] Based on the network topology data, the environmental parameters, and the historical fault data, regional correlation factors that affect the fault propagation range of the first and second adjacent nodes are extracted.

[0008] Based on the operational fluctuations of the target detection node in different historical periods, predict the operational status of the target detection node in the current period and determine the node status information;

[0009] When the node status information indicates that the target detection node has a fault warning in the current period, the faults of the first adjacent node and the second adjacent node in the current period are predicted based on the historical fault association, and the current fault prediction probability value is obtained.

[0010] Based on the regional correlation factor and the current fault prediction probability value, the fault influence domain boundary of the target detection node is determined;

[0011] Based on the current fault prediction probability value, the fault cascading risk situation within the boundary of the fault influence domain is predicted to obtain the fault cascading risk value, and fault warning information is output based on the fault cascading risk value.

[0012] Optionally, the step of acquiring network topology data, environmental parameters, and historical fault data of the target area of ​​the smart distribution network within a historical period, and extracting the target detection node and adjacent first and second adjacent nodes from the target area, includes:

[0013] Identify the target area in the intelligent power distribution network that is to be analyzed for faults and contains target detection nodes;

[0014] Collect network topology data, environmental parameters, and historical fault data for the target area within a historical period;

[0015] Extract the target detection node from the target region, as well as the first adjacent node and the second adjacent node that are adjacent to the target detection node.

[0016] Optionally, the step of predicting the operating status of the target detection node in the current period based on the operating fluctuations of the target detection node in different historical periods, and determining the node status information, includes:

[0017] The current state prediction value is obtained by predicting the operation status of the target detection node in the current period based on the operation fluctuation of the target detection node in different historical periods.

[0018] If the current state prediction value is greater than the preset standard prediction threshold, it is determined that the target detection node has a fault warning in the current period, and node status information is generated.

[0019] If the current state prediction value is less than or equal to the preset standard prediction threshold, it is determined that there is no fault warning for the target detection node in the current period, and node status information is generated.

[0020] Optionally, the step of predicting the current state prediction value by forecasting the operating status of the target detection node in the current period based on the operating fluctuations of the target detection node in different historical periods includes:

[0021] Collect historical operating parameters of the target detection nodes and historical environmental parameters of the target during the operation of the target area within the historical period;

[0022] The ratio between the variation range of the historical operating parameters and the variation range of the target historical environmental parameters is calculated to obtain the operating parameter fluctuation coefficient;

[0023] The operating status of the target detection node in the current period is predicted based on the fluctuation coefficient of the operating parameters, and the predicted value of the current status is obtained.

[0024] Optionally, the step of predicting the faults of the first and second adjacent nodes in the current period based on historical fault associations to obtain the current fault prediction probability value includes:

[0025] Collect historical fault data associated with the first and second adjacent nodes during the operation of the target area within a historical period;

[0026] Collect first historical feature association data during the operation of the target area within a historical period, wherein the first historical feature association data includes historical topological changes and historical environmental disturbances;

[0027] The ratio between the variation range of the historical fault data in different historical periods and the variation range of the first historical feature-related data is calculated to obtain the fault association feature coefficient.

[0028] Based on the fault association feature coefficients, the faults of the first and second adjacent nodes in the current period are predicted to obtain the current fault prediction probability value.

[0029] Optionally, the step of predicting the faults of the first neighboring node and the second neighboring node in the current period based on the fault association feature coefficient to obtain the current fault prediction probability value includes:

[0030] Collect the current operating parameters of the first adjacent node and the second adjacent node during the operation of the target area in the current cycle;

[0031] Collect the current topology state and current environmental disturbances in the target area during the current cycle.

[0032] Based on the current operating parameters, the current topology state, and the current environmental disturbances, the failure probabilities of the first and second adjacent nodes in the current period are predicted to obtain the current failure prediction probability value.

[0033] Optionally, determining the fault influence domain boundary of the target detection node based on the regional correlation factor and the current fault prediction probability value includes:

[0034] The current topology state, the current environmental disturbance, and the current fault prediction probability value are combined to obtain the current feature association data;

[0035] Based on the current feature association data and the regional association factor, the fault propagation range of the first adjacent node and the second adjacent node is predicted to obtain the preprocessed propagation range.

[0036] The fault influence domain boundary of the target detection node is determined based on the preprocessing diffusion range.

[0037] Optionally, the step of extracting regional correlation factors affecting the fault propagation range of the first and second neighboring nodes based on the network topology data, the environmental parameters, and the historical fault data includes:

[0038] The network topology data, the environmental parameters, and the historical fault data are combined to obtain the second historical feature association data;

[0039] Based on the second historical feature association data, the historical fault propagation range of the first adjacent node and the second adjacent node is collected;

[0040] The regional correlation factor is obtained by calculating the ratio between the variation range of the historical fault propagation range in different historical periods and the variation range of the second historical feature correlation data.

[0041] Optionally, the step of predicting the cascading risk of faults within the boundary of the fault influence domain based on the current fault prediction probability value to obtain a cascading risk value, and outputting fault warning information based on the cascading risk value, includes:

[0042] Pre-determine the fault cascade monitoring area based on the boundary of the fault impact domain;

[0043] The fault propagation probability is predicted based on the current fault prediction probability value and the preprocessing propagation range;

[0044] A preset simulation period is used to predict the fault cascading risk value within the boundary of the fault influence domain based on the fault cascading monitoring area and the fault propagation probability.

[0045] When the cascading fault risk value exceeds a preset risk threshold, a fault warning message is output.

[0046] A second aspect of the present invention provides a fault analysis system for smart power distribution networks, comprising:

[0047] The acquisition module is used to acquire network topology data, environmental parameters and historical fault data of the target area of ​​the smart power distribution network in the historical period, and extract the target detection node and the adjacent first and second adjacent nodes from the target area.

[0048] The extraction module is used to extract regional correlation factors that affect the fault propagation range of the first adjacent node and the second adjacent node based on the network topology data, the environmental parameters and the historical fault data.

[0049] The first prediction module is used to predict the operation of the target detection node in the current period based on the operation fluctuation of the target detection node in different historical periods, and to determine the node status information.

[0050] The second prediction module is used to predict the faults of the first adjacent node and the second adjacent node in the current period based on the historical fault association situation when the node status information indicates that there is a fault warning for the target detection node in the current period, and obtain the current fault prediction probability value.

[0051] The determination module is used to determine the fault influence domain boundary of the target detection node based on the regional correlation factor and the current fault prediction probability value.

[0052] The early warning module is used to predict the cascading risk of faults within the boundary of the fault influence domain based on the current fault prediction probability value, and output fault early warning information based on the cascading risk value.

[0053] As can be seen from the above technical solutions, the present invention has the following advantages:

[0054] This invention acquires historical network topology, environmental parameters, and fault data of a target area in a smart distribution network, extracting the target and adjacent nodes. First, it predicts the fault warning status based on the operational fluctuations of the target node. Then, it analyzes the current fault probability of adjacent nodes in conjunction with historical fault correlation analysis. Simultaneously, it accurately defines the boundary of the fault impact domain by combining regional correlation factors extracted from the topology and environmental parameters. Finally, it quantifies and predicts the cascading risks within the impact domain based on the fault probability and outputs a warning. This invention overcomes the limitations of traditional technologies that rely on single device parameters by integrating multi-dimensional historical data and real-time status. Through adjacent node correlation analysis and dynamic topology environment adaptation, it solves the problem of insufficient integration between fault propagation patterns and real-time network status. Relying on regional correlation factors and cascading risk quantification assessment, it achieves accurate definition of the fault impact domain and early warning of chain reactions, forming a closed loop from node fault warning to global cascading prevention and control. This comprehensively improves the accuracy of fault prediction, constructing a full-process, multi-dimensional guarantee mechanism for the safe and stable operation of the smart distribution network, enhancing fault prediction accuracy and ensuring the safe and stable operation of the smart distribution network. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 A flowchart illustrating the steps of a fault analysis method for smart power distribution networks provided in an embodiment of the present invention;

[0057] Figure 2 This is a structural block diagram of a fault analysis system for smart power distribution networks provided in an embodiment of the present invention. Detailed Implementation

[0058] This invention provides a fault analysis method and system for smart distribution networks, which addresses the technical problem of improving the accuracy of fault prediction in smart distribution networks and ensuring the safe and stable operation of the distribution network.

[0059] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0060] This invention comprehensively collects multi-dimensional data on network topology, environmental parameters, and historical faults in the target area. Utilizing key indicators such as operational parameter fluctuation coefficients and fault correlation characteristic coefficients, it comprehensively considers historical and current operational conditions to accurately analyze the operational status and fault probability of the target detection node and its neighboring nodes. For example, for the target detection node, it predicts the current status by combining the historical operational parameter and environmental parameter variation relationships; for neighboring nodes, it predicts the fault probability based on historical fault correlations and current operational, topological, and environmental factors, significantly improving the accuracy and reliability of fault prediction.

[0061] By leveraging regional correlation factors and combining current topology status, environmental disturbances, and fault prediction probability values, the boundary of the fault impact domain is determined. The correlation between topology, environment, and fault propagation is mined from historical data, and the fault propagation range is accurately predicted based on the current situation. This allows maintenance personnel to clearly understand the areas that may be affected by the fault, providing clear direction for fault prevention and handling, avoiding blind troubleshooting, and improving maintenance efficiency.

[0062] Based on factors such as the fault cascading monitoring area and fault propagation probability, a scientific risk assessment model is used to calculate the fault cascading risk value, which is then compared with a preset threshold to output early warning information. This process fully considers factors such as the importance and vulnerability of equipment within the monitoring area and the network topology, enabling a quantitative assessment of fault cascading risks. This allows maintenance personnel to intuitively understand the degree of risk, formulate response strategies in advance, effectively reduce losses caused by fault chain reactions, and ensure the safe and stable operation of the power distribution network.

[0063] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a fault analysis method for smart power distribution networks provided in this embodiment of the invention.

[0064] This invention provides a fault analysis method for smart power distribution networks, comprising:

[0065] Step 101: Obtain network topology data, environmental parameters and historical fault data of the target area of ​​the smart power distribution network within the historical period, and extract the target detection node and the first and second adjacent nodes from the target area.

[0066] Further, step 101 may include the following sub-steps:

[0067] S11. Determine the target area in the intelligent power distribution network that is to be analyzed for faults and contains target detection nodes.

[0068] S12. Collect network topology data, environmental parameters, and historical fault data of the target area within the historical period.

[0069] S13. Extract the target detection node from the target area, as well as the first adjacent node and the second adjacent node that are adjacent to the target detection node distribution.

[0070] In this embodiment of the invention, the target area in the smart power distribution network that requires fault analysis and contains the target detection node is first determined. Network topology data, environmental parameters and historical fault data of the target area are collected within a historical period. The network topology data includes network structure information such as describing the node connection relationship. The environmental parameters include parameters such as temperature and humidity. The historical fault data includes the time, type and location of previous faults. Then, the target detection node and the first and second adjacent nodes that are adjacent to the target detection node are extracted from the target area.

[0071] Step 102: Based on network topology data, environmental parameters, and historical fault data, extract the regional correlation factors that affect the fault propagation range of the first and second adjacent nodes.

[0072] In this embodiment of the invention, network topology data, environmental parameters and historical fault data are integrated, and a regional correlation factor is obtained by analyzing the ratio of the historical fault propagation range to the magnitude of these data changes. The regional correlation factor reflects the relevant factors affecting the fault propagation range of adjacent nodes.

[0073] Furthermore, step 102 may include the following sub-steps:

[0074] S21. Combine network topology data, environmental parameters, and historical fault data to obtain second historical feature association data.

[0075] In this embodiment of the invention, network topology data, environmental parameters, and historical fault data are combined into second historical feature associated data.

[0076] Create a database table with different fields to store various types of data. For example, set a "timestamp" field to record the data collection time; a "network topology description" field to record network topology information, such as node connections and line status; an "environmental parameter value" field to record specific values ​​of environmental parameters such as temperature, humidity, and air pressure; and a "historical fault details" field to record historical fault information such as the fault occurrence time, fault type, and fault location. Enter this data into the table one by one according to time sequence or event association to form a second historical feature-based associated data.

[0077] S22. Collect the historical fault propagation range of the first and second adjacent nodes based on the second historical feature association data.

[0078] In this embodiment of the invention, network topology data presents structural information such as the connection relationships between nodes and the route of lines in the intelligent power distribution network; environmental parameters cover external conditions affecting equipment operation such as temperature, humidity, and wind speed; historical fault data records the location, type, and time of past faults. Combining these three types of data into second historical feature association data is to comprehensively reflect the operating background and historical status of the power distribution network.

[0079] By analyzing the specific circumstances of historical fault occurrences based on the second historical feature correlation data, the geographical scope of the fault propagation from the first and second adjacent nodes in different historical periods can be determined. For example, it can be clarified which other nodes and lines were affected after the fault occurred.

[0080] S23. Calculate the ratio between the variation range of historical fault propagation in different historical periods and the variation range of the second historical characteristic correlation data to obtain the regional correlation factor.

[0081] In this embodiment of the invention, the ratio of the variation range of historical fault propagation in different historical periods to the variation range of the second historical feature-related data is calculated. This ratio (regional correlation factor) reflects the degree of influence of changes in network topology, environmental parameters, and historical fault conditions on the fault propagation range, revealing the intrinsic relationship between these three factors and fault propagation.

[0082] Suppose that in a certain smart power distribution network, there is a target area consisting of multiple substations, transmission lines and distribution transformers.

[0083] Network topology data records which transmission lines connect the various substations and how the distribution transformers are connected to the lines; environmental parameters record the average temperature and humidity for each month over the past few years; historical fault data shows that a certain transmission line experienced two faults due to lightning strikes in the past three years, and a transformer at a nearby substation experienced one fault due to overload. These data are integrated to form the second historical feature correlation data.

[0084] For example, when a transmission line fails due to a lightning strike, not only does the power supply area where the line is located go out, but it also causes two downstream distribution transformers connected to it to stop working due to power loss. In this case, the range of lines and transformers involved in the fault spread is determined as the historical fault spread range. Another transformer overload fault caused voltage fluctuations in the surrounding area, affecting the normal operation of three adjacent transmission lines. This is another type of historical fault spread range.

[0085] Assume that from year one to year two, the scope of historical fault propagation increased from affecting 3 devices to affecting 5 devices, a change of 2. Simultaneously, the combined change in network topology, environmental parameters, and historical fault data covered by the second historical feature correlation data is quantified as 4. By calculating 2 ÷ 4 = 0.5, the regional correlation factor is obtained as 0.5. This means that for every unit change in the second historical feature correlation data, the fault propagation scope changes by an average of 0.5 units.

[0086] Step 103: Predict the operating status of the target detection node in the current period based on the operating fluctuations of the target detection node in different historical periods, and determine the node status information.

[0087] Furthermore, step 103 may include the following sub-steps:

[0088] S31. Based on the fluctuations in the operation of the target detection node in different historical periods, predict the operation of the target detection node in the current period to obtain the current state prediction value.

[0089] In this embodiment of the invention, historical operating parameters (such as voltage, current, etc.) and historical environmental parameters of the target are collected, and the ratio of their variation amplitudes is calculated to obtain the operating parameter fluctuation coefficient. Then, the current operating parameters and current environmental parameters are combined to predict the operating status of the target detection node in the current cycle to obtain the current state prediction value.

[0090] Furthermore, S31 may include the following sub-steps:

[0091] S311. Collect historical operating parameters of the target detection nodes and historical environmental parameters of the target during the operation of the target area within the historical period.

[0092] In this embodiment of the invention, historical operating parameters of the target detection node during its operation within a historical period are collected, such as voltage, current, and power, which reflect the electrical operating status of the node; at the same time, historical environmental parameters of the target are collected, such as temperature, humidity, and air pressure, which may affect the operation of the node.

[0093] S312. Calculate the ratio between the variation range of historical operating parameters and the variation range of target historical environmental parameters to obtain the operating parameter fluctuation coefficient.

[0094] In this embodiment of the invention, the variation range of the collected historical operating parameters and target historical environmental parameters is calculated separately. For example, historical operating parameters may have different values ​​in different historical periods; the variation range is measured by calculating the parameter difference between adjacent periods or within a specific period. The same applies to the target historical environmental parameters. Then, the variation range of the historical operating parameters is divided by the variation range of the target historical environmental parameters to obtain the operating parameter fluctuation coefficient. This coefficient reflects the degree of influence of historical environmental parameter changes on operating parameter changes, demonstrating the correlation between the two.

[0095] For example, the fluctuation range can be calculated on a monthly time interval. For instance, in January, the average output voltage of the transformer changes from 230V at the beginning of the month to 232V at the end, a fluctuation of 2V; simultaneously, the average ambient temperature in January rises from 10℃ at the beginning of the month to 12℃ at the end, a fluctuation of 2℃. The fluctuation range for each month is calculated using the same method, and then the sum of the historical operating parameter fluctuation ranges for all months of the year is divided by the sum of the target historical environmental parameter fluctuation ranges. Assuming the sum of the historical operating parameter fluctuation ranges is 30V and the sum of the target historical environmental parameter fluctuation ranges is 15℃, then the operating parameter fluctuation coefficient = 30 ÷ 15 = 2. This means that for every 1℃ change in ambient temperature, the transformer output voltage changes by an average of 2V.

[0096] S313. Based on the fluctuation coefficient of the operating parameters, predict the operating status of the target detection node in the current cycle to obtain the predicted value of the current status.

[0097] In this embodiment of the invention, the obtained operating parameter fluctuation coefficient is used in conjunction with the target detection node's operating parameters and current environmental parameters for the current period. Based on the historical correlation between the environment and operating parameters (reflected by the fluctuation coefficient), the operating state of the target detection node for the current period is inferred, thereby deriving a predicted value for the current state. This is used to determine whether the node is likely to experience anomalies or malfunctions in the current period.

[0098] At the beginning of this month (current cycle), the transformer output voltage was 233V and the ambient temperature was 15℃. According to the weather forecast, the temperature is expected to rise by 5℃ this month. Since the operating parameter fluctuation coefficient is 2, it can be predicted that the transformer output voltage will increase by 5×2=10V. Therefore, the predicted transformer output voltage at the end of this month is 233+10=243V. This 243V is the predicted value under the current conditions.

[0099] S32. When the current state prediction value is greater than the preset standard prediction threshold, it is determined that there is a fault warning for the target detection node in the current period, and node status information is generated.

[0100] The preset standard prediction threshold refers to the critical value determined by combining the physical characteristics of the target detection node's equipment, the statistical range of historical data from long-term stable operation, and the safety operation specifications of the smart power distribution network. It is determined by analyzing the extreme values ​​and distribution characteristics of the target detection node's operating parameters under normal operating conditions within a historical period, and then verifying it through expert rules or calibrating it using machine learning algorithms.

[0101] In this embodiment of the invention, when the current state prediction value derived by the correlation between the operating parameter fluctuation coefficient and the environmental parameter exceeds the upper limit of the preset standard prediction threshold, it is determined that the target detection node in the current period has a tendency to develop a fault, and node state information containing fault warning is generated.

[0102] S33. When the current state prediction value is less than or equal to the preset standard prediction threshold, it is determined that there is no fault warning for the target detection node in the current cycle, and node status information is generated.

[0103] In this embodiment of the invention, if the current state prediction value is within a reasonable range of the preset standard prediction threshold, the target detection node is determined to be in a stable operating state, and node state information of "no fault warning and current parameter compliance verification result" is generated.

[0104] Node status information refers to structured data generated to accurately describe the operating status of a target detection node in a fault analysis scenario of a smart power distribution network, based on the comparison between the current periodic operating status prediction value and the preset standard prediction threshold.

[0105] Step 104: When the node status information indicates that there is a fault warning for the target detection node in the current period, the faults of the first and second adjacent nodes in the current period are predicted based on the historical fault association, and the current fault prediction probability value is obtained.

[0106] In this embodiment of the invention, when a fault warning exists for a target detection node in the current period, the fault prediction probability value of the first and second adjacent nodes in the current period is obtained by predicting the faults of the first and second adjacent nodes in the current period based on the fault association between the first and second adjacent nodes in different historical periods. Specifically, historical fault data of the first and second adjacent nodes, as well as historical topology changes and historical environmental disturbances (combined into first historical feature association data), are collected, and the ratio of the change amplitude of historical fault data to that of the first historical feature association data is calculated to obtain the fault association feature coefficient. When a fault warning exists for a target detection node, the current fault prediction probability value of the adjacent node is predicted based on the fault association feature coefficient, combined with the operating parameters, topology status, and environmental disturbances of the adjacent nodes in the current period.

[0107] Furthermore, step 104 may include the following sub-steps:

[0108] S41. Collect historical fault data associated with the first and second adjacent nodes during the operation of the target area within the historical period.

[0109] In this embodiment of the invention, historical fault data associated with the first and second adjacent nodes during the operation of the target area within a historical period is collected. Collecting historical fault data can help us understand the frequency and type of past faults. The first historical feature associated data, composed of historical topology changes (such as changes in line connection methods, additions or subtractions of equipment) and historical environmental disturbances (such as extreme weather, surrounding construction, etc.), is collected because topology and environmental changes can affect the occurrence of faults.

[0110] Assume the target detection node is a critical switch in the power distribution network, with its first adjacent node being an upstream transformer and its second adjacent node being a downstream transmission line segment. Record historical fault data over the past 5 years (historical cycle), including 3 transformer trips due to overload and 2 insulation aging failures; also record line modifications (historical topology changes) and annual summer high-temperature weather (historical environmental disturbances).

[0111] S42. Collect the first historical feature association data during the operation of the target area within the historical period, wherein the first historical feature association data includes historical topological changes and historical environmental disturbances.

[0112] In this embodiment of the invention, historical topological changes and historical environmental disturbances during the operation of the target area within a historical period are collected; wherein, historical topological changes and historical environmental disturbances are combined to form first historical feature associated data.

[0113] In practical implementation, historical topology change data (such as the timing of network structure adjustments and changes in node connectivity) and historical environmental disturbance data (such as the timing and specific parameter values ​​of environmental factors like weather changes and temperature fluctuations) are organized into the same data record or dataset according to chronological order or the order of events. For example, a table is constructed where each row records a specific moment or time period, and the corresponding columns record descriptions of the topology changes at that time, environmental disturbance parameter values, and other relevant information. This combination of data yields the first historical feature-related data.

[0114] S43. Calculate the ratio between the variation range of historical fault data and the variation range of the first historical feature correlation data in different historical periods to obtain the fault correlation feature coefficient.

[0115] In this embodiment of the invention, the ratio of the variation amplitude of historical fault data to the variation amplitude of the first historical feature-related data is used to obtain the fault correlation feature coefficient. This fault correlation feature coefficient reflects the degree of influence and correlation between historical topological and environmental changes and the occurrence of faults.

[0116] According to annual statistics, the number of transformer failures in the past 5 years was 2 in the first year and 3 in the second year, with a variation of 1. Due to factors such as line upgrades and changes in the number of high-temperature days, the overall variation of the first historical characteristic correlation data is assumed to be 2 (this can be measured according to established quantitative standards). The variation for each year is calculated, and then the sum of the variation of historical failure data for all years is divided by the sum of the variation of the first historical characteristic correlation data. Assuming the sum of the variation of historical failure data is 5 and the sum of the variation of the first historical characteristic correlation data is 10, then the failure correlation coefficient = 5 ÷ 10 = 0.5, indicating that for every 1 unit change in the first historical characteristic correlation data, the average change in failure data is 0.5 units.

[0117] The probability value of the current fault is predicted based on the fault correlation characteristic coefficient combined with the operating parameters, topology status and environmental disturbances of the adjacent nodes in the current period.

[0118] It should be noted that the transformer load rate is high in the current cycle (operating parameters), and there has been a recent connection of a new line (change in current topology), while the summer heat has arrived early (current environmental disturbance). Based on a fault correlation characteristic coefficient of 0.5 and a quantitative assessment of these current factors, the probability of transformer failure in the current cycle is predicted. Assuming that through comprehensive analysis and model calculation, the current fault prediction probability is 20%, that is, based on historical correlation patterns and the current state, it is determined that there is a 20% chance that the transformer will fail.

[0119] The fault correlation characteristic coefficient is calculated by comparing the variation range of historical fault data across different historical periods with the variation range of the first historical characteristic correlation data (a combination of historical topology changes and historical environmental disturbances). This coefficient reflects the degree of influence and correlation between historical topology changes and environmental disturbances on fault occurrence. For example, a fault correlation characteristic coefficient of 0.6 means that for every 1 unit change in the first historical characteristic correlation data, the average fault data changes by 0.6 units.

[0120] This refers to the operating status parameters of the first and second adjacent nodes in the current cycle. For example, for adjacent transformer nodes, current operating parameters may include real-time load rate, winding temperature, etc.; for transmission line nodes, there are real-time current, voltage values, etc.

[0121] Current topology status: This refers to the network topology within the target area during the current period. For example, whether there are new line connections or equipment switching.

[0122] Current environmental disturbances: These are factors in the environment of the target area during the current cycle that may affect the operation of the nodes. These include factors such as whether there is severe weather such as heavy rain or strong winds, and whether there is construction affecting the surrounding area.

[0123] The changes in current operating parameters, current topology status, and current environmental disturbances relative to historical average levels or normal conditions are quantified. For example, if the current transformer load rate is 20% higher than the historical average load rate, it can be quantified as a change of 20%; if a new line connection is added to the current topology status, its degree of change is quantified; if a rainstorm occurs in the current environmental disturbance, it is quantified as a corresponding value based on the degree of impact.

[0124] The quantified changes in current operating parameters, current topology state, and current environmental disturbance are combined. Different weights can be assigned to each factor based on their importance before summing; for example, the weight of current operating parameters could be 0.5, current topology state 0.3, and current environmental disturbance 0.2. Assuming the quantified change in current operating parameters is 10, the quantified change in current topology state is 5, and the quantified change in current environmental disturbance is 3, then the combined change value = 10 × 0.5 + 5 × 0.3 + 3 × 0.2 = 5 + 1.5 + 0.6 = 7.1.

[0125] The change in fault prediction is calculated based on the fault correlation characteristic coefficient and the comprehensive change value. The calculation formula is: Fault prediction change value = Fault correlation characteristic coefficient × Comprehensive change value. If the fault correlation characteristic coefficient is 0.6 and the comprehensive change value is 7.1, then the fault prediction change value = 0.6 × 7.1 = 4.26.

[0126] By combining historical failure probability baselines (which can be derived from historical failure data statistics), the failure prediction variation is converted into a probability value and adjusted accordingly. For example, if the historical average failure probability of this adjacent node is 10%, the failure prediction variation of 4.26 is converted into a probability increase (e.g., each variation corresponds to a 1% probability increase). Therefore, the current failure prediction probability value = 10% + 4.26% = 14.26%. This method yields the current failure prediction probability values ​​for the first and second adjacent nodes in the current period.

[0127] S44. Based on the fault association characteristic coefficient, predict the faults of the first and second adjacent nodes in the current period to obtain the current fault prediction probability value.

[0128] Furthermore, S44 may include the following sub-steps:

[0129] S441. Collect the current operating parameters of the first and second adjacent nodes during the operation of the target area in the current cycle.

[0130] In this embodiment of the invention, the current operating parameters of the first and second adjacent nodes are collected during the operation of the target area in the current cycle.

[0131] S442. Collect the current topology state and current environmental disturbances in the target area during the current cycle.

[0132] In this embodiment of the invention, the current topological state of the target area during the current cycle and the current environmental disturbances in the environment are collected.

[0133] S443. Based on the current operating parameters, current topology status, and current environmental disturbances, predict the failure probability of the first and second adjacent nodes in the current period to obtain the current failure prediction probability value.

[0134] In this embodiment of the invention, the failure probability of the first and second adjacent nodes in the current period is predicted based on the current operating parameters, the current topology state, and the current environmental disturbances to obtain the current failure prediction probability value.

[0135] In practice, collecting current operating parameters can provide information on the actual operating status of the first and second adjacent nodes. For example, abnormal voltage and current can reflect problems with the electrical performance of the equipment. Collecting the current topology status can clarify changes in the network connection structure. New connections or removal of equipment will change the power transmission path and node load. Collecting current environmental disturbances can help understand the impact of the external environment on the nodes. Severe weather and construction in the surrounding area may increase the risk of failure.

[0136] Based on the fault correlation characteristic coefficient (reflecting the degree of correlation between historical topology, environmental changes, and faults), and combined with three newly collected current data, the probability of faults occurring in current neighboring nodes is assessed to derive the current fault prediction probability value.

[0137] Assume that the target detection node in the smart power distribution network is the main substation, the first adjacent node is a distribution transformer, and the second adjacent node is a section of transmission line.

[0138] Current operating parameters: For the first and second adjacent nodes, various parameters reflecting their operating status are collected. For example, for power equipment nodes, these include voltage, current, power, temperature, vibration frequency, etc. Assuming the first adjacent node is a transformer, its current operating parameters are: load rate reaches 80% (normal range is generally 30%-70%), and winding temperature is 75℃ (close to its heat resistance limit).

[0139] Current topology status: Assess the current state of the network topology within the target area. This includes checking for new line connections, equipment removal, or changes in connection methods. For example, if a new transmission line connects to the busbar of the first adjacent node (transformer), it increases the load on that node.

[0140] Current environmental disturbances: Understand the factors affecting node operation in the current environment, such as weather conditions (high temperature, heavy rain, strong winds, etc.), surrounding construction, electromagnetic interference, etc. Assume that the current environmental disturbance is continuous high temperature weather, with the outdoor temperature reaching 40℃, which exacerbates the difficulty of heat dissipation for the transformer.

[0141] Quantification parameters: These parameters quantify the collected data. For example, load rate and temperature are compared to standard ranges and converted into quantified deviation values. A load rate of 80%, exceeding the upper limit of the normal range by 10%, is quantified as +10; a winding temperature of 75℃, compared to the upper limit of normal operating temperature (assuming 65℃), is quantified as +10. New line connections are quantified based on their impact on nodes, such as being set to +8; continuous high-temperature weather is quantified as +6.

[0142] Weights were assigned to different types of data based on a large amount of experimental data. Generally, operating parameters have a greater impact on equipment failure, so the weight is set to 0.5; topology state weight is set to 0.3; and environmental disturbance weight is set to 0.2.

[0143] The comprehensive impact value is calculated based on the quantified parameters and weights. The calculation formula is: Comprehensive impact value = Quantified value of operating parameters × Weight of operating parameters + Quantified value of topology state × Weight of topology state + Environmental disturbance value × Weight of environmental disturbance. Using the example data above: Comprehensive impact value = (10 × 0.5 + 8 × 0.3 + 6 × 0.2) = 5 + 2.4 + 1.2 = 8.6.

[0144] The fault correlation characteristic coefficient reflects the degree of correlation between historical changes in topology, environment, and other factors and the occurrence of faults. Assuming the fault correlation characteristic coefficient is known to be 0.6, multiplying the comprehensive impact value by the fault correlation characteristic coefficient yields the fault tendency value: Fault tendency value = Comprehensive impact value × Fault correlation characteristic coefficient = 8.6 × 0.6 = 5.16.

[0145] Establish a mapping relationship between fault tendency values ​​and fault prediction probability values. This can be determined through historical data statistics. For example, after analyzing a large number of historical fault cases, the following mapping relationship was obtained: when the fault tendency value is 0-2, the fault prediction probability value is 5%-10%; when it is 2-4, it is 10%-20%; when it is 4-6, it is 20%-30%, and so on. Based on the previously calculated fault tendency value of 5.16, the corresponding current fault prediction probability value is 25%, which is the fault prediction probability value of the first adjacent node in the current period. Similarly, the fault prediction probability value of the second adjacent node can be calculated.

[0146] Step 105: Determine the boundary of the fault influence domain of the target detection node based on the regional correlation factor and the current fault prediction probability value.

[0147] In this embodiment of the invention, the current topology state, current environmental disturbance and current fault prediction probability value are combined into current feature association data, and the fault spread range of adjacent nodes is predicted by combining regional association factors (preprocessed spread range), thereby determining the fault influence domain boundary of the target detection node.

[0148] Furthermore, step 105 may include the following sub-steps:

[0149] S51. Combine the current topology state, current environmental disturbance, and current fault prediction probability value to obtain the current feature association data.

[0150] In this embodiment of the invention, the current topology state, the current environmental disturbance, and the current fault prediction probability value are combined into current feature association data.

[0151] In the specific implementation, a database table is designed, containing a "current topology status field" (which can store text descriptions, graphical data, etc. of the topology structure), a "current environmental disturbance field" (which stores environmental parameters such as temperature, humidity, and wind speed), and a "current fault prediction probability value field" (which stores specific probability values). The corresponding data is then inserted into the same record to achieve data combination.

[0152] The current topology reflects the network connection layout, such as the addition or removal of lines and changes in node connections; the current environmental disturbances reflect external influences, such as severe weather and surrounding construction. The current topology, current environmental disturbances, and current fault prediction probability values ​​are combined to form the current feature-related data.

[0153] S52. Based on the current feature association data and regional association factors, predict the fault propagation range of the first and second adjacent nodes to obtain the preprocessed propagation range.

[0154] In this embodiment of the invention, fault propagation range prediction involves a regional correlation factor that reflects the degree of correlation between historical topological, environmental, and other factors and the fault propagation range. By combining current feature correlation data with the regional correlation factor, and considering the current network state, environmental impact, and fault probability, the potential propagation range of faults from adjacent nodes is predicted, resulting in a preprocessed propagation range.

[0155] S53. Determine the boundary of the fault influence domain of the target detection node based on the preprocessing diffusion range.

[0156] In this embodiment of the invention, the current feature association data is composed of the current topology state, the current environmental disturbance, and the current fault prediction probability value. The current topology state reflects the changes in the connection relationship between nodes and lines in the smart distribution network, such as whether a new line has been connected or whether equipment has been removed; the current environmental disturbance reflects the impact of the external environment on the network, such as weather conditions like high temperatures, heavy rain, and strong winds, or human factors such as nearby construction; the current fault prediction probability value indicates the likelihood of the first adjacent node and the second adjacent node experiencing a fault in the current period.

[0157] The regional correlation factor is calculated by comparing the variation range of historical fault propagation with the variation range of historical characteristic correlation data (composed of network topology data, environmental parameters, and historical fault data). This factor reflects the combined influence of factors such as network topology, environmental parameters, and fault conditions on the fault propagation range throughout history.

[0158] Changes in the current topology are quantified. For example, adding an important line can be quantified as +5 (the value is set according to the importance of the line and the degree of change to the network structure), while removing a device is quantified as -3, and so on.

[0159] The current environmental disturbances are quantified according to their degree of impact. For example, heavy rain is quantified as +4, and minor construction impact is quantified as +1, etc.

[0160] The current fault prediction probability value is itself quantitative data and can be used directly. Assume that the fault prediction probability value of the first adjacent node is 0.3 (30%) and that of the second adjacent node is 0.2 (20%).

[0161] The above quantified values ​​are calculated by combining them according to certain weights. Assuming the topology state weight is 0.4, the environmental disturbance weight is 0.3, and the fault prediction probability weight is 0.3, then the current feature-related data quantified value = topology state quantified value × 0.4 + environmental disturbance value × 0.3 + fault prediction probability value × 0.3.

[0162] The quantized current feature correlation data is combined with the regional correlation factor. Assuming the regional correlation factor is 0.6, a quantized value is calculated using the formula: Preprocessed diffusion range quantized value = Current feature correlation data quantized value × Regional correlation factor.

[0163] According to pre-defined mapping rules, the preprocessed diffusion range quantization value is converted into the actual fault diffusion range. For example, when the quantization value is between 0 and 2, the fault diffusion range is 1-2 adjacent nodes; when the quantization value is between 2 and 4, the fault diffusion range is 2-4 adjacent nodes and related lines, etc. In this way, the preprocessed diffusion range of the first and second adjacent nodes is obtained, clarifying the area that may be affected when the target detection node fails.

[0164] Determine the boundary of the fault impact area: Based on the preprocessing diffusion range, clarify the boundary of the area affected when the target detection node fails, define the range of affected nodes and lines, and provide a clear range definition for fault prevention and handling.

[0165] The pre-processed diffusion range indicates the area that a fault may affect. First, it's necessary to determine which specific power distribution network elements are included within this range, such as which nodes (transformers, switches, distribution boxes, etc.) and lines are within it. For example, the pre-processed diffusion range might indicate that a fault could affect 3 transformers, 2 transmission lines, and several distribution boxes.

[0166] Clarify the connections between these nodes and lines to define the power transmission paths. For example, determine which transmission lines each of the three transformers is connected to, and from which nodes the distribution box receives its power. Clear connections help determine the path of fault propagation and the boundaries of its potential impact.

[0167] Different types of faults exhibit different propagation characteristics in power distribution networks. For example, a short-circuit fault may instantly affect directly connected nodes and lines; while faults caused by equipment aging may gradually spread. Based on the fault type and its propagation characteristics, the actual impact boundary of the fault within the pre-processing diffusion range is further determined. If it is a short-circuit fault, it spreads rapidly along directly connected lines from the fault point, with the impact relatively concentrated on directly connected equipment; if it is an equipment aging fault, it will affect related equipment within the power supply radius over time.

[0168] Based on the overall topology of the smart power distribution network, factors such as network redundancy and zoning are considered. If the network has redundant lines or backup power supplies, some areas may not be affected when a fault occurs because there are alternative power supply paths, thus reducing the boundary of the fault impact area. If the network has clear zoning, the fault may be confined to a certain zoning, which also helps to define the boundary.

[0169] The boundaries of the area formed by the nodes and lines that will actually be affected by the fault within the pre-processing diffusion range must be determined. These nodes and lines can be located on a map using a Geographic Information System (GIS) to clearly define their boundaries; alternatively, a detailed list of the affected nodes and lines, including their names and numbers, can be used to define the boundaries of the fault-affected area.

[0170] For example: Suppose that the target detection node in the smart power distribution network is a critical substation.

[0171] Current topology status: Recently, a new transmission line has been connected to one of the busbars of this substation, changing the power transmission path.

[0172] Current environmental disturbances: The local meteorological department has issued a blue gale warning, which may affect outdoor power transmission lines.

[0173] Current fault prediction probability values: The calculated fault prediction probability value for the first adjacent node (a distribution transformer) connected to the substation is 30%, and the fault prediction probability value for the second adjacent node (a section of transmission line) is 20%. This information is combined into the current feature association data.

[0174] Fault propagation range prediction: The previously calculated regional correlation factor is known to be 0.5. The current feature correlation data is input into the fault propagation prediction model. Combined with the regional correlation factor, the model considers factors such as changes in power distribution due to the access of new lines and the possibility of line faults due to strong winds. It predicts that if a substation or adjacent node fails, the fault may propagate to three surrounding distribution transformers and two transmission lines, thus obtaining the preprocessed propagation range.

[0175] Determine the boundary of the fault impact domain: Based on the above preprocessing diffusion range, determine the boundary of the fault impact domain of the target detection node (substation), which includes the area containing these 3 distribution transformers and 2 transmission lines. Maintenance personnel can focus on monitoring and prevention in this area.

[0176] Step 106: Based on the current fault prediction probability value, predict the fault cascading risk situation within the fault influence domain boundary to obtain the fault cascading risk value, and output fault warning information based on the fault cascading risk value.

[0177] In this embodiment of the invention, a fault cascading monitoring area is preset within the boundary of the fault influence domain. The fault propagation probability is calculated based on the current fault prediction probability value and the pre-processed propagation range, and the fault cascading risk value is predicted. The predicted fault cascading risk value is compared with a preset risk threshold, and a fault warning message is output when the threshold is exceeded.

[0178] Furthermore, step 106 may include the following sub-steps:

[0179] S61. Preset the fault cascade monitoring area according to the boundary of the fault impact domain.

[0180] S62. Predict the fault propagation probability based on the current fault prediction probability value and the pre-processing propagation range.

[0181] S63. Preset simulation period and predict the fault cascading risk value within the fault influence domain boundary based on the fault cascading monitoring area and fault propagation probability.

[0182] Pre-defined fault cascading monitoring area: The boundary of the fault impact domain defines the range of impact of a fault at the target detection node. Based on this range, a dedicated fault cascading monitoring area is delineated, focusing monitoring on distribution network elements such as nodes and lines within this range to facilitate subsequent targeted assessment of fault cascading risks.

[0183] Predicting fault propagation probability: The current predicted fault probability value reflects the likelihood of faults occurring in adjacent nodes, while the preprocessed propagation range defines the area where the fault may propagate. Combining these two factors, the probability of a fault propagating from the initial fault point to other nodes or lines within the fault influence domain boundary is predicted, quantifying the likelihood of fault propagation.

[0184] The current fault prediction probability value reflects the probability that the first and second adjacent nodes will fail in the current cycle. For example, if the current fault prediction probability value of the first adjacent node (such as a transformer) is 30%, it means that under the current operating conditions, topology, and environmental conditions, the transformer has a 30% probability of failing.

[0185] Preprocessing diffusion range: refers to the area to which the fault will spread, predicted based on the regional correlation factor and the current feature correlation data, which determines which nodes and lines may be affected by the fault.

[0186] In this embodiment of the invention, the connection relationships between nodes (such as transformers, switches, distribution boxes, etc.) and lines within the boundary of the fault impact domain are clarified. The power transmission path and the mutual influence between nodes are defined. For example, if a transmission line connects two transformers, when one transformer fails, the fault may be transmitted to the other transformer through this line.

[0187] Consider the electrical characteristics between nodes, such as impedance matching and load distribution. If there is an impedance mismatch between nodes, voltage fluctuations are more likely to occur when a fault happens, thus increasing the possibility of fault propagation.

[0188] Review historical failure propagation patterns under similar topologies, environmental conditions, and failure types. For example, what was the propagation range and probability of a certain type of equipment failure under similar high-temperature weather conditions in the past?

[0189] Analyze the factors influencing fault propagation in historical fault data, such as the degree of equipment aging and maintenance conditions. Severely aged equipment is more likely to trigger a chain reaction when a fault occurs, expanding the scope of the fault propagation.

[0190] Rules are set based on factors such as the tightness of connections between nodes and the magnitude of current fault prediction probabilities. For example, it is stipulated that if the current fault prediction probabilities of two adjacent nodes are both high (e.g., both exceeding 20%), and they are directly connected, then the probability of a fault spreading between them increases by a certain percentage (e.g., 20%).

[0191] Machine learning algorithms, such as Bayesian networks and decision trees, can also be used. The current fault prediction probability, node connection information within the preprocessed propagation range, and historical fault data are used as input features. By learning from and training on historical data, a model capable of predicting the probability of fault propagation can be constructed.

[0192] Taking Bayesian networks as an example, first define the nodes in the network (such as individual device nodes, fault event nodes, etc.) and the conditional probability relationships between them. Then, input data such as the current fault prediction probability value into the network, and calculate the probability of the fault spreading within the preprocessing diffusion range through probabilistic inference.

[0193] Based on the constructed model, the current fault prediction probability value and pre-processing diffusion range information are input, and the model calculates the fault diffusion probability. For example, the model calculates that under the current circumstances, the probability of the fault spreading within the pre-processing diffusion range is 40%, meaning that there is a 40% chance that the fault will further spread within this range.

[0194] Predicting the risk value of cascading failures: A simulation period is preset (such as the next few hours or days). Within this period, based on information such as the network structure and equipment status in the fault cascading monitoring area, as well as the previously predicted failure propagation probability, a risk assessment model is used to calculate the risk value of a fault triggering a cascading failure (one failure triggering a series of subsequent failures) in the area, measuring the probability and severity of the fault chain reaction.

[0195] Fault Cascade Monitoring Area: This is a specific area pre-defined based on the boundary of the fault impact domain, encompassing power distribution network components such as nodes (e.g., transformers, switches, distribution boxes) and lines that may be affected by the fault. Information such as the number, type, importance, and interconnections of these components forms the basis for subsequent analysis. For example, the monitoring area may contain 5 critical transformers and 10 transmission lines, exhibiting a ring network structure with some lines serving as backups for each other.

[0196] Fault propagation probability: Calculated using information such as the current fault prediction probability value and the pre-processed propagation range, it represents the likelihood of a fault spreading from its initial fault point to other nodes or lines. Assuming a fault propagation probability of 60%, it means there is a 60% probability that the fault will further propagate within the monitored area.

[0197] Equipment Importance: Assess the importance of each device within the fault cascading monitoring area to the normal operation of the entire power distribution network. For example, assign a higher importance weight to the main transformer of a substation because its failure could lead to a large-scale power outage; while some small distribution boxes have relatively lower importance weights. Weights can be determined using methods such as the analytic hierarchy process (AHP), for example, setting the importance weight of the main transformer to 0.8 and that of small distribution boxes to 0.2.

[0198] Equipment vulnerability: This considers the condition of the equipment itself, such as its aging level and the presence of potential hazards. Severely aged equipment that frequently malfunctions is relatively more vulnerable and is more easily affected when a failure occurs, potentially triggering a cascading failure. Vulnerability can be quantified based on information such as the equipment's operating years and maintenance records. For example, equipment that has been in operation for more than 10 years and undergoes frequent maintenance has a vulnerability score of 8 out of 10.

[0199] Network topology: Analyze the impact of the network topology within the monitoring area on fault propagation. For example, a ring network structure may complicate fault propagation paths and increase the risk of fault cascading; while a radial structure is relatively simple and may have a lower risk of fault cascading. Appropriate risk coefficients can be set based on the complexity of the topology, such as a risk coefficient of 1.2 for a ring network structure and 1 for a radial structure.

[0200] Constructing a risk assessment model: A simple model can be built using a weighted summation approach. Assuming the fault cascading risk value is R, the equipment importance weight is w1, the equipment vulnerability score is v, the network topology risk coefficient is k, and the fault propagation probability is p, the model formula can be expressed as: R = w1 × v × k × p. Alternatively, a more complex machine learning model, such as a neural network, can be used. By using various attributes of equipment within the fault cascading monitoring area (importance, vulnerability, etc.) and the fault propagation probability as input features, and training on a large amount of historical fault data, the model can automatically learn the mapping relationship between these factors and the fault cascading risk value.

[0201] For the weighted summation model, the relevant parameters of each device are substituted into the formula for calculation. For example, if there are multiple devices in a monitoring area, the risk value corresponding to each device is calculated separately, and then the values ​​are summed to obtain the cascading risk value of the entire fault influence domain boundary. Assuming that a certain device has w1=0.6, v=7, k=1.2, and p=0.6, then the risk value corresponding to this device is 0.6×7×1.2×0.6=3.024. The final cascading risk value of the fault is obtained by summing the risk values ​​of all devices.

[0202] S64. When the cascading fault risk value is greater than the preset risk threshold, a fault warning message will be output.

[0203] In this embodiment of the invention, the fault cascading risk value is compared with a preset risk threshold. When it is determined that the fault cascading risk value exceeds the preset risk threshold, a fault warning message is output.

[0204] The fault cascading risk value reflects the degree of risk that a fault in a target detection node in a smart distribution network will trigger a chain reaction within its influence domain boundary. This value comprehensively considers factors such as the fault propagation probability and the distribution of nodes within the fault influence domain.

[0205] The preset risk threshold is a standard value that represents the upper limit of acceptable cascading failure risk. It is determined through extensive experiments based on factors such as the security and reliability requirements of the power distribution network and its operation and maintenance capabilities.

[0206] The calculated cascading fault risk value is compared with a preset risk threshold. If the cascading fault risk value is less than or equal to the preset risk threshold, it indicates that the risk of the current fault causing a cascading fault is within a controllable range, and no warning is needed at this time. If the cascading fault risk value exceeds the preset risk threshold, it means that the fault is more likely to cause a chain of faults, which may have a significant impact on the power distribution network. In this case, a fault warning message is output to remind maintenance personnel to take measures to prevent the fault from escalating.

[0207] Suppose a critical transformer in a substation within a smart power distribution network is the target detection node. Its fault cascading risk value is calculated to be 0.7 (range 0-1, with higher values ​​indicating higher risk).

[0208] Based on the actual conditions of the power distribution network, such as the degree of equipment aging and redundant configuration, the power operation and maintenance department sets a preset risk threshold of 0.6.

[0209] The fault cascading risk value of 0.7 is compared with the preset risk threshold of 0.6. It is found that 0.7 > 0.6, meaning the fault cascading risk value exceeds the preset risk threshold. At this point, the system immediately outputs a fault warning message, such as sending a text message to the maintenance personnel's mobile phone stating, "[Distribution Network Fault Warning] The fault cascading risk value of a transformer in a certain substation exceeds the threshold, which may trigger a chain reaction of faults. Please investigate and handle it immediately!" After receiving the warning, the maintenance personnel will promptly inspect and maintain the distribution network in that area to reduce the risk of the fault escalating.

[0210] Please see Figure 2 , Figure 2 This is a structural block diagram of a fault analysis system for smart power distribution networks provided in an embodiment of the present invention.

[0211] This invention provides a fault analysis system for smart power distribution networks, comprising:

[0212] The acquisition module 301 is used to acquire network topology data, environmental parameters and historical fault data of the target area of ​​the smart power distribution network in the historical period, and extract the target detection node and the adjacent first and second adjacent nodes from the target area.

[0213] The extraction module 302 is used to extract regional correlation factors that affect the fault propagation range of the first and second adjacent nodes based on network topology data, environmental parameters and historical fault data.

[0214] The first prediction module 303 is used to predict the operation of the target detection node in the current period based on the operation fluctuation of the target detection node in different historical periods, and to determine the node status information.

[0215] The second prediction module 304 is used to predict the faults of the first and second adjacent nodes in the current period based on historical fault associations when the node status information indicates that there is a fault warning for the target detection node in the current period, and obtain the current fault prediction probability value.

[0216] The determination module 305 is used to determine the boundary of the fault influence domain of the target detection node based on the regional correlation factor and the current fault prediction probability value.

[0217] The early warning module 306 is used to predict the cascading risk of faults within the boundary of the fault influence domain based on the current fault prediction probability value, and output fault early warning information based on the cascading risk value.

[0218] Furthermore, the acquisition module 301 includes:

[0219] The target area submodule is used to determine the target area in the smart power distribution network that is to be analyzed for faults and contains target detection nodes.

[0220] The data acquisition submodule is used to collect network topology data, environmental parameters, and historical fault data of the target area within a historical period.

[0221] The extraction and processing submodule is used to extract the target detection node from the target area, as well as the first adjacent node and the second adjacent node that are adjacent to the target detection node distribution.

[0222] Furthermore, the first prediction module 303 includes:

[0223] The current state prediction submodule is used to predict the current state prediction value of the target detection node in the current period based on the fluctuation of the target detection node's operation in different historical periods.

[0224] The first determination submodule is used to determine that there is a fault warning for the target detection node in the current period when the current state prediction value is greater than the preset standard prediction threshold, and to generate node status information.

[0225] The second determination submodule is used to determine that there is no fault warning for the target detection node in the current period when the current state prediction value is less than or equal to the preset standard prediction threshold, and to generate node status information.

[0226] Furthermore, the current state prediction submodule includes:

[0227] The parameter acquisition unit is used to collect historical operating parameters of the target detection node and historical environmental parameters of the target during the operation of the target area within the historical period.

[0228] The operating parameter fluctuation coefficient unit is used to calculate the ratio between the variation range of historical operating parameters and the variation range of target historical environmental parameters to obtain the operating parameter fluctuation coefficient.

[0229] The state prediction unit is used to predict the operating state of the target detection node in the current cycle based on the fluctuation coefficient of the operating parameters, and obtain the current state prediction value.

[0230] Furthermore, the second prediction module 304 includes:

[0231] The historical fault data submodule is used to collect historical fault data associated with the first and second adjacent nodes during the operation of the target area within a historical period.

[0232] The first historical feature association data submodule is used to collect the first historical feature association data during the operation of the target area in the historical period. The first historical feature association data includes historical topological changes and historical environmental disturbances.

[0233] The fault association feature coefficient submodule is used to calculate the ratio between the variation amplitude of historical fault data and the variation amplitude of the first historical feature association data in different historical periods to obtain the fault association feature coefficient.

[0234] The fault prediction submodule is used to predict the faults of the first and second adjacent nodes in the current period based on the fault association feature coefficients, and obtain the current fault prediction probability value.

[0235] Furthermore, the fault prediction submodule includes:

[0236] The current operating parameter unit is used to collect the current operating parameters of the first and second adjacent nodes during the operation of the target area in the current cycle;

[0237] The environmental disturbance acquisition unit is used to acquire the current topology state and the current environmental disturbance in the target area during the current cycle.

[0238] The current fault prediction probability value unit is used to predict the fault probability of the first and second adjacent nodes in the current period based on the current operating parameters, the current topology state, and the current environmental disturbances, and to obtain the current fault prediction probability value.

[0239] Furthermore, module 305 is determined to include:

[0240] The current feature association data submodule is used to combine the current topology state, current environmental disturbance, and current fault prediction probability value to obtain the current feature association data;

[0241] The preprocessing diffusion range submodule is used to predict the fault diffusion range of the first adjacent node and the second adjacent node based on the current feature association data and regional association factors, so as to obtain the preprocessing diffusion range.

[0242] The fault impact domain boundary submodule is used to determine the fault impact domain boundary of the target detection node based on the preprocessing diffusion range.

[0243] Furthermore, the extraction module 302 includes:

[0244] The second historical feature association data submodule is used to combine network topology data, environmental parameters and historical fault data to obtain the second historical feature association data.

[0245] The historical fault propagation range submodule is used to collect the historical fault propagation range of the first and second adjacent nodes based on the association data of the second historical features.

[0246] The regional correlation factor submodule is used to calculate the ratio between the variation range of historical fault propagation in different historical periods and the variation range of second historical feature correlation data to obtain the regional correlation factor.

[0247] Furthermore, the early warning module 306 includes:

[0248] The fault cascading monitoring area submodule is used to preset the fault cascading monitoring area according to the boundary of the fault influence domain;

[0249] The fault propagation probability submodule is used to predict the fault propagation probability based on the current fault prediction probability value and the preprocessed propagation range.

[0250] The fault cascading risk value submodule is used to preset the simulation period and predict the fault cascading risk value within the boundary of the fault influence domain based on the fault cascading monitoring area and the fault propagation probability.

[0251] The fault warning information submodule is used to output fault warning information when the cascading fault risk value is greater than the preset risk threshold.

[0252] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0253] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0254] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0255] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0256] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0257] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fault analysis method for smart power distribution networks, characterized in that, include: Acquire network topology data, environmental parameters and historical fault data of the target area of ​​the smart power distribution network within a historical period, and extract the target detection node and the adjacent first and second adjacent nodes from the target area; Based on the network topology data, the environmental parameters, and the historical fault data, regional correlation factors that affect the fault propagation range of the first and second adjacent nodes are extracted. Based on the operational fluctuations of the target detection node in different historical periods, predict the operational status of the target detection node in the current period and determine the node status information; When the node status information indicates that the target detection node has a fault warning in the current period, the faults of the first adjacent node and the second adjacent node in the current period are predicted based on the historical fault association, and the current fault prediction probability value is obtained. Based on the regional correlation factor and the current fault prediction probability value, the fault influence domain boundary of the target detection node is determined; Based on the current fault prediction probability value, the fault cascading risk situation within the boundary of the fault influence domain is predicted to obtain the fault cascading risk value, and fault warning information is output based on the fault cascading risk value.

2. The fault analysis method for intelligent power distribution networks according to claim 1, characterized in that, The process of acquiring network topology data, environmental parameters, and historical fault data of a target area in a smart power distribution network within a historical period, and extracting target detection nodes and adjacent first and second neighboring nodes from the target area, includes: Identify the target area in the intelligent power distribution network that is to be analyzed for faults and contains target detection nodes; Collect network topology data, environmental parameters, and historical fault data for the target area within a historical period; Extract the target detection node from the target region, as well as the first adjacent node and the second adjacent node that are adjacent to the target detection node.

3. The fault analysis method for intelligent power distribution networks according to claim 1, characterized in that, The step of predicting the operational status of the target detection node in the current period based on the operational fluctuations of the target detection node in different historical periods, and determining the node status information, includes: The current state prediction value is obtained by predicting the operation status of the target detection node in the current period based on the operation fluctuation of the target detection node in different historical periods. If the current state prediction value is greater than the preset standard prediction threshold, it is determined that the target detection node has a fault warning in the current period, and node status information is generated. If the current state prediction value is less than or equal to the preset standard prediction threshold, it is determined that there is no fault warning for the target detection node in the current period, and node status information is generated.

4. The fault analysis method for intelligent power distribution networks according to claim 3, characterized in that, The step of predicting the current state prediction value by forecasting the target detection node's operation in the current period based on the operational fluctuations of the target detection node in different historical periods includes: Collect historical operating parameters of the target detection nodes and historical environmental parameters of the target during the operation of the target area within the historical period; The ratio between the variation range of the historical operating parameters and the variation range of the target historical environmental parameters is calculated to obtain the operating parameter fluctuation coefficient; The operating status of the target detection node in the current period is predicted based on the fluctuation coefficient of the operating parameters, and the predicted value of the current status is obtained.

5. The fault analysis method for intelligent power distribution networks according to claim 1, characterized in that, The step of predicting the faults of the first and second adjacent nodes in the current period based on historical fault associations to obtain the current fault prediction probability value includes: Collect historical fault data associated with the first and second adjacent nodes during the operation of the target area within a historical period; Collect first historical feature association data during the operation of the target area within a historical period, wherein the first historical feature association data includes historical topological changes and historical environmental disturbances; The ratio between the variation range of the historical fault data in different historical periods and the variation range of the first historical feature-related data is calculated to obtain the fault association feature coefficient. Based on the fault association feature coefficients, the faults of the first and second adjacent nodes in the current period are predicted to obtain the current fault prediction probability value.

6. The fault analysis method for intelligent power distribution networks according to claim 5, characterized in that, The step of predicting the faults of the first and second adjacent nodes in the current period based on the fault association feature coefficients to obtain the current fault prediction probability value includes: Collect the current operating parameters of the first adjacent node and the second adjacent node during the operation of the target area in the current cycle; Collect the current topology state and current environmental disturbances in the target area during the current cycle. Based on the current operating parameters, the current topology state, and the current environmental disturbances, the failure probabilities of the first and second adjacent nodes in the current period are predicted to obtain the current failure prediction probability value.

7. The fault analysis method for intelligent power distribution networks according to claim 6, characterized in that, Determining the fault influence domain boundary of the target detection node based on the regional correlation factor and the current fault prediction probability value includes: The current topology state, the current environmental disturbance, and the current fault prediction probability value are combined to obtain the current feature association data; Based on the current feature association data and the regional association factor, the fault propagation range of the first adjacent node and the second adjacent node is predicted to obtain the preprocessed propagation range. The fault influence domain boundary of the target detection node is determined based on the preprocessing diffusion range.

8. The fault analysis method for intelligent power distribution networks according to claim 1, characterized in that, The step of extracting regional correlation factors affecting the fault propagation range of the first and second adjacent nodes based on the network topology data, the environmental parameters, and the historical fault data includes: The network topology data, the environmental parameters, and the historical fault data are combined to obtain the second historical feature association data; Based on the second historical feature association data, the historical fault propagation range of the first adjacent node and the second adjacent node is collected; The regional correlation factor is obtained by calculating the ratio between the variation range of the historical fault propagation range in different historical periods and the variation range of the second historical feature correlation data.

9. The fault analysis method for intelligent power distribution networks according to any one of claims 1-8, characterized in that, The step of predicting the cascading risk of faults within the boundary of the fault influence domain based on the current fault prediction probability value, and outputting fault warning information based on the cascading risk value, includes: Pre-determine the fault cascade monitoring area based on the boundary of the fault impact domain; The fault propagation probability is predicted based on the current fault prediction probability value and the preprocessing propagation range; A preset simulation period is used to predict the fault cascading risk value within the boundary of the fault influence domain based on the fault cascading monitoring area and the fault propagation probability. When the cascading fault risk value exceeds a preset risk threshold, a fault warning message is output.

10. A fault analysis system for intelligent power distribution networks, characterized in that, include: The acquisition module is used to acquire network topology data, environmental parameters and historical fault data of the target area of ​​the smart power distribution network in the historical period, and extract the target detection node and the adjacent first and second adjacent nodes from the target area. The extraction module is used to extract regional correlation factors that affect the fault propagation range of the first adjacent node and the second adjacent node based on the network topology data, the environmental parameters and the historical fault data. The first prediction module is used to predict the operation of the target detection node in the current period based on the operation fluctuation of the target detection node in different historical periods, and to determine the node status information. The second prediction module is used to predict the faults of the first adjacent node and the second adjacent node in the current period based on the historical fault association situation when the node status information indicates that there is a fault warning for the target detection node in the current period, and obtain the current fault prediction probability value. The determination module is used to determine the fault influence domain boundary of the target detection node based on the regional correlation factor and the current fault prediction probability value. The early warning module is used to predict the cascading risk of faults within the boundary of the fault influence domain based on the current fault prediction probability value, and output fault early warning information based on the cascading risk value.

Images