Multi-source data fusion electric power facility operation abnormity identification method and system

Abstract

The invention discloses a multi-source data fusion electric power facility operation abnormity identification method and system, and relates to the technical field of electric power facilities, and the method comprises the steps: constructing a tree-shaped electric power network topology which comprises a plurality of electric power nodes; respectively deploying a power monitor at each power node; detecting a node power consumption difference value of each power node layer by layer, and progressively positioning an abnormal power distribution line section along the power distribution path based on the node power consumption difference value of each power node; and performing multi-source data verification on the abnormal distribution line section, and determining an electricity stealing distribution line section according to a verification result. The problem that an existing electric power facility operation abnormity identification method depends on single data for analysis, abnormal conditions are difficult to accurately identify, and the abnormity identification efficiency is low is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power facilities, in particular to a power facility operation anomaly identification method and system based on multi-source data fusion. BACKGROUND

[0002] With the continuous development and expansion of the power system, the number and complexity of power facilities are increasing. The stable operation of power facilities is crucial to ensure the normal production and life of society. However, power facilities may have various abnormal situations, such as electricity stealing. Electricity stealing not only causes economic losses to power companies, but also affects the safe and stable operation of the power system.

[0003] The existing power facility operation anomaly identification method relies on a single type of data for analysis, which is difficult to accurately identify abnormal situations, resulting in low anomaly identification efficiency and inability to quickly locate the anomaly. SUMMARY

[0004] The embodiments of the present application provide a power facility operation anomaly identification method and system based on multi-source data fusion, which solves the technical problem that the existing power facility operation anomaly identification method relies on a single type of data for analysis, which is difficult to accurately identify abnormal situations, resulting in low anomaly identification efficiency.

[0005] The technical solution of the present application to solve the above technical problems is as follows: In a first aspect, the present application provides a power facility operation anomaly identification method based on multi-source data fusion, comprising: constructing a tree-shaped power network topology of a target area, the tree-shaped power network topology comprising a plurality of power nodes, the plurality of power nodes comprising a power entry node and a plurality of power distribution branch nodes; deploying a power monitor at each power node, the power monitor being configured to monitor the actual power consumption value of the corresponding node and calculate the theoretical power consumption value of the corresponding node; when the difference between the actual power consumption value and the theoretical power consumption value of the power entry node exceeds a preset power consumption difference value, detecting the node power consumption difference value of each power node layer by layer from the power entry node, and locating the abnormal power distribution line segment along the power distribution path based on the node power consumption difference value of each power node; verifying the abnormal power distribution line segment based on multi-source data, and determining the electricity stealing power distribution line segment based on the verification result.

[0006] In a second aspect, the present application provides a power facility operation anomaly identification system based on multi-source data fusion, comprising: a topology construction module configured to construct a tree-shaped power network topology of a target area, the tree-shaped power network topology comprising a plurality of power nodes, the plurality of power nodes comprising a power entry node and a plurality of power distribution branch nodes; The power consumption calculation module is configured to deploy power monitors at each power node, and the power monitors are configured to monitor actual power consumption values of the corresponding nodes and calculate theoretical power consumption values of the corresponding nodes. The anomaly positioning module is configured to detect node power consumption difference values of each power node layer by layer starting from the power inlet node when the difference between the actual power consumption value and the theoretical power consumption value of the power inlet node exceeds the preset power consumption difference value, and to position the abnormal power distribution line segment along the power distribution path based on the node power consumption difference values of each power node. The result verification module is configured to verify the abnormal power distribution line segment based on multi-source data, and to determine the electricity stealing power distribution line segment according to the verification result.

[0007] The present application provides one or more technical solutions, at least having the following technical effects or advantages: The power facility operation anomaly identification method and system provided by the embodiments of the present application first construct a tree-shaped power network topology to present the structure and hierarchical relationship of the power network, thereby providing a basic framework for subsequent monitoring and anomaly positioning. Second, power monitors are deployed at each power node to obtain actual power consumption values and theoretical power consumption values of each node, and preliminary judgment is made by comparison. Third, comprehensive analysis of video data and infrared thermal imaging data can more comprehensively and accurately determine whether the abnormal power distribution line segment has electricity stealing behavior, thereby determining the electricity stealing power distribution line segment. Finally, the multi-source data verification of the abnormal power distribution line segment is performed according to the verification result to determine the electricity stealing power distribution line segment, effectively avoiding the misjudgment and omission caused by a single data source, and improving the accuracy and efficiency of anomaly identification.

[0008] Through the above technical solutions, multi-source data reflects the operation state of the power facility from different angles, and is mutually complementary and verified, thereby improving the accuracy and efficiency of anomaly identification. The method can quickly and accurately locate the electricity stealing power distribution line segment, which helps power enterprises to take timely measures to reduce economic losses and ensure the safe and stable operation of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0010] Figure 1 is a flowchart of the power facility operation anomaly identification method provided by the embodiments of the present application; Figure 2 is a structural schematic diagram of the power facility operation anomaly identification system provided by the embodiments of the present application.

[0011] The components represented by each number in the attached diagram are explained below: Topology construction module 11, power consumption calculation module 12, anomaly location module 13, and result verification module 14. Detailed Implementation

[0012] This application provides a method and system for identifying power facility operation anomalies by fusing multi-source data. This addresses the technical problem that existing power facility operation anomaly identification methods rely on single data for analysis, making it difficult to accurately identify anomalies and resulting in low anomaly identification efficiency.

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0014] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0015] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid unnecessarily obscuring the description of this application. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0016] Example 1, as Figure 1 As shown in the embodiments of this application, a method for identifying operational anomalies in power facilities based on multi-source data fusion is provided, including: S10: Construct a tree-shaped power network topology for the target area. The tree-shaped power network topology includes multiple power nodes, which include power inlet nodes and multi-level distribution branch nodes. In the embodiments of the present application, first, by collecting information such as power line layout, device connection relationship, etc., a tree-shaped power network topology is constructed by using graph theory and other related knowledge. The topology structure shows the transmission path of power from the entry node to each branch node, facilitating subsequent monitoring and analysis of the operation state of the power facility.

[0017] The tree-shaped power network topology includes a plurality of power nodes, including a power entry node and a plurality of power distribution branch nodes. The power entry node is the starting point of the power entering the target area, and the plurality of power distribution branch nodes are branch points in the power distribution process. In constructing the tree-shaped power network topology, a power network modeling software is used to visually display each power node and the connection line. At the same time, a unique identifier is assigned to each power node for subsequent data collection and processing.

[0018] Exemplarily, the power entry node is taken as the root node of the tree, and each layer of power distribution branch nodes is constructed as each level of branch of the tree according to the actual connection relationship, with each node representing a specific power device or line connection point.

[0019] Specifically, step S10 in the method includes: determining the power distribution facilities in the target area, identifying the total incoming line power distribution facility and each level of power distribution facility, and identifying the connection relationship of the power distribution line between each power distribution facility; setting the total incoming line power distribution facility as the power entry node; according to the power distribution level of each level of power distribution facility, sequentially setting each level of power distribution facility as the first layer of power distribution branch node to the Nth layer of power distribution branch node to obtain the plurality of layers of power distribution branch nodes; based on the connection relationship of the power distribution line between each power distribution facility, connecting the power entry node and the plurality of layers of power distribution branch nodes to form the tree-shaped power network topology.

[0020] In the embodiments of the present application, first, the target area is surveyed and data is collected to determine all power distribution facilities in the area, record the total incoming line power distribution facility, and identify each level of power distribution facility and the connection relationship of the power distribution line between each power distribution facility, laying a foundation for constructing the tree-shaped power network topology.

[0021] Secondly, the total incoming line power distribution facility is explicitly set as the power entry node as the starting point of the entire power network. According to the power distribution level of each level of power distribution facility, each level of power distribution facility is sequentially set as the first layer of power distribution branch node to the Nth layer of power distribution branch node, thereby obtaining the plurality of layers of power distribution branch nodes, reflecting the distribution path and hierarchical structure of power from the entry to each branch.

[0022] Thirdly, based on the connection relationship between the power distribution facilities that have been identified, the power entry node and the multi-layer power distribution branch nodes are connected. In the connection process, the actual power transmission logic and line direction are followed to ensure that the tree-shaped power network topology formed can truly and accurately reflect the power network structure of the target area.

[0023] The tree-shaped power network topology constructed in the above manner provides a basic framework for subsequent monitoring of the operation state of power facilities, abnormality identification, and fault positioning. At the same time, in the construction process, with the aid of geographic information system technology, the power network topology is combined with geographic information to achieve more intuitive and efficient management and analysis.

[0024] S20: deploying a power monitor at each power node, the power monitor being configured to monitor an actual power consumption value of the corresponding node and calculate a theoretical power consumption value of the corresponding node; In the embodiment of the present application, a power monitor is installed at each power node to obtain the actual power consumption value of the corresponding node in real time. At the same time, the power monitor also calculates the theoretical power consumption value of the corresponding node according to the type, number, operation time, and rated power of the power equipment connected to the node.

[0025] Specifically, the deployment of the power monitor is different for different types of power nodes. For example, for the power entry node, since it is a key node for the power entering the target area, a power monitor with higher performance and higher stability is deployed to ensure that the large-scale power input can be accurately monitored. For the multi-layer power distribution branch node, a power monitor with appropriate specifications is selected and deployed according to the level and load of the node.

[0026] By deploying a power monitor at each power node, the power consumption of each node in the power network can be comprehensively and real-timely mastered, providing data support for subsequent abnormality identification and positioning.

[0027] Specifically, the power monitor is deployed at each power node in the following manner: The power monitor is deployed at each node in the order from the Nth layer power distribution branch node to the power entry node; Each power monitor includes a power consumption monitoring unit and a power consumption prediction unit. The power consumption monitoring unit is configured to monitor the actual power consumption value of the corresponding node, and the power consumption prediction unit is configured to calculate the theoretical power consumption value of the corresponding node.

[0028] In the embodiment of the present application, first, the power monitor is deployed in the reverse order from the Nth layer power distribution branch node to the power entry node. The end nodes of the power network are monitored first, and the monitoring is gradually pushed up to ensure that the entire power network is comprehensively covered.

[0029] Further, for each power monitor, the power consumption monitoring unit it contains will continuously monitor the actual power consumption value of the corresponding node. This unit uses high-precision sensors that can accurately measure parameters such as current and voltage, and then calculate the actual power consumption. At the same time, the power consumption prediction unit will calculate the theoretical power consumption value of the corresponding node according to the relevant parameters of the power equipment connected to the node, such as equipment type, number, running time, and rated power, combined with historical data and advanced algorithm models.

[0030] For example, for a power distribution branch node connected to multiple industrial equipment, the power consumption prediction unit will consider factors such as the rated power of the equipment, the running time, and the usage frequency of different time periods, and predict the theoretical power consumption of the node by establishing a mathematical model. In actual operation, the power monitor will compare the actual power consumption value monitored in real time with the predicted theoretical power consumption value, and once the difference between the two exceeds a certain range, it will promptly issue a warning signal to provide a basis for subsequent anomaly identification and processing.

[0031] Further, in order from the Nth layer power distribution branch node to the power inlet node, power monitors are deployed in sequence at each node, including: For the Nth layer power distribution branch node, the power consumption monitoring unit collects the power consumption values of the load equipment directly connected to the Nth layer power distribution branch node to obtain the actual power consumption value, and the power consumption prediction unit obtains the theoretical power consumption value according to the equipment operation parameters of the load equipment directly connected to the Nth layer power distribution branch node; For the i-th layer power distribution branch node, the power consumption monitoring unit aggregates the actual power consumption values of each subordinate node in the i+1th layer and the power consumption values of the load equipment directly connected to the i-th layer power distribution branch node to obtain the actual power consumption value of the i-th layer power distribution branch node, and the power consumption prediction unit aggregates the theoretical power consumption values of each subordinate node in the i+1th layer and the theoretical power consumption value obtained according to the equipment operation parameters of the load equipment directly connected to the i-th layer power distribution branch node to obtain the theoretical power consumption value of the i-th layer power distribution branch node, where 1≤i<N; For the power inlet node, the power consumption monitoring unit aggregates the actual power consumption values of the first layer power distribution branch nodes to obtain the actual power consumption value of the power inlet node, and the power consumption prediction unit aggregates the theoretical power consumption values of the first layer power distribution branch nodes to obtain the theoretical power consumption value of the power inlet node.

[0032] In the embodiments of the present application, firstly, for the Nth layer power distribution branch node, since it is at the end of the power network and directly connected with the load device, the power consumption monitoring unit collects the power consumption value of the load device as the actual power consumption value of the node. The power consumption prediction unit calculates the theoretical power consumption value according to the device operating parameters of the load device, such as the working mode, running time, load rate and so on, in combination with the rated power calculation theory of the device.

[0033] Exemplarily, the actual power consumption of the Nth layer power distribution branch node is the actual power consumption of the device, such as the actual running power consumption of the motor; the theoretical power consumption is equal to the normal power consumption value predicted based on the device parameters.

[0034] Secondly, for the ith layer power distribution branch node, and 1≤i

[0035] Exemplarily, the ith layer power distribution branch node needs to process two parts of data, and needs to collect the actual power consumption value and the theoretical power consumption value of all the lower-level child nodes of the i+1th layer downwards, and collect the power consumption of the device directly connected with the node at the same time, such as the lighting, fan and other auxiliary devices of the power distribution cabinet. Further, the calculation formula is: the actual power consumption of the ith layer = Σ (the actual power consumption of each child node of the i+1th layer) + the actual power consumption of the device at the same level; the theoretical power consumption of the ith layer = Σ (the theoretical power consumption of each child node of the i+1th layer) + the theoretical power consumption of the device at the same level.

[0036] Further, the power entry node is the starting point of the power entering the target area, and the power consumption monitoring unit collects the actual power consumption value of the first layer power distribution branch node as the actual power consumption value of the power entry node. The power consumption prediction unit collects the theoretical power consumption value of the first layer power distribution branch node to obtain the theoretical power consumption value of the power entry node.

[0037] Exemplarily, the power entry node is located at the top layer and is a data collection point. The monitoring unit is used to collect the actual power consumption values of all the power distribution branch nodes of the first layer to obtain the actual total power consumption of the whole network; the prediction unit is used to collect the theoretical power consumption values of all the power distribution branch nodes of the first layer to obtain the theoretical total power consumption of the whole network. The difference between the actual power consumption and the theoretical power consumption is the total loss of the system. When the total loss exceeds the preset threshold, the abnormal detection process is triggered.

[0038] By the above-mentioned manner, the power consumption of the power network is grasped as a whole, complete power consumption monitoring from the local to the global is realized, the overall situation can be grasped, and the specific abnormal node can be accurately located when needed. The monitor of each layer needs to monitor the devices at the current level and aggregate the data of the lower level to form a complete monitoring network, thereby providing a basis for subsequent judgment on whether the power facility is abnormal.

[0039] During the deployment of the power monitor, regular maintenance and calibration are needed. The maintenance work includes checking the hardware status of the device to ensure that the sensors, lines and the like are in normal operation; the calibration is to ensure the accuracy of the measurement data of the power monitor, so that it can truly reflect the power consumption of the power node. At the same time, in order to improve the reliability and security of the data, data encryption and backup and other technical means are also used to prevent data loss or tampering.

[0040] Among them, the theoretical power consumption value is obtained according to the device operation parameters of the load device directly connected to the Nth layer power distribution branch node, including: A first branch node is determined from the Nth layer power distribution branch node, and a power consumption prediction model of the first branch node is constructed according to the load device directly connected to the first branch node; The device operation parameters of the load device directly connected to the first branch node are called, and the power consumption prediction model of the first branch node is inputted to obtain the theoretical power consumption value of the first branch node; The theoretical power consumption values of the remaining branch nodes are obtained in the same way as the theoretical power consumption value of the first branch node, and the theoretical power consumption values of each branch node in the Nth layer power distribution branch node are obtained.

[0041] In the embodiments of the present application, first, a first branch node is determined from the Nth layer power distribution branch node. The selection of the first branch node is based on factors such as the importance of the node, the type of load or the number of connected devices. For example, a node connected to a key load device, a device type with representative or a large number of connected devices is selected as the first branch node.

[0042] Secondly, a power consumption prediction model is constructed according to the load device directly connected to the first branch node. When constructing the model, various characteristics of the load device are considered, such as the rated power, operating efficiency, working mode, starting characteristics and the like of the device. For example, a machine learning algorithm is used, such as based on neural network, combined with historical power consumption data for training, to improve the accuracy and generalization ability of the model.

[0043] Then, the device operation parameters of the load device directly connected to the first branch node are called, including the real-time operation state, operation time, load rate and the like of the device. The obtained parameters are inputted into the constructed power consumption prediction model of the first branch node, and the model is calculated and analyzed to obtain the theoretical power consumption value of the first branch node.

[0044] Then, the remaining branch nodes in the Nth layer distribution branch node are processed in the same way of obtaining the theoretical power consumption value of the first branch node. That is, the remaining branch nodes are determined in turn, a corresponding power consumption prediction model is constructed for each branch node, the operating parameters of the load devices directly connected to the branch node are called and input into the model, and thus the theoretical power consumption value of the remaining branch nodes is obtained. Finally, the theoretical power consumption values of the branch nodes in the Nth layer distribution branch node are obtained.

[0045] Through the above manner, the theoretical power consumption of each branch node in the Nth layer distribution branch node is predicted, which provides an important data basis for subsequent comparison of actual power consumption and theoretical power consumption and identification of power facility operation abnormities.

[0046] Further, the power consumption prediction model of the first branch node is constructed according to the load devices directly connected to the first branch node, and the power consumption prediction model of the first branch node comprises: According to the load devices directly connected to the first branch node, a load device list of the first branch node is constructed, and the load device list comprises a plurality of node load device models. Based on the plurality of node load device models, a plurality of device power consumption prediction components are called from a device power consumption prediction component library. The plurality of device power consumption prediction components are integrated to obtain the power consumption prediction model of the first branch node.

[0047] In the embodiment of the application, first, a load device list is constructed according to the load devices directly connected to the first branch node, and various load device models connected to the node are recorded. Since different load device models have different power consumption characteristics, the same model devices are combined to form a list containing a plurality of node load device models. For example, if the first branch node is connected to motors, lighting devices, air conditioners and the like with different powers, the specific models of the devices will be recorded in the list respectively.

[0048] Then, based on the plurality of node load device models in the load device list, a plurality of device power consumption prediction components are called from a device power consumption prediction component library. The device power consumption prediction component library is a pre-constructed resource library containing power consumption prediction components for various different load device models.

[0049] Then, the multiple device power consumption prediction components retrieved are integrated, that is, the output results of the multiple device power consumption prediction components are summed. Since the load devices connected to the first branch node can be of multiple different models, the power consumption prediction components corresponding to each model are integrated to form a model that can comprehensively consider the power consumption conditions of all the load devices. In the integration process, the cooperative working mechanism among the components is considered to ensure that the model can accurately reflect the overall power consumption condition of the first branch node. For example, different weights can be assigned to the components according to the importance and frequency of use of different load devices in the first branch node, so that the model is more accurate and reasonable in prediction.

[0050] The power consumption prediction model of the first branch node is constructed in the above manner, fully considering the characteristics of various load devices connected to the node, and laying a foundation for accurately predicting the theoretical power consumption value of the first branch node.

[0051] The construction of the device power consumption prediction component library includes the following steps: Obtain a set of load device models in the target area; Iterate through the set of load device models to determine a first load device model; Obtain a first set of device operation records based on the first load device model, each device operation record in the first set of device operation records including historical device operation parameters and historical device power consumption values; Construct a first set of sample device operation parameters based on the historical device operation parameters in the first set of device operation records, and construct a first set of sample device power consumption values based on the historical device power consumption values in the first set of device operation records; Take the first set of sample device operation parameters as input features and the first set of sample device power consumption values as supervision labels to train and generate a first device power consumption prediction component for the first load device model; Store the first load device model and the first device power consumption prediction component in the device power consumption prediction component library.

[0052] In the embodiments of the present application, a set of load device models in the target area is first obtained, which can be completed by conducting a general survey and registration of all power devices in the target area to ensure that various different types and specifications of load devices are covered. For example, the following records are provided: Motors: A brand 1LE1001-1CB03-4AA4, ABB M3BP112M; Lighting: B brand LED panel light TBS168, C brand LED tube; Air conditioners: D brand KFR-50LW, E brand MDV-140W; Production equipment: numerical control machine tool CK6140, injection molding machine HTF160X1.

[0053] Next, traverse the load device model set to determine the first load device model. When determining the first load device model, select a specific device model from the device model set as a training target, which can be selected according to the alphabetical order of the model, the importance or frequency of use of the device, etc., such as selecting "A brand 1LE1001-1CB03-4AA4 motor" as the first load device model.

[0054] Secondly, after determining the first load device model, collect the first device operation record set based on the model. Extract from the historical data of the power system, which can come from power monitors, device-embedded recording systems, etc. Each device operation record in the first device operation record set contains historical device operation parameters and historical device power consumption values. Historical device operation parameters may include device start time, stop time, running time, load rate, voltage, current, etc. Historical device power consumption values are the amount of electricity consumed by the device in the corresponding operating state.

[0055] For example, for a device of a specific model, its historical operation data is collected, and each record contains historical device operation parameters and historical device power consumption values. For example, historical device operation parameters include motor speed 1450 rpm, load rate 75%, ambient temperature 25℃, running time 8 hours, voltage level 380V, and historical device power consumption values, including actual power consumption 7.2kW under corresponding operating parameters.

[0056] Then, according to the historical device operation parameters in the first device operation record set, the first sample device operation parameter set is constructed, the parameters are sorted and classified, invalid data and outliers are removed to ensure the accuracy and reliability of the data. At the same time, according to the historical device power consumption values in the first device operation record set, the first sample device power consumption value set is constructed.

[0057] Further, using the first sample device operation parameter set as input features and the first sample device power consumption value set as supervised labels, a machine learning algorithm is used for training, such as linear regression, decision tree, support vector machine, neural network, etc., to generate a first device power consumption prediction component for the first load device model.

[0058] Finally, the first load device model and the first device power consumption prediction component are stored in the device power consumption prediction component library. When storing, use the form of a database, store the device model as the primary key and the corresponding device power consumption prediction component as the related data, which facilitates subsequent query and call, and gradually completes the construction of the device power consumption prediction component library, providing support for accurately predicting the theoretical power consumption values of each node.

[0059] S30: when the difference between the actual power consumption value and the theoretical power consumption value of the power entry node exceeds the preset power consumption difference value, detecting the node power consumption difference value of each power node layer by layer from the power entry node, and locating the abnormal power distribution line section along the power distribution path based on the node power consumption difference value of each power node; In the embodiments of the present application, when the difference between the actual power consumption value and the theoretical power consumption value of the power entry node exceeds the preset power consumption difference value, it indicates that the power network may have abnormal operation conditions. Starting from the power entry node, each power node is detected layer by layer according to the hierarchical structure of the power network. For each power node, the difference between its actual power consumption value and theoretical power consumption value, i.e. the node power consumption difference value, is calculated.

[0060] Further, the running state of each node is continuously checked by progressive detection along the power distribution path. For example, whether the working mode of the device is normal, whether the running time is as expected, whether the load rate is within a reasonable range, etc. At the same time, the working state of the power monitor is also checked to ensure that the collected power consumption data is accurate and reliable.

[0061] When locating the abnormal power distribution line section, if the node power consumption difference values of multiple nodes on a certain line segment are found to be abnormal and show a certain regularity, such as gradually increasing or decreasing along the line direction, it can be preliminarily judged that the line segment has an abnormality. Further, the line segment is checked in detail, including checking whether the connection of the line is loose, whether there is a short circuit or leakage, etc.

[0062] Through the method of layer-by-layer detection and progressive location, starting from the power entry node, the abnormal range is gradually narrowed, and the abnormal power distribution line section is finally accurately found. This provides support for timely repairing power facility faults and ensuring the stable operation of the power network.

[0063] Specifically, step S30 in the method includes: calculating the node power consumption difference value of each power distribution branch node in the first layer through the power monitor of the first layer power distribution branch node; determining an abnormal node in the first layer power distribution branch node whose node power consumption difference value exceeds the preset abnormal threshold value of the corresponding node; when the abnormal node has a lower level child node, calculating the node power consumption difference value of the lower level child node through the power monitor of the lower level child node of the abnormal node; progressively detecting layer by layer downward until it is determined that there is no lower level child node or the power consumption difference value of the lower level child node does not exceed the preset abnormal threshold value of the corresponding node; based on the power distribution connection relationship between the abnormal node and its upper level node, determining the abnormal power distribution line section.

[0064] In the embodiments of the present application, first, the power monitor of the first layer power distribution branch node calculates the node power consumption difference. The power monitor can collect the actual power consumption value of each power distribution branch node, and subtract the theoretical power consumption value obtained previously to obtain the node power consumption difference.

[0065] Secondly, the abnormal nodes in the first layer power distribution branch nodes whose node power consumption difference exceeds the corresponding node preset abnormal threshold are determined. The preset abnormal threshold is set according to a large amount of historical data and actual operation experience, and different types and importance of nodes may have different thresholds. Since the nodes exceeding the threshold may have abnormal operation conditions, they are marked and focused on.

[0066] When the abnormal node has a lower level child node, the node power consumption difference of the lower level child node is calculated through the power monitor of the lower level child node of the abnormal node. Similarly, when calculating the node power consumption difference of the lower level child node, the normal operation of the monitor and the reliability of the data are also ensured.

[0067] Then, the detection is carried out layer by layer and proceeds downward, and the conditions of the lower level child nodes are continuously checked until it is determined that there is no lower level child node or the power consumption difference of the lower level child node does not exceed the preset abnormal threshold of the corresponding node. In the process, a complete detection process and a recording mechanism are established to record the detection results of each layer node for subsequent analysis and processing.

[0068] Finally, the abnormal power distribution line section is determined based on the power distribution connection relationship between the abnormal node and its upper node. By analyzing the position and connection mode of the abnormal node in the power network and combining the detection results of the previous nodes, the line section that may have problems is determined. For example, if there are multiple abnormal nodes on the line between a certain abnormal node and its upper node, and the power consumption difference of the abnormal nodes shows a certain trend, it is preliminarily determined that the line section has an abnormality. Then, detailed on-site inspection is carried out on the line section, including checking the insulation condition of the line, whether the joint is firm, whether there are aging and damage problems, etc., so as to accurately find the abnormal reason and repair it, and ensure the stable operation of the power network.

[0069] S40: verifying the abnormal power distribution line section by multi-source data, and determining the electricity stealing power distribution line section according to the verification result.

[0070] In the embodiments of the present application, after the abnormal power distribution line section is determined, multi-source data verification is carried out. The multi-source data includes but is not limited to the metering data of the power system, the monitoring video data, the user electricity consumption behavior data, etc.

[0071] Firstly, the metering data of the power system is analyzed. The data contains real-time power, electricity, voltage, current and other information of the line. By comparing the metering data of the abnormal power distribution line section at different time periods, whether there is a power mutation, an abnormal increase or decrease of electricity, etc. is observed.

[0072] For example, if the power of a line suddenly drops significantly within a certain time period, which is not during the normal low power consumption period, this situation may be a sign of electricity stealing behavior. At the same time, by analyzing the voltage and current changes in the metering data, it can be determined whether there are abnormal fluctuations or deviations.

[0073] Secondly, the monitoring video data around the abnormal power distribution line section is retrieved to check whether there are suspicious persons nearby or whether there are illegal operations on the line.

[0074] By comprehensively analyzing multi-source data and determining the electricity stealing power distribution line section according to the verification results, if the multi-source data all point to an abnormal situation in a certain line section, and the abnormal situation meets the characteristics of electricity stealing behavior, the line section is preliminarily determined as an electricity stealing power distribution line section.

[0075] After that, further on-site investigation and investigation are needed to collect more data to ensure the accuracy of the judgment. For the determined electricity stealing power distribution line section, appropriate measures such as cutting off the power supply, notifying relevant departments for processing, etc. are taken in time to maintain the normal operation of the power system and the fair electricity environment.

[0076] Among them, the multi-source data verification is performed on the abnormal power distribution line section, and the electricity stealing power distribution line section is determined according to the verification results, including: activating the video monitoring device and the infrared thermal imaging device of the abnormal power distribution line section to obtain video data and infrared thermal imaging data of the abnormal power distribution line section; determining the electricity stealing power distribution line section in the abnormal power distribution line section according to the video data and the infrared thermal imaging data.

[0077] In the embodiments of the present application, when the abnormal power distribution line section is located through power consumption difference analysis, the video monitoring device and the infrared thermal imaging device deployed near the line section are activated immediately. The video monitoring device starts recording the real-time picture of the line section, and the infrared thermal imaging device starts collecting the temperature distribution data of the line section.

[0078] The acquired video data is analyzed by intelligent image recognition, focusing on detecting whether there are illegal line connections or private connection phenomena, whether there are abnormal external device connections to the power distribution line, and whether the line direction is consistent with the original design.

[0079] The infrared thermal imaging data is analyzed for temperature anomalies, mainly to identify whether there are abnormal heating at the line connection point, whether there are hot spots exceeding the normal working temperature, whether the temperature distribution is significantly different from the normal operating state, and the heating situation of the newly added connection point.

[0080] Secondly, the video analysis results and infrared analysis results are comprehensively evaluated. If the video detects physical alteration and the infrared detects abnormal heating, it is highly certain that electricity theft exists. If only one of them detects an anomaly, further analysis is conducted or the observation time is extended. When an anomaly is confirmed, the abnormal power distribution line segment is marked as an electricity theft power distribution line segment.

[0081] The aforementioned dual verification mechanism ensures the accuracy of electricity theft identification, avoids misjudgments that may occur if power consumption analysis is relied upon alone, and provides visual evidence support.

[0082] In summary, compared with existing technologies, this application combines electrical analysis with physical verification, which can quickly locate abnormal areas through power consumption differences and confirm specific locations of electricity theft through video and thermal imaging. At the same time, the strategy of activating video devices on demand effectively reduces system operating costs and improves monitoring efficiency.

[0083] In summary, the embodiments of this application have at least the following technical effects: This application provides a method for identifying operational anomalies in power facilities through multi-source data fusion. First, a tree-structured power network topology is constructed to present the network's structure and hierarchical relationships, providing a foundational framework for subsequent monitoring and anomaly localization. Second, power monitors are deployed at each power node to obtain the actual and theoretical power consumption values, allowing for preliminary assessment of any anomalies. Third, comprehensive analysis of video and infrared thermal imaging data enables a more comprehensive and accurate determination of whether electricity theft occurs in abnormal distribution line segments, thus identifying the theft-related distribution lines. Finally, multi-source data verification is performed on the abnormal distribution line segments, and the theft-related distribution lines are identified based on the verification results. This effectively avoids misjudgments and omissions that may arise from a single data source, improving the accuracy and efficiency of anomaly identification. Through this technical solution, multi-source data reflects the operational status of power facilities from different perspectives, complementing and verifying each other, thereby improving the accuracy and efficiency of anomaly identification. This method can quickly and accurately locate theft-related distribution line segments, helping power companies to take timely measures to reduce economic losses and ensure the safe and stable operation of the power system.

[0084] Example 2, as Figure 2 As shown, based on the same inventive concept as the power facility operation anomaly identification method using multi-source data fusion provided in Embodiment 1, this application also provides a power facility operation anomaly identification system using multi-source data fusion, including: Topology building module 11 is used to build a tree-shaped power network topology for the target area. The tree-shaped power network topology includes multiple power nodes, which include power inlet nodes and multi-level distribution branch nodes. a power consumption calculation module 12, configured to deploy a power monitor at each power node respectively, the power monitor being configured to monitor an actual power consumption value of the corresponding node and calculate a theoretical power consumption value of the corresponding node; an anomaly positioning module 13, configured to, when a difference between the actual power consumption value and the theoretical power consumption value of the power inlet node exceeds a preset power consumption difference value, detect a node power consumption difference value of each power node layer by layer starting from the power inlet node, and progressively locate an abnormal power distribution line section along a power distribution path based on the node power consumption difference value of each power node; a result verification module 14, configured to perform multi-source data verification on the abnormal power distribution line section, and determine a power stealing power distribution line section according to a verification result.

[0085] In one embodiment, the topology construction module 11 is specifically configured to: determine power distribution facilities in the target area, identify a total incoming line power distribution facility and power distribution facilities at different levels, and identify a power distribution line connection relationship between the power distribution facilities; set the total incoming line power distribution facility as a power inlet node; according to power distribution levels of the power distribution facilities at different levels, set the power distribution facilities at different levels as first-layer power distribution branch nodes to N-layer power distribution branch nodes in sequence, to obtain a plurality of layers of power distribution branch nodes; based on the power distribution line connection relationship between the power distribution facilities, connect the power inlet node and the plurality of layers of power distribution branch nodes, to form the tree-shaped power network topology.

[0086] Further, in one application embodiment, deploying a power monitor at each power node includes: deploying the power monitor at each node in sequence according to an order from the N-layer power distribution branch node to the power inlet node; Each of the power monitors includes a power consumption monitoring unit and a power consumption prediction unit, the power consumption monitoring unit being configured to monitor an actual power consumption value of the corresponding node, and the power consumption prediction unit being configured to calculate a theoretical power consumption value of the corresponding node.

[0087] Further, in one application embodiment, deploying the power monitor at each node in sequence according to an order from the N-layer power distribution branch node to the power inlet node includes: for the N-layer power distribution branch node, the power consumption monitoring unit collects a power consumption value of a load device directly connected to the N-layer power distribution branch node, to obtain an actual power consumption value, and the power consumption prediction unit obtains a theoretical power consumption value according to a device operation parameter of the load device directly connected to the N-layer power distribution branch node; For the i-th layer power distribution branch node, the power consumption monitoring unit aggregates the actual power consumption values of the i+1-th layer subordinate nodes and the power consumption values of the load devices directly connected to the i-th layer power distribution branch node to obtain the actual power consumption value of the i-th layer power distribution branch node, and the power consumption prediction unit aggregates the theoretical power consumption values of the i+1-th layer subordinate nodes and the theoretical power consumption values obtained according to the device operation parameters of the load devices directly connected to the i-th layer power distribution branch node to obtain the theoretical power consumption value of the i-th layer power distribution branch node, wherein 1≤i<N; For the power inlet node, the power consumption monitoring unit aggregates the actual power consumption values of the first layer power distribution branch nodes to obtain the actual power consumption value of the power inlet node, and the power consumption prediction unit aggregates the theoretical power consumption values of the first layer power distribution branch nodes to obtain the theoretical power consumption value of the power inlet node.

[0088] Further, in an application embodiment, the theoretical power consumption value is obtained according to the device operation parameters of the load devices directly connected to the N-th layer power distribution branch node, comprising: determining a first branch node from the N-th layer power distribution branch node, and constructing a power consumption prediction model of the first branch node according to the load devices directly connected to the first branch node; calling the device operation parameters of the load devices directly connected to the first branch node, inputting the power consumption prediction model of the first branch node, and obtaining the theoretical power consumption value of the first branch node; obtaining the theoretical power consumption values of the remaining branch nodes in the same manner as obtaining the theoretical power consumption value of the first branch node to obtain the theoretical power consumption values of the branch nodes in the N-th layer power distribution branch node.

[0089] Further, the power consumption prediction model of the first branch node is constructed according to the load devices directly connected to the first branch node, comprising: constructing a load device list of the first branch node according to the load devices directly connected to the first branch node, wherein the load device list comprises a plurality of node load device models; based on the plurality of node load device models, calling a plurality of device power consumption prediction components in a device power consumption prediction component library; integrating the plurality of device power consumption prediction components to obtain the power consumption prediction model of the first branch node.

[0090] In an application embodiment, the construction of the device power consumption prediction component library comprises: obtaining a set of load device models in the target area; traversing the set of load device models to determine a first load device model; obtain a first device running record set based on the first load device model, each device running record in the first device running record set comprising a historical device running parameter and a historical device power consumption value; construct a first sample device running parameter set according to the historical device running parameter in the first device running record set, and construct a first sample device power consumption value set according to the historical device power consumption value in the first device running record set; train a first device power consumption prediction component of the first load device model by taking the first sample device running parameter set as an input feature and taking the first sample device power consumption value set as a supervision label; store the first load device model and the first device power consumption prediction component into the device power consumption prediction component library.

[0091] In one embodiment, the anomaly positioning module 13 is specifically configured to: calculate a node power consumption difference value of each power distribution branch node in the first layer through the power monitor of the first layer power distribution branch node; determine an anomaly node in the first layer power distribution branch node whose node power consumption difference value exceeds a preset anomaly threshold of the corresponding node; when the anomaly node has a lower-level child node, calculate a node power consumption difference value of the lower-level child node through the power monitor of the lower-level child node of the anomaly node; progressively detect layer by layer downward until it is determined that there is no lower-level child node or the power consumption difference values of the lower-level child nodes do not all exceed the preset anomaly threshold of the corresponding node; determine an anomaly power distribution line segment based on the power distribution connection relationship between the anomaly node and its upper node.

[0092] In one embodiment, the result verification module 14 is specifically configured to: perform multi-source data verification on the anomaly power distribution line segment, and determine a power stealing power distribution line segment according to the verification result, comprising: activate the video monitoring device and the infrared thermal imaging device of the anomaly power distribution line segment to obtain video data and infrared thermal imaging data of the anomaly power distribution line segment; determine a power stealing power distribution line segment in the anomaly power distribution line segment according to the video data and the infrared thermal imaging data.

[0093] It should be noted that the above sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes a specific embodiment of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0094] The above description is only the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0095] The specification and drawings are only exemplary and illustrative of the present application and are to be considered within the scope of the present application. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and equivalent technology, the present application is intended to include these modifications and variations.

Claims

1. A method for identifying operational anomalies in power facilities based on multi-source data fusion, characterized in that, The method includes: Construct a tree-shaped power network topology for the target area. The tree-shaped power network topology includes multiple power nodes, which include power inlet nodes and multi-level distribution branch nodes. Power monitors are deployed at each power node to monitor the actual power consumption of the corresponding node and calculate the theoretical power consumption of the corresponding node. When the difference between the actual power consumption value and the theoretical power consumption value of the power inlet node exceeds the preset power consumption difference value, the node power consumption difference of each power node is detected layer by layer starting from the power inlet node, and the abnormal power distribution line segment is located progressively along the power distribution path based on the node power consumption difference value of each power node. Multi-source data verification is performed on the abnormal power distribution line sections, and the power theft power distribution line sections are determined based on the verification results, including: Activate the video monitoring equipment and infrared thermal imaging equipment of the abnormal power distribution line segment to acquire video data and infrared thermal imaging data of the abnormal power distribution line segment; Based on the video data and the infrared thermal imaging data, the electricity theft section is identified in the abnormal power distribution line section.

2. The method according to claim 1, characterized in that, Construct a tree-shaped power network topology for the target area, the tree-shaped power network topology including power inlet nodes and multi-layer distribution branch nodes, including: Identify the power distribution facilities within the target area, identify the main incoming power distribution facilities and power distribution facilities at all levels, and identify the power distribution line connection relationships between the power distribution facilities; Set the main incoming power distribution facility as the power inlet node; Based on the power distribution level of the power distribution facilities at each level, the power distribution facilities at each level are sequentially set as the first-level power distribution branch node up to the Nth-level power distribution branch node, thus obtaining a multi-level power distribution branch node; Based on the power distribution line connection relationship between the various power distribution facilities, the power inlet node and the multi-layer power distribution branch node are connected to form the tree-shaped power network topology.

3. The method according to claim 2, characterized in that, Power monitoring devices are deployed at each power node, including: Power monitors are deployed sequentially at each node, from the Nth layer distribution branch node to the power inlet node. Each of the power monitors includes a power consumption monitoring unit and a power consumption prediction unit. The power consumption monitoring unit is used to monitor the actual power consumption value of the corresponding node, and the power consumption prediction unit is used to calculate the theoretical power consumption value of the corresponding node.

4. The method according to claim 3, characterized in that, Following the order from the Nth-level distribution branch node to the power inlet node, power monitors are deployed sequentially at each node, including: For the Nth layer power distribution branch node, the power consumption monitoring unit collects the power consumption value of the load equipment directly connected to the Nth layer power distribution branch node to obtain the actual power consumption value, and the power consumption prediction unit obtains the theoretical power consumption value based on the equipment operating parameters of the load equipment directly connected to the Nth layer power distribution branch node. For the i-th layer distribution branch node, the power consumption monitoring unit summarizes the actual power consumption values ​​of each lower-level sub-node of the (i+1)-th layer and the power consumption values ​​of the load devices directly connected to the i-th layer distribution branch node to obtain the actual power consumption value of the i-th layer distribution branch node. The power consumption prediction unit summarizes the theoretical power consumption values ​​of each lower-level sub-node of the (i+1)-th layer and the theoretical power consumption values ​​obtained based on the equipment operating parameters of the load devices directly connected to the i-th layer distribution branch node to obtain the theoretical power consumption value of the i-th layer distribution branch node, where 1≤i <N; For the power inlet node, the power consumption monitoring unit summarizes the actual power consumption values ​​of the first-level distribution branch nodes to obtain the actual power consumption value of the power inlet node, and the power consumption prediction unit summarizes the theoretical power consumption values ​​of the first-level distribution branch nodes to obtain the theoretical power consumption value of the power inlet node.

5. The method according to claim 4, characterized in that, Based on the equipment operating parameters of the load devices directly connected to the Nth layer power distribution branch node, the theoretical power consumption value is obtained, including: The first branch node is determined from the Nth layer power distribution branch nodes, and a power consumption prediction model of the first branch node is constructed based on the load devices directly connected to the first branch node. Retrieve the device operating parameters of the load device directly connected to the first branch node, input the power consumption prediction model of the first branch node, and obtain the theoretical power consumption value of the first branch node; By obtaining the theoretical power consumption value of the first branch node in the same way as obtaining the theoretical power consumption value of the remaining branch nodes, the theoretical power consumption value of each branch node in the Nth layer of power distribution branch nodes is obtained.

6. The method according to claim 5, characterized in that, Based on the load devices directly connected to the first branch node, a power consumption prediction model for the first branch node is constructed, including: Based on the load devices directly connected to the first branch node, a load device list for the first branch node is constructed, and the load device list includes multiple node load device models; Based on the multiple node load device models, multiple device power prediction components are retrieved from the device power prediction component library; The power consumption prediction components of the multiple devices are integrated and processed to obtain the power consumption prediction model of the first branch node.

7. The method according to claim 6, characterized in that, The steps for constructing the device power consumption prediction component library include: Obtain the set of load device models within the target area; Traverse the set of load device models to determine the first load device model; A first device operation record set is obtained based on the first load device model. Each device operation record in the first device operation record set includes historical device operation parameters and historical device power consumption values. A first sample set of device operating parameters is constructed based on the historical device operating parameters in the first device operating record set, and a first sample set of device power consumption values ​​is constructed based on the historical device power consumption values ​​in the first device operating record set. Using the first sample device operating parameter set as input features and the first sample device power consumption value set as supervision labels, a first device power consumption prediction component for the first load device model is trained and generated. The first load device model is associated with the first device power consumption prediction component and stored in the device power consumption prediction component library.

8. The method according to claim 2, characterized in that, Starting from the power inlet node, the power consumption difference of each power node is detected layer by layer. Based on the power consumption difference of each power node, abnormal power distribution line segments are located progressively along the distribution path, including: The power consumption difference between each distribution branch node in the first layer is calculated using the power monitor of the first layer distribution branch node. Identify abnormal nodes in each power distribution branch node of the first layer whose power consumption difference exceeds the preset abnormal threshold of the corresponding node; When the abnormal node has a subordinate child node, the node power consumption difference of the subordinate child node is calculated through the power monitor of the subordinate child node of the abnormal node. The detection proceeds layer by layer downwards until it is determined that there are no lower-level child nodes or that the power consumption difference of the lower-level child nodes does not exceed the preset abnormal threshold of the corresponding node. Based on the power distribution connection relationship between the abnormal node and its superior node, the abnormal power distribution line segment is determined.

9. A power facility operation anomaly identification system based on multi-source data fusion, characterized in that, For performing the method according to any one of claims 1-8, comprising: The topology building module is used to build a tree-shaped power network topology for the target area. The tree-shaped power network topology includes multiple power nodes, which include power inlet nodes and multi-level distribution branch nodes. A power consumption calculation module is used to deploy power monitors at each power node. The power monitors are used to monitor the actual power consumption of the corresponding node and calculate the theoretical power consumption of the corresponding node. The anomaly location module is used to detect the node power consumption difference of each power node layer by layer starting from the power entry node when the difference between the actual power consumption value and the theoretical power consumption value of the power entry node exceeds the preset power consumption difference value. Based on the node power consumption difference value of each power node, the abnormal power distribution line segment is located progressively along the power distribution path. The result verification module is used to perform multi-source data verification on the abnormal power distribution line segment, and determine the power theft power distribution line segment based on the verification results, including: Activate the video monitoring equipment and infrared thermal imaging equipment of the abnormal power distribution line segment to acquire video data and infrared thermal imaging data of the abnormal power distribution line segment; Based on the video data and the infrared thermal imaging data, the electricity theft section is identified in the abnormal power distribution line section.

Images