A data anomaly detection method for live detection robots in substations

Abstract

The present invention relates to the field of electric power detection technology, and discloses a data anomaly detection method for a live detection robot in a substation, including: obtaining data information of an area to be detected, performing three-dimensional modeling of the area, identifying target equipment for equipment classification; generating detection tasks, and planning the optimal detection path of the live detection robot; real-time collection of detection data of the target equipment by the live detection robot, building a normal operation model of the equipment, and identifying abnormal conditions of the target equipment; performing equipment anomaly analysis and early warning reporting, and updating detection tasks. The present invention uses live detection robots to conduct equipment inspections, and formulates according to the actual conditions of the equipment, which can greatly improve detection efficiency and detection accuracy. Through real-time data collection and anomaly detection, potential problems of the equipment can be discovered early, the occurrence of failures can be prevented, and the safe and stable operation of the substation can be guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power detection, and in particular to a data anomaly detection method for a live detection robot in a substation. Background Art

[0002] As the scale of the power system continues to expand and the complexity of electrical equipment increases, the safe operation of the power grid is constantly facing new challenges. How to improve the operating reliability of equipment to ensure the safety and stability of the power system has become a research topic that has received much attention in the power industry. With the development of the economy, the improvement of science and technology and informatization, and the rapid expansion of the power grid, the workload of regular maintenance has increased dramatically, and there is a shortage of maintenance personnel. As a result, the traditional maintenance model cannot meet the needs of the current power grid development due to its lack of pertinence. Therefore, live detection technology has become a hot topic for power users and scientific researchers. Live detection uses effective detection methods and analysis and diagnosis technologies to timely and accurately grasp the operating status of power equipment and ensure the safe and reliable operation of power equipment. The high-tech of live detection equipment has become a guarantee for completing the preventive test tasks of power equipment, and it will definitely become a development trend in the field of power equipment detection in the future.

[0003] For power equipment, by detecting abnormal values, abnormal conditions in the operation of the power system can be discovered, and timely remediation and repair can be carried out to ensure the normal operation of power equipment and facilities. Therefore, detecting abnormal values ​​in the data generated by substation operation and maintenance inspections is of great significance to power grid security. The existing manual inspection work is inefficient and has low accuracy, or traditional intelligent detection technology has problems such as low abnormal detection efficiency and accuracy, and low real-time detection. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a data anomaly detection method of a live detection robot in a substation to solve the problems of how to improve the efficiency of anomaly detection and the low accuracy and low real-time detection performance.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a data anomaly detection method for a live detection robot in a substation, comprising: acquiring data information of an area to be detected, performing three-dimensional modeling of the area, identifying target equipment in the area to be detected, and classifying the equipment;

[0008] Based on the equipment classification result and equipment detection characteristics of the target equipment, a detection task is generated, the detection points of the target equipment are determined, and the optimal detection path of the live detection robot is planned;

[0009] Based on the optimal detection path, the live detection robot collects detection data of the target device in real time, builds a normal operation model of the device according to the historical operation data of the target device, and identifies abnormal conditions of the target device through a data anomaly detection algorithm;

[0010] Based on the abnormal situation of the target device, perform device abnormality analysis and early warning reporting, and update the detection task.

[0011] As a preferred solution of the data anomaly detection method of the live substation detection robot described in the present invention, the regional three-dimensional modeling and target equipment classification include:

[0012] Determine the area to be inspected for regional scanning, obtain terrain data of the area to be inspected, construct a regional three-dimensional space map with the center of the area as the origin, divide the substation area to be inspected into several sub-areas, determine the boundaries of each sub-area in the regional three-dimensional space map, and mark the actual coordinate position of each target device in the area;

[0013] Acquire image data of a target detection device in a region, preprocess the image data, identify the target device through a deep learning model, identify the type of the device, and classify devices of the same type to generate a first device classification set;

[0014] Obtain basic data of the target device, including device age, priority, historical fault data, and equipment inspection items. The historical fault data includes the cause of equipment failure, faulty component, number of failures, and failure time, and calculate the failure rate and frequency of the device;

[0015] Based on the first device classification set, the failure rate of the target device is compared with a preset failure rate threshold, the target device in the first device classification set that exceeds the threshold or has the highest device priority is selected, a second device classification set is generated, and a device classification result of the target device is obtained.

[0016] As a preferred solution of the data anomaly detection method of the live detection robot of the substation described in the present invention, the generating detection task includes:

[0017] Based on the basic data of the target device, evaluate the health index of the device and calculate the detection frequency of the target device;

[0018] According to the preset equipment detection cycle, the equipment detection cycle corresponding to the detection frequency is judged, the detection time and detection items of the target equipment are determined, and the detection task corresponding to the target equipment in each detection cycle is generated.

[0019] As a preferred solution of the data anomaly detection method of the live detection robot of the substation described in the present invention, wherein: the optimal detection path of the live detection robot is planned including:

[0020] Based on the detection task, counting target devices in the current detection task and at least one task sub-area where the target devices are located;

[0021] The target device weight values ​​of the task sub-areas are counted, where the target device weight values ​​represent the weights of the target devices to be detected in the current detection task and belonging to the second device classification set to the total target devices to be detected;

[0022] Sorting the target device weight values ​​to obtain a detection order of task sub-areas in the current detection task;

[0023] Based on the detection order of the task sub-area and the actual coordinate position of the target device in the task sub-area, the coordinate information of the detection point is identified according to the detection point information corresponding to the target device, and the optimal detection path for the target device to be detected by the live detection robot is generated with the shortest path distance as the goal.

[0024] As a preferred solution of the data anomaly detection method of the live detection robot of the substation described in the present invention, the real-time collection of detection data of the target device includes:

[0025] The live detection robot reaches the coordinate position of the target device in the area according to the detection order of the task sub-area according to the optimal detection path;

[0026] The detection points of the target equipment are tested according to the corresponding detection items, the detection data is collected in real time, the data is uploaded, and the data is analyzed and stored.

[0027] As a preferred solution of the data anomaly detection method of the live detection robot for substations of the present invention, identifying the abnormal situation of the target device includes:

[0028] Perform data preprocessing based on the detection data collected by the substation detection robot;

[0029] Acquire historical operating data of the target device under normal conditions, extract characteristic indicators for data anomaly detection, build a normal operating model of the device through a machine learning algorithm, and perform model training;

[0030] The detection data is input into the normal operation model of the equipment, the model calculation result is output, and compared with the preset abnormal threshold. If the model calculation result is greater than the preset abnormal threshold, it is marked as an abnormal data point of the equipment.

[0031] As a preferred solution of the data anomaly detection method of the live detection robot of the substation described in the present invention, the update detection task includes:

[0032] Obtain the abnormal time, abnormal type and abnormal degree corresponding to the abnormal data point of the equipment, compare and analyze with historical data and fault library, obtain the fault analysis data of equipment detection, and generate a detection data abnormality report;

[0033] Based on the fault analysis data, the historical fault data of the device is updated in real time, the health index of the device is re-evaluated, the detection frequency of the target device is calculated, and the detection task of the current area to be detected is updated.

[0034] In a second aspect, the present invention provides a data anomaly detection system for a live detection robot in a substation, comprising:

[0035] A classification module is used to obtain data information of the area to be detected, perform three-dimensional modeling of the area, identify target devices in the area to be detected, and classify the devices;

[0036] A path planning module, used to generate a detection task, determine the detection points of the target device, and plan the optimal detection path of the live detection robot based on the device classification result and device detection characteristics of the target device;

[0037] An anomaly detection module is used to collect detection data of the target device in real time through the live detection robot based on the optimal detection path, build a normal operation model of the device according to the historical operation data of the target device, and identify the abnormal situation of the target device through a data anomaly detection algorithm;

[0038] The output update module is used to perform device abnormality analysis and early warning reporting based on the abnormal conditions of the target device, and update the detection task.

[0039] In a third aspect, the present invention provides an electronic device, comprising:

[0040] Memory and processor;

[0041] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the data anomaly detection method of the substation live detection robot are implemented.

[0042] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for detecting data anomalies of a live substation inspection robot.

[0043] Compared with the prior art, the present invention has the following beneficial effects: the present invention uses live detection robots to conduct equipment inspections and formulates them according to the actual conditions of the equipment, which can greatly improve the detection efficiency and accuracy. Through multi-sensor fusion technology and advanced machine learning algorithms, the accuracy and robustness of equipment detection are improved; automated detection path planning and task scheduling reduce the need for manual intervention and improve detection efficiency; through real-time data collection and anomaly detection, potential problems with equipment can be discovered early to prevent the occurrence of failures; timely detection and handling of equipment anomalies helps to ensure the safe and stable operation of substations and reduce power outages caused by equipment failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0045] Figure 1 It is a schematic diagram of the overall process of a method for detecting data anomalies of a live detection robot in a substation according to an embodiment of the present invention;

[0046] Figure 2 The figure is a schematic diagram of the overall structure of a data anomaly detection system for a live substation detection robot according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0050] Example 1, reference Figure 1-Figure 2 , is an embodiment of the present invention, and provides a method for detecting abnormal data of a live detection robot in a substation, such as Figure 1 ,include:

[0051] S100, acquiring data information of the area to be detected, performing three-dimensional modeling of the area, identifying target devices in the area to be detected, and classifying the devices;

[0052] Preferably, the regional three-dimensional modeling and target device classification include:

[0053] Determine the area to be inspected for regional scanning, obtain the terrain data of the area to be inspected, build a regional three-dimensional space map with the center of the area as the origin, and divide the substation area to be inspected into several sub-areas, determine the boundaries of each sub-area in the regional three-dimensional space map, and mark the actual coordinate position of each target device in the area;

[0054] Exemplarily, in this embodiment, a device such as a laser radar (LiDAR) or a 3D scanner can be used to scan the substation, collect spatial geometric information, process the collected point cloud data, remove noise points, perform registration, generate a complete point cloud map, and then convert the processed point cloud into a three-dimensional space map; at the same time, according to the generated three-dimensional space map, the area length or radius is set, and the target devices with close geographical locations are grouped into a sub-area, which can reduce the distance that the live detection robot moves and improve the detection efficiency.

[0055] Acquire image data of target detection devices in the area, preprocess the image data, identify the type of target devices through a deep learning model, and classify devices of the same type to generate a first device classification set;

[0056] Exemplarily, the system will pre-store data information and fault data corresponding to various equipment types in the equipment library. Through a deep learning model, exemplarily a convolutional neural network based on architectures such as CNN, ResNet, and Inception, multimodal feature learning can be performed in combination with sensor data (such as temperature distribution, vibration patterns, etc.) to improve classification accuracy, perform feature matching based on the equipment library and the collected image data, complete device identification, and determine the device type of the target device.

[0057] Obtain the basic data of the target equipment, including equipment age, priority, historical fault data, and equipment inspection items. The historical fault data includes the cause of equipment failure, faulty components, number of failures, and failure time. Calculate the failure rate and frequency of the equipment.

[0058] Based on the first device classification set, the failure rate of the target device is compared with a preset failure rate threshold, and the target device in the first device classification set that exceeds the preset failure rate threshold or has the highest device priority is selected to generate a second device classification set, and obtain a device classification result of the target device;

[0059] Specifically, the above-mentioned preset failure rate threshold can be defined according to the actual detection situation, and relatively important equipment and equipment with a higher failure rate can be divided into the second equipment set to facilitate the planning of subsequent live detection robot detection tasks.

[0060] It should be noted that this step can accurately determine the location of the equipment and the surrounding environment through regional three-dimensional modeling, providing an accurate reference coordinate system for subsequent inspections; identifying and classifying equipment can understand the equipment type and its distribution in advance, provide a basis for generating inspection tasks, and improve the targeted nature of inspections. In addition, equipment classification helps in path planning in subsequent steps, ensuring that the inspection robot can arrive at each inspection point in a priority and efficient manner to complete the inspection.

[0061] S200, based on the equipment classification result and equipment detection characteristics of the target equipment, generating a detection task, determining the detection point of the target equipment, and planning the optimal detection path of the live detection robot;

[0062] Preferably, generating a detection task includes:

[0063] Based on the basic data of the target device, evaluate the health indicators of the device and calculate the detection frequency of the target device;

[0064] For example, the physical characteristics of the equipment and historical data analysis results are combined to evaluate the current health status of the equipment and the health indicators of the equipment. It is expressed as:

[0065] ;

[0066] in, Indicates the age of the equipment. represents the equipment failure rate, Indicates the equipment operation cycle. Indicates the number of historical faults. Indicates the equipment workload. , , , , Indicates the weight coefficient of each indicator;

[0067] Exemplarily, based on the evaluation results, the detection frequency of each target device is calculated , expressed as:

[0068] ;

[0069] in, Indicates the device priority. Indicates the highest priority of the device. Indicates the basic detection frequency.

[0070] According to the preset equipment detection cycle, the equipment detection cycle corresponding to the detection frequency is determined, the detection time and detection items of the target equipment are determined, and the detection tasks corresponding to the target equipment in each detection cycle are generated.

[0071] Specifically, this embodiment can define each interval of the detection cycle corresponding to each interval of the detection frequency, where the detection cycle can be weekly, daily, etc. All target devices belonging to the same detection cycle generate a detection task, and the detection time corresponding to each detection cycle interval is obtained, and then the detection task for each detection time period is generated. The detection task includes but is not limited to the target device that needs to be detected, the inspection items corresponding to the equipment (such as temperature measurement, vibration inspection), the inspection point and other information.

[0072] Preferably, planning the optimal detection path of the live detection robot includes:

[0073] Based on the detection task, counting the target devices in the current detection task and at least one task sub-area where the target devices are located;

[0074] Exemplarily, the specific target device to be detected within the current detection time in the detection task, the total number of devices, and the task sub-area where each device is located, that is, the sub-area to which the device belongs, are obtained, wherein the detection task detects at least one target device in at least one sub-area.

[0075] The target device weight values ​​of the task sub-areas are counted, where the target device weight values ​​represent the weight of the target devices to be detected in the second device classification set in the current detection task to the total number of target devices to be detected;

[0076] Sort the target device weights to obtain the detection order of the task sub-areas in the current detection task;

[0077] Based on the detection order of the task sub-area and the actual coordinate position of the target device in the task sub-area, the coordinate information of the detection point is identified according to the detection point information corresponding to the target device, and the optimal detection path for the target device to be detected by the live detection robot is generated with the shortest path distance as the goal.

[0078] Specifically, path planning can use path planning algorithms such as the A* algorithm or RRT (Rapidly-exploring Random Trees), combined with a three-dimensional space map, to generate the optimal path for the live inspection robot to reach each inspection point in the order of sub-area inspections. At the same time, obstacle avoidance and safety must be considered, and obstacles in the path must be identified to ensure that the robot does not collide with obstacles.

[0079] It should be noted that this embodiment automatically generates inspection tasks based on the equipment classification results, reduces the tedious process of manually compiling tasks, improves work efficiency, reduces unnecessary movement of the inspection robot through the planning of the optimal inspection path, saves time and energy consumption, and rationally allocates inspection resources according to the equipment's priority level or historical fault conditions to ensure that key equipment is inspected in a timely manner.

[0080] S300, based on the optimal detection path, collects the detection data of the target device in real time through the live detection robot, builds the normal operation model of the device according to the historical data of the target device, and identifies the abnormal situation of the target device through the data anomaly detection algorithm;

[0081] Preferably, real-time collection of detection data of the target device includes:

[0082] The live inspection robot reaches the coordinate position of the target equipment in the area according to the inspection order of the task sub-areas based on the optimal inspection path;

[0083] The detection points of the target equipment are tested according to the corresponding detection items, the detection data is collected in real time, the data is uploaded, and the data is analyzed and stored.

[0084] Specifically, during the inspection process of the live inspection robot, different inspection components are used for different inspection items to obtain real-time data, which may be temperature data (temperature data on the surface or inside of the equipment), vibration data (vibration frequency and amplitude data when the equipment is running), electrical parameters (electrical signal data such as current and voltage), sound data (sound data when the equipment is running, used to analyze whether there is abnormal noise), image data, etc. (image or video data of the equipment, used for visual inspection and recognition).

[0085] Preferably, identifying an abnormal situation of the target device includes:

[0086] Perform data preprocessing based on the detection data collected by the substation detection robot;

[0087] Exemplarily, data preprocessing may be data cleaning, data standardization, and other processing methods, which can remove invalid data, fill or delete missing values, ensure data quality, or standardize different types of sensor data.

[0088] Obtain the normal historical operation data of the target device, extract the characteristic indicators for data anomaly detection, build the normal operation model of the device through machine learning algorithm and perform model training;

[0089] The detection data is input into the normal operation model of the equipment, the model calculation result is output, and compared with the preset abnormal threshold. If the model calculation result is greater than the preset abnormal threshold, it is marked as an abnormal data point of the equipment.

[0090] Specifically, the normal operation model of the equipment can be constructed through machine or deep learning algorithms, such as local outlier factor (LOF), random forest (RF) and other algorithms.

[0091] Exemplarily, a normal operation model of the equipment is constructed by an optimized local outlier factor (LOF) algorithm, including:

[0092] Calculate the k nearest neighbors. For each data point, find the nearest Neighbors.

[0093] Calculate the local reachability density for the data point , whose accessibility density is defined as:

[0094] ;

[0095] in, Represents data points of The nearest neighbor set, the neighbor points are sorted by distance Arrange in order from smallest to largest, Represents data points , The distance between them (can be Euclidean distance, Manhattan distance, etc.), Represents data points No. The distance to the nearest neighbor;

[0096] Calculate the local outlier factor, expressed as:

[0097] ;

[0098] in, Represents data points of The reachability density of the nearest neighbors and the data point The average value of the ratio of its own reachability density.

[0099] Use the historical data of the equipment under normal operating conditions to train the equipment normal operating model. For new detection data, calculate its LOF score and compare it with the preset abnormal threshold. If it is greater than the preset abnormal threshold, it is marked as an abnormal equipment data point.

[0100] Specifically, the preset anomaly threshold can be defined based on business requirements or historical data analysis;

[0101] Exemplarily, by analyzing the LOF score distribution of the device under normal operating conditions, a quantile that meets the requirements is selected as a threshold; or if the LOF scores under normal operating conditions are mostly concentrated in a certain range, the upper limit of the range can be selected as the threshold.

[0102] It should be noted that by modeling the behavior of equipment under normal working conditions, behaviors that deviate from these patterns can be identified, thereby detecting potential problems and predicting possible equipment failures in advance, providing a basis for preventive maintenance and reducing unexpected downtime.

[0103] This step can detect equipment abnormalities in a timely manner and improve fault response speed by collecting and analyzing data in real time; building a normal equipment operation model through historical data provides a benchmark for anomaly detection, improves detection accuracy, and can accurately identify equipment abnormalities.

[0104] S400, based on the abnormal situation of the target device, perform device abnormality analysis and early warning reporting, and update the detection task.

[0105] Preferably, the abnormal time, abnormal type and abnormal degree corresponding to the abnormal data point of the equipment are obtained, compared and analyzed with historical data and the fault library, the fault analysis data of the equipment detection is obtained, and a detection data abnormality report is generated, wherein the detection data abnormality report includes but is not limited to data information such as the cause of the equipment failure, the fault component and the recommended measures;

[0106] Based on the fault analysis data, the historical fault data of the equipment is updated in real time, the health indicators of the equipment are re-evaluated, the detection frequency of the target equipment is calculated, and the detection tasks of the current area to be detected are updated.

[0107] It should be noted that in-depth analysis of detected abnormalities can help quickly locate the root cause of the problem and develop effective solutions. Through early warning, relevant personnel can be notified in a timely manner to ensure that abnormalities are handled in a timely manner and reduce losses caused by failures. Fault data can be obtained based on data abnormalities, and detection tasks can be dynamically adjusted to ensure that key equipment receives timely and sufficient attention, thereby improving the effectiveness of maintenance work.

[0108] The present invention uses live detection robots to conduct equipment inspections and formulates plans based on the actual conditions of the equipment, which can greatly improve detection efficiency and detection accuracy. Through multi-sensor fusion technology and advanced machine learning algorithms, the accuracy and robustness of equipment detection are improved; automated detection path planning and task scheduling reduce the need for manual intervention and improve detection efficiency; through real-time data collection and anomaly detection, potential problems with equipment can be discovered early to prevent the occurrence of failures; timely detection and processing of equipment anomalies helps to ensure the safe and stable operation of substations and reduce power outages caused by equipment failures.

[0109] The above is a schematic scheme of a data anomaly detection method for a live substation detection robot in this embodiment. It should be noted that the technical scheme of the data anomaly detection system for the live substation detection robot and the technical scheme of the data anomaly detection method for the live substation detection robot are of the same concept. For details not described in detail in the technical scheme of the data anomaly detection system for the live substation detection robot in this embodiment, please refer to the description of the technical scheme of the data anomaly detection method for the live substation detection robot.

[0110] In this embodiment, the data anomaly detection system of the live detection robot of the substation is as follows: Figure 2 ,include:

[0111] The classification module is used to obtain data information of the area to be detected, perform three-dimensional modeling of the area, identify target devices in the area to be detected, and classify the devices;

[0112] The path planning module is used to generate the inspection task, determine the inspection points of the target equipment, and plan the optimal inspection path of the live inspection robot based on the equipment classification results and equipment inspection characteristics of the target equipment;

[0113] The anomaly detection module is used to collect the detection data of the target equipment in real time through the live detection robot based on the optimal detection path, build a normal operation model of the equipment according to the historical operation data of the target equipment, and identify the abnormal situation of the target equipment through the data anomaly detection algorithm;

[0114] The output update module is used to analyze device anomalies and issue early warning reports based on the abnormal conditions of the target device, and to update the detection tasks.

[0115] This embodiment also provides an electronic device, which is applicable to data anomaly detection of a live detection robot in a substation, and includes:

[0116] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the data anomaly detection method of the substation live detection robot as proposed in the above embodiment.

[0117] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for detecting data anomalies of a live substation detection robot as proposed in the above embodiment is implemented.

[0118] The storage medium proposed in this embodiment and the method for realizing data anomaly detection of live substation detection robot proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0119] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various embodiments of the present invention.

[0120] Example 2, referring to Table 1, is an embodiment of the present invention, which provides a data anomaly detection method for a live substation detection robot. In order to verify its beneficial effects, comparison results of two solutions are provided.

[0121] By implementing the method of the present invention to carry out simulation experiments, an area to be detected in a certain substation is selected, and the area contains a variety of typical target equipment, such as transformers, circuit breakers, disconnectors, etc. It is assumed that there are 50 devices in total, of which 20 devices have abnormal conditions to varying degrees. The method of the present invention dynamically adjusts the detection frequency according to the equipment health indicators, while the traditional method performs detection according to a fixed period.

[0122] Data collection preparation is carried out, including a certain amount of normal operation data and abnormal data. Data during a normal operation period is generated for each device, including temperature, vibration, current, voltage, etc.; data containing abnormal conditions is generated for abnormal devices, such as excessive temperature, abnormal vibration, etc. The data is divided into a training set and a test set. The training set is used to build a normal operation model, and the test set is used to evaluate the detection performance. The data is cleaned and standardized. This method uses the training set data to build a normal operation model and identifies anomalies through the LOF algorithm. Finally, the two methods use their own detection data for detection. The comparison indicators of the following two methods are selected to comprehensively evaluate the performance of the two methods, including detection accuracy, detection efficiency, detection frequency, false alarm rate, and missed alarm rate, and the detection results are recorded. The detection results are shown in Table 1.

[0123] Table 1: Comparative test results

[0124] Comparison indicators Method of the present invention Other traditional methods Detection accuracy (%) 95 85 Detection efficiency (min) 10 19 Testing frequency (times / unit) 3 5 False alarm rate (%) 5 16 Missing reporting rate (%) 2 10

[0125] Specifically, the detection accuracy rate indicates the average proportion of correctly identified abnormal devices, the detection efficiency indicates the average time required to complete the detection, the detection frequency indicates the average number of detections for each device, the false alarm rate indicates the average proportion of normal devices identified as abnormal devices, and the missed alarm rate indicates the average proportion of abnormal devices that are not correctly identified. From the results, the detection accuracy of the method is 10% higher than that of the traditional method, indicating that the method can more accurately identify the abnormal conditions of the equipment, and the detection efficiency is significantly better than the traditional method. At the same time, the detection frequency can be dynamically adjusted according to the equipment health indicators, which reduces the number of unnecessary detections and improves the detection efficiency. The false alarm rate and missed alarm rate also illustrate the better detection effect of the present invention. The present invention significantly improves the performance of live detection of substations by introducing advanced live detection robots' abnormality detection and dynamic adjustment of detection tasks.

[0126] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for detecting data anomalies of a live detection robot in a substation, characterized in that: include: Acquire data information of the area to be detected, perform three-dimensional modeling of the area, identify target devices in the area to be detected, and classify the devices; Based on the equipment classification result and equipment detection characteristics of the target equipment, a detection task is generated, the detection points of the target equipment are determined, and the optimal detection path of the live detection robot is planned; Based on the optimal detection path, the live detection robot collects detection data of the target device in real time, builds a normal operation model of the device according to the historical operation data of the target device, and identifies abnormal conditions of the target device through a data anomaly detection algorithm; Based on the abnormal situation of the target device, perform device abnormality analysis and early warning reporting, and update the detection task; Among them, regional 3D modeling and target equipment classification include: Performing a regional scan on the area to be inspected, acquiring terrain data of the area to be inspected, constructing a regional three-dimensional spatial map with the center of the area as the origin, dividing the substation area to be inspected into a number of sub-areas, determining the boundaries of each sub-area in the regional three-dimensional spatial map, and marking the actual coordinate position of each target device in the area; Acquire image data of target devices in the area, preprocess the image data, identify the type of the target devices through a deep learning model, and classify devices of the same type to generate a first device classification set; Obtaining basic data of the target device, including device age, priority, historical fault data, and device inspection items, wherein the historical fault data includes the cause of device failure, faulty component, number of failures, and failure time, and calculating the failure rate and frequency of the device; Based on the first device classification set, the failure rate of the target device is compared with a preset failure rate threshold, and the target device in the first device classification set that exceeds the preset failure rate threshold or has the highest device priority is selected to generate a second device classification set to obtain a device classification result for the target device.

2. The data anomaly detection method of the live detection robot of the substation as claimed in claim 1 is characterized in that: The generation detection task includes: Based on the basic data of the target device, evaluate the health index of the device and calculate the detection frequency of the target device; According to the preset equipment detection cycle, the equipment detection cycle corresponding to the detection frequency is judged, the detection time and detection items of the target equipment are determined, and the detection task corresponding to the target equipment in each detection cycle is generated.

3. The data anomaly detection method of the live detection robot of the substation as claimed in claim 2 is characterized in that: The optimal detection path of the live detection robot is planned to include: Based on the detection task, counting target devices in the current detection task and at least one task sub-area where the target devices are located; The target device weight values ​​of the task sub-areas are counted, where the target device weight values ​​represent the weights of the target devices to be detected in the current detection task and belonging to the second device classification set to the total target devices to be detected; Sorting the target device weight values ​​to obtain a detection order of task sub-areas in the current detection task; Based on the detection order of the task sub-area and the actual coordinate position of the target device in the task sub-area, the coordinate information of the detection point is identified according to the detection point information corresponding to the target device, and the optimal detection path for the target device to be detected by the live detection robot is generated with the shortest path distance as the goal.

4. The data anomaly detection method of the live detection robot of the substation as claimed in claim 3 is characterized in that: The real-time acquisition of detection data of the target device includes: The live detection robot reaches the coordinate position of the target device in the area according to the detection order of the task sub-area according to the optimal detection path; The detection points of the target equipment are tested according to the corresponding detection items, the detection data is collected in real time, the data is uploaded, and the data is analyzed and stored.

5. The data anomaly detection method of a live substation detection robot according to claim 1 or 4, characterized in that: Identifying abnormal conditions of the target device includes: Perform data preprocessing based on the detection data collected by the live detection robot; Acquire historical operating data of the target device under normal conditions, extract characteristic indicators for data anomaly detection, build a normal operating model of the device through a machine learning algorithm, and perform model training; The detection data is input into the normal operation model of the equipment, the model calculation result is output, and compared with the preset abnormal threshold. If the model calculation result is greater than the preset abnormal threshold, it is marked as an abnormal data point of the equipment.

6. The data anomaly detection method of the live detection robot of the substation as claimed in claim 5, characterized in that: The update detection task includes: Obtain the abnormal time, abnormal type and abnormal degree corresponding to the abnormal data point of the equipment, compare and analyze with historical data and fault library, obtain the fault analysis data of equipment detection, and generate a detection data abnormality report; Based on the fault analysis data, the historical fault data of the device is updated in real time, the health index of the device is re-evaluated, the detection frequency of the target device is calculated, and the detection task of the current area to be detected is updated.

7. A data anomaly detection system for a live substation detection robot, applied to the data anomaly detection method for a live substation detection robot as claimed in any one of claims 1 to 6, characterized in that: include: A classification module is used to obtain data information of the area to be detected, perform three-dimensional modeling of the area, identify target devices in the area to be detected, and classify the devices; A path planning module, used to generate a detection task, determine the detection points of the target device, and plan the optimal detection path of the live detection robot based on the device classification result and device detection characteristics of the target device; An anomaly detection module is used to collect detection data of the target device in real time through the live detection robot based on the optimal detection path, build a normal operation model of the device according to the historical operation data of the target device, and identify the abnormal situation of the target device through a data anomaly detection algorithm; An output update module is used to perform device abnormality analysis and early warning reporting based on the abnormal conditions of the target device, and update the detection task; Among them, regional 3D modeling and target equipment classification include: Performing a regional scan on the area to be inspected, acquiring terrain data of the area to be inspected, constructing a regional three-dimensional spatial map with the center of the area as the origin, dividing the substation area to be inspected into a number of sub-areas, determining the boundaries of each sub-area in the regional three-dimensional spatial map, and marking the actual coordinate position of each target device in the area; Acquire image data of target devices in the area, preprocess the image data, identify the type of the target devices through a deep learning model, and classify devices of the same type to generate a first device classification set; Obtaining basic data of the target device, including device age, priority, historical fault data, and device inspection items, wherein the historical fault data includes the cause of device failure, faulty component, number of failures, and failure time, and calculating the failure rate and frequency of the device; Based on the first device classification set, the failure rate of the target device is compared with a preset failure rate threshold, and the target device in the first device classification set that exceeds the preset failure rate threshold or has the highest device priority is selected to generate a second device classification set to obtain a device classification result for the target device.

8. An electronic device, characterized in that: include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method for detecting data anomalies of a live substation inspection robot described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: It stores computer executable instructions, which, when executed by a processor, can implement the steps of the data anomaly detection method for a live substation inspection robot as described in any one of claims 1 to 6.

Images