Power equipment identification and state monitoring method and system

Abstract

The invention discloses a power equipment identification and state monitoring method and system, and belongs to the field of power equipment monitoring, and the method comprises the steps: basic data collection and equipment identification; classifying and storing the data; performing equipment state analysis and anomaly detection; performing abnormal alarm and maintenance early warning; and recording and optimizing data. The device state is monitored in real time, video data and sensor data are used for comprehensive analysis, abnormity or damage of the device can be found in time, the operation efficiency and safety of the power grid device can be improved, and shutdown time and loss caused by device faults are reduced; and the accuracy and the intelligent level of the system can be gradually improved through a continuous optimization algorithm and data analysis subsequently.

Description

Technical Field

[0001] The present invention relates to the field of power equipment monitoring, and in particular to a method and system for power equipment identification and status monitoring. Background Art

[0002] Monitoring of power equipment plays an extremely important role in the power system, which is directly related to the safe, stable and efficient operation of the power system and the reliability of energy supply. It is of great significance to ensure the safe and stable operation of the power system, improve energy efficiency, reduce operating costs and promote the construction of smart grids. By real-time monitoring of the operating status of power equipment, potential faults or abnormalities can be discovered in time, so that preventive measures or timely maintenance can be taken to avoid production interruptions and losses caused by sudden equipment shutdown, which helps to reduce power outages caused by equipment failures and improve power supply reliability. Failure or improper operation of power equipment may cause safety accidents such as fire and electric shock. Effective monitoring can timely discover safety hazards such as overheating and leakage of equipment, and take measures to deal with them, thereby protecting the safety of staff and the public. The large amount of data generated by power equipment monitoring provides managers with a rich source of information. By analyzing these data, managers can understand the overall condition, operating trends and potential problems of the equipment.

[0003] At present, the monitoring of power equipment still has the following difficulties and challenges in identifying abnormalities: High-dimensional and large-scale data: The data generated by the power system is usually high-dimensional and large-scale, including multiple variables such as current, voltage, and power. Processing these large-scale data sets requires efficient algorithms and computing resources.

[0004] Nonlinear relationships: The behavior of devices in power systems often exhibits nonlinear relationships, which means that traditional linear models may not be able to capture the complexity of the system well.

[0005] Real-time requirements: The power system needs to respond to equipment anomalies in a timely manner to prevent accidents and failures. Therefore, the anomaly identification system needs to work in real time or quasi-real time.

[0006] Multi-source heterogeneous data: The power system involves multiple devices and sensors, each of which may have different data sources and data formats. Integrating and processing these heterogeneous data is a challenge.

[0007] Potentially hidden: Some equipment anomalies may not appear noticeable in the data or may vary slightly in normal operations, which increases the difficulty of detecting these anomalies.

[0008] Complex topology: The power system has a complex network topology, including substations, transmission lines, transformers, etc. Anomalies may propagate in these topologies, increasing the difficulty of anomaly identification.

[0009] External interference: The power system is susceptible to interference from external factors, such as weather changes and power supply fluctuations, which may cause data fluctuations and increase the risk of misjudgment of anomaly detection. Summary of the invention

[0010] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a method and system for identifying and monitoring the status of electric equipment.

[0011] The objective of the present invention is achieved through the following technical solutions: A method for identifying and monitoring the status of electric power equipment, comprising: Basic data collection and equipment identification: Set up cameras in key areas of power equipment to collect video streams, identify equipment through object detection and recognition algorithms, and establish a basic identification model for equipment; collect power grid equipment status data in real time through sensors, and associate the real-time data collected by sensors with video streams to achieve the fusion of video streams and equipment status data; Equipment status analysis and anomaly detection: Combine equipment identification data and equipment status data to establish a baseline status model for equipment operation; identify and analyze abnormal situations, and establish analysis models according to regional divisions; analyze the equipment operation status by using machine learning or deep learning algorithms, and use multi-model anomaly recognition algorithms to identify abnormal situations that do not match the baseline status; Abnormal alarm and maintenance warning: When an abnormal state is detected, an alarm is immediately issued and the alarm message is notified to the message push platform. The message push platform pushes the alarm to the relevant operation and maintenance personnel, safety supervision and biotechnology departments by SMS or email according to the alarm level and notification object; and generates an abnormal report providing detailed information; Data recording and optimization: Record and analyze historical data and continuously optimize the recognition algorithm.

[0012] Furthermore, the key area of ​​power equipment is around the power grid infrastructure, including transformers, cable lines and transmission towers.

[0013] Furthermore, the abnormal conditions include equipment displacement, equipment deformation, covering on the equipment, equipment leakage, equipment oil pollution, open flames and smoke, and biological intrusion.

[0014] Furthermore, the analysis model includes a common analysis model, a line analysis model, a tower analysis model and a transformer analysis model.

[0015] Furthermore, the reference state model includes: Calculate the mean: , in, is the mean, n is the sample size, is the ith observation; Calculate the standard deviation: , Where σ is the standard deviation, is the mean, n is the sample size, is the ith observation; Threshold Statistics: , in, is the statistical threshold, d is the historical threshold, and c is the coefficient of deviation; Detecting outliers: , Among them, p is the abnormal result.

[0016] Furthermore, the use of a multi-model anomaly recognition algorithm to identify anomalies that are inconsistent with a baseline state is specifically as follows: a random tree is constructed in a high-dimensional space, an anomaly score is calculated by the length of the path, and then a situation where the device state is inconsistent with the baseline state model is identified.

[0017] Furthermore, the path length calculation formula is: , Among them, l(x) is the path length of sample point x in the tree, m is the number of leaf nodes in the tree, L is the expected value of the average path length, and h(x, T) is the path length of sample point x in the Tth tree; The anomaly score calculation formula is: , Among them, S(x) is the anomaly score, E(h(x)) is the average path length of sample point x in all trees; h(x) is the path length of sample point x in the tree.

[0018] Furthermore, the collected video stream is based on the Internet of Things network and the intranet dedicated line. The device image is captured from the video stream at a regular interval. Computer vision technology is used to identify the device using object detection and recognition algorithms to establish a basic recognition model for the device.

[0019] Furthermore, the data obtained from basic data collection and device identification are classified and stored. Specifically, the data classification and storage is as follows: the device image data, video stream data, and analysis samples are stored in an unstructured database in a file format, the sensor real-time collected data is stored in a structured database, and data cleaning is performed on a rolling basis to provide data support.

[0020] A power equipment identification and status monitoring system comprises a basic acquisition layer, a data storage layer, a model analysis layer and an abnormal warning and push layer, wherein the basic acquisition layer collects image data of power equipment and power grid equipment status data; the data storage layer stores the image data and the power grid equipment status data respectively; the model analysis layer identifies abnormal situations by means of an identification model, and identifies abnormal situations that are inconsistent with a reference state by means of the analysis model; when the abnormal warning and push layer identifies an abnormal situation that is inconsistent with a reference state, an alarm is immediately sounded, and the alarm message is notified to a message push platform, and the message push platform pushes the alarm to relevant personnel by SMS or email according to the alarm level and the notification object.

[0021] Beneficial effects of the present invention: The present invention monitors the equipment status in real time and uses video data and sensor data for comprehensive analysis, which can promptly detect equipment abnormalities or damage, help improve the operating efficiency and safety of power grid equipment, and reduce downtime and losses caused by equipment failures; subsequently, the accuracy and intelligence level of the system can be gradually improved through continuous optimization of algorithms and data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0023] Figure 1 It is a flow chart of the method of the present invention.

[0024] Figure 2 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0025] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] Embodiment 1, as Figure 1 As shown, a method for identifying and monitoring the status of electric equipment includes: 1. Basic data collection and equipment identification: Cameras are set up in key areas of power equipment to collect video streams. The collected video streams are based on the Internet of Things network and the intranet dedicated line. Equipment images are captured from the video stream at regular intervals. Computer vision technology is used to identify equipment using object detection and recognition algorithms, and a basic recognition model for the equipment is established.

[0028] The power grid equipment status data is collected in real time through sensors, and the real-time data collected by the sensors is associated with the video stream to achieve the fusion of the video stream and the equipment status data.

[0029] Among them, the key areas of power equipment are around the power grid infrastructure, including transformers, cable lines, transmission towers and other infrastructure; the power grid equipment status data includes status parameters such as temperature, current, and voltage, which are transmitted to the data storage layer through a dedicated network.

[0030] 2. Data classification storage: Device image data, video stream data, and analysis samples are stored in an unstructured database in file format. Real-time sensor data is stored in a structured database, and data cleaning is performed on a rolling basis to provide data support.

[0031] 3. Equipment status analysis and anomaly detection: Combine equipment identification data and equipment status data to establish a baseline state model for equipment operation; identify and analyze abnormal conditions such as equipment displacement, equipment deformation, equipment covering, equipment leakage, equipment oil pollution, open flame smoke, and biological intrusion, and establish a public analysis model, line analysis model, tower analysis model, and transformer analysis model according to regional divisions; analyze the equipment operating status by using machine learning or deep learning algorithms, and use a multi-model anomaly recognition algorithm to identify abnormal conditions that do not match the baseline state.

[0032] Among them, the baseline state model is a model used to represent the normal operation state of the equipment. It uses a moving average or an exponentially weighted moving average to capture the trend of the equipment state. It assumes that the data of the normal operation of the model should be near the mean, and most of the data should fall within a standard deviation range. Outliers are usually defined as data points that deviate from the mean by more than a certain threshold. The model calculation process is as follows: Calculate the mean: , in, is the mean, n is the sample size, is the ith observation; Calculate the standard deviation: , Where σ is the standard deviation, is the mean, n is the sample size, is the ith observation; Threshold Statistics: , in, is the statistical threshold, d is the historical threshold, and c is the deviation coefficient. The outlier is defined as a standard deviation that deviates from the mean by a certain multiple. The deviation coefficient is adjusted reasonably according to the accuracy of the anomaly detection result by taking the arithmetic average from the historical threshold and weighting the deviation coefficient, ranging from 0.5 to 20.

[0033] Detecting outliers: , Among them, p is the abnormal result.

[0034] Among them, the multi-model anomaly recognition algorithm is used to identify abnormal situations that do not match the baseline state. Specifically, a random tree is constructed in a high-dimensional space, and the anomaly score is calculated by the length of the path, thereby identifying situations where the device state does not match the baseline state model, that is, the abnormal state. By constructing a random tree to divide the data, the anomaly points require fewer segmentation steps. In this algorithm, the path of the outlier in the tree is shorter, and the path of the normal value in the tree is longer.

[0035] The path length calculation formula is: , Among them, l(x) is the path length of sample point x in the tree, m is the number of leaf nodes in the tree, L is the expected value of the average path length, and h(x, T) is the path length of sample point x in the Tth tree; The anomaly score calculation formula is: , Among them, S(x) is the anomaly score, E(h(x)) is the average path length of sample point x in all trees; h(x) is the path length of sample point x in the tree.

[0036] 4. Abnormal alarm and maintenance warning: When an abnormal state is detected, an alarm is immediately issued and the alarm message is notified to the message push platform. The message push platform pushes the alarm to relevant operation and maintenance personnel, safety supervision and biotechnology departments by SMS or email according to the alarm level and notification object; and generates a detailed abnormal report, including the location of the abnormal equipment, possible problem types and maintenance suggestions.

[0037] 5. Data recording and optimization: Record and analyze historical data to provide data support for optimization of recognition algorithms.

[0038] Embodiment 2, as Figure 2 As shown, a power equipment identification and status monitoring system includes a basic acquisition layer, a data storage layer, a model analysis layer, and an abnormal warning and push layer.

[0039] The basic collection layer collects image data of power equipment and status data of power grid equipment. Specifically, cameras are set up around the power grid infrastructure to cover key areas, including transformers, cable lines, transmission towers and other locations. The collection sensors or monitoring equipment of transformers, cable lines and transmission towers collect real-time operating data of power grid equipment, such as temperature, current, voltage and other status parameters, and transmit them to the data storage layer through a dedicated network.

[0040] The data storage layer stores image data and power grid equipment status data separately; specifically, equipment image data, video stream data, and analysis samples are stored in an unstructured database in file format, real-time collected data is stored in a structured database, and data cleaning is performed on a rolling basis to provide data support for the model analysis layer.

[0041] The model analysis layer identifies abnormal situations through the recognition model, and identifies abnormal situations that are inconsistent with the baseline state through the analysis model; specifically, the real-time data collected by the sensor is associated with the video stream to achieve the fusion of the video stream and the equipment status data. Combine object recognition data and status monitoring data to establish a baseline state model for equipment operation. Identify and analyze equipment displacement, equipment deformation, equipment covering, equipment leakage, equipment oil pollution, open flames and smoke, biological intrusion, etc., and establish a public analysis model, line analysis model, tower analysis model, and transformer analysis model; use machine learning or deep learning algorithms to analyze the equipment operation status and identify abnormal situations that are inconsistent with the baseline state.

[0042] When the abnormal warning and push layer identifies an abnormal situation that is inconsistent with the baseline state, an alarm will be issued immediately and the alarm message will be notified to the message push platform. The message push platform will push the alarm to relevant personnel by SMS or email based on the alarm level and notification object.

[0043] The present invention aims at the difficulties and challenges of abnormal identification of power equipment and makes the following improvements: 1. High-dimensional and large-scale data processing: The algorithm effectively processes large-scale and high-dimensional data generated by the power system through efficient data processing technology and computing resource management. For example, unstructured databases are used to store device image data and video stream data, and structured databases are used to store real-time collected power grid equipment operation data. At the same time, data cleaning is performed on a rolling basis to optimize data storage and processing.

[0044] 2. Identification of nonlinear relationships: Using machine learning or deep learning algorithms, this technology can capture the nonlinear relationships of equipment behavior in the power system, thereby improving the ability to identify and analyze complex system behavior.

[0045] 3. Meeting real-time requirements: The algorithm design takes into account real-time or quasi-real-time working conditions, and ensures that the power system can respond to equipment anomalies in a timely manner through fast data processing and anomaly detection technology.

[0046] 4. Integration of multi-source heterogeneous data: By associating the real-time data collected by sensors with the video stream, the fusion of video stream and device status data is achieved, effectively integrating and processing heterogeneous data from different devices and sensors.

[0047] 5. Identification of potentially hidden anomalies: Using complex models based on machine learning or deep learning, algorithms are able to identify anomalies in the data that are not significant, even if these anomalies are small changes in normal operations.

[0048] 6. The problem of obtaining labeled samples: By continuously collecting and analyzing historical data, the algorithm can optimize the recognition model and improve the accuracy of anomaly recognition even when labeled samples are relatively difficult to obtain.

[0049] 7. Consideration of complex topological structures: The algorithm considers the complex network topological structures of the power system by establishing a baseline state model of equipment operation and adopting a multi-model anomaly recognition algorithm to cope with the possible propagation of anomalies in these structures.

[0050] 8. Handling of external interference: The algorithm can reduce the impact of data fluctuations caused by external factors such as weather changes and power supply fluctuations on anomaly detection by continuously optimizing and adjusting the accuracy of anomaly detection results and predicting potential failures by analyzing historical data.

[0051] It should be noted that, for the above-mentioned various method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0052] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0053] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a ROM, a RAM, etc.

[0054] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for identifying and monitoring the status of electric power equipment, characterized in that: include: Basic data collection and equipment identification: Set up cameras in key areas of power equipment to collect video streams, identify equipment through object detection and recognition algorithms, and establish a basic identification model for equipment; The sensors are used to collect the status data of power grid equipment in real time, and the real-time data collected by the sensors is associated with the video stream to achieve the fusion of the video stream and the equipment status data; Equipment status analysis and anomaly detection: Combine equipment identification data and equipment status data to establish a baseline status model for equipment operation; identify and analyze abnormal situations, and establish analysis models according to regional divisions; analyze the equipment operation status by using machine learning or deep learning algorithms, and use multi-model anomaly recognition algorithms to identify abnormal situations that do not match the baseline status; Abnormal alarm and maintenance warning: When an abnormal state is detected, an alarm is immediately issued and the alarm message is notified to the message push platform. The message push platform pushes the alarm to the relevant operation and maintenance personnel, safety supervision and biotechnology departments by SMS or email according to the alarm level and notification object; and generates an abnormal report providing detailed information; Data recording and optimization: Record and analyze historical data to provide data support for optimization of recognition algorithms.

2. A method for identifying and monitoring the state of electric power equipment according to claim 1, characterized in that: The key areas of power equipment are around the power grid infrastructure, including transformers, cable lines and transmission towers.

3. A method for identifying and monitoring the state of electric equipment according to claim 1, characterized in that: The abnormal conditions include equipment displacement, equipment deformation, covering on equipment, equipment leakage, equipment oil pollution, open flames and smoke, and biological intrusion.

4. A method for identifying and monitoring the state of electric power equipment according to claim 1, characterized in that: The analysis model includes a common analysis model, a line analysis model, a tower analysis model and a transformer analysis model.

5. The method for identifying and monitoring the state of electric power equipment according to claim 1, characterized in that: The reference state model includes: Calculate the mean: , in, is the mean, n is the sample size, is the ith observation; Calculate the standard deviation: , Where σ is the standard deviation, is the mean, n is the sample size, is the ith observation; Threshold Statistics: , in, is the statistical threshold, d is the historical threshold, and c is the coefficient of deviation; Detecting outliers: , Among them, p is the abnormal result.

6. A method for identifying and monitoring the state of electric power equipment according to claim 1, characterized in that: The multi-model anomaly recognition algorithm is used to identify anomalies that are inconsistent with the baseline state. Specifically, a random tree is constructed in a high-dimensional space, an anomaly score is calculated by the length of the path, and then the situation where the device state is inconsistent with the baseline state model is identified.

7. A method for identifying and monitoring the state of electric power equipment according to claim 6, characterized in that: The path length calculation formula is: ， Among them, l(x) is the path length of sample point x in the tree, m is the number of leaf nodes in the tree, L is the expected value of the average path length, and h(x, T) is the path length of sample point x in the Tth tree; The anomaly score calculation formula is: ， Among them, S(x) is the anomaly score, E(h(x)) is the average path length of sample point x in all trees; h(x) is the path length of sample point x in the tree.

8. A method for identifying and monitoring the state of electric power equipment according to claim 1, characterized in that: The collected video stream is based on the Internet of Things network and the intranet dedicated line. The device image is captured from the video stream at a regular interval. Computer vision technology is used to identify the device using object detection and recognition algorithms to establish a basic recognition model for the device.

9. A method for identifying and monitoring the state of electric power equipment according to claim 1, characterized in that: The data obtained from basic data collection and device identification are classified and stored. Specifically, the device image data, video stream data, and analysis samples are stored in an unstructured database in file format, the sensor real-time collected data is stored in a structured database, and data cleaning is performed on a rolling basis to provide data support.

10. A power equipment identification and status monitoring system, applying a power equipment identification and status monitoring method according to any one of claims 1 to 9, characterized in that: It includes a basic acquisition layer, a data storage layer, a model analysis layer and an abnormal warning and push layer. The basic acquisition layer collects image data of power equipment and power grid equipment status data; the data storage layer stores image data and power grid equipment status data respectively; the model analysis layer identifies abnormal situations through recognition models, and identifies abnormal situations that are inconsistent with the baseline status through analysis models; when the abnormal warning and push layer identifies abnormal situations that are inconsistent with the baseline status, it immediately issues an alarm and notifies the alarm message to the message push platform. The message push platform pushes the alarm to relevant personnel by SMS or email according to the alarm level and notification object.

Images