Robot dog system and method for power equipment inspection

Abstract

The invention discloses a robot dog system and method for power equipment inspection, and relates to the technical field of power equipment maintenance, and the method comprises the steps: issuing an inspection task to a robot dog, including the position information, the equipment type, the inspection priority and the task requirement of target power equipment, scanning the surrounding environment, and constructing initial three-dimensional map data; based on the position of the target power equipment and the initial three-dimensional map data, defining a comprehensive cost function, calculating path node cost, and generating an optimal path; and based on the health diagnosis report of the target power equipment, generating a task summary report by judging an abnormal state, an abnormal state severity degree and an abnormal reason summary. According to the invention, by defining the comprehensive cost function to generate the optimal path and combining multi-modal data acquisition with intelligent analysis, efficient, safe and accurate inspection of the power equipment is realized.

Description

Technical Field

[0001] This invention relates to the field of power equipment maintenance technology, and in particular to a robot dog system and method for power equipment inspection. Background Technology

[0002] Power equipment inspection, as a crucial link in ensuring the stable operation of the power system, has become increasingly important in recent years with the development of smart grids and technological advancements. Traditional manual inspection methods are not only inefficient but also pose safety hazards, especially in high-voltage environments where inspectors face significant risks. Therefore, automated inspection technology has gradually become a research and application hotspot. Currently, various robot- or drone-based inspection systems have been developed both domestically and internationally, which can, to some extent, replace manual labor in repetitive and highly dangerous tasks. However, existing inspection systems mostly rely on preset paths or simple navigation algorithms, exhibiting limitations in complex environments and failing to achieve efficient and accurate inspection operations. Furthermore, most existing systems assess the health status of power equipment primarily based on single-type sensor data, lacking the ability to comprehensively analyze multimodal data, resulting in incomplete and inaccurate diagnostic results.

[0003] With the rapid development of emerging technologies such as Artificial Intelligence (AI), the Internet of Things (IoT), and big data analytics, intelligent inspection methods have been further improved. In particular, the application of machine learning and deep learning algorithms has enabled mobile platforms such as robot dogs to adapt more flexibly to various working scenarios and improve fault detection accuracy through multi-sensor fusion. Nevertheless, existing technologies still have the following shortcomings: First, traditional inspection systems often fail to fully consider dynamic changes in the actual environment when planning paths, such as the appearance of obstacles or changes in equipment location, which may lead to suboptimal inspection paths or even failure to complete the intended task. Second, for assessing the health status of power equipment, existing methods typically focus on processing a specific type of data, ignoring the correlation between different modalities of data, thus limiting the accuracy of the overall assessment. To address these issues, this invention proposes a novel robot dog system and method for power equipment inspection, aiming to overcome the shortcomings of existing technologies by introducing advanced path planning algorithms and multimodal data analysis models, significantly improving the efficiency and reliability of inspection work. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a robot dog method for power equipment inspection, which solves the problems of unintelligent path planning and single health status assessment in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for a robot dog to inspect power equipment, comprising: issuing an inspection task to the robot dog, including the location information, equipment type, inspection priority, and task requirements of the target power equipment; scanning the surrounding environment to construct initial three-dimensional map data; defining a comprehensive cost function based on the location of the target power equipment and the initial three-dimensional map data; calculating the cost of path nodes and generating an optimal path; according to the optimal path, the robot dog scans the equipment number and edge features of the equipment using a visual sensor, compares them with the equipment information in the inspection task, and confirms the target power equipment information; based on the target power equipment information, activating multimodal data acquisition, the robot dog collects multimodal data from the target power equipment, and generates a multimodal data package; based on the multimodal data package, calculating the equipment health score through a health status assessment model, using AI algorithms for data fusion and intelligent analysis, and generating a health diagnosis report for the target power equipment; and based on the health diagnosis report of the target power equipment, generating a task summary report by judging abnormal states, the severity of abnormal states, and summarizing the causes of abnormalities.

[0007] As a preferred embodiment of the robot dog method for power equipment inspection according to the present invention, the steps of issuing inspection tasks to the robot dog include the location information, equipment type, inspection priority, and task requirements of the target power equipment, and scanning the surrounding environment to construct initial three-dimensional map data. The specific steps are as follows: The power equipment backend issues inspection tasks to the robot dog. The target power equipment task includes the geographical coordinates of the target power equipment, equipment type, inspection priority and task requirements. The robot dog activates SLAM, scanning the surrounding environment with its onboard LiDAR, depth camera, and inertial measurement unit to build real-time 3D map data.

[0008] As a preferred embodiment of the robot dog method for power equipment inspection described in this invention, the steps of defining a comprehensive cost function based on the target power equipment location and initial 3D map data, calculating path node costs, and generating the optimal path are as follows: Based on the location of the target power equipment Based on the initial 3D map data, the robot dog's current position is determined. The starting point is the location of the target power equipment, and the ending point is the location of the target power equipment. Define a comprehensive cost function to calculate the cost of each path node; the expression is: ; in, Represents the cost of path nodes. Indicates the path from the starting point to the node. Path distance, Represents path nodes Heuristic distance estimation to the target point Represents path nodes Obstacle costs, Represents path nodes The cost of terrain complexity Weights representing path distances Weights representing the proximity to the target This indicates the weight of the impact of obstacles. Weights representing terrain complexity. Indicates a path node; From the starting point Start by gradually expanding the adjacent nodes. Calculate the value of each node Select the node with the lowest cost as the next expansion point when the target power equipment is located. When expanded, backtrack the path nodes to generate the optimal path.

[0009] In a preferred embodiment of the robot dog method for power equipment inspection according to the present invention, the robot dog scans the equipment number and edge features of the equipment using a visual sensor based on the optimal path, compares them with the equipment information in the inspection task, and confirms the target power equipment information. The specific steps are as follows. According to the optimal path, reach the target power equipment area. The robot dog activates its vision sensor to perform a preliminary scan of the target power equipment, acquires image data of the target area, and extracts the edge features of the target power equipment image. Based on the edge features of the target power equipment image, the equipment number region is located using a target detection algorithm, and the equipment number is extracted using optical character recognition. Based on equipment images in the background task data of power equipment, define the edge features of standard equipment images; The edge features of the extracted target power equipment image are matched with the edge features of the standard equipment image. A structural similarity index is calculated using the structural similarity method, expressed as: ; in, Represents the structural similarity index. This represents the edge features extracted from the image of the target power equipment. Represents the edge features of a standard device image. Image representing target power equipment The average pixel value, Represents standard equipment images The average pixel value, Image representing target power equipment and standard equipment images covariance, This indicates the weight used to adjust the average brightness. Indicates the weights used to adjust the covariance; Based on the combined edge features and device number, the target power equipment information is confirmed using the following expression: ; in, This indicates the confirmation result of the target power equipment. express and exactly the same, express and They are completely dissimilar.

[0010] In a preferred embodiment of the robot dog method for power equipment inspection described in this invention, the step of activating multimodal data acquisition based on target power equipment information, and the robot dog acquiring multimodal data from the target power equipment to generate a multimodal data packet, includes the following specific steps: Multimodal sensors include vision sensors, thermal imaging sensors, vibration sensors, sound sensors, and gas sensors; Based on the target power equipment information, high-definition images of the target power equipment are acquired through a visual sensor; Crack features are extracted using an edge detection algorithm, and crack density is calculated using the following expression: ; in, Indicates crack density. Indicates the number of cracked areas. Indicates the first Area of ​​each crack region Indicates the total surface area of ​​the equipment; Based on the target power equipment information, the surface temperature distribution of the target power equipment is collected using a thermal imaging sensor; It also detects abnormal hot spots and calculates the temperature anomaly coefficient, expressed as: ; in, Indicates the temperature anomaly coefficient. Indicates the highest temperature. Indicates the average temperature. Indicates the standard deviation of temperature; Vibration signals are recorded using vibration sensors based on information about the target power equipment. Frequency domain signal extracted using Fast Fourier Transform And calculate the clock frequency, the expression is: ; in, Indicates the clock speed. Indicates the amplitude of vibration in the frequency domain. Indicates frequency; Vibration anomaly detection involves calculating vibration anomaly indices, expressed as: ; in, Indicators of abnormal vibration Indicates the threshold frequency; Based on the target power equipment information, sound signals of the target power equipment's operating noise are collected using sound sensors; The frequency domain signal is extracted using FFT, and the noise energy distribution is calculated. The expression is as follows: ; in, Indicates the noise energy distribution. Represents frequency Frequency domain signal, This indicates the upper limit of the integral. This represents the lower limit of integration. Differential unit representing frequency; Based on the target power equipment information, the composition of the surrounding gas is collected using gas sensors; Gas leak detection, calculation of gas leak index The expression is: ; in, Indicates the gas leakage index. Indicates gas composition, Indicates the first The concentration of the gas, Indicates the total number of gas types detected; High-definition images, crack density, temperature distribution, temperature anomaly coefficient, vibration signal, dominant frequency, vibration anomaly index, sound signal, noise energy distribution, gas composition, and gas leakage index are collected by vision, thermal imaging, vibration, sound, sound, and gas sensors to generate multimodal data packets.

[0011] As a preferred embodiment of the robot dog method for power equipment inspection described in this invention, the steps of calculating equipment health scores based on multimodal data packets and a health status assessment model, and generating a health diagnosis report for the target power equipment using AI algorithms for data fusion and intelligent analysis are as follows: Extract the raw data from the multimodal data packets and normalize the data. Normalized data The device health score is calculated using a supervised learning algorithm, expressed as: ; in, This indicates the health score of the target electrical equipment. Represents the exponentially decaying function. Indicates the temperature sensitivity coefficient. This represents the normalized crack density. This represents the normalized temperature anomaly coefficient. This represents the normalized dominant vibration frequency. This represents a normalized vibration anomaly index. This represents the normalized noise energy distribution. The normalized gas leakage index; Based on the analysis results of historical equipment operation data, the dividing points for the classification rules are defined as A and B; Based on the health score of the target power equipment, the health status of the equipment is classified using an AI classification algorithm; when At that time, the health score of the target power equipment is considered to be in a normal state; when If so, the health score of the target electrical equipment is considered to be slightly abnormal; when If so, the health score of the target electrical equipment is considered to be severely abnormal; For minor or severe anomalous states, the anomalous contribution of each mode is calculated using the following expression: ; in, Indicates the first The anomaly contribution of each modality Represents the normalized i-th One modal value, This represents the normalized sum of all modal values; Based on health scores and health status, generate a health diagnosis report for the target electrical equipment.

[0012] As a preferred embodiment of the robot dog method for power equipment inspection according to the present invention, the step of generating a task summary report based on the health diagnosis report of the target power equipment by judging abnormal states, the severity of abnormal states, and the summary of abnormal causes is as follows: Based on the abnormal contribution and health score of each mode in the health diagnosis report of the target power equipment, the severity of the abnormal state is generated, expressed as follows: ; in, Indicates the severity of the abnormal state. This represents the weight of each mode, reflecting the degree of influence of the corresponding mode on equipment anomalies. This indicates the degree of influence of the smoothed health score; Based on the severity of the abnormal state and the contribution of each modality to the abnormality, a summary of the causes of the abnormality is generated; Based on the abnormal status, the severity of the abnormal status, and the cause of the abnormality, a task summary report is generated.

[0013] Secondly, this invention provides a robot dog system for power equipment inspection, comprising: a map data construction module, an optimal path generation module, an information confirmation module, a multimodal data packet generation module, a health diagnosis report generation module, and a task summary report generation module; the map data construction module is used to issue inspection tasks to the robot dog, including the location information, equipment type, inspection priority, and task requirements of the target power equipment, and to scan the surrounding environment to construct initial three-dimensional map data; the optimal path generation module is used to define a comprehensive cost function based on the location of the target power equipment and the initial three-dimensional map data, calculate the cost of path nodes, and generate the optimal path; the information confirmation module is used to, according to the optimal path, have the robot dog scan the surrounding environment using a visual sensor. The system identifies the equipment number and edge features, compares them with the equipment information in the inspection task, and confirms the target power equipment information. A multimodal data packet generation module activates multimodal data acquisition based on the target power equipment information, allowing the robot to collect multimodal data from the target power equipment and generate a multimodal data packet. A health diagnosis report generation module calculates the equipment health score based on the multimodal data packet using a health status assessment model, and uses AI algorithms for data fusion and intelligent analysis to generate a health diagnosis report for the target power equipment. A task summary report generation module generates a task summary report based on the target power equipment's health diagnosis report, by judging abnormal states, the severity of abnormal states, and summarizing the causes of abnormalities.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the robot dog method for power equipment inspection as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the robot dog method for power equipment inspection as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by defining a comprehensive cost function to generate the optimal path and combining multimodal data acquisition with intelligent analysis, efficient, safe, and accurate power equipment inspection is achieved. Optimal path planning ensures the safety and energy efficiency of the inspection process, while multimodal data analysis provides a comprehensive and accurate equipment health assessment, significantly improving the reliability of fault detection and the targeting of maintenance work, effectively solving the problems of unintelligent path planning and singular health status assessment in existing technologies. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the robot dog method for power equipment inspection in Example 1.

[0019] Figure 2 This is a schematic diagram of the robot dog system for power equipment inspection in Example 1. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] 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 phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1, referring to Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a method for a robot dog to inspect power equipment, including the following steps: S1. Issue inspection tasks to the robot dog, including the location information, equipment type, inspection priority and task requirements of the target power equipment, and scan the surrounding environment to build initial 3D map data. The power equipment backend issues inspection tasks to the robot dog. The target power equipment task includes the geographical coordinates of the target power equipment, equipment type, inspection priority and task requirements. It should be noted that the equipment category (such as transformer, distribution box, high-voltage switch, etc.) provides basic information for the formulation of inspection strategy. The priority of tasks (such as emergency inspection, regular inspection, etc.) is set according to the importance of the equipment or abnormal situation to ensure reasonable allocation of resources. Specific inspection objectives or indicators (such as checking the appearance of the equipment, monitoring the temperature, detecting vibration or recording other abnormal parameters) are also specified.

[0024] The robot dog activates SLAM and scans the surrounding environment using its onboard LiDAR, depth camera, and inertial measurement unit to build real-time 3D map data. It should be noted that SLAM is a technology that allows robots to create maps and determine their own location while moving in unknown environments. For power equipment inspection tasks, SLAM can ensure that robot dogs can accurately locate themselves even in complex power facility environments, thereby achieving precise navigation.

[0025] S2. Based on the location of the target power equipment and the initial 3D map data, define a comprehensive cost function, calculate the cost of path nodes, and generate the optimal path; Based on the location of the target power equipment Based on the initial 3D map data, the robot dog's current position is determined. The starting point is the location of the target power equipment, and the ending point is the location of the target power equipment. It should be noted that the initial 3D map data was constructed by the robot dog after SLAM was activated by scanning the surrounding environment using LiDAR, depth camera, and inertial measurement unit. This map contains information such as the geometry of the environment and the distribution of obstacles, providing basic data support for path planning.

[0026] Define a comprehensive cost function to calculate the cost of each path node; the expression is: ; in, Represents the cost of path nodes. Indicates the path from the starting point to the node. Path distance, Represents path nodes Heuristic distance estimation to the target point Represents path nodes Obstacle costs, Represents path nodes The cost of terrain complexity Weights representing path distances Weights representing the proximity to the target This indicates the weight of the impact of obstacles. Weights representing terrain complexity. Indicates a path node; It should be noted that, It is a key indicator in determining whether a node is selected as the next expansion point; This ensures that the actual length of the path is as short as possible, thereby improving inspection efficiency; It provides an estimate of the remaining path length, helping the algorithm approach the target faster; This parameter reflects the presence and extent of obstacles around the node, ensuring that the path avoids potentially dangerous areas. This includes information such as slope and surface friction to assess the difficulty and safety of the route; Used to adjust the importance of path length in the total cost; Emphasizing the importance of heuristic distance estimation helps to quickly find a path close to the target; Emphasize the importance of avoiding obstacles and ensure the safety of the path; It reflects the impact of terrain complexity on the total cost and ensures that environmental conditions are fully considered when selecting routes.

[0027] From the starting point Start by gradually expanding the adjacent nodes. Calculate the value of each node Select the node with the lowest cost as the next expansion point when the target power equipment is located. When expanded, backtrack the path nodes to generate the optimal path; It should be noted that from the target point Begin by tracing back along the parent node chain of the records until you return to the starting point. This yields an optimal path consisting of a series of nodes.

[0028] S3. Based on the optimal path, the robot dog scans the device number and edge features of the device using a vision sensor, compares them with the device information in the inspection task, and confirms the target power equipment information. According to the optimal path, reach the target power equipment area. The robot dog activates its vision sensor to perform a preliminary scan of the target power equipment, acquires image data of the target area, and extracts the edge features of the target power equipment image. It should be noted that the acquired raw image data may contain noise or unwanted background information, thus requiring preprocessing. This step typically includes operations such as image denoising, contrast enhancement, and color correction to improve image quality and the accuracy of subsequent processing.

[0029] Based on the edge features of the target power equipment image, the equipment number region is located using a target detection algorithm, and the equipment number is extracted using optical character recognition. It should be noted that among all detected candidate regions, the region most likely to be the number is further filtered out using some rules (e.g., aspect ratio, area size, etc.).

[0030] Based on equipment images in the background task data of power equipment, define the edge features of standard equipment images; It should be noted that the extracted edge features are stored in a standardized feature database, with each device corresponding to a feature set. Each feature is annotated in detail, including feature type (such as contour, numbered region, specific part), location information, and related attributes (such as size, orientation).

[0031] The edge features of the extracted target power equipment image are matched with the edge features of the standard equipment image. A structural similarity index is calculated using the structural similarity method, expressed as: ; in, Represents the structural similarity index. This represents the edge features extracted from the image of the target power equipment. Represents the edge features of a standard device image. Image representing target power equipment The average pixel value, Represents standard equipment images The average pixel value, Image representing target power equipment and standard equipment images covariance, This indicates the weight used to adjust the average brightness. Indicates the weights used to adjust the covariance; It should be noted that, The value range is [-1, 1], where 1 represents complete correlation, 0 represents no correlation, and negative values ​​represent inverse correlation. To avoid a denominator of zero, Ensure numerical stability.

[0032] Based on the combined edge features and device number, the target power equipment information is confirmed using the following expression: ; in, This indicates the confirmation result of the target power equipment. express and exactly the same, express and Completely dissimilar; It should be noted that, This indicates that the device has been identified as the target electrical equipment. This indicates that the target electrical equipment could not be identified and may be abnormal or require further inspection. The structural similarity index is used to measure the similarity of images of target power equipment. Image of standard equipment The similarity between edge features, according to the previous definition, when At that time, it was assumed that the edge features of the two were highly consistent; This indicates the device number extracted from the target power equipment image.

[0033] S4. Based on the target power equipment information, activate multimodal data acquisition. The robot dog collects multimodal data from the target power equipment and generates multimodal data packets. Multimodal sensors include vision sensors, thermal imaging sensors, vibration sensors, sound sensors, and gas sensors; It should be noted that visual sensors capture details of the equipment's appearance and key components; thermal imaging sensors monitor the surface temperature distribution of the equipment and detect overheated areas or abnormal hot spots; vibration sensors record vibration signals during equipment operation and analyze the working status of mechanical components; sound sensors collect equipment operating noise and analyze its frequency and energy distribution; and gas sensors detect the gas composition in the environment surrounding the equipment, especially any potential leaks of harmful or flammable gases.

[0034] Based on the target power equipment information, high-definition images of the target power equipment are acquired through a visual sensor; It should be noted that, based on the information of the target power equipment (such as location, type, etc.), the robot dog activates its onboard vision sensors (such as high-definition cameras or depth cameras) and automatically adjusts the camera's exposure time, aperture size, focal length, and other parameters according to ambient light conditions and the characteristics of the equipment surface to ensure image quality.

[0035] Crack features are extracted using an edge detection algorithm, and crack density is calculated using the following expression: ; in, Indicates crack density. Indicates the number of cracked areas. Indicates the first Area of ​​each crack region Indicates the total surface area of ​​the equipment; It should be noted that, This reflects the proportion of cracks on the equipment surface. That is, the total number of independent regions in the image that are identified as cracks. The unit can be pixels or other appropriate units of area. It can usually be calculated by converting the image resolution to the actual size of the device.

[0036] Based on the target power equipment information, the surface temperature distribution of the target power equipment is collected using a thermal imaging sensor; It should be noted that, based on the information of the target power equipment (such as location, type, etc.), the robot dog activates its onboard thermal imaging sensor and adjusts the sensor parameters to adapt to the current environmental conditions, including factors such as temperature range, humidity, and background radiation, to ensure measurement accuracy.

[0037] It also detects abnormal hot spots and calculates the temperature anomaly coefficient, expressed as: ; in, Indicates the temperature anomaly coefficient. Indicates the highest temperature. Indicates the average temperature. Indicates the standard deviation of temperature; It should be noted that, This reflects the degree of anomaly in the temperature distribution; higher... The value indicates the presence of a significant temperature anomaly; That is, the maximum temperature value found in the collected temperature data.

[0038] Vibration signals are recorded using vibration sensors based on information about the target power equipment. It should be noted that, based on the information of the target power equipment (such as location, type, etc.), the robot dog activates its onboard vibration sensor and adjusts the sensor's sampling frequency, range, and other parameters to adapt to different types of equipment and working environments, ensuring measurement accuracy.

[0039] Frequency domain signal extracted using Fast Fourier Transform And calculate the clock frequency, the expression is: ; in, Indicates the clock speed. Indicates the amplitude of vibration in the frequency domain. Indicates frequency; It should be noted that, That is, the frequency component with the largest amplitude in the vibration signal; It reflects the vibration intensity at each frequency; The unit is usually Hertz; This represents the amplitude of the frequency domain vibration. The frequency corresponding to the maximum value .

[0040] Vibration anomaly detection involves calculating vibration anomaly indices, expressed as: ; in, Indicators of abnormal vibration Indicates the threshold frequency; It should be noted that, Reflects the frequency exceeding the threshold The proportion of vibration energy to total vibration energy; It is usually set to a reasonable upper limit to distinguish between normal vibration and abnormal vibration; It represents the vibration amplitude in the frequency domain, reflecting the vibration intensity at each frequency; This represents all frequencies exceeding the threshold. The sum of the frequency domain vibration amplitudes; It represents the sum of the frequency domain vibration amplitudes at all frequencies, i.e., the total vibration energy.

[0041] Based on the target power equipment information, sound signals of the target power equipment's operating noise are collected using sound sensors; It should be noted that, based on the information of the target power equipment (such as location, type, etc.), the robot dog activates its onboard sound sensor and adjusts the sensor parameters to adapt to the current environmental conditions, including background noise level, sound source distance, and other factors, to ensure measurement accuracy.

[0042] The frequency domain signal is extracted using FFT, and the noise energy distribution is calculated. The expression is as follows: ; in, Indicates the noise energy distribution. Represents frequency Frequency domain signal, This indicates the upper limit of the integral. This represents the lower limit of integration. Differential unit representing frequency; It should be noted that, This reflects the range of frequencies within the specified frequency range (from arrive The total amount of sound energy; It reflects the sound intensity at various frequencies; It is usually set as the starting frequency of the band of interest; It is usually set to the end frequency of the band of interest; Used for integration operations.

[0043] Based on the target power equipment information, the composition of the surrounding gas is collected using gas sensors; It should be noted that, based on the information of the target electrical equipment (such as location, type, etc.), the robot dog activates its onboard gas sensor and adjusts the sensor parameters to adapt to the current environmental conditions, including factors such as temperature and humidity, to ensure measurement accuracy.

[0044] Gas leak detection, calculation of gas leak index The expression is: ; in, Indicates the gas leakage index. Indicates gas composition, Indicates the first The concentration of the gas, Indicates the total number of gas types detected; It should be noted that, This reflects the total gas leakage level in the environment surrounding the equipment; That is, the sum of the gas concentrations that should exist under normal conditions, used as a reference benchmark; It represents different kinds of gases.

[0045] High-definition images, crack density, temperature distribution, temperature anomaly coefficient, vibration signal, dominant frequency, vibration anomaly index, sound signal, noise energy distribution, gas composition, and gas leakage index are collected by vision, thermal imaging, vibration, sound, sound, and gas sensors to generate multimodal data packets. It should be noted that multimodal data packets are used to comprehensively assess the overall health of the equipment and identify potential faults or safety hazards.

[0046] S5. Based on the multimodal data packet, calculate the equipment health score through the health status assessment model, use AI algorithms for data fusion and intelligent analysis, and generate a health diagnosis report for the target power equipment. Extract the raw data from the multimodal data packets and normalize the data. It should be noted that normalization should be performed on each class of data according to the selected method. Ensure that the necessary statistics (such as minimum, maximum, mean, and standard deviation) are calculated across the entire dataset, and then apply these statistics to transform the data.

[0047] Normalized data The device health score is calculated using a supervised learning algorithm, expressed as: ; in, This indicates the health score of the target electrical equipment. Represents the exponentially decaying function. Indicates the temperature sensitivity coefficient. This represents the normalized crack density. This represents the normalized temperature anomaly coefficient. This represents the normalized dominant vibration frequency. This represents a normalized vibration anomaly index. This represents the normalized noise energy distribution. The normalized gas leakage index; It should be noted that, This reflects the overall health status of the equipment; Controlling temperature anomaly coefficient The extent of its impact on health scores; The base of the natural logarithm (approximately 2.718); It reflects the severity of surface cracks in the equipment; The degree of abnormal surface temperature of the quantified equipment; Reflects the main vibration frequency during equipment operation; Quantify the degree of abnormal vibration; Quantify noise energy at different frequencies; Quantify the extent of gas leakage.

[0048] Based on the analysis results of historical equipment operation data, the dividing points for the classification rules are defined as A and B; It should be noted that the dividing point A is defined as the threshold that separates the normal state from the slightly abnormal state. It is typically chosen in the lower part of the health score distribution, but still higher than the score for most normal states. The dividing point B is defined as the threshold that separates the slightly abnormal state from the severely abnormal state. It is typically chosen in the higher part of the health score distribution, but still lower than the score for most severely abnormal states.

[0049] Based on the health score of the target power equipment, the health status of the equipment is classified using an AI classification algorithm; when At that time, the health score of the target power equipment is considered to be in a normal state; when If so, the health score of the target electrical equipment is considered to be slightly abnormal; when If so, the health score of the target electrical equipment is considered to be severely abnormal; It should be noted that a large amount of device operation data was collected from historical records, and each data point was labeled (normal, slightly abnormal, severely abnormal). The AI ​​classification algorithm used was the random forest algorithm.

[0050] For minor or severe anomalous states, the anomalous contribution of each mode is calculated using the following expression: ; in, Indicates the first The anomaly contribution of each modality Represents the normalized i-th One modal value, This represents the normalized sum of all modal values; It should be noted that each mode includes crack density, temperature anomaly coefficient, dominant vibration frequency, vibration anomaly index, noise energy distribution, and gas leakage index.

[0051] Based on health scores and health status, generate a health diagnosis report for the target electrical equipment; It should be noted that the health diagnosis report of the target power equipment includes the equipment name, model, location, installation date, the specific date and time of this inspection, and a list of the sensor types used for data acquisition (such as vision, thermal imaging, vibration, sound, and gas sensors).

[0052] S6. Based on the health diagnosis report of the target power equipment, generate a task summary report by judging the abnormal status, the severity of the abnormal status and the cause of the abnormality; Based on the abnormal contribution and health score of each mode in the health diagnosis report of the target power equipment, the severity of the abnormal state is generated, expressed as follows: ; in, Indicates the severity of the abnormal state. This represents the weight of each mode, reflecting the degree of influence of the corresponding mode on equipment anomalies. This indicates the degree of influence of the smoothed health score; It should be noted that, It reflects the degree of impact of the corresponding mode on equipment malfunctions; Ensure numerical stability and avoid division by zero errors; for minor or severe anomalies, the anomaly contribution of each mode must also be provided. This is to quantify the contribution of each mode to the overall abnormal state.

[0053] Based on the severity of the abnormal state and the contribution of each modality to the abnormality, a summary of the causes of the abnormality is generated; It should be noted that abnormal crack density is caused by material aging, stress concentration, and external impact; abnormal temperature coefficient is caused by poor heat dissipation, overload operation, and insulation problems; abnormal vibration frequency is caused by imbalance, looseness, and bearing wear; abnormal vibration index is caused by mechanical failure, improper installation, and resonance; abnormal noise energy distribution is caused by friction, looseness, and damage to internal components; and abnormal gas leakage index is caused by seal failure, pipeline damage, and valve failure.

[0054] Based on the abnormal status, the severity of the abnormal status, and the cause of the abnormality, generate a task summary report; It should be noted that, in conjunction with the task summary report, a comprehensive assessment of the overall health status of the equipment can be conducted to identify the main problems. Combined with other sensor data, the task summary report can serve as an important indicator of the equipment's health status, helping to predict potential mechanical failures or material aging issues. The task summary report should be included in the inspection record and provided to maintenance personnel for reference so that timely measures can be taken.

[0055] This embodiment also provides a robot dog system for power equipment inspection, including: a map data construction module, an optimal path generation module, an information confirmation module, a multimodal data packet generation module, a health diagnosis report generation module, and a task summary report generation module; the map data construction module is used to issue inspection tasks to the robot dog, including the location information, equipment type, inspection priority, and task requirements of the target power equipment, and to scan the surrounding environment to construct initial three-dimensional map data; the optimal path generation module is used to define a comprehensive cost function based on the location of the target power equipment and the initial three-dimensional map data, calculate the cost of path nodes, and generate the optimal path; the information confirmation module is used to scan the equipment using a visual sensor according to the optimal path. The system compares the equipment number and edge features with the equipment information in the inspection task to confirm the target power equipment information; the multimodal data packet generation module is used to activate multimodal data acquisition based on the target power equipment information, and the robot dog collects multimodal data from the target power equipment to generate multimodal data packets; the health diagnosis report generation module is used to calculate the equipment health score based on the multimodal data packets through a health status assessment model, and use AI algorithms for data fusion and intelligent analysis to generate a health diagnosis report for the target power equipment; the task summary report generation module is used to generate a task summary report based on the health diagnosis report of the target power equipment by judging abnormal states, the severity of abnormal states, and the summary of abnormal causes.

[0056] This embodiment also provides a computer device applicable to the robot dog method for power equipment inspection, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the robot dog method for power equipment inspection as proposed in the above embodiment.

[0057] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0058] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for implementing a robot dog for power equipment inspection as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0059] In summary, this invention achieves high efficiency, safety, and accuracy in power equipment inspection by defining a comprehensive cost function to generate the optimal path and combining multimodal data acquisition with intelligent analysis. Optimal path planning ensures the safety and energy efficiency of the inspection process, while multimodal data analysis provides a comprehensive and accurate equipment health assessment, significantly improving the reliability of fault detection and the targeted nature of maintenance work. This effectively solves the problems of unintelligent path planning and singular health status assessment in existing technologies.

[0060] 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for a robot dog to inspect power equipment, characterized in that: include, The robot dog is given inspection tasks, including the location information of the target power equipment, equipment type, inspection priority and task requirements, and the surrounding environment is scanned to build initial 3D map data. Based on the location of the target power equipment and the initial 3D map data, a comprehensive cost function is defined to calculate the cost of path nodes and generate the optimal path. According to the optimal path, the robot dog scans the device number and edge features of the device using a visual sensor, compares them with the device information in the inspection task, and confirms the target power equipment information. Based on the target power equipment information, multimodal data acquisition is activated, and the robot dog collects multimodal data from the target power equipment to generate multimodal data packets; Based on multimodal data packets, a health status assessment model is used to calculate the equipment health score. AI algorithms are then used for data fusion and intelligent analysis to generate a health diagnosis report for the target power equipment. Based on the health diagnosis report of the target power equipment, a task summary report is generated by judging the abnormal status, the severity of the abnormal status, and the cause of the abnormality.

2. The method for inspecting power equipment using a robotic dog as described in claim 1, characterized in that: The process of issuing inspection tasks to the robot dog includes the location information, equipment type, inspection priority, and task requirements of the target power equipment, as well as scanning the surrounding environment to construct initial 3D map data. The specific steps are as follows. The power equipment backend issues inspection tasks to the robot dog. The target power equipment task includes the geographical coordinates of the target power equipment, equipment type, inspection priority and task requirements. The robot dog activates SLAM, scanning the surrounding environment with its onboard LiDAR, depth camera, and inertial measurement unit to build real-time 3D map data.

3. The method for inspecting power equipment using a robotic dog as described in claim 2, characterized in that: Based on the location of the target power equipment and initial 3D map data, a comprehensive cost function is defined to calculate the cost of path nodes and generate the optimal path. The specific steps are as follows. Based on the location of the target power equipment Based on the initial 3D map data, the robot dog's current position is determined. The starting point is the location of the target power equipment, and the ending point is the location of the target power equipment. Define a comprehensive cost function to calculate the cost of each path node; the expression is: ； in, Represents the cost of path nodes. Indicates the path from the starting point to the node. Path distance, Represents path nodes Heuristic distance estimation to the target point Represents path nodes Obstacle costs, Represents path nodes The cost of terrain complexity Weights representing path distances Weights representing the proximity to the target This indicates the weight of the impact of obstacles. Weights representing terrain complexity. Indicates a path node; From the starting point Start by gradually expanding the adjacent nodes. Calculate the value of each node Select the node with the lowest cost as the next expansion point when the target power equipment is located. When expanded, backtrack the path nodes to generate the optimal path.

4. The method for inspecting power equipment using a robotic dog as described in claim 3, characterized in that: According to the optimal path, the robot dog scans the device number and edge features using visual sensors, compares them with the device information in the inspection task, and confirms the target power equipment information. The specific steps are as follows. According to the optimal path, reach the target power equipment area. The robot dog activates its vision sensor to perform a preliminary scan of the target power equipment, acquires image data of the target area, and extracts the edge features of the target power equipment image. Based on the edge features of the target power equipment image, the equipment number region is located using a target detection algorithm, and the equipment number is extracted using optical character recognition. Based on equipment images in the background task data of power equipment, define the edge features of standard equipment images; The edge features of the extracted target power equipment image are matched with the edge features of the standard equipment image. A structural similarity index is calculated using the structural similarity method, expressed as: ； in, Represents the structural similarity index. This represents the edge features extracted from the image of the target power equipment. Represents the edge features of a standard device image. Image representing target power equipment The average pixel value, Represents standard equipment images The average pixel value, Image representing target power equipment and standard equipment images covariance, This indicates the weight used to adjust the average brightness. Indicates the weights used to adjust the covariance; Based on the combined edge features and device number, the target power equipment information is confirmed using the following expression: ； in, This indicates the confirmation result of the target power equipment. express and exactly the same, express and They are completely dissimilar.

5. The method for inspecting power equipment using a robotic dog as described in claim 4, characterized in that: Based on the target power equipment information, multimodal data acquisition is activated. The robot dog collects multimodal data from the target power equipment and generates multimodal data packets. The specific steps are as follows: Multimodal sensors include vision sensors, thermal imaging sensors, vibration sensors, sound sensors, and gas sensors; Based on the target power equipment information, high-definition images of the target power equipment are acquired through a visual sensor; Crack features are extracted using an edge detection algorithm, and crack density is calculated using the following expression: ； in, Indicates crack density. Indicates the number of cracked areas. Indicates the first Area of ​​each crack region Indicates the total surface area of ​​the equipment; Based on the target power equipment information, the surface temperature distribution of the target power equipment is collected using a thermal imaging sensor; It also detects abnormal hot spots and calculates the temperature anomaly coefficient, expressed as: ； in, Indicates the temperature anomaly coefficient. Indicates the highest temperature. Indicates the average temperature. Indicates the standard deviation of temperature; Vibration signals are recorded using vibration sensors based on information about the target power equipment. Frequency domain signal extracted using Fast Fourier Transform And calculate the clock frequency, the expression is: ； in, Indicates the clock speed. Indicates the amplitude of vibration in the frequency domain. Indicates frequency; Vibration anomaly detection involves calculating vibration anomaly indices, expressed as: ； in, Indicators of abnormal vibration Indicates the threshold frequency; Based on the target power equipment information, sound signals of the target power equipment's operating noise are collected using sound sensors; The frequency domain signal is extracted using FFT, and the noise energy distribution is calculated. The expression is as follows: ； in, Indicates the noise energy distribution. Represents frequency Frequency domain signal, This indicates the upper limit of the integral. This represents the lower limit of integration. Differential unit representing frequency; Based on the target power equipment information, the composition of the surrounding gas is collected using gas sensors; Gas leak detection, calculation of gas leak index The expression is: ； in, Indicates the gas leakage index. Indicates gas composition, Indicates the first The concentration of the gas, Indicates the total number of gas types detected; High-definition images, crack density, temperature distribution, temperature anomaly coefficient, vibration signal, dominant frequency, vibration anomaly index, sound signal, noise energy distribution, gas composition, and gas leakage index are collected by vision, thermal imaging, vibration, sound, sound, and gas sensors to generate multimodal data packets.

6. The method for inspecting power equipment using a robotic dog as described in claim 5, characterized in that: The process involves calculating a device health score based on multimodal data packets using a health status assessment model, then employing AI algorithms for data fusion and intelligent analysis to generate a health diagnostic report for the target power equipment. The specific steps are as follows: Extract the raw data from the multimodal data packets and normalize the data. Normalized data The device health score is calculated using a supervised learning algorithm, expressed as: ； in, This indicates the health score of the target electrical equipment. Represents the exponentially decaying function. Indicates the temperature sensitivity coefficient. This represents the normalized crack density. This represents the normalized temperature anomaly coefficient. This represents the normalized dominant vibration frequency. This represents a normalized vibration anomaly index. This represents the normalized noise energy distribution. The normalized gas leakage index; Based on the analysis results of historical equipment operation data, the dividing points for the classification rules are defined as A and B; Based on the health score of the target power equipment, the health status of the equipment is classified using an AI classification algorithm; when At that time, the health score of the target power equipment is considered to be in a normal state; when If so, the health score of the target electrical equipment is considered to be slightly abnormal; when If so, the health score of the target electrical equipment is considered to be severely abnormal; For minor or severe anomalous states, the anomalous contribution of each mode is calculated using the following expression: ； in, Indicates the first The anomaly contribution of each modality Represents the normalized i-th One modal value, This represents the normalized sum of all modal values; Based on health scores and health status, generate a health diagnosis report for the target electrical equipment.

7. The method for inspecting power equipment using a robotic dog as described in claim 6, characterized in that: The health diagnosis report based on the target power equipment generates a task summary report by judging abnormal states, the severity of abnormal states, and summarizing the causes of abnormalities. The specific steps are as follows. Based on the abnormal contribution and health score of each mode in the health diagnosis report of the target power equipment, the severity of the abnormal state is generated, expressed as follows: ； in, Indicates the severity of the abnormal state. This represents the weight of each mode, reflecting the degree of influence of the corresponding mode on equipment anomalies. This indicates the degree of influence of the smoothed health score; Based on the severity of the abnormal state and the contribution of each modality to the abnormality, a summary of the causes of the abnormality is generated; Based on the abnormal status, the severity of the abnormal status, and the cause of the abnormality, a task summary report is generated.

8. A robot dog system for power equipment inspection, based on the robot dog method for power equipment inspection according to any one of claims 1 to 7, characterized in that: It includes a map data construction module, an optimal route generation module, an information confirmation module, a multimodal data packet generation module, a health diagnosis report generation module, and a task summary report generation module; The map data construction module is used to issue inspection tasks to the robot dog, including the location information of the target power equipment, equipment type, inspection priority and task requirements, and to scan the surrounding environment to build initial 3D map data. The optimal path generation module is used to define a comprehensive cost function, calculate the cost of path nodes, and generate the optimal path based on the location of the target power equipment and the initial 3D map data. The information confirmation module is used to scan the device number and edge features of the device using a visual sensor according to the optimal path, and compare them with the device information in the inspection task to confirm the target power equipment information. The multimodal data packet generation module is used to activate multimodal data acquisition based on the target power equipment information. The robot dog collects multimodal data from the target power equipment and generates multimodal data packets. The health diagnosis report generation module is used to calculate the equipment health score based on multimodal data packets and a health status assessment model, and to generate a health diagnosis report for the target power equipment by using AI algorithms for data fusion and intelligent analysis. The task summary report generation module is used to generate a task summary report based on the health diagnosis report of the target power equipment by judging the abnormal status, the severity of the abnormal status, and the cause of the abnormality.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the robot dog method for inspecting power equipment as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the robot dog method for inspecting power equipment as described in any one of claims 1 to 7.

Images