Distribution network feeder terminal debugging operation and maintenance system based on multi-source data model

Abstract

The invention discloses a distribution network feeder terminal debugging operation and maintenance system based on a multi-source data model, and relates to the technical field of power systems. Through distributed sensing elements, FTU operation data and environment data are collected in real time, a topological structure and a connection relation of a power distribution network are obtained, and a multi-source sensing data calibration unit is matched, so that real-time collection and deviation calibration of the FTU operation data and the environment data are realized, the problems that traditional data collection deviation is large, and sensor abnormity is difficult to recognize are solved, and the reliability of the system is improved. According to the method, equipment operation states are modeled, a multi-dimensional model system is constructed, fault prediction, anomaly detection and health assessment are realized, faults are early warned in advance, the misjudgment rate is reduced, the equipment state management and control perspectiveness is improved, a fault point is quickly positioned in combination with power grid topology and real-time data, the fault processing time is shortened, and the influence of power failure of a distribution network is reduced. The operation and maintenance efficiency and the power supply reliability of the distribution network are remarkably improved, and the debugging and operation and maintenance intelligence of distribution network equipment is realized.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a distribution network feeder terminal commissioning and maintenance system based on a multi-source data model. Background Technology

[0002] Currently, the operation and maintenance of distribution network feeder terminals largely relies on manual inspections and post-fault repairs. Traditional operation and maintenance methods can no longer meet the requirements for efficient and precise operation and maintenance. Chinese patent application CN116647051A discloses an integrated control system for distribution automation and distribution network automation, including distribution network automation, distribution automation, and a control software platform. The distribution automation includes a fault location module, a local feeder module, a centralized feeder automation module, an integrated dispatching module, an information acquisition module, and a power grid operation and maintenance module. The automation master station module includes a distribution fault diagnosis and recovery and distribution network application software subsystem (DAS), a distribution SCADA master station system, and a distribution AM / FM / GIS application subsystem (DMS). This patent application, through the coordinated operation of distribution network automation, distribution automation, and the control software platform, can not only quickly implement remote control, isolate faulty lines, and provide power to users as soon as possible to restore normal power supply to the circuit, but also achieves a high level of automation, reliable power supply performance, and low maintenance costs.

[0003] However, while the aforementioned patents have solved some of the problems related to remote control and fault isolation in traditional manual operation and maintenance, the following issues still exist: 1. The data acquisition method is singular, relying heavily on the FTU's own basic monitoring. The integration of environmental and topology data is insufficient, and the sensors are prone to data deviation due to aging and environmental interference, resulting in low data reliability. 2. Equipment status assessment often uses fixed thresholds for judgment, lacks dynamic model support, and cannot adapt to scenarios such as equipment aging and load fluctuations. Fault warning is delayed and the rate of false alarms is high. 3. Fault location is difficult to locate quickly and accurately using simple topology data. The handling plan needs to be formulated manually, resulting in low resource scheduling efficiency, long fault handling cycle, and serious impact on the reliability of power distribution network. Summary of the Invention

[0004] The purpose of this invention is to provide a distribution network feeder terminal debugging and maintenance system based on a multi-source data model. Through multi-sensor integration and calibration units, it identifies sensor anomalies, accurately predicts faults, detects anomalies, quantifies health status, adapts to dynamic changes in equipment, realizes intelligent upgrade of distribution network feeder terminal operation and maintenance, significantly improves distribution network operation and maintenance efficiency and power supply reliability, and provides strong support for the full life cycle management of distribution network terminals, thereby solving the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The distribution network feeder terminal debugging and maintenance system based on a multi-source data model includes a data acquisition module, an equipment status assessment module, and a fault handling module. The data acquisition module, equipment status assessment module, and fault handling module interact with each other via a wireless network. The data acquisition module is used to collect FTU operation data and environmental data in real time, and at the same time, acquire the topology and connection relationship of the distribution network, and preprocess the collected FTU operation data, environmental data and distribution network topology and connection relationship data. The equipment status assessment module is used to acquire preprocessed multi-source data and, in combination with the topology and connection relationships of the distribution network, to model the equipment operating status, including fault prediction model, equipment normal behavior model and equipment health index model. The fault handling module combines the power grid topology with real-time data to quickly locate the fault point when a fault or anomaly is detected, and generates a corresponding fault handling plan based on the fault type and severity, while sending a fault notification to the operation and maintenance personnel.

[0006] Furthermore, in the data acquisition module, voltage sensors, current sensors, power calculation modules, and temperature sensors are integrated into the FTU device to collect FTU operating data, and environmental data is collected by deploying temperature and humidity sensors in the surrounding environment and connecting to a meteorological data interface.

[0007] Furthermore, the equipment status assessment module includes: The data parsing unit is used to acquire preprocessed multi-source data, extract and parse key information fields from the multi-source data, and classify and store the key information fields based on the parsing results. The condition assessment model construction unit is used to build fault prediction models, equipment normal behavior models and equipment health index models based on the parsed multi-source data, forming a multi-dimensional condition assessment model system. The equipment status assessment unit is used to generate fault prediction results, anomaly detection results, and health assessment results of the equipment status based on a multi-dimensional status assessment model system and the acquired multi-source data, and transmits them to the fault handling module via a wireless network.

[0008] Furthermore, the equipment condition assessment unit generates fault prediction results, anomaly detection results, and health assessment results. The specific process is as follows: The real-time parsed FTU operation data and environmental data are input into the fault prediction model to obtain the fault probability of each fault type in the future prediction period. When the fault probability exceeds the preset fault rate threshold, a fault warning is issued and the key parameters corresponding to the warning are recorded. The real-time parsed FTU operating data is compared with the baseline of the normal behavior model of the equipment. If the parameter exceeds the normal threshold range or the parameter change trend deviates from the normal model, it is determined that the equipment is operating abnormally and the abnormal parameter type is located. The equipment's operating years, historical failure counts, real-time parameter deviations, and environmental data are input into the equipment health index model to calculate the current equipment health index. Based on the calculation results, the corresponding health level is determined. The fault prediction results, anomaly detection results, and health assessment results are standardized and converted into a format, then transmitted in real time to the fault handling module via a wireless network, and simultaneously stored in the historical assessment database.

[0009] Furthermore, the equipment status assessment module also includes: The fault prediction model, equipment normal behavior model and equipment health index model are trained using the training set. The parameters of each model are adjusted based on the validation set until the evaluation indicators of each model reach the preset standard. Among them, the fault prediction model uses prediction accuracy and recall as evaluation indicators, the normal equipment behavior model uses the abnormal data misjudgment rate as evaluation indicator, and the equipment health index model uses the matching degree between health level and actual equipment status as evaluation indicator. Historical evaluation results and corresponding actual operating status data are extracted from the historical evaluation database at preset time intervals to determine the deviation data between the historical evaluation results and the actual operating status data of each model. The accuracy of equipment condition assessment is calculated based on the multi-dimensional condition assessment model system using deviation data. When the assessment accuracy is lower than the preset accuracy threshold, newly collected deviation data and newly added fault data are added to the training set, and the parameters of each model are adjusted.

[0010] Furthermore, the fault handling module includes: The fault location unit is used to obtain fault prediction results and anomaly detection results, combine the distribution network topology and connection relationship data to perform preliminary location of the fault point, and obtain the fault location coordinates by real-time operation data of adjacent equipment. The fault handling unit is used to query the fault handling solution database based on the fault location coordinates and the corresponding fault type, determine at least one fault handling solution, filter the fault handling solutions in combination with the real-time load status of the distribution network and the operation and maintenance resources, and perform corresponding fault handling based on the filtering results.

[0011] Furthermore, the data acquisition module also includes a multi-source sensor data calibration unit, used to perform deviation calibration and anomaly identification on the FTU operating data and environmental data collected by each sensor. The specific process is as follows: The system acquires raw monitoring data from each sensor in real time, performs deviation calibration on the raw monitoring data based on a preset sensor calibration model, and calculates the percentage of data calibration amplitude for each type of sensor during the calibration process. The calculated calibration amplitude ratio of each sensor data is compared with the preset amplitude ratio threshold, and all abnormal correction amplitude ratios exceeding the preset amplitude ratio threshold are extracted. A set of abnormal correction amplitude ratios is generated based on the sensor type. The data volume of the percentage of abnormal correction magnitudes included in each abnormal correction magnitude percentage set is counted, and the data volume is compared with the preset data number threshold. When the amount of data in the abnormal correction magnitude ratio set corresponding to any sensor type exceeds the preset data number threshold, the data deviation of that type of sensor is determined to be abnormal, and the lower limit of the preset similarity threshold range is dynamically adjusted based on the abnormal correction magnitude ratio.

[0012] Furthermore, the multi-source sensor data calibration unit also includes: Each time the lower limit of the preset similarity threshold range is dynamically adjusted, the calibration amplitude ratio of all data contained in the corresponding abnormal correction amplitude ratio set is reset, and calibration data exceeding the preset amplitude ratio threshold is re-recorded. The lower limit of the preset similarity threshold range is monitored in real time and compared with the preset lower limit threshold. When the lower limit is lower than the preset lower limit threshold, the sensor is determined to be in abnormal operating condition and an abnormal warning is issued.

[0013] Furthermore, the equipment condition assessment unit includes: The first assessment subunit is used to generate fault prediction results, anomaly detection results, and health assessment results of equipment status based on the multi-dimensional status assessment model system and the acquired multi-source data. The filtering sub-unit is used for: Test data is constructed by randomly selecting fault prediction results, anomaly detection results, and health assessment results of equipment status. Obtain information on several current wireless network transmission channels; The test data is transmitted separately based on each wireless network transmission channel, and the packet loss rate corresponding to each wireless network transmission channel is calculated. ; in, This represents the packet loss rate corresponding to the j-th wireless network transmission channel; This indicates the total number of data packets in the test data; This represents the number of data packets received by the j-th wireless network transmission channel; Based on the test data, calculate the propagation energy consumption, queuing energy consumption, and total energy consumption of each wireless network transmission channel; ; in, This represents the propagation energy consumption of the i-th data packet in the j-th channel; This represents the energy consumption for processing the i-th data packet; This represents the queuing energy consumption of the i-th data packet in the j-th channel; Let represent the real-time load rate of the j-th channel; c and η are device constants, and c>0, η>0; d is a constant, and d>0. The total energy consumption for transmitting test data on the j-th channel; Represents the signal strength of the j-th channel; e represents the natural constant; Calculate the transmission evaluation value of each wireless network transmission channel based on the packet loss rate corresponding to each wireless network transmission channel; ; in, This represents the transmission evaluation value of the j-th wireless network transmission channel; , , These represent weight coefficients, all of which are greater than 0 and have a sum of 1. This represents the actual stability score of the j-th wireless network transmission channel; This indicates the preset standard stability score; The total energy consumption for transmitting test data on the j-th channel; This indicates the preset standard energy consumption threshold; Sort the transmission evaluation values ​​of several wireless network transmission channels in descending order, and select N data transmission channels as target data transmission channels; The transmission subunit is used to package the fault prediction results, anomaly detection results, and health assessment results into several data packets; and to transmit these data packets to the fault handling module based on the target data transmission channel.

[0014] Furthermore, the method for constructing a fault handling solution database includes: Obtain the commissioning and maintenance data of the distribution network feeder terminal, preprocess the commissioning and maintenance data of the distribution network feeder terminal, and obtain the target dataset; Obtain a preset fault scenario set corresponding to the fault type. The fault scenario set includes multiple first fault scenarios. Each first fault scenario is associated with a fault type feature code, fault location topological coordinates, and fault location spatial environment label. Each first fault scenario corresponds to at least one fault handling solution. Obtain historical application data, operation and maintenance cost data, fault type adaptation data, and fault location adaptation data corresponding to the first fault scenario in the target dataset; determine the comprehensive value of the fault handling solution based on the historical application data, operation and maintenance cost data, fault type adaptation data, and fault location adaptation data; if the comprehensive value is greater than or equal to a preset comprehensive value threshold, pair the corresponding fault handling solution with the first fault scenario associated with the fault type feature code, fault location topological coordinates, and fault location spatial environment label to obtain a first pairing item; store each of the first pairing items in a blank database to obtain a fault scenario-handling solution value library; Based on the FTU operating status data, historical fault data and fault location spatial environment labels in the target dataset, a fault probability factor calculation model is constructed; the fault probability factor calculation model and parameters are stored in a blank database to obtain a fault factor dynamic calibration library. Define FTU model nodes, fault type nodes, fault location nodes, handling step nodes, and maintenance personnel nodes, and determine the relationships between each node; add attribute information to each node; construct an maintenance knowledge graph based on the nodes, relationships, and attribute information; store the maintenance knowledge graph in a blank database to obtain the maintenance knowledge graph library; Based on the aforementioned fault scenario-handling solution value library, fault factor dynamic calibration library, and operation and maintenance knowledge graph library, a fault handling solution database is constructed in the distribution network feeder terminal debugging and operation and maintenance system.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By using distributed sensing elements, real-time FTU operating data and environmental data are collected, along with the topology and connectivity of the distribution network. Combined with a multi-source sensor data calibration unit, real-time acquisition and deviation calibration of FTU operating and environmental data are achieved. This solves the problems of large data deviations and difficulty in identifying sensor anomalies in traditional methods. The system models the equipment's operating status, constructs a multi-dimensional model system, and enables fault prediction, anomaly detection, and health assessment. This provides early warning of faults, reduces false alarm rates, and improves the foresight of equipment status management. By combining grid topology and real-time data, fault points can be quickly located, shortening fault handling time, reducing the impact of power outages, and significantly improving the efficiency of distribution network operation and maintenance and power supply reliability. This achieves intelligent commissioning and operation and maintenance of distribution network equipment. Attached Figure Description

[0016] Figure 1 This is a block diagram of the distribution network feeder terminal debugging and maintenance system based on a multi-source data model according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] To address the technical problems of existing technologies, such as single data acquisition methods, lack of integration of environmental and topology data, absence of sensor data calibration mechanisms, low data reliability, lack of dynamic models, inability to adapt to equipment aging and load fluctuations, delayed fault warnings, and high false alarm rates, please refer to [link to relevant documentation]. Figure 1 This embodiment provides the following technical solution: The distribution network feeder terminal debugging and maintenance system based on a multi-source data model includes a data acquisition module, an equipment status assessment module, and a fault handling module. These modules interact via a wireless network. The data acquisition module integrates multiple types of sensors and topology data interfaces to simultaneously collect and preprocess data. The equipment status assessment module constructs and dynamically optimizes a multi-model system, combining topology data to achieve fault prediction, anomaly detection, and quantitative health assessment, overcoming the limitations of insufficient adaptability of fixed models. The fault handling module combines topology and real-time data to achieve rapid fault location, intelligent solution generation, and resource scheduling, shortening fault handling time and improving fault handling efficiency. In this embodiment, voltage sensors, current sensors, power calculation modules, and temperature sensors are integrated into the FTU device to collect FTU operating data, and environmental data is collected by deploying temperature and humidity sensors in the surrounding environment and connecting to a meteorological data interface. The data acquisition module is used to collect FTU operating data in real time, including electrical parameters such as voltage, current, and power of the FTU, as well as real-time operating parameters such as the temperature of the FTU itself, and environmental data, including temperature, humidity, and meteorological conditions. At the same time, it acquires the topology and connection relationship of the distribution network by reading the topology file of the distribution network and the equipment connection relationship database. The module also preprocesses the collected FTU operating data, environmental data, and distribution network topology and connection relationship data. The equipment status assessment module is used to acquire preprocessed multi-source data, combine the topology and connection relationship of the distribution network, and model the equipment operating status, including fault prediction model, equipment normal behavior model and equipment health index model. It constructs a fault prediction model based on historical fault data and real-time operating data, establishes equipment normal behavior model through learning from normal operating data, constructs equipment health index model, and quantitatively assesses the health status of equipment to achieve fault prediction, anomaly detection and health assessment. The fault handling module combines power grid topology and real-time data to quickly locate fault points when faults or anomalies are detected. Based on the fault type and severity, it generates corresponding fault handling solutions and simultaneously sends fault notifications to maintenance personnel, including detailed information such as fault location and type. This shortens fault handling time and improves the reliability of the distribution network. Features include: The fault location unit is used to obtain fault prediction results and anomaly detection results, combine the distribution network topology and connection relationship data to perform preliminary location of the fault point, and obtain the fault location coordinates by real-time operation data of adjacent equipment. The fault handling unit is used to query the fault handling solution database based on the fault location coordinates and the corresponding fault type, determine at least one fault handling solution, filter the fault handling solutions in combination with the real-time load status of the distribution network and the operation and maintenance resources, and perform corresponding fault handling based on the filtering results.

[0019] In this embodiment, the device status assessment module includes: The data parsing unit is used to acquire preprocessed multi-source data, extract and parse key information fields from the multi-source data, such as FTU device identifier, data acquisition timestamp, electrical parameter values, topology node connection relationship, etc., and classify and store the key information fields based on the parsing results; The condition assessment model building unit is used to construct fault prediction models, equipment normal behavior models, and equipment health index models based on the parsed multi-source data, forming a multi-dimensional condition assessment model system and providing an algorithmic basis for equipment condition assessment. The equipment status assessment unit is used to generate fault prediction results, anomaly detection results and health assessment results of the equipment status based on a multi-dimensional status assessment model system and the acquired multi-source data, and transmit them to the fault handling module via a wireless network. In this embodiment, the device status assessment module further includes: The fault prediction model, equipment normal behavior model and equipment health index model are trained using the training set. The parameters of each model are adjusted based on the validation set until the evaluation indicators of each model reach the preset standard. Among them, the fault prediction model uses prediction accuracy and recall as evaluation indicators, the normal equipment behavior model uses the abnormal data misjudgment rate as evaluation indicator, and the equipment health index model uses the matching degree between health level and actual equipment status as evaluation indicator. Historical evaluation results and corresponding actual operating status data are extracted from the historical evaluation database at preset time intervals. Deviation data between the historical evaluation results of each model and the actual operating status data are determined, including failures that were not predicted by the fault prediction model, normal data that were misjudged by the normal behavior model, and cases where the evaluation level of the health index model does not match the actual status. The accuracy of equipment condition assessment is calculated based on the multi-dimensional condition assessment model system using deviation data. When the assessment accuracy is lower than the preset accuracy threshold, newly collected deviation data and newly added fault data are added to the training set, and the parameters of each model are adjusted to ensure that the model is adapted to dynamic scenarios such as equipment aging and environmental changes.

[0020] In this embodiment, a training set is used to train the fault prediction model, the normal behavior model, and the health index model. Based on the validation set, the parameters of each trained model are adjusted until the prediction accuracy and recall of the fault prediction model, the false positive rate of abnormal data in the normal behavior model, and the matching degree between the health level and the actual equipment state of the health index model all reach the preset standards. The evaluation accuracy of the multi-dimensional state assessment model system is calculated based on the deviation data. When the evaluation accuracy is lower than the preset accuracy threshold, newly collected deviation data and newly added fault data are included in the training set, and the parameters of each model are dynamically adjusted to ensure that the model can adapt to dynamic scenarios such as equipment aging and environmental changes, and continuously ensure the accuracy and reliability of equipment state assessment.

[0021] In this embodiment, the equipment status assessment unit generates fault prediction results, anomaly detection results, and health assessment results. The specific process is as follows: The real-time parsed FTU operation data and environmental data are input into the fault prediction model to obtain the fault probability of each fault type in the future prediction period. When the fault probability exceeds the preset fault rate threshold, a fault warning is issued and the key parameters corresponding to the warning are recorded, such as the current value and temperature value that caused the warning. The real-time analyzed FTU operating data is compared with the benchmark of the normal behavior model of the equipment. If the parameters exceed the normal threshold range or the parameter change trend deviates from the normal model, such as voltage deviation rate > ±5% or current with load growth slope abnormal, the equipment is judged to be abnormal and the abnormal parameter type is located, such as voltage abnormality or temperature abnormality. The equipment's operating years, historical failure counts, real-time parameter deviations, and environmental data are input into the equipment health index model to calculate the current equipment health index. Based on the calculation results, the corresponding health level is determined. The fault prediction results, anomaly detection results, and health assessment results are standardized and converted into a format, and then transmitted to the fault handling module in real time via wireless network. At the same time, they are stored in the historical assessment database to provide data support for subsequent model optimization. In this embodiment, the fault prediction results include high-risk fault types, fault probabilities, and key parameters; the anomaly detection results include anomaly types and anomaly parameters; and the health assessment results include health indices and health levels. Based on real-time data input and output, the probability of each fault type is triggered and an early warning is generated. Combined with key parameter records, risks are identified earlier than traditional post-fault responses, reducing the fault outbreak rate. By comparing with the normal behavior model benchmark, parameters exceeding thresholds or abnormal trends are accurately determined, the anomaly type is clarified, and mishandling caused by ambiguous anomaly judgments is avoided. The health index and level are calculated based on multi-dimensional data, and the assessment results are synchronously transmitted to the fault handling module and archived in the database. This provides a basis for fault handling and accumulates samples for model optimization, adapting to dynamic changes in equipment and continuously improving the accuracy of assessment. Overall, it significantly enhances the foresight and reliability of distribution network equipment status management.

[0022] In this embodiment, the data acquisition module further includes a multi-source sensor data calibration unit, used to perform deviation calibration and anomaly identification on the FTU operating data and environmental data collected by each sensor. The specific process is as follows: The system acquires raw monitoring data from each sensor in real time, including voltage values ​​from voltage sensors and FTU temperature values ​​from temperature sensors. It performs deviation calibration on the raw monitoring data based on a preset sensor calibration model and calculates the percentage of data calibration amplitude for each type of sensor during the calibration process. In this embodiment, the data calibration amplitude ratio is calculated by the difference between the calibrated data value and the original data value and its ratio to the original data value. This is used to quantify the calibration amplitude of each sensor data and ensure that the data error after calibration is within a reasonable range corresponding to the preset error threshold. The calculated calibration amplitude ratio of each sensor data is compared with the preset amplitude ratio threshold. All abnormal correction amplitude ratios exceeding the preset amplitude ratio threshold are extracted, i.e. abnormal calibration data whose calibration amplitudes exceed the reasonable range. A set of abnormal correction amplitude ratios is generated based on the sensor type. The data volume of the percentage of abnormal correction magnitudes included in each abnormal correction magnitude percentage set is counted, and the data volume is compared with the preset data number threshold. When the amount of data in the abnormal correction magnitude ratio set corresponding to any sensor type exceeds the preset data number threshold, the data deviation of that type of sensor is determined to be abnormal, and the lower limit of the preset similarity threshold range is dynamically adjusted based on the abnormal correction magnitude ratio. In a preferred embodiment, the specific adjustment logic for dynamically adjusting the lower limit of the preset similarity threshold range is as follows: The adjusted lower limit is based on the lower limit before adjustment, and the comprehensive characteristic value of the data calibration amplitude proportion set is calculated. In this embodiment, a comprehensive calculation is performed by combining the data volume of the set, the dispersion index of the proportion of abnormal correction amplitude, the sum of the absolute deviations of each abnormal correction amplitude proportion from the set mean, and the set mean. The larger the comprehensive feature value, the more drastic the fluctuation of the sensor data calibration amplitude and the more unstable the deviation. Based on the nonlinear mapping characteristics, the comprehensive feature value is converted into an adjustment coefficient with a value range of [-1, 1], and the adjustment range of the lower limit value is nonlinearly controlled based on the adjustment coefficient. When the adjustment coefficient is close to 1, the result based on the decrease of the difference from 1 will result in a significant reduction in the optimized dynamic lower limit value compared to the initial lower limit value, in order to adapt to scenarios where the data drift of the sensing terminal is more severe. When the adjustment coefficient is close to -1, the result based on the difference from 1 approaches 2, and the optimization amplitude is slightly reduced to ensure threshold stability. Each time the lower limit of the preset similarity threshold range is dynamically adjusted, the calibration amplitude ratio of all data contained in the corresponding abnormal correction amplitude ratio set is reset, and the calibration data exceeding the preset amplitude ratio threshold is re-recorded to ensure that each adjustment is based on the latest sensor deviation data. The lower limit of the preset similarity threshold range is monitored in real time and compared with the preset lower limit threshold. When the lower limit is lower than the preset lower limit threshold, the sensor is determined to be in abnormal operating condition and an abnormal warning is issued.

[0023] In this embodiment, the device status assessment unit includes: The first assessment subunit is used to generate fault prediction results, anomaly detection results, and health assessment results of equipment status based on the multi-dimensional status assessment model system and the acquired multi-source data. The filtering sub-unit is used for: Test data is constructed by randomly selecting fault prediction results, anomaly detection results, and health assessment results of equipment status. Obtain information on several current wireless network transmission channels; The test data is transmitted separately based on each wireless network transmission channel, and the packet loss rate corresponding to each wireless network transmission channel is calculated. ; in, This represents the packet loss rate corresponding to the j-th wireless network transmission channel; This indicates the total number of data packets in the test data; This represents the number of data packets received by the j-th wireless network transmission channel; Based on the test data, calculate the propagation energy consumption, queuing energy consumption, and total energy consumption of each wireless network transmission channel; ; in, This represents the propagation energy consumption of the i-th data packet in the j-th channel; This represents the energy consumption for processing the i-th data packet; This represents the queuing energy consumption of the i-th data packet in the j-th channel; Let represent the real-time load rate of the j-th channel; c and η are device constants, and c>0, η>0; d is a constant, and d>0. The total energy consumption for transmitting test data on the j-th channel; Represents the signal strength of the j-th channel; e represents the natural constant; Calculate the transmission evaluation value of each wireless network transmission channel based on the packet loss rate corresponding to each wireless network transmission channel; ; in, This represents the transmission evaluation value of the j-th wireless network transmission channel; , , These represent weight coefficients, all of which are greater than 0 and have a sum of 1. This represents the actual stability score of the j-th wireless network transmission channel; This indicates the preset standard stability score; The total energy consumption for transmitting test data on the j-th channel; This indicates the preset standard energy consumption threshold; Sort the transmission evaluation values ​​of several wireless network transmission channels in descending order, and select N data transmission channels as target data transmission channels; The transmission subunit is used to package the fault prediction results, anomaly detection results, and health assessment results into several data packets; and to transmit these data packets to the fault handling module based on the target data transmission channel.

[0024] In this embodiment, propagation energy consumption refers to the energy consumed when data is transmitted in the channel; processing energy consumption refers to the energy consumed by the gateway / router in parsing and verifying the data; queuing energy consumption refers to the energy consumed when data is waiting to be transmitted in the buffer.

[0025] In this embodiment, the actual stability score is determined based on historical data of the wireless network transmission channel.

[0026] In this embodiment, N is a preset number, that is, the first N data transmission channels are sorted. For example, N=5, that is, the first 5 data transmission channels are selected as the target data transmission channels according to the transmission evaluation value from largest to smallest.

[0027] The working principle and beneficial effects of the above technical solution are as follows: A multi-dimensional state assessment model system and multi-source data generate fault prediction, anomaly detection, and health assessment results, enabling a comprehensive and accurate understanding of the equipment status and providing a reliable basis for subsequent processing. By randomly sampling and constructing test data, the packet loss rate and transmission evaluation value of each wireless network transmission channel are calculated, and the channel with the highest transmission evaluation value is selected as the target transmission channel. This ensures the efficiency and reliability of data transmission, reduces the risk of data loss, and improves data transmission quality. After packaging the evaluation results, they are transmitted to the fault handling module based on the target channel. Utilizing the selected high-quality channel, fast and stable data transmission can be achieved, enabling the fault handling module to obtain information promptly, respond quickly to equipment faults and anomalies, and ensure the stable operation of the equipment.

[0028] In this embodiment, the method for constructing a fault handling scheme database includes: Obtain the commissioning and maintenance data of the distribution network feeder terminal, preprocess the commissioning and maintenance data of the distribution network feeder terminal, and obtain the target dataset; Obtain a preset fault scenario set corresponding to the fault type. The fault scenario set includes multiple first fault scenarios. Each first fault scenario is associated with a fault type feature code, fault location topological coordinates, and fault location spatial environment label. Each first fault scenario corresponds to at least one fault handling solution. Obtain historical application data, operation and maintenance cost data, fault type adaptation data, and fault location adaptation data corresponding to the first fault scenario in the target dataset; determine the comprehensive value of the fault handling solution based on the historical application data, operation and maintenance cost data, fault type adaptation data, and fault location adaptation data; if the comprehensive value is greater than or equal to a preset comprehensive value threshold, pair the corresponding fault handling solution with the first fault scenario associated with the fault type feature code, fault location topological coordinates, and fault location spatial environment label to obtain a first pairing item; store each of the first pairing items in a blank database to obtain a fault scenario-handling solution value library; Based on the FTU operating status data, historical fault data and fault location spatial environment labels in the target dataset, a fault probability factor calculation model is constructed; the fault probability factor calculation model and parameters are stored in a blank database to obtain a fault factor dynamic calibration library. Define FTU model nodes, fault type nodes, fault location nodes, handling step nodes, and maintenance personnel nodes, and determine the relationships between each node; add attribute information to each node; construct an maintenance knowledge graph based on the nodes, relationships, and attribute information; store the maintenance knowledge graph in a blank database to obtain the maintenance knowledge graph library; Based on the aforementioned fault scenario-handling solution value library, fault factor dynamic calibration library, and operation and maintenance knowledge graph library, a fault handling solution database is constructed in the distribution network feeder terminal debugging and operation and maintenance system.

[0029] In this embodiment, the commissioning and maintenance data of the distribution network feeder terminal includes real-time operating data of the distribution network feeder terminal, historical maintenance work order data, equipment manufacturer's fault handling manual data, third-party related data, and GPS coordinate data of the fault location; preprocessing includes outlier removal, feature standardization, and scene labeling.

[0030] In this embodiment, a fault recovery effectiveness index is calculated based on the historical application data; an operation and maintenance cost controllability index is calculated based on the operation and maintenance cost data; a type adaptability index is calculated based on the fault type adaptation data, and a fault type adaptation matrix is ​​constructed; a location adaptability index is calculated based on the fault location adaptation data, and a fault location adaptation matrix is ​​constructed; the comprehensive value of the fault handling solution is determined based on the fault recovery effectiveness index, operation and maintenance cost controllability index, type adaptability index, and location adaptability index. The fault recovery effectiveness index includes the reciprocal of the ratio of fault recovery time to standard recovery time and the fault recovery success rate. The calculation formula is: Fault recovery effectiveness index = (1 - (actual fault recovery time / standard fault recovery time)) × 0.6 + (number of successful recovery attempts / total number of attempts) × 0.4; where the standard fault recovery time is statistically derived from historical data based on FTU model, fault type, and fault location environmental characteristics; the fault recovery success rate is calculated statistically from the fault handling result field in historical operation and maintenance work orders. The controllability indicators of operation and maintenance costs include the ratio of actual cost to budgeted cost and resource utilization rate. The calculation formula is: Controllability indicator of operation and maintenance costs = (1 - (actual cost / budgeted cost)) × 0.7 + (number of resources used effectively / total number of resources called up) × 0.3. Wherein, the budgeted cost is determined based on the historical average value of FTU model, fault type, and fault location environmental characteristics; the resource utilization rate refers to the proportion of tools, equipment and other resources called up on site that are actually used. The method for constructing the fault type adaptation matrix is ​​as follows: Extract the association records of FTU model, fault type feature code and handling solution from historical maintenance work order data; Calculate the success rate of each handling solution under each FTU model-fault type combination as the basic matching degree; Combine the fault handling manual data of the equipment manufacturer to perform weighted correction on the basic matching degree; Normalize the weighted matching degree to the [0,1] interval as the final matching degree; Type adaptability index = min(1, matching degree × 1.25), ensure that the index is 1 when the matching degree is ≥80%, and reduce the index value by 0.125 for every 10% lower.

[0031] The method for constructing the fault location adaptation matrix is ​​as follows: Define environmental feature vectors based on the spatial environmental labels of fault locations (steep slopes in mountainous areas, commercial areas in urban areas, etc.); extract the association records between environmental features of fault locations and treatment solutions from historical maintenance work order data; calculate the matching degree based on the similarity of environmental features: matching degree = Σ(weight of environmental feature i × success rate of treatment solution under this feature). The weight of environmental features is dynamically determined through regression analysis of historical fault data (e.g., the initial weight of steep slopes in mountainous areas is 0.3, and it is updated monthly based on the actual treatment effect); the location adaptability index = min(1, matching degree × 1.25), ensuring that the index is 1 when the matching degree is ≥80%, and the index value is reduced by 0.125 for every 10% decrease; update the environmental feature weights and matching degree statistics monthly based on newly added historical data.

[0032] In this embodiment, the fault location feature set includes: topological features: line segment ID, tower number, straight-line distance from substation, spacing between adjacent towers, and network connectivity; environmental features: spatial environment labels (steep slopes in mountainous areas, commercial areas in urban areas, river crossings, areas surrounding chemical industrial parks, areas surrounding substations, etc.), meteorological conditions, and transportation accessibility.

[0033] In this embodiment, the calculation formula for the fault probability factor calculation model is as follows: ; in, This represents the failure probability factor; Indicates the aging factor of the equipment; This indicates the current operating status value of the FTU; Indicates the FTU's rated state value; Indicates the influence coefficient of the spatial environment at the fault location; Indicates the current runtime; Indicates the design life of the FTU; This indicates the historical number of occurrences of the fault type-fault location combination within a preset time period; This indicates the total number of failures within the preset time period; , , The preset weights are greater than 0 and have a sum of 1.

[0034] In this embodiment, the fault location node includes: topological coordinates: line segment ID, tower number, straight-line distance from the substation, and spacing between adjacent towers; spatial environment labels: steep slopes in mountainous areas, commercial areas in urban areas, river crossings, areas surrounding chemical industrial parks, areas surrounding substations, etc.; maintenance area: responsible unit, service radius, and accessibility score; and adjacent resources: distance to the nearest maintenance point, list of available tools and equipment, and spare parts inventory status.

[0035] In this embodiment, the association relationships include: the matching relationship between fault type and fault location: based on the matching degree between the fault location feature set and the fault type; the relationship between the preconditions of fault location and processing steps: based on the necessary conditions determined by spatial environment tags; the matching relationship between fault location and the skills of maintenance personnel: based on the location environment operation qualifications in the node attributes of maintenance personnel; the association relationship between fault processing solutions and comprehensive value: using the comprehensive value in the fault scenario-processing solution value library as the attribute value of the processing solution node; and the association relationship between fault probability and fault location: using the fault probability P in the fault factor dynamic calibration library as the dynamic attribute of the fault location node.

[0036] In this embodiment, the node attributes include: FTU model node attributes: model specifications, rated parameters, manufacturer, applicable environment level, historical failure rate; maintenance personnel node attributes: skill level, number of times of handling experience, location, location environment operation qualifications, average handling efficiency; fault location node attributes: topology coordinates, spatial environment label, adjacent resources, historical failure probability, average recovery time; fault type node attributes: feature code, typical manifestation, standard handling process, average impact range; handling steps node attributes: operation content, required tools, safety requirements, average time consumption, comprehensive value score.

[0037] In this embodiment, the following functions are performed based on the operation and maintenance knowledge graph: Recommending suitable operation and maintenance personnel: Matching the location environment and operational qualifications of operation and maintenance personnel nodes according to the spatial environment tags of the fault location nodes, and calculating the matching degree based on the skill level and location of the operation and maintenance personnel, recommending personnel with a matching degree ≥ 85%; Prompting preconditions for processing steps: Automatically associating the precondition requirements of processing step nodes with the environmental characteristics of the fault location nodes, and issuing an early warning when the conditions are not met; Warning of abnormal situations: Triggering a high-level warning when the real-time fault probability P of the fault location node exceeds a dynamic threshold; Recommending the optimal processing solution: Comprehensively considering the comprehensive value score in the fault scenario-processing solution value library and the current resource availability, recommending the feasible solution with the highest comprehensive value score.

[0038] In this embodiment, the formula for calculating the comprehensive value is: Comprehensive Value = ×Normalization (fault recovery effectiveness index)+ × Normalized (1 - Operation and maintenance cost controllability index) + ×Type compatibility index+ × Location adaptability index; where: =0.3, =0.25, =0.2, =0.25; all indicators are first normalized to the [0,1] interval.

[0039] In this embodiment, the fault handling solution database integration logic uses the comprehensive value score in the fault scenario-handling solution value library as the core attribute of the handling solution node; it uses the fault probability P in the fault factor dynamic calibration library as the dynamic attribute of the fault location node and establishes a probability association with the fault type node; it adds a "handling solution-comprehensive value" relationship edge to the operation and maintenance knowledge graph, with the edge weight being the comprehensive value score; it adds a "fault location-fault probability" relationship edge, with the edge weight being the dynamically calculated P value; it constructs an indexing mechanism to support fast retrieval based on fault type feature encoding, fault location topological coordinates, and spatial environment labels; and it enables multi-condition combined queries, prioritizing the return of solutions with high comprehensive value scores and high fault probability P values.

[0040] The working principle and beneficial effects of the above technical solutions are as follows: By combining multi-dimensional data to evaluate the comprehensive value of fault handling solutions, solutions suitable for specific fault scenarios are selected, improving the accuracy and efficiency of fault handling while reducing operation and maintenance costs; Models are built based on FTU operation, historical faults, and environmental data to achieve dynamic calibration of fault probability, providing early warning of potential faults, making the solutions more suitable for different environmental conditions; Operation and maintenance knowledge graphs integrate operation and maintenance knowledge, clarify node relationships, promote knowledge sharing, and improve the collaboration efficiency of the operation and maintenance team; Multiple databases are integrated to build a comprehensive database, improving the intelligence and informatization level of the distribution network feeder terminal debugging and operation and maintenance system, ensuring stable and reliable power supply.

[0041] In this embodiment, the original sensor data is accurately corrected using a preset calibration model. The calibration effect is quantified by the proportion of data calibration amplitude, ensuring that the data error is within a reasonable range. Sensor deviation is determined based on the abnormal correction amplitude proportion set. The lower limit of the similarity threshold is dynamically adjusted by combining comprehensive feature values ​​and nonlinear mapping to adapt to data drift scenarios and avoid misjudgments caused by fixed thresholds. The lower limit of the threshold is monitored in real time, and an abnormal warning is triggered when it is lower than the threshold. At the same time, the data set is reset to ensure the timeliness of the adjustment, improve the data credibility from the source, provide reliable data support for subsequent equipment evaluation and fault handling, and significantly enhance the stability and accuracy of the system's data acquisition process.

[0042] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A distribution network feeder terminal commissioning and maintenance system based on a multi-source data model, characterized in that, It includes a data acquisition module, an equipment status assessment module, and a fault handling module. The data acquisition module, the equipment status assessment module, and the fault handling module interact with each other via a wireless network. The data acquisition module is used to collect FTU operation data and environmental data in real time. At the same time, it acquires the topology and connection relationship of the distribution network and preprocesses the collected FTU operation data, environmental data and distribution network topology and connection relationship data. The equipment status assessment module is used to acquire preprocessed multi-source data and, in combination with the topology and connection relationships of the distribution network, to model the equipment operating status, including fault prediction model, equipment normal behavior model and equipment health index model. The fault handling module combines the power grid topology with real-time data to quickly locate the fault point when a fault or anomaly is detected, and generates a corresponding fault handling plan based on the fault type and severity, while sending a fault notification to the operation and maintenance personnel.

2. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 1, characterized in that, In the data acquisition module, voltage sensors, current sensors, power calculation modules, and temperature sensors are integrated into the FTU device to collect FTU operating data, and environmental data is collected by deploying temperature and humidity sensors in the surrounding environment and connecting to a meteorological data interface.

3. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 2, characterized in that, The equipment status assessment module includes: The data parsing unit is used to acquire preprocessed multi-source data, extract and parse key information fields from the multi-source data, and classify and store the key information fields based on the parsing results. The condition assessment model construction unit is used to build fault prediction models, equipment normal behavior models and equipment health index models based on the parsed multi-source data, forming a multi-dimensional condition assessment model system. The equipment status assessment unit is used to generate fault prediction results, anomaly detection results, and health assessment results of the equipment status based on a multi-dimensional status assessment model system and the acquired multi-source data, and transmits them to the fault handling module via a wireless network.

4. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 3, characterized in that, The equipment condition assessment unit generates fault prediction results, anomaly detection results, and health assessment results. The specific process is as follows: The real-time parsed FTU operation data and environmental data are input into the fault prediction model to obtain the fault probability of each fault type in the future prediction period. When the fault probability exceeds the preset fault rate threshold, a fault warning is issued and the key parameters corresponding to the warning are recorded. The real-time parsed FTU operating data is compared with the baseline of the normal behavior model of the equipment. If the parameter exceeds the normal threshold range or the parameter change trend deviates from the normal model, it is determined that the equipment is operating abnormally and the abnormal parameter type is located. The equipment's operating years, historical failure counts, real-time parameter deviations, and environmental data are input into the equipment health index model to calculate the current equipment health index. Based on the calculation results, the corresponding health level is determined. The fault prediction results, anomaly detection results, and health assessment results are standardized and converted into a format, then transmitted in real time to the fault handling module via a wireless network, and simultaneously stored in the historical assessment database.

5. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 4, characterized in that, The equipment status assessment module also includes: The fault prediction model, equipment normal behavior model and equipment health index model are trained using the training set. The parameters of each model are adjusted based on the validation set until the evaluation indicators of each model reach the preset standard. Among them, the fault prediction model uses prediction accuracy and recall as evaluation indicators, the normal equipment behavior model uses the abnormal data misjudgment rate as evaluation indicator, and the equipment health index model uses the matching degree between health level and actual equipment status as evaluation indicator. Historical evaluation results and corresponding actual operating status data are extracted from the historical evaluation database at preset time intervals to determine the deviation data between the historical evaluation results and the actual operating status data of each model. The accuracy of equipment condition assessment is calculated based on the multi-dimensional condition assessment model system using deviation data. When the assessment accuracy is lower than the preset accuracy threshold, newly collected deviation data and newly added fault data are added to the training set, and the parameters of each model are adjusted.

6. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 5, characterized in that, The fault handling module includes: The fault location unit is used to obtain fault prediction results and anomaly detection results, combine the distribution network topology and connection relationship data to perform preliminary location of the fault point, and obtain the fault location coordinates by real-time operation data of adjacent equipment. The fault handling unit is used to query the fault handling solution database based on the fault location coordinates and the corresponding fault type, determine at least one fault handling solution, filter the fault handling solutions in combination with the real-time load status of the distribution network and the operation and maintenance resources, and perform corresponding fault handling based on the filtering results.

7. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 1, characterized in that, The data acquisition module also includes a multi-source sensor data calibration unit, used to perform deviation calibration and anomaly identification on the FTU operating data and environmental data collected by each sensor. The specific process is as follows: The system acquires raw monitoring data from each sensor in real time, performs deviation calibration on the raw monitoring data based on a preset sensor calibration model, and calculates the percentage of data calibration amplitude for each type of sensor during the calibration process. The calculated calibration amplitude ratio of each sensor data is compared with the preset amplitude ratio threshold, and all abnormal correction amplitude ratios exceeding the preset amplitude ratio threshold are extracted. A set of abnormal correction amplitude ratios is generated based on the sensor type. The data volume of the percentage of abnormal correction magnitudes included in each abnormal correction magnitude percentage set is counted, and the data volume is compared with the preset data number threshold. When the amount of data in the abnormal correction magnitude ratio set corresponding to any sensor type exceeds the preset data number threshold, the data deviation of that type of sensor is determined to be abnormal, and the lower limit of the preset similarity threshold range is dynamically adjusted based on the abnormal correction magnitude ratio.

8. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 7, characterized in that, The multi-source sensor data calibration unit also includes: Each time the lower limit of the preset similarity threshold range is dynamically adjusted, the calibration amplitude ratio of all data contained in the corresponding abnormal correction amplitude ratio set is reset, and calibration data exceeding the preset amplitude ratio threshold is re-recorded. The lower limit of the preset similarity threshold range is monitored in real time and compared with the preset lower limit threshold. When the lower limit is lower than the preset lower limit threshold, the sensor is determined to be in abnormal operating condition and an abnormal warning is issued.

9. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 3, characterized in that, The equipment condition assessment unit includes: The first assessment subunit is used to generate fault prediction results, anomaly detection results, and health assessment results of equipment status based on the multi-dimensional status assessment model system and the acquired multi-source data. The filtering sub-unit is used for: Test data is constructed by randomly selecting fault prediction results, anomaly detection results, and health assessment results of equipment status. Obtain information on several current wireless network transmission channels; The test data is transmitted separately based on each wireless network transmission channel, and the packet loss rate corresponding to each wireless network transmission channel is calculated. ； in, This represents the packet loss rate corresponding to the j-th wireless network transmission channel; This indicates the total number of data packets in the test data; This represents the number of data packets received by the j-th wireless network transmission channel; Based on the test data, calculate the propagation energy consumption, queuing energy consumption, and total energy consumption of each wireless network transmission channel; ； in, This represents the propagation energy consumption of the i-th data packet in the j-th channel; This represents the energy consumption for processing the i-th data packet; This represents the queuing energy consumption of the i-th data packet in the j-th channel; Let represent the real-time load rate of the j-th channel; c and η are device constants, and c>0, η>0; d is a constant, and d>0. The total energy consumption for transmitting test data on the j-th channel; Represents the signal strength of the j-th channel; e represents the natural constant; Calculate the transmission evaluation value of each wireless network transmission channel based on the packet loss rate corresponding to each wireless network transmission channel; ； in, This represents the transmission evaluation value of the j-th wireless network transmission channel; , , These represent weight coefficients, all of which are greater than 0 and have a sum of 1. This represents the actual stability score of the j-th wireless network transmission channel; This indicates the preset standard stability score; The total energy consumption for transmitting test data on the j-th channel; This indicates the preset standard energy consumption threshold; Sort the transmission evaluation values ​​of several wireless network transmission channels in descending order, and select N data transmission channels as target data transmission channels; The transmission subunit is used to package the fault prediction results, anomaly detection results, and health assessment results into several data packets; and to transmit these data packets to the fault handling module based on the target data transmission channel.

10. The distribution network feeder terminal debugging and maintenance system based on a multi-source data model as described in claim 6, characterized in that, The method for constructing a fault handling solution database includes: Obtain the commissioning and maintenance data of the distribution network feeder terminal, preprocess the commissioning and maintenance data of the distribution network feeder terminal, and obtain the target dataset; Obtain a preset fault scenario set corresponding to the fault type. The fault scenario set includes multiple first fault scenarios. Each first fault scenario is associated with a fault type feature code, fault location topological coordinates, and fault location spatial environment label. Each first fault scenario corresponds to at least one fault handling solution. Obtain historical application data, operation and maintenance cost data, fault type adaptation data, and fault location adaptation data corresponding to the first fault scenario in the target dataset; determine the comprehensive value of the fault handling solution based on the historical application data, operation and maintenance cost data, fault type adaptation data, and fault location adaptation data; if the comprehensive value is greater than or equal to a preset comprehensive value threshold, pair the corresponding fault handling solution with the first fault scenario associated with the fault type feature code, fault location topological coordinates, and fault location spatial environment label to obtain a first pairing item; store each of the first pairing items in a blank database to obtain a fault scenario-handling solution value library; Based on the FTU operating status data, historical fault data and fault location spatial environment labels in the target dataset, a fault probability factor calculation model is constructed; the fault probability factor calculation model and parameters are stored in a blank database to obtain a fault factor dynamic calibration library. Define FTU model nodes, fault type nodes, fault location nodes, handling step nodes, and maintenance personnel nodes, and determine the relationships between each node; add attribute information to each node; construct an maintenance knowledge graph based on the nodes, relationships, and attribute information; store the maintenance knowledge graph in a blank database to obtain the maintenance knowledge graph library; Based on the aforementioned fault scenario-handling solution value library, fault factor dynamic calibration library, and operation and maintenance knowledge graph library, a fault handling solution database is constructed in the distribution network feeder terminal debugging and operation and maintenance system.

Images