A metering box fault analysis system based on a correlation relationship

Abstract

The application relates to a meter box fault analysis system based on a correlation relationship, which comprises a correlation model training subsystem, a fault monitoring subsystem, and a carrier wave generating unit and a carrier wave receiving unit arranged in the meter box; through the correlation relationship between the meter boxes, fault analysis is carried out by means of a carrier wave circuit, the actual carrier wave can be used to judge the different influences of lines, interfaces or internal meters on carrier wave information, so that different fault types can be analyzed, and through a pre-constructed model, the relationship between the actual carrier wave communication and the actual power transmission waveform between the meter boxes is deeply learned, so that the meter box fault can be quickly self-checked, the meter box fault analysis has a more dimensional data basis through the content of the correlation information, and the fault automatic analysis is separated from the troubleshooting mode of manual maintenance.

Description

Technical Field

[0001] This invention relates to the field of meter box failure analysis, and more specifically, to a meter box failure analysis system based on correlation. Background Technology

[0002] Metering boxes are tools used in power grids to assist in the metering of electrical energy. They include energy meters, voltage and current transformers and their secondary circuits, energy metering panels, cabinets, and boxes. Generally, the installation location of metering boxes is determined by the needs of the metered parties. When the locations of the metered parties are relatively dispersed, the installation of metering boxes is also relatively dispersed. Since some metering boxes are installed outdoors, they are very prone to damage and malfunctions. Besides damage to the appearance and box body caused by human intervention, more serious electrical faults are caused by damage to components, windings, and contacts. Traditionally, manual inspection is used, but this method is inefficient and also poses a risk to the inspectors. The professional requirements for maintenance personnel are also relatively high. CN115856756A discloses a fault assessment method for electricity metering boxes, which uses neural networks combined with stress cross-analysis to achieve fault assessment of metering boxes, reducing the professional requirements for electricity metering box maintenance personnel and the safety hazards that may arise from manual maintenance. Although it has been optimized based on the fault analysis system of an electricity metering device disclosed in CN114879124A, and can obtain more accurate fault analysis conclusions when the data volume is large, the basic information for fault assessment still needs to be obtained by maintenance personnel. Thus, the problem of low frequency and inefficiency of manual maintenance remains unsolved. For example, CN112782639B discloses a fault intelligent indication method for low-voltage metering boxes and the cause analysis technology of the indicator, and CN108764598A discloses a fault risk assessment method for low-voltage metering boxes. These methods can analyze the faults of metering boxes through data, but they all require basic data from humans or users as intervention variables. This is because the basic data of metering boxes is relatively simple, and the models and communication protocols are different, making it difficult to use as the data basis for analyzing metering box faults. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a meter box fault analysis system based on correlation.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is: a metering box fault analysis system based on correlation, including a correlation model training subsystem, a fault monitoring subsystem, and a carrier generation unit and a carrier receiving unit configured in the metering box;

[0005] The correlation model training subsystem includes a topology management module, a sample simulation module, and a correlation marking module. The topology management module generates correlation information based on the power topology relationships between metering boxes. The sample simulation module generates carrier simulation commands based on the correlation information. When a metering box receives a simulated carrier command through the sample simulation module, it sends a transmitting-side response and transmits the carrier simulation command to other metering boxes with correlation relationships through its carrier generation unit. When a metering box receives a simulated carrier command through its carrier receiving unit, it generates a receiving-side response. The sample simulation module compares the transmitting-side response and the receiving-side response to generate a baseline loss marker. The correlation marking module configures the baseline loss marker based on the correlation information to generate a metering correlation model.

[0006] The fault monitoring subsystem includes a detection configuration module and a fault analysis module. The detection configuration module generates detection tasks and sends them to the corresponding metering boxes. When a metering box receives a detection task, it sends detection data to the metering box at the target address of the detection task through a carrier generation unit. The fault analysis module is connected to the carrier receiving unit and receives detection data from the carrier receiving unit. It compares the data with the carrier reference data corresponding to the detection task to generate carrier deviation information and inputs the carrier deviation information into the metering correlation model to generate fault analysis results.

[0007] Furthermore, the carrier simulation command includes topology simulation data, the sample simulation module is configured with a topology association feature table, the topology association feature table stores several different topology simulation data with topology association features as indexes, the sample simulation module determines the topology association features of the corresponding metering box according to the association information, and calls the corresponding topology simulation data from the topology association feature table according to the topology association features;

[0008] The transmitting-side response information includes the topology simulation data, and the receiving-side response information includes the topology simulation data;

[0009] The sample simulation module compares the deviations in the topology simulation data between the transmitting-side response information and the receiving-side response information to generate a solid-state loss deviation term and a delay deviation term. The solid-state loss deviation term is determined by the numerical deviation in the topology simulation data, and the delay deviation term is determined by the time delay deviation in the topology simulation data.

[0010] Furthermore, the sample simulation module is also configured with a dynamic simulation strategy to respond to carrier simulation commands, the dynamic simulation strategy including...

[0011] Step A1: Synchronize the time base of the metering box according to the delay deviation item, and determine the intercept time period by analyzing the historical metering data of the metering box;

[0012] Step A2: Obtain real-time metering data based on the captured time period to generate dynamic metering waveforms;

[0013] Step A3: Obtain the parameter information of the metering box, generate a metering reference waveform based on the parameter information, compare the deviation between the metering reference waveform and the dynamic metering waveform to generate a metering deviation waveform. The sending side response information includes the dynamic metering waveform and the metering deviation waveform, and the receiving side response information includes the dynamic metering waveform and the metering deviation waveform.

[0014] Furthermore, the sample simulation module is configured with a waveform loss algorithm, which is configured as follows: Among them, G c For the dynamic loss bias term, α1 is the preset loss bias sensitivity coefficient, and g fd (x) represents the dynamic metering waveform in the sending-side response information, g jd (x) represents the dynamic metering waveform in the receiver's response information, g fp (x) represents the measurement deviation waveform in the sending-side response information, g jp (x) represents the measurement deviation waveform in the receiver response information, T c To extract the duration of a time period;

[0015] The benchmark loss label includes a dynamic loss deviation term and a solid loss deviation term.

[0016] Furthermore, the association tagging module is configured with an association tagging strategy, which includes...

[0017] Step B1: Establish a coordinate system for a measuring box;

[0018] Step B2: Determine a metering box as the initial coordinate based on the power topology;

[0019] Step B3: Starting from the initial coordinates, determine the measurement vector corresponding to each piece of related information in sequence. G is a metric vector. s This is the solid loss deviation term;

[0020] Step B4: Determine the measurement coordinates of each measurement box in the measurement coordinate system based on the measurement vector;

[0021] Step B5: Generate the measurement correlation model.

[0022] Furthermore, the detection configuration module is configured with an active detection strategy for generating the detection task, the active detection strategy being...

[0023] Step C1: Obtain the basic information of the metering box, retrieve the corresponding detection task items, and generate the basic inspection level according to the corresponding detection task items. Each detection task item has a corresponding task sub-load.

[0024] Step C2: Determine the corresponding inspection frequency based on the basic inspection level, and sum the task sub-loads to generate the task load for the metering box;

[0025] Step C3: Input the inspection frequency into the inspection load function to calculate the inspection load for each metering box. The inspection load function is configured as follows: There is U d For inspection load, R d For the current workload of the metering box, W s t represents the current inspection frequency of the metering box. e For the proposed inspection time, t e-1 For the actual time of one inspection, t n R represents the planned inspection time for the nth meter box that has associated information with this meter box. n The task load is the nth metering box that has associated information with the current metering box, k is the total number of metering boxes that have associated information with the current metering box, β1 is the preset inspection delay weight, and β2 is the preset inspection impact weight.

[0026] Step C4: Determine the planned inspection time for each metering box to minimize the total inspection load of the metering boxes.

[0027] Furthermore, in step C1, a task index table is pre-configured, which stores several detection task items. Each detection task item is indexed by basic information, and each detection task item is pre-configured with a task sub-load.

[0028] Furthermore, the detection task items in the task index table are indexed by detection requests, and the detection configuration module is configured with a passive detection strategy for generating the detection tasks. The passive detection strategy includes...

[0029] Step D1: Receive the detection request and determine the corresponding detection task item;

[0030] Step D2: Input the coordinates of the metering box corresponding to the detection request into the metering association model to obtain the target metering box corresponding to the detection task item;

[0031] Step D3: Associate the target metering box and the detection task item to generate the detection task.

[0032] Furthermore, the fault analysis module is configured with a fault analysis strategy for generating the fault analysis results, the fault analysis strategy including...

[0033] Step E1: A carrier deviation database is configured, wherein the carrier deviation database is configured with carrier deviation features, and each carrier deviation feature is pre-configured with fault anomaly sub-values ​​and associated anomaly sub-values;

[0034] Step E2: Process the carrier offset information by detecting the task item to obtain several carrier offset data;

[0035] Step E3: Identify carrier deviation features from carrier deviation data using the carrier deviation database, and extract the corresponding feature identification similarity values;

[0036] Step E4: Send the associated anomaly sub-values ​​corresponding to the carrier deviation characteristics to the metering box as the target metering box according to the metering association model;

[0037] Step E5: An anomaly analysis association table is configured. The anomaly analysis association table stores several fault anomaly items. Each fault anomaly item is associated with several different carrier deviation features. The similarity value of the features is used as the weight, and the fault anomaly sub-values ​​are weighted and summed to obtain the fault anomaly value of each fault anomaly item.

[0038] Step E6: If the fault anomaly value is greater than the preset anomaly baseline threshold for the fault anomaly item, then the fault anomaly item is output as the fault analysis result.

[0039] Furthermore, step E4 also includes retrieving the transmission path length between metering boxes from the metering correlation model, where p c =r d χ d / S a , where p c For the fault anomaly sub-value of the metering box as the target, r d χ represents the associated anomaly sub-value of the current metering box. d S is the preset transmission attenuation coefficient. a This represents the length of the transmission path between the two metering boxes.

[0040] The main technical effects of this invention are reflected in the following aspects: By setting it up in this way, fault analysis is performed based on the correlation between metering boxes and the carrier circuit. The actual carrier can be used to determine the different effects of lines, interfaces, or internal components on carrier information, thereby analyzing different fault types. By pre-building a model and using deep learning to determine the relationship between the actual carrier communication and the actual power transmission waveform between metering boxes, a rapid self-check of metering box faults can be performed. The correlation information provides a more multi-dimensional data foundation for metering box fault analysis, supporting automatic fault analysis that is no longer dependent on manual maintenance. Attached Figure Description

[0041] Figure 1This invention provides a power topology relationship model for a metering box fault analysis system based on correlation.

[0042] Figure 2 This invention presents a schematic diagram of a meter box fault analysis system architecture based on correlation relationships.

[0043] Reference numerals: 1. Metering box; 11. Carrier generator unit; 12. Carrier receiver unit; 2. Power supply terminal; 3. Power supply terminal; 100. Model training subsystem; 110. Topology management module; 120. Sample simulation module; 130. Association tagging module; 200. Fault monitoring subsystem; 210. Detection configuration module; 220. Fault analysis module. Detailed Implementation

[0044] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, so that the technical solution of the present invention can be more easily understood and mastered.

[0045] Referring to the illustration, a fault analysis system for a metering box 1 based on correlation includes a correlation model training subsystem 100, a fault monitoring subsystem 200, and a carrier generation unit 11 and a carrier receiving unit 12 configured in the metering box 1. The difference between this invention and a traditional metering box 1 lies in the installation of the carrier generation unit 11 and carrier receiving unit 12 on the traditional metering box 1. The core principle is that changes in the waveform of electricity consumption or current waveform have various causes and are not easily correlated to a specific fault condition due to a single current change. This limited information makes fault analysis difficult. However, by transmitting and receiving carrier signals, the electricity consumption information passing through the metering box 1 can be correlated and analyzed, leading to more effective and accurate conclusions in fault analysis. The carrier communication technology will not be elaborated upon here. The carrier generation unit 11 and carrier receiving unit 12 are located between the primary and secondary sides inside the metering box 1.

[0046] The association model training subsystem 100 includes a topology management module 110, a sample simulation module 120, and an association labeling module 130.

[0047] The topology management module 110 generates association information based on the power consumption topology relationships between the metering boxes 1. First, given that the power consumption relationships of the power-consuming terminals 2 are known, the topology management module 110 connects the metering boxes 1 according to the topology relationships and records the scale characteristics of each metering box 1, thus forming the association information. When a new metering box 1 is connected to the management network, the association information can be quickly collected by updating the topology management module 110. More importantly, the format of the association information allows the topology management module 110 to format the parameter information, protocols, scale, and other information of different metering boxes 1, making the association information corresponding to different metering boxes 1 usable and processable. The sample simulation module 120 uses... Based on the associated information, a carrier simulation command is generated. This command determines the content carried by the carrier information. The carrier information needs to be analyzed in two dimensions during the sample training logic. One dimension is the loss caused by the data transmission itself under the actual power consumption relationship, which we define as the solid-state loss deviation term. The other dimension is the transmission loss caused by the current readings of different metering boxes 1 under the power consumption relationship, which we define as the dynamic loss deviation term. Since both of these data will be lost during transmission, if we assume that each metering box 1 is fault-free and operating well during training, then if we can determine these two loss terms, we can determine the data transmission deviation under normal conditions and the actual power transmission deviation. Therefore, through training the model... Theoretically, this model can obtain deviations under conditions such as interface failure, dial short circuit / open circuit, and component oxidation. By analyzing deviation characteristics, the fault type can be identified and diagnosed. Specifically, the carrier simulation command includes topology simulation data. The sample simulation module 120 is configured with a topology association feature table. The topology association feature table stores several different topology simulation data based on topology association features. The sample simulation module 120 determines the topology association features of the corresponding metering box 1 based on the association information. Topology association features can be concentrated, such as the level of the meter in the power topology. For example, the metering box 1 is directly connected to the power supply terminal 3 in the power network without an upstream metering box 1, or the metering box 1 is in the power grid. In the network, the connection to the power user 2 is direct, without any downstream metering boxes 1. The quantity of upstream and downstream metering boxes 1 within the metering box 1 can all serve as topology association features. Topology association features can also include information such as the size and type of the metering boxes 1. The different number of items required for carrier transmission between different metering boxes 1 can also serve as topology association features. A topology association feature table can be pre-constructed based on these features, using them as indexes to retrieve corresponding topology simulation data. The topology simulation data can be retrieved from the topology association feature table based on the topology association features. The transmitting-side response information includes the topology simulation data, and the receiving-side response information also includes the topology simulation data.In this way, the actual transmission deviation can be determined by the deviation between sending and receiving topology simulation data. When metering box 1 receives the simulated carrier command through sample simulation module 120, it sends a transmission-side response information and sends the carrier simulation command to related metering boxes 1 through carrier generation unit 11. The sample simulation module 120 compares the deviation of the topology simulation data in the transmission-side response information and the reception-side response information to generate a solid loss deviation term and a delay deviation term. The solid loss deviation term is determined by the numerical deviation in the topology simulation data. For example, if there is transmission line loss, the impact of line loss on the data voltage value should show a regular change in the topology simulation data, such as the voltage loss ratio. The solid loss deviation term can be determined by this changing deviation, which reflects the actual loss relationship in data transmission. The delay deviation term is determined by the time delay deviation in the topology simulation data. The delay deviation term is designed based on the fundamental principle that the carrier signal transmission time and carrier signal reception time are the same in theory. If a delay occurs, it may be due to carrier signal transmission delay or different timing references between the two metering boxes 1. If a correlation analysis is performed directly, the asynchronous timing of the delay will lead to a large deviation in the final data. Therefore, by determining the delay deviation term corresponding to the time delay deviation, time synchronization between different metering boxes 1 can be achieved. The sample simulation module 120 is also configured with a dynamic simulation strategy to respond to carrier simulation commands. The dynamic simulation strategy includes:

[0048] Step A1: Synchronize the time base of metering box 1 according to the delay deviation term, and determine the intercept time period by analyzing the historical metering data of metering box 1. The delay deviation term can be used to synchronize the metering time base of different metering boxes 1. After synchronization, it is necessary to determine the intercept time period. The purpose of determining the intercept time period is that if the electricity consumption data of the intercept time period changes regularly, such as the electricity consumption of a certain period showing a stable value, then by analyzing the historical data, the regularity of subsequent data can be analyzed, and the period that can be statistically analyzed can be selected as the intercept time period. For example, during periods when there is no large electricity demand on the electricity consumption side, only the basic electricity consumption units continuously consume electricity, then the electricity consumption on the electricity consumption side has a strong regularity. Or, during periods of more intensive electricity consumption, the electricity consumption will also have a strong regularity. In this way, data from different metering boxes 1 under the above conditions can be collected.

[0049] Step A2: Obtain real-time metering data based on the captured time period to generate a dynamic metering waveform; Since the time base is the same, the metering data generated by metering box 1 is different on the sending and receiving sides, which may reflect the loss caused by line loss or other metering losses when the actual power consumption is transmitted. Therefore, the calculation of the dynamic metering waveform can be used as a basis for comparison.

[0050] Step A3: Obtain the parameter information of metering box 1, and generate a metering reference waveform based on the parameter information. Compare the deviation between the metering reference waveform and the dynamic metering waveform to generate a metering deviation waveform. The sending-side response information includes the dynamic metering waveform and the metering deviation waveform, and the receiving-side response information also includes the dynamic metering waveform and the metering deviation waveform. Obtaining the parameter information of metering box 1 allows the generation of the metering reference waveform. Based on the parameter information of metering box 1 itself and the corresponding intercepted time period, a theoretically reliable metering reference waveform can be obtained. The metering reference waveform is the weighted average of historical metering waveforms within the same time period. Both the sending and receiving sides generate dynamic metering waveforms and metering deviation waveforms for the topology simulation data.

[0051] The sample simulation module is configured with a waveform loss algorithm, which is configured as follows: Among them, G c For the dynamic loss bias term, α1 is the preset loss bias sensitivity coefficient, and g fd (x) represents the dynamic metering waveform in the sending-side response information, g jd (x) represents the dynamic metering waveform in the receiver's response information, g fp (x) represents the measurement deviation waveform in the sending-side response information, g jp (x) represents the measurement deviation waveform in the receiver response information, T c To extract the duration of the time period, a waveform loss algorithm is used to calculate the dynamic loss deviation term. This allows us to determine the losses in the power waveform caused by line loss and component impedance during power transmission. The difference between the dynamically measured waveform on the sending and receiving sides is divided by the data reliability values ​​on the sending and receiving sides. The data reliability value is obtained by calculating the measurement deviation waveform. The larger the measurement deviation waveform, the less reliable the data, and therefore the smaller the dynamic loss deviation term. Conversely, the larger the dynamic loss deviation term, the larger the value of the measurement deviation loss. By calculating the magnitude of this measurement deviation loss and integrating the entire function, the larger the dynamic loss deviation term, the greater the actual power transmission loss between the two metering boxes, and vice versa.

[0052] When metering box 1 receives a simulated carrier command through carrier receiving unit 12, it generates receiving-side response information. The sample simulation module 120 compares the transmitting-side response information and the receiving-side response information to generate a baseline loss marker. The baseline loss marker includes a dynamic loss deviation term and a solid-state loss deviation term. By marking the baseline loss marker in the corresponding correlation, the entire model can be generated. However, a problem remains: since the baseline loss marker is recorded using values ​​from two different dimensions, there is still a certain degree of complexity in data processing. The computational complexity of the model will affect the efficiency of the system's fault analysis. Therefore, the following methods are used to reprocess the data to generate a simpler model:

[0053] The association labeling module 130 configures a baseline loss label based on the association information to generate the association model of the metering bin 1; the association labeling module 130 is configured with an association labeling strategy, the association labeling strategy including

[0054] Step B1: Establish a coordinate system for a measuring box 1;

[0055] Step B2: Determine a meter box 1 as the initial coordinate based on the power topology; for example, the meter box 1 that is centered in the topology can be used as the initial coordinate, or the meter box 1 that is directly connected to the power supply side in the power topology can be used as the initial coordinate.

[0056] Step B3: Starting from the initial coordinates, determine the measurement vector corresponding to each piece of related information in sequence. G is a metric vector. s This is the solid loss deviation term;

[0057] Step B4: Determine the measurement coordinates of each measurement box 1 in the measurement coordinate system based on the measurement vector; by generating the measurement vector, the coordinates of all measurement boxes 1 in the measurement box 1 coordinate system can be determined sequentially.

[0058] Step B5: Generate the association model of metering box 1.

[0059] After the metering correlation model is generated, it is analyzed by the fault monitoring subsystem 200. The fault monitoring subsystem 200 includes a detection configuration module 210 and a fault analysis module 220. The detection configuration module 210 is used to generate detection tasks and send the detection tasks to the corresponding metering box 1. When the metering box 1 receives the detection task, it sends detection data to the metering box 1 at the target address corresponding to the detection task through the carrier generation unit 11. Different metering boxes 1 have different detection tasks under different circumstances. At the same time, the generation of detection tasks is also determined according to the coordinate relationship of the metering box 1 in the metering correlation model. The detection configuration module 210 is configured with an active detection strategy to generate the detection tasks. The active detection strategy is as follows:

[0060] Step C1: Obtain the basic information of metering box 1 and retrieve the corresponding detection task items. First, the detection task is generated based on the basic information of metering box 1. The basic information of metering box 1 reflects the number, type, and scale of equipment inside metering box 1. This information determines the items that need to be detected in metering box 1. For different numbers and types of equipment, the detection items are different. The detection items are generated by generating corresponding data. Therefore, step C1 also pre-configures a task index table, which stores several detection task items. Each detection task item is indexed by the basic information, and each detection task item is pre-configured with a task sub-load. A basic inspection level is generated based on the corresponding detection task item. Each detection task item corresponds to a task sub-load. After determining the corresponding detection task item, the corresponding task sub-load can be generated based on the detection task item to determine the load that needs to be generated by executing these tasks. This task sub-load reflects the amount of data processing resources that the detection task needs to call.

[0061] Step C2: Determine the corresponding inspection frequency based on the basic inspection level, and sum the task sub-loads to generate the task load of meter box 1; if the basic inspection level is higher, it means that the inspection demand is greater, which means that the detection item is more likely to make a mistake. Therefore, the basic inspection level can determine the inspection frequency, and then the task load can be calculated by summing.

[0062] Step C3: Substitute the inspection frequency into the inspection load function to calculate the inspection load for each metering box 1. The inspection load function is configured as follows: There is U d The inspection load, R, reflects the impact of setting up an inspection at a certain inspection time. d This represents the current workload of the metering box. The larger the workload, the greater the resulting load. W s This represents the current inspection frequency of the metering box. A higher inspection frequency means more inspections are required, resulting in longer time intervals and a greater load. e For the proposed inspection time, te-1 This represents the actual time interval of a single inspection. A longer inspection interval indicates a greater impact from not setting up inspections, and vice versa. n This refers to the planned inspection time for the nth meter box that has associated information with this meter box. On the other hand, the determination of the inspection time for a particular meter box is also related to the inspection times of other meter boxes. Theoretically, the closer the inspection times of meter boxes with associated information are, the smaller the inspection load. R n The task load of the nth meter box with associated information is defined as follows: the larger the task load of other meter boxes, the longer the inspection time is required, and the greater the impact. k is the total number of meter boxes with associated information with the current meter box, β1 is the preset inspection delay weight, and β2 is the preset inspection impact weight. The relationship between the inspection setting time and the corresponding inspection load can be determined through the inspection load function.

[0063] Step C4: Determine the planned inspection time for each metering box to minimize the total inspection load. When the total inspection load is minimized, the time allocation for each metering box's inspection task achieves optimal overall efficiency, enabling the system's inspection task settings to have overall analytical capabilities and enhancing the system's dynamic adaptability.

[0064] In addition to the proactively configured inspection plan, the detection task items in the task index table are also indexed by detection requests. The detection configuration module is configured with a passive detection strategy to generate the detection tasks. The passive detection strategy includes...

[0065] Step D1: Receive the detection request and determine the corresponding detection task. For example, the detection request is to determine the short-circuit time of a metering box in order to trace the erroneous data and complete the information. If a traditional metering box does not have the ability to record short-circuit time, it is necessary to determine whether the fault is a collective fault. If it is a collective fault, the short-circuit time of other metering boxes with short-circuit time recording function can be used as the basis. If it is not a collective fault, the numerical change of related metering boxes can be analyzed to see if it matches the characteristics of the short circuit of the metering box. If it matches, the time at that time can be used as the short-circuit time for data restoration. Therefore, the detection request can also be uploaded by the user or maintenance personnel.

[0066] Step D2: Input the coordinates of the metering box corresponding to the detection request into the metering association model to obtain the target metering box corresponding to the detection task item; the target metering box can be obtained by inputting the coordinates of the metering box into the association model according to the detection request.

[0067] Step D3: Associate the target metering box with the testing task item to generate the testing task. Associating the target metering box with the testing task item generates the testing task.

[0068] Finally, the fault analysis module, a core inventive point of this invention, is connected to the carrier receiving unit. It receives detection data from the carrier receiving unit and compares it with carrier reference data corresponding to the detection task to generate carrier deviation information. This carrier deviation information is then input into a metrological correlation model to generate fault analysis results. Specifically, the fault analysis module is configured with a fault analysis strategy to generate the fault analysis results. The fault analysis strategy includes...

[0069] Step E1: A carrier deviation database is configured, which is configured with carrier deviation features. This carrier deviation database is pre-configured, and each carrier deviation feature is pre-configured with fault anomaly sub-values ​​and associated anomaly sub-values. Because if a deviation occurs, it may be caused by a fault in the metering box itself, or it may be caused by a fault in a metering box with a related relationship. Therefore, the fault anomaly sub-values ​​indicate the degree of correlation between the feature and the local metering box, and the associated anomaly sub-values ​​indicate the degree of correlation between the feature and the related metering box. The assignment of the feature and the corresponding anomaly sub-value is obtained by referring to the fault situations in big data information and the theoretical deviations in historical metering data.

[0070] Step E2: Process the carrier deviation information through the detection task to obtain several carrier deviation data; the carrier deviation information obtained in each detection task can obtain carrier deviation data under different detection tasks. For example, the carrier deviation data corresponding to the detection task of detecting whether the current transformer is working properly reflects the working status of the current transformer.

[0071] Step E3: Identify carrier deviation features from carrier deviation data through the carrier deviation database and extract the corresponding feature identification similarity values; analyze the corresponding fault type by identifying carrier deviation features. Because the data dimension is single, it is not possible to analyze the fault type based on a single waveform change. It is only possible to estimate several possibilities. Therefore, by comparing the waveform change with the preset waveform features, the higher the similarity, the greater the possibility.

[0072] Step E4: The associated anomaly sub-values ​​corresponding to the carrier deviation characteristics are sent to the target metering box according to the metering association model as the fault anomaly sub-values ​​of the target metering box; the specific design is as follows: it also includes retrieving the transmission path length between metering boxes from the metering association model, where p c =r d χ d / S a , where p cFor the fault anomaly sub-value of the metering box as the target, r d χ represents the associated anomaly sub-value of the current metering box. d S is the preset transmission attenuation coefficient. a This represents the transmission path length between the two metering boxes. The transmission path length between the two metering boxes is determined based on the distance between their coordinates. A longer path indicates greater loss and lower correlation, while a shorter path indicates higher correlation. This allows the associated abnormal sub-values ​​of one metering box to be converted into fault abnormal sub-values ​​of other metering boxes.

[0073] Step E5: An anomaly analysis association table is configured. The anomaly analysis association table stores several fault anomaly items. Each fault anomaly item is associated with several different carrier deviation features. The similarity value of the features is used as the weight, and the fault anomaly sub-values ​​are weighted and summed to obtain the fault anomaly value of each fault anomaly item. Since the fault anomaly item is not represented by a single waveform feature, the fault anomaly value is obtained by weighted summing of the fault anomaly sub-values.

[0074] Step E6: If the fault anomaly value exceeds the preset anomaly threshold for that fault anomaly item, then that fault anomaly item is output as the fault analysis result. Therefore, through the above scheme, the fault status of the metering box can be automatically determined. Furthermore, the data from the fault analysis results, after verification as true or false, can be used as training samples and historical data. With a large amount of data, learning can improve accuracy.

[0075] Of course, the above are just typical examples of the present invention. In addition, the present invention may have many other specific embodiments. All technical solutions formed by equivalent substitution or equivalent transformation fall within the scope of protection claimed by the present invention.

Claims

1. A fault analysis system for metering boxes based on correlation, characterized in that: It includes an association model training subsystem, a fault monitoring subsystem, and a carrier generation unit and a carrier receiving unit configured in the metering box; The correlation model training subsystem includes a topology management module, a sample simulation module, and a correlation marking module. The topology management module generates correlation information based on the power topology relationships between metering boxes. The sample simulation module generates carrier simulation commands based on the correlation information. When a metering box receives a simulated carrier command through the sample simulation module, it sends a transmitting-side response and transmits the carrier simulation command to other metering boxes with correlation relationships through its carrier generation unit. When a metering box receives a simulated carrier command through its carrier receiving unit, it generates a receiving-side response. The sample simulation module compares the transmitting-side response and the receiving-side response to generate a baseline loss marker. The correlation marking module configures the baseline loss marker based on the correlation information to generate a metering correlation model. The fault monitoring subsystem includes a detection configuration module and a fault analysis module. The detection configuration module generates detection tasks and sends them to the corresponding metering boxes. When a metering box receives a detection task, it sends detection data to the metering box at the target address of the detection task through a carrier generation unit. The fault analysis module is connected to the carrier receiving unit and receives detection data from the carrier receiving unit. It compares the data with the carrier reference data corresponding to the detection task to generate carrier deviation information and inputs the carrier deviation information into the metering correlation model to generate fault analysis results.

2. The meter box fault analysis system based on correlation as described in claim 1, characterized in that: The carrier simulation command includes topology simulation data. The sample simulation module is configured with a topology association feature table. The topology association feature table stores several different topology simulation data with topology association features as indexes. The sample simulation module determines the topology association features of the corresponding metering box according to the association information and calls the corresponding topology simulation data from the topology association feature table according to the topology association features. The transmitting-side response information includes the topology simulation data, and the receiving-side response information includes the topology simulation data; The sample simulation module compares the deviations in the topology simulation data between the transmitting-side response information and the receiving-side response information to generate a solid-state loss deviation term and a delay deviation term. The solid-state loss deviation term is determined by the numerical deviation in the topology simulation data, and the delay deviation term is determined by the time delay deviation in the topology simulation data.

3. The meter box fault analysis system based on correlation as described in claim 2, characterized in that: The sample simulation module is also configured with a dynamic simulation strategy to respond to carrier simulation commands. The dynamic simulation strategy includes... Step A1: Synchronize the time base of the metering box according to the delay deviation item, and determine the intercept time period by analyzing the historical metering data of the metering box; Step A2: Obtain real-time metering data based on the captured time period to generate dynamic metering waveforms; Step A3: Obtain the parameter information of the metering box, generate a metering reference waveform based on the parameter information, compare the deviation between the metering reference waveform and the dynamic metering waveform to generate a metering deviation waveform. The sending side response information includes the dynamic metering waveform and the metering deviation waveform, and the receiving side response information includes the dynamic metering waveform and the metering deviation waveform.

4. The meter box fault analysis system based on correlation as described in claim 3, characterized in that: The sample simulation module is configured with a waveform loss algorithm, which is configured as follows: Among them, G c For the dynamic loss bias term, α1 is the preset loss bias sensitivity coefficient, and g fd (x) represents the dynamic metering waveform in the sending-side response information, g jd (x) represents the dynamic metering waveform in the receiver's response information, g fp (x) represents the measurement deviation waveform in the sending-side response information, g jp (x) represents the measurement deviation waveform in the receiver response information, T c To extract the duration of a time period; The benchmark loss label includes a dynamic loss deviation term and a solid loss deviation term.

5. The meter box fault analysis system based on correlation as described in claim 4, characterized in that: The association tagging module is configured with an association tagging strategy, which includes: Step B1: Establish a coordinate system for a measuring box; Step B2: Determine a metering box as the initial coordinate based on the power topology; Step B3: Starting from the initial coordinates, determine the measurement vector corresponding to each piece of related information in sequence. G is a metric vector. s This is the solid loss deviation term; Step B4: Determine the measurement coordinates of each measurement box in the measurement coordinate system based on the measurement vector; Step B5: Generate the measurement correlation model.

6. The meter box fault analysis system based on correlation as described in claim 1, characterized in that: The detection configuration module is configured with an active detection strategy for generating the detection task. The active detection strategy is as follows: Step C1: Obtain the basic information of the metering box, retrieve the corresponding detection task items, and generate the basic inspection level according to the corresponding detection task items. Each detection task item has a corresponding task sub-load. Step C2: Determine the corresponding inspection frequency based on the basic inspection level, and sum the task sub-loads to generate the task load for the metering box; Step C3: Input the inspection frequency into the inspection load function to calculate the inspection load for each metering box. The inspection load function is configured as follows: U d For inspection load, R d For the current workload of the metering box, W s t represents the current inspection frequency of the metering box. e For the proposed inspection time, t e-1 For the actual time of one inspection, t n R represents the planned inspection time for the nth meter box that has associated information with this meter box. n The task load is the nth metering box that has associated information with the current metering box, k is the total number of metering boxes that have associated information with the current metering box, β1 is the preset inspection delay weight, and β2 is the preset inspection impact weight. Step C4: Determine the planned inspection time for each metering box to minimize the total inspection load of the metering boxes.

7. The meter box fault analysis system based on correlation as described in claim 6, characterized in that: In step C1, a task index table is pre-configured. The task index table stores several detection task items. Each detection task item is indexed by basic information, and each detection task item is pre-configured with a task sub-load.

8. The meter box fault analysis system based on correlation as described in claim 7, characterized in that: The detection task items in the task index table are also indexed by detection requests. The detection configuration module is configured with a passive detection strategy for generating the detection tasks. The passive detection strategy includes... Step D1: Receive the detection request and determine the corresponding detection task item; Step D2: Input the coordinates of the metering box corresponding to the detection request into the metering association model to obtain the target metering box corresponding to the detection task item; Step D3: Associate the target metering box and the detection task item to generate the detection task.

9. The meter box fault analysis system based on correlation as described in claim 1, characterized in that: The fault analysis module is configured with a fault analysis strategy to generate the fault analysis results. The fault analysis strategy includes... Step E1: A carrier deviation database is configured, wherein the carrier deviation database is configured with carrier deviation features, and each carrier deviation feature is pre-configured with fault anomaly sub-values ​​and associated anomaly sub-values; Step E2: Process the carrier offset information by detecting the task item to obtain several carrier offset data; Step E3: Identify carrier deviation features from carrier deviation data using the carrier deviation database, and extract the corresponding feature identification similarity values; Step E4: Send the associated anomaly sub-values ​​corresponding to the carrier deviation characteristics to the metering box as the target metering box according to the metering association model; Step E5: An anomaly analysis association table is configured. The anomaly analysis association table stores several fault anomaly items. Each fault anomaly item is associated with several different carrier deviation features. The similarity value of the features is used as the weight, and the fault anomaly sub-values ​​are weighted and summed to obtain the fault anomaly value of each fault anomaly item. Step E6: If the fault anomaly value is greater than the preset anomaly baseline threshold for the fault anomaly item, then the fault anomaly item is output as the fault analysis result.

10. The meter box fault analysis system based on correlation as described in claim 9, characterized in that: Step E4 also includes retrieving the transmission path length between metering boxes from the metering correlation model, where p c =r d χ d / S a , where p c For the fault anomaly sub-value of the metering box as the target, r d χ represents the associated anomaly sub-value of the current metering box. d S is the preset transmission attenuation coefficient. a This represents the length of the transmission path between the two metering boxes.

Images