Power distribution network inspection robot system and method

Abstract

The invention discloses a power distribution network inspection robot system and method, and relates to the technical field of intelligent inspection, and the system comprises a data collection module, a data registration module, an anomaly detection module, a space verification module, a time sequence analysis module and a comprehensive judgment module. The data acquisition module controls the robot to acquire multi-modal data and record space-time positioning information; the data registration module registers the multi-modal data to an equipment coordinate system; the anomaly detection module extracts anomaly features and calculates confidence; the space verification module carries out space consistency verification and physical logic consistency verification according to the physical rule base and evaluates the abnormal level; the time sequence analysis module evaluates a defect evolution state; and the comprehensive judgment module performs defect judgment and generates a processing strategy. According to the method, single-mode false detection and environmental interference are effectively eliminated through a dual verification mechanism, comprehensive evaluation of the current state and the development trend of the defect is realized in combination with time sequence evolution analysis, and the accuracy and the reliability of defect detection are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent inspection, in particular to a power distribution network inspection robot system and method. BACKGROUND

[0002] As an important part of the power system, the operation state of the power distribution network is directly related to the power supply reliability and power safety. With the continuous expansion of the power distribution network scale and the continuous growth of the number of equipment, the traditional manual inspection method has been difficult to meet the growing inspection demand. In recent years, the inspection robot technology has developed rapidly, which can realize the automatic detection of power distribution equipment by carrying various sensors such as visible light cameras, infrared thermal imagers, and sound sensors, and improve the inspection efficiency.

[0003] However, the existing inspection robot technology still has deficiencies in defect detection accuracy. Single sensor is easily disturbed by environmental factors, leading to false detection, and multi-sensor data lacks effective correlation verification mechanism, making it difficult to accurately determine the real defect. At the same time, the existing technology relies on single detection result for judgment, lacks analysis of defect evolution trend, and cannot effectively distinguish between stable defects and rapidly deteriorating defects, affecting the scientificity and pertinence of maintenance decision. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a power distribution network inspection robot system and method to solve the technical problems of lack of effective correlation verification of multi-modal data, insufficient defect detection accuracy, and lack of defect evolution trend analysis in the prior art.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a power distribution network inspection robot system, comprising: a data acquisition module for reading a three-dimensional model of a power distribution equipment and historical inspection data, controlling the robot to move to a detection point, collecting multi-modal data of multiple sensors and recording spatio-temporal positioning information; a data registration module for registering the multi-modal data into a unified equipment coordinate system based on the three-dimensional model and the spatio-temporal positioning information; an anomaly detection module for extracting abnormal features from the registered modal data and performing anomaly judgment to obtain anomaly judgment results and confidence levels of each modal; a spatial verification module for performing spatial cross-verification on the anomaly judgment results according to a physical rule library, analyzing the spatial consistency and physical logic consistency of the abnormal features of different modalities, and evaluating the anomaly level of the verification results to obtain a spatial verification result; a time sequence analysis module configured to perform time sequence comparison between the feature parameters extracted in the current inspection and historical inspection data, calculate the change trend and change rate of the feature parameters, and evaluate the evolution state of the defect; a comprehensive judgment module configured to make a defect judgment based on the spatial verification result and the evolution state, and perform corresponding operations according to the judgment result.

[0007] As a preferred scheme of the power distribution network inspection robot system, the data acquisition module comprises: reading a three-dimensional model of the power distribution equipment and historical inspection records; analyzing the three-dimensional model, extracting the spatial position relationship of the equipment components, and calculating the coordinates of the detection points and the observation angles of the sensors; driving the robot to move to the detection points, and adjusting the posture of the robot to align the multiple sensors with the power distribution equipment; collecting multi-modal data of the power distribution equipment through the multiple sensors, wherein the multi-modal data comprises visible light images, infrared thermal images, sound signals and environmental parameters; recording the collection time stamps of the sensors, the spatial pose of the robot, and the installation pose of the sensors relative to the robot body.

[0008] As a preferred scheme of the power distribution network inspection robot system, the data registration module comprises: calculating the spatial transformation matrix of each sensor relative to the equipment coordinate system according to the spatial pose of the robot and the installation pose of the sensors in the space-time positioning information; extracting three-dimensional coordinates and normal vector information of the equipment surface from the three-dimensional model; transforming each modal data into the equipment coordinate system according to the spatial transformation matrix; correlating the transformed modal data to the corresponding positions on the equipment surface according to the three-dimensional coordinates, and establishing the spatial correspondence relationship between different modal data.

[0009] As a preferred scheme of the power distribution network inspection robot system, the anomaly detection module comprises: performing image processing on the registered visible light images, extracting appearance features and identifying appearance anomalies, and recording the spatial positions of the appearance anomalies; performing temperature analysis on the registered infrared thermal images, extracting temperature features and identifying temperature anomalies, and recording the spatial positions and temperature values of the temperature anomalies; performing frequency spectrum analysis on the registered sound signals, extracting acoustic features and identifying acoustic anomalies, and recording the feature parameters of the acoustic anomalies; environmentally compensating the identified anomaly features according to the environmental parameters; The abnormal recognition results of the various modalities are integrated to establish a correlation between the abnormal features and the spatial positions, and to obtain abnormal judgment results and confidence levels of the various modalities.

[0010] As a preferred scheme of the power distribution network inspection robot system, the spatial verification module comprises: The spatial position coordinates of the abnormal features in the device coordinate system are extracted from the abnormal judgment results of the various modalities. The spatial distances between the abnormal features of different modalities are calculated according to the spatial position coordinates, the spatial coincidence relationship of the abnormal features is determined, and a spatial position matching result is obtained. The multi-modality feature correlation rules corresponding to the abnormal feature types are queried from the physical rule library. The physically logically consistent verification of the spatially coincident abnormal features is performed according to the multi-modality feature correlation rules, and a physically logically consistent determination result is obtained. The abnormal features that pass the physically logically consistent verification are evaluated, and the abnormal level is determined according to the abnormal feature intensity and the multi-modality consistency degree. The spatial verification result containing the abnormal level is output.

[0011] As a preferred scheme of the power distribution network inspection robot system, the time sequence analysis module comprises: The historical feature parameter sequence and the historical detection time sequence of the target device are extracted from the historical inspection data. The feature parameters are extracted from the current abnormal determination result, and compared with the historical feature parameter sequence to calculate the change amount of the feature parameters. The time interval between adjacent detections is calculated according to the historical detection time sequence and the current detection time. The change rate of the feature parameters is calculated according to the change amount of the feature parameters and the time interval. The evolution state of the abnormal features is determined according to the change amount and the change rate.

[0012] As a preferred scheme of the power distribution network inspection robot system, the comprehensive determination module comprises: The evolution state and the abnormal level in the spatial verification result are obtained. The defect type and the defect severity are determined according to the abnormal level and the evolution state. The device health score is calculated according to the defect type and the defect severity. The processing strategy is generated according to the device health score and the evolution state, and the inspection report and the response signal are generated according to the processing strategy. The defect determination result, the device health score and the current detection data are updated to the device health file.

[0013] In a second aspect, the present application provides a power distribution network inspection robot method, comprising: reading a three-dimensional model of a power distribution device and historical inspection data, controlling the robot to move to a detection point, collecting multi-modal data of multiple sensors and recording spatio-temporal positioning information; based on the three-dimensional model and the spatio-temporal positioning information, registering the multi-modal data into a unified device coordinate system; extracting abnormal features from the registered modal data and performing abnormality determination to obtain abnormality determination results and confidence levels of each modality; performing spatial cross-validation on the abnormality determination results according to a physical rule library, performing spatial consistency analysis and physical logic consistency analysis on abnormal features of different modalities, and performing abnormality level evaluation on the validation results to obtain spatial validation results; performing time sequence comparison between the extracted feature parameters and the historical inspection data, calculating the change trend and rate of the feature parameters, and evaluating the evolution state of the defect; comprehensively determining the defect based on the spatial validation results and the evolution state, and performing corresponding operations according to the determination results.

[0014] In a third aspect, the present application provides a computer device comprising a memory and a processor, the memory storing a computer program, wherein the computer program is executed by the processor to implement any step of the power distribution network inspection robot system of the first aspect of the present application.

[0015] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement any step of the power distribution network inspection robot system of the first aspect of the present application.

[0016] The present application has the following advantages: the present application realizes collaborative cross-validation of multi-modal abnormal features through three-dimensional model guided multi-modal data accurate registration, combined with the dual verification mechanism of spatial consistency verification and physical logic consistency verification, effectively eliminates single modality false positives and environmental interference. At the same time, the temporal evolution information of the defect is integrated with the spatial verification result to realize comprehensive evaluation of the current state and development trend of the defect. Through multi-dimensional verification constrained by physical rules, the accuracy and reliability of defect identification are significantly improved, and difference processing strategies are generated based on defect severity and evolution trend, providing reliable technical support for intelligent operation and maintenance of power distribution networks. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without any creative effort.

[0018] Fig. 1 A module connection diagram of the power distribution network inspection robot system.

[0019] Fig. 2 A process diagram for generating a processing strategy of the power distribution network inspection robot system.

[0020] Fig. 3 A flowchart of the power distribution network inspection robot method. DETAILED DESCRIPTION

[0021] In order to make the above objectives, features and advantages of the present application more apparent and comprehensible, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0022] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0023] Secondly, the "one embodiment" or "embodiment" referred to herein means that a specific feature, structure or characteristic can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor does it mean an embodiment that is separate or selectively excluded from other embodiments.

[0024] Reference Figs. 1-3 For one embodiment of the present application, the embodiment provides a power distribution network inspection robot system, a module connection diagram as shown in Fig. 1 includes: A data acquisition module for reading a three-dimensional model of a power distribution device and historical inspection data, controlling the robot to move to a detection point, collecting multi-modal data of multiple sensors and recording spatio-temporal positioning information; A data registration module for registering multi-modal data into a unified device coordinate system based on the three-dimensional model and spatio-temporal positioning information; An anomaly detection module for extracting abnormal features from the registered modal data and performing anomaly determination to obtain anomaly determination results and confidence levels of each modality; The spatial verification module is used to perform spatial cross-validation on the anomaly judgment results based on the physical rule base, perform spatial consistency analysis and physical logic consistency analysis on the anomaly features of different modalities, evaluate the anomaly level of the verification results, and obtain the spatial verification results. The time series analysis module is used to compare the feature parameters extracted in this inspection with the historical inspection data in terms of execution time series, calculate the changing trend and rate of change of the feature parameters, and evaluate the evolution status of the defects. The comprehensive judgment module is used to determine defects by combining the spatial verification results and the evolution status, and to perform corresponding operations based on the judgment results.

[0025] Specifically, the data acquisition module includes: reading the three-dimensional model of the power distribution equipment and historical inspection records. The historical inspection data includes the multimodal characteristic values ​​of the equipment in the previous N inspections and the environmental parameters at the time of inspection.

[0026] Further, the 3D model is analyzed to extract the spatial relationships of equipment components, and the coordinates of detection points and sensor observation angles are calculated. The steps are as follows: Analyze the 3D model to identify the types of key components and their spatial coordinates. Extract the connection relationships and relative positional relationships between each key component. Determine candidate detection points for each key component based on the observation requirements and field of view of each sensor. Prioritize candidate detection points based on historical inspection records. Perform reachability verification on candidate detection points. Calculate the 3D coordinates of the detection points based on the spatial relationship between the candidate detection points and the components. Calculate the optimal observation angle for different sensors to determine the collaborative observation pose of multiple sensors. Generate robot motion commands based on the detection point coordinates and observation angles.

[0027] The priority ranking is determined based on the frequency of defects and the importance level of components in the historical inspection records, while accessibility verification is used to eliminate points that are blocked by obstacles or cannot be reached by the robot.

[0028] In one embodiment, a robot is driven to the detection point, and its posture is adjusted to align multiple sensors with the power distribution equipment. Multimodal data from the power distribution equipment is collected through a combination of multiple sensors. This multimodal data includes visible light images, infrared thermal images, sound signals, and environmental parameters, with the environmental parameters including at least ambient temperature. The multiple sensors achieve synchronous acquisition via hardware trigger signals, and the timestamp deviation of each modality's data is controlled within milliseconds.

[0029] Further, recording the collection time stamp of each sensor, the spatial pose of the robot, and the installation pose of each sensor relative to the robot body includes: synchronously acquiring the system time stamp when each sensor starts collecting data. Reading the three-dimensional position coordinates of the robot in the site coordinate system output by the robot positioning system, the three-axis attitude angle of the robot output by the robot attitude sensor, and the installation position and installation attitude parameters of each sensor relative to the robot body. Filtering processing is performed on the three-dimensional position coordinates and the three-axis attitude angle. The time stamp, three-dimensional position coordinates, three-axis attitude angle and installation parameters are stored in association to form a space-time positioning information record, and a data quality flag is added.

[0030] It should be noted that the data quality flag includes a time stamp exception flag, a position jump flag, an attitude exception flag and a sensor failure flag. When detecting data quality exception, corresponding processing is performed according to the exception type: if the position jump exceeds the set threshold, the repositioning process is performed; if the sensor fails, the sensor data is marked as invalid and it is judged whether the remaining sensors meet the collection requirements.

[0031] Preferably, the precise positioning of the detection point and the collaborative observation of the multi-sensor are realized by the three-dimensional model guidance, the spatial consistency of the multi-modal data is ensured by combining the space-time positioning information record, which lays a foundation for the subsequent multi-modal data registration and abnormal feature space cross verification, and improves the accuracy and reliability of defect detection. At the same time, based on the point priority sorting and data quality control mechanism of the historical inspection data, the inspection efficiency and data effectiveness are improved.

[0032] Specifically, the data registration module includes: calculating the spatial transformation matrix of each sensor relative to the device coordinate system according to the robot spatial pose and the sensor installation pose in the space-time positioning information, the steps are as follows: obtaining the spatial pose of the robot in the site coordinate system from the space-time positioning information, including three-dimensional position coordinates and three-axis attitude angle. Reading the installation pose parameters of each sensor relative to the robot body. According to the robot spatial pose and the installation pose parameters, the spatial pose of each sensor in the site coordinate system is calculated through coordinate transformation. Reading the position and attitude of the device in the site coordinate system from the three-dimensional model. According to the spatial pose of each sensor in the site coordinate system and the position and attitude of the device, the rotation matrix and translation vector of each sensor relative to the device coordinate system are calculated, and the spatial transformation matrix is obtained by combination.

[0033] Further, the three-dimensional coordinates and normal vector information of the device surface are extracted from the three-dimensional model. According to the spatial transformation matrix, each modal data is transformed into the device coordinate system.

[0034] Preferably, the transformed modal data is associated to corresponding positions on the equipment surface according to three-dimensional coordinates, establishing spatial correspondence between different modal data, including the following steps: dividing the equipment surface into spatial grid cells according to three-dimensional coordinates and spatial dimensions of equipment components. Traversing the transformed modal data, determining the grid cell to which each data point belongs according to the spatial coordinates and normal vector information of each data point. Organizing the modal data by grid cells, establishing association between different modal data in the same grid cell. Establishing a multi-modal data index for each grid cell, recording the modal data contained in the cell and their corresponding relationship.

[0035] In the embodiment, the above data registration scheme realizes accurate registration and spatial association of multi-modal data collected by different sensors in the equipment coordinate system through the spatial transformation matrix and the grid cell index, laying a foundation for spatial cross-verification of multi-modal abnormal features.

[0036] Specifically, the anomaly detection module includes: image processing of the registered visible light image, extracting appearance features and identifying appearance anomalies, recording the spatial position of the appearance anomaly. Wherein, the appearance feature parameters include edge contour features, color distribution features, texture features and shape parameters, and the appearance anomalies include surface discoloration, rust, cracks, deformation, contamination and mechanical damage.

[0037] In one embodiment, the appearance anomaly identification adopts a rule-based method or a deep learning-based image recognition method. Crack anomaly is identified through edge detection and morphological analysis, discoloration anomaly is identified through color distribution deviation analysis, and contamination anomaly is identified through texture feature change. The confidence calculation comprehensively considers the feature matching degree, the abnormal feature saliency and the image quality, and the confidence is higher when the abnormal feature is clear and the matching degree with the known pattern is high, and the confidence is lower when the feature is not obvious or the image quality is poor.

[0038] Further, temperature analysis is performed on the registered infrared thermal image, temperature features are extracted and temperature anomalies are identified, and the spatial position and temperature value of the temperature anomaly are recorded, including the following steps: temperature calibration is performed on the infrared thermal image, the pixel value is converted into the actual temperature value, and the equipment temperature distribution map is generated. Extracting temperature feature parameters, the temperature feature parameters include the maximum temperature value, the average temperature value, the temperature gradient, the hot spot area and the temperature distribution uniformity. Identifying temperature anomalies based on the extracted temperature feature parameters, the temperature anomalies include local high temperature points, abnormal heating areas and temperature distribution anomalies. Recording the spatial position and temperature value of the temperature anomaly.

[0039] In one embodiment, the temperature anomaly identification is based on threshold judgment or anomaly detection model. The device temperature is compared with the standard working temperature range, and the area exceeding the normal range is marked as temperature anomaly; the local hot spot is identified through temperature gradient analysis; the temperature rising trend is identified through comparison of historical temperature data. The confidence calculation comprehensively considers the temperature rise amplitude, temperature distribution characteristics and environmental temperature influence. For example, the greater the temperature rise amplitude and the more concentrated the temperature distribution, the higher the confidence; at the same time, the interference of environmental factors (such as direct sunlight) needs to be excluded, and for the temperature rise possibly caused by environmental factors, the confidence is correspondingly reduced.

[0040] Further, the registered sound signal is subjected to spectrum analysis, acoustic features are extracted and acoustic anomalies are identified, and the feature parameters of the acoustic anomalies are recorded, the steps being as follows: the sound signal is preprocessed, including noise reduction and signal enhancement. The spectrum analysis is performed by short-time Fourier transform, wavelet transform or mel frequency cepstral coefficient method. The acoustic feature parameters are extracted, including time domain feature parameters and frequency domain feature parameters. The acoustic anomalies are identified based on the extracted acoustic feature parameters, including abnormal discharge sound, arc sound, mechanical vibration anomaly and insulation defect sound. The feature parameters of the acoustic anomalies are recorded.

[0041] In one embodiment, the acoustic anomaly identification is based on feature pattern matching or voiceprint recognition model. The discharge sound is identified by high-frequency pulse feature, the arc sound is identified by energy mutation in a specific frequency band, and the mechanical vibration anomaly is identified by low-frequency periodic signal. The confidence calculation comprehensively considers signal intensity, feature matching degree and signal-to-noise ratio. For example, when the detected abnormal sound signal intensity is large, the feature is obvious and highly matched with the known abnormal pattern, the confidence is high; when the signal is weak or the background noise is large, the confidence is correspondingly reduced.

[0042] Further, the identified abnormal features are subjected to environmental compensation according to environmental parameters, including the following steps: reading the environmental parameters such as environmental temperature, humidity and wind speed at the time of collection; performing environmental temperature compensation on the temperature anomaly to eliminate the temperature rise caused by environmental temperature change or solar radiation; performing environmental noise compensation on the acoustic anomaly to eliminate the interference of environmental background noise; adjusting the abnormal judgment result and confidence according to the compensation result.

[0043] Preferably, the abnormal identification results of each modality are integrated to establish the correlation between the abnormal features and the spatial positions, to obtain the abnormal judgment result and the confidence of each modality, including the following steps: extracting the spatial position coordinates, abnormal type, feature parameters and confidence of the abnormal features from the abnormal identification results of each modality, to establish an abnormal feature list. The spatial distance between any two abnormal features in the abnormal feature list is calculated, and the abnormal features with a spatial distance less than a distance threshold are classified into an abnormal feature group, the distance threshold being determined based on the sensor positioning accuracy and the device component size.

[0044] For each abnormal feature group, the number of modalities contained in the group and the confidence of each modality abnormality are counted, and the comprehensive confidence of the abnormal feature group is calculated according to the weighted combination of the number of modalities and the confidence. The abnormal type combination, spatial position, feature parameter combination and comprehensive confidence of each abnormal feature group are output as the abnormal judgment result of each modality.

[0045] Among them, the determination of the distance threshold comprehensively considers the positioning accuracy of the sensor and the spatial scale of the equipment component, a smaller distance threshold is adopted for a smaller component, and a larger distance threshold is adopted for a larger component, so as to ensure that the multi-modal abnormalities of the same defect can be correctly grouped, and the abnormalities of different defects are avoided to be wrongly aggregated. The calculation of the comprehensive confidence adopts a weighted fusion method, the more the number of modalities, the higher the weight, and the higher the comprehensive confidence of each modality confidence.

[0046] In this embodiment, through the spatial clustering of abnormal features and the fusion of multi-modal confidence, the abnormal performance of the same defect in different modalities is grouped into a unified abnormal feature group, the correspondence between the abnormal type combination and the spatial position is established, the structured input is provided for the subsequent spatial cross verification and physical logic consistency analysis, the redundancy and dispersion of multi-modal abnormal information are avoided, and the efficiency and accuracy of defect judgment are improved.

[0047] Specifically, the spatial verification module includes: extracting the spatial position coordinates of the abnormal features in the equipment coordinate system from the abnormal judgment results of each modality.

[0048] Further, according to the spatial position coordinates, the spatial distance between different modality abnormal features is calculated, the spatial coincidence relationship of the abnormal features is determined, and the spatial position matching result is obtained, and the steps are as follows: calculating the three-dimensional spatial distance between any two different modality abnormal features. The different modality abnormal features with a spatial distance less than the spatial matching threshold are determined to be spatially coincident. The spatially coincident abnormal features are indexed to establish a spatial correlation index, and the correlated modality type, abnormal type and spatial position are recorded. The spatial position matching result is output, including the combination of spatially coincident abnormal features and the spatial distance.

[0049] Among them, the spatial matching threshold is determined according to the positioning accuracy of the sensor and the spatial scale of the equipment component, a smaller matching threshold is adopted for a sensor combination with higher positioning accuracy, and a larger matching threshold is adopted for a sensor combination with lower positioning accuracy.

[0050] Further, the multi-modal feature correlation rule corresponding to the abnormal feature type is queried from the physical rule library, including: determining the rule type to be queried according to the abnormal type in the combination of spatially coincident abnormal features. Retrieving the matched multi-modal feature correlation rule from the physical rule library. Extracting the physical logic judgment condition and feature parameter constraint relationship in the correlation rule.

[0051] In one embodiment, the construction process of the physical rule base is as follows: collect historical defect data of power distribution equipment, including the performance characteristics and characteristic parameters of different defect types under various modalities. Analyze the physical mechanism of different defect types, and establish the correlation between the physical characteristics of defects and the multi-modal detection characteristics. According to the physical mechanism and historical data, formulate multi-modal feature correlation rules, including the constraint relationship and logical judgment condition of the characteristic parameters. Organize the multi-modal feature correlation rules into a structured physical rule base to support rule query and matching based on abnormal type combination.

[0052] Preferably, the spatially coincident abnormal features are verified for physical and logical consistency according to the multi-modal feature correlation rules, and a physical and logical consistency determination result is obtained, the steps being as follows: extracting the characteristic parameters of the spatially coincident abnormal features. According to the physical and logical judgment conditions in the multi-modal feature correlation rules, verify whether the correlation between different modal characteristic parameters conforms to the physical law. Calculate the physical and logical consistency score, and the higher the score, the stronger the physical and logical consistency of the multi-modal abnormal features. Output the physical and logical consistency determination result, including the abnormal feature combination that passes the verification and its consistency score.

[0053] Illustratively, the physical and logical consistency verification includes the following typical rules: for joint heating defects, verify whether the high-temperature area of the infrared thermal image and the discoloration area of the visible light image are spatially coincident, and whether there is a positive correlation between the temperature value and the discoloration degree. For insulation defects, verify whether the crack or stain in the visible light image and the discharge feature of the sound signal are spatially coincident, and whether there is a positive correlation between the crack size or stain degree and the discharge signal strength. For mechanical looseness defects, verify whether the deformation or displacement in the visible light image and the vibration feature of the sound signal are spatially coincident, and whether there is a positive correlation between the deformation degree and the vibration amplitude.

[0054] Further, the abnormality of the physical and logical consistency verification is evaluated, and the abnormality level is determined according to the abnormal feature intensity and the multi-modal consistency degree, specifically including: extracting the abnormal feature intensity index, including the temperature deviation amplitude, the appearance defect area, the acoustic signal energy, etc. According to the abnormal feature intensity index and the physical and logical consistency score, calculate the abnormal comprehensive score. According to the abnormal comprehensive score, divide the abnormality level, including serious abnormality, general abnormality and slight abnormality. Output the abnormality level evaluation result, including the abnormal type, the spatial position, the abnormality level and the comprehensive score.

[0055] Through the technical solution, the spatial cross-validation module realizes collaborative gain through a double verification mechanism of spatial consistency verification and physical logic consistency verification: the spatial consistency verification ensures that different modal anomalies point to the same physical location, eliminating spatially isolated false detections; the physical logic consistency verification further confirms that the multi-modal anomaly features that coincide in space conform to the physical law constraints, eliminating false judgments that are physically and logically contradictory. The two layers of verification form a verification chain that filters layer by layer, effectively reducing false judgments caused by single-modal false detections and environmental interference, and significantly improving the accuracy and reliability of defect detection. At the same time, the level evaluation mechanism based on anomaly feature intensity and multi-modal consistency degree realizes the quantitative evaluation of defect severity, providing a reliable basis for maintenance decisions.

[0056] Specifically, the time series analysis module includes: extracting a historical feature parameter sequence and a historical detection time sequence of the target device from historical inspection data.

[0057] Further, the feature parameters are extracted from the current anomaly determination result, and compared with the historical feature parameter sequence to calculate the change amount of the feature parameters, including: extracting the feature parameters from the current anomaly determination result, including temperature value, defect area, signal energy, etc. Extracting the historical values corresponding to the current feature parameters from the historical feature parameter sequence. Calculate the difference between the current feature parameter and the historical feature parameter to obtain the absolute change amount of the feature parameter. Calculate the change proportion of the current feature parameter relative to the historical feature parameter to obtain the relative change amount of the feature parameter.

[0058] Further, according to the historical detection time sequence and the current detection time, the time interval between adjacent detections is calculated. According to the change amount and the time interval of the feature parameters, the change rate of the feature parameters is calculated.

[0059] Preferably, according to the change amount and the change rate, the evolution state of the anomaly feature is determined, including: setting the change amount threshold and the change rate threshold according to the defect type and the importance of the component. Compare the change amount of the feature parameter with the change amount threshold, and compare the change rate with the change rate threshold. Determine the evolution state according to the comparison result and output the evolution state determination result, including the anomaly type, the evolution state and the evolution trend parameter.

[0060] In one embodiment, the determination rule of the evolution state is as follows: when the change amount is less than the change amount threshold and the change rate is close to zero, it is determined as a stable state; when the change amount is greater than the change amount threshold but the change rate is lower than the change rate threshold, it is determined as a slow deterioration state; when the change amount is greater than the change amount threshold and the change rate is higher than the change rate threshold, it is determined as a rapid deterioration state; when the change amount is negative, it is determined as an improved state; when the historical feature parameter sequence does not exist for the anomaly feature, it is determined as a new defect.

[0061] Based on the above technical scheme, the time sequence analysis module realizes quantitative evaluation of the abnormal feature evolution trend by extracting the historical feature parameter sequence and calculating the change amount and change rate. The scheme introduces the time dimension evolution information of the defect into the judgment process, can distinguish between stable defects and rapidly deteriorating defects, provides a basis for maintenance priority ranking and preventive maintenance decision, avoids misjudgment caused by relying only on single detection results, and improves the forward-looking and effectiveness of defect management.

[0062] Specifically, the comprehensive judgment module includes: obtaining the evolution state and the abnormal level in the spatial verification result. According to the abnormal level and the evolution state, the defect judgment is carried out to determine the defect type and the defect severity, and the steps are as follows: reading the abnormal level in the spatial verification result and the evolution state in the time sequence analysis result. Establish the mapping relationship between the abnormal level and the evolution state, and determine the defect severity according to the mapping relationship. For defects with the same abnormal level, the evolution state is rapidly deteriorating, the severity level is improved, the evolution state is stable or improved, and the severity level is maintained or reduced. Output the defect judgment result, including the defect type, the defect severity and the spatial position.

[0063] Further, according to the defect type and the defect severity, the device health score is calculated, including the following steps: counting the number and severity distribution of each type of defect on the device. According to the defect type, the defect severity and the component importance, the weighted score of each defect is calculated. The weighted scores of each defect are summarized to calculate the comprehensive health score of the device. According to the health score, the device health level is divided.

[0064] Further, according to the device health score and the evolution state, a processing strategy is generated, and the processing strategy generation flow chart is as shown in Fig. 2 According to the processing strategy, an inspection report and a response signal are generated.

[0065] Among them, the processing strategy includes: immediate processing strategy, suitable for the case that the defect severity is serious and the evolution state is rapidly deteriorating, generating an emergency alarm signal and notifying the operation and maintenance personnel to handle immediately. The planned processing strategy is suitable for the case that the defect severity is general or the evolution state is slowly deteriorating, generating a maintenance plan and arranging the subsequent processing time. The continuous monitoring strategy is suitable for the case that the defect severity is slight and the evolution state is stable, increasing the inspection frequency of the device and continuously tracking. The normal inspection strategy is suitable for the case that the device health condition is good, and the inspection is carried out according to the normal cycle.

[0066] Further, the defect judgment result, the device health score and the detection data of this time are updated to the device health file.

[0067] Through the above scheme, the spatial verification result and the timing evolution state are combined by the comprehensive judgment module for defect judgment, the comprehensive evaluation of the current state and development trend of the defect is realized, and the differentiated processing strategy is generated according to the defect severity and evolution trend, and the scientificity and pertinence of the maintenance decision are improved.

[0068] The embodiment also provides a power distribution network inspection robot method, a flow chart is as shown in Fig. 3 The embodiment also provides a power distribution network inspection robot method, a flow chart is as shown in The three-dimensional model and the historical inspection data of the power distribution equipment are read, the robot is controlled to move to the detection point, the multi-modal data of the multi-sensor are collected and the space-time positioning information is recorded; Based on the three-dimensional model and the space-time positioning information, the multi-modal data are registered into a unified equipment coordinate system; Abnormal features are extracted from the registered modal data and abnormal judgment is performed, abnormal judgment results and confidence levels of the modal data are obtained; According to the physical rule library, spatial cross verification is performed on the abnormal judgment results, spatial consistency analysis and physical logic consistency analysis are performed on the abnormal features of different modal data, abnormal level evaluation is performed on the verification results, and spatial verification results are obtained; The feature parameters extracted in this inspection are compared with the historical inspection data in time sequence, the change trend and the change rate of the feature parameters are calculated, and the evolution state of the defect is evaluated; The spatial verification result and the evolution state are combined for defect judgment, and corresponding operations are performed according to the judgment result.

[0069] The embodiment also provides a computer device suitable for the power distribution network inspection robot system, including a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the power distribution network inspection robot system provided by the above embodiment.

[0070] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is configured to perform wired or wireless communication with an external terminal. The wireless communication can be achieved by WIFI, an operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0071] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the power distribution network inspection robot system according to the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic storage, a flash memory, a magnetic disk or an optical disk.

[0072] To sum up, the application realizes the collaborative cross-verification of multi-modal abnormal features by the three-dimensional model guided multi-modal data accurate registration, the dual verification mechanism of spatial consistency verification and physical logic consistency verification, effectively eliminates single modal false detection and environmental interference. Meanwhile, the temporal evolution information of defects is fused with the spatial verification result to realize the comprehensive evaluation of the current state and development trend of defects. The multi-dimensional verification constrained by physical rules significantly improves the accuracy and reliability of defect identification, and generates a differentiated processing strategy based on the defect severity and evolution trend, which provides reliable technical support for intelligent operation and maintenance of power distribution networks.

[0073] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A power distribution network inspection robot system, characterized in that: include: The data acquisition module is used to read the 3D model of the power distribution equipment and historical inspection data, control the robot to move to the detection point, collect multimodal data from multiple sensors, and record spatiotemporal positioning information; The data registration module is used to register multimodal data to a unified device coordinate system based on the three-dimensional model and the spatiotemporal positioning information. The anomaly detection module is used to extract abnormal features from the registered modal data and make anomaly judgments, and obtain the anomaly judgment results and confidence levels for each modality. The spatial verification module is used to perform spatial cross-validation on the anomaly judgment results based on the physical rule base, perform spatial consistency analysis and physical logic consistency analysis on the anomaly features of different modalities, evaluate the anomaly level of the verification results, and obtain the spatial verification results. The time series analysis module is used to compare the feature parameters extracted in this inspection with the historical inspection data in terms of execution time series, calculate the changing trend and rate of change of the feature parameters, and evaluate the evolution status of the defects. The comprehensive judgment module is used to determine defects by combining the spatial verification results and the evolution state, and to perform corresponding operations based on the judgment results.

2. The power distribution network inspection robot system as described in claim 1, characterized in that: The data acquisition module includes: Read the 3D model and historical inspection records of the power distribution equipment; Analyze the 3D model, extract the spatial relationships of the equipment components, and calculate the coordinates of the detection points and the sensor observation angles; Drive the robot to the detection point and adjust the robot's posture to align multiple sensors with the power distribution equipment; Multimodal data of power distribution equipment is collected by combining multiple sensors, including visible light images, infrared thermal images, sound signals, and environmental parameters; Record the data acquisition timestamps of each sensor, the robot's spatial pose, and the installation pose of each sensor relative to the robot body.

3. The power distribution network inspection robot system as described in claim 1, characterized in that: The data registration module includes: Based on the robot's spatial pose and sensor installation pose in the spatiotemporal positioning information, calculate the spatial transformation matrix of each sensor relative to the device coordinate system; Extract the three-dimensional coordinates and normal vector information of the device surface from the three-dimensional model; The modal data are transformed into the device coordinate system according to the spatial transformation matrix; Based on the three-dimensional coordinates, the transformed modal data are associated with the corresponding positions on the device surface, establishing a spatial correspondence between different modal data.

4. The power distribution network inspection robot system as described in claim 1, characterized in that: The anomaly detection module includes: Image processing is performed on the registered visible light image to extract appearance features and identify appearance anomalies, and the spatial location of appearance anomalies is recorded. Temperature analysis is performed on the registered infrared thermal images to extract temperature features and identify temperature anomalies, and the spatial location and temperature value of the temperature anomalies are recorded. Spectral analysis is performed on the registered sound signal to extract acoustic features and identify acoustic anomalies, and the characteristic parameters of the acoustic anomalies are recorded. Environmental compensation is applied to the identified anomalous features based on environmental parameters; By integrating the anomaly identification results of each modality, the correlation between anomaly features and spatial location is established, and the anomaly judgment results and confidence levels of each modality are obtained.

5. The power distribution network inspection robot system as described in claim 1, characterized in that: The space verification module includes: Extract the spatial coordinates of the abnormal features in the device coordinate system from the abnormality determination results of each mode; Calculate the spatial distance between different modal anomaly features based on the spatial location coordinates, determine the spatial overlap relationship of the anomaly features, and obtain the spatial location matching result; Query the multimodal feature association rules corresponding to the anomaly feature types from the physical rule base; Based on the multimodal feature association rules, the physical and logical consistency of spatially overlapping abnormal features is verified to obtain the physical and logical consistency judgment result. Anomalies that pass physical logic consistency verification are evaluated for level, and the anomaly level is determined based on the anomaly characteristic intensity and multimodal consistency degree. The output includes spatial verification results with anomaly levels.

6. The power distribution network inspection robot system as described in claim 1, characterized in that: The time series analysis module includes: Extract the historical characteristic parameter sequence and historical detection time sequence of the target equipment from historical inspection data; Feature parameters are extracted from the current anomaly detection result and compared with the historical feature parameter sequence to calculate the change in the feature parameters; Calculate the time interval between adjacent detections based on the historical detection time series and the current detection time; The rate of change of the characteristic parameter is calculated based on the amount of change of the characteristic parameter and the time interval. The evolutionary state of the abnormal feature is determined based on the amount of change and the rate of change.

7. The power distribution network inspection robot system as described in claim 1, characterized in that: The comprehensive determination module includes: Obtain the anomaly level in the evolutionary state and spatial verification results; Defects are determined based on the anomaly level and the evolutionary state to identify the defect type and severity. Calculate the equipment health score based on the defect type and the defect severity; A processing strategy is generated based on the equipment health score and the evolution status, and an inspection report and response signal are generated based on the processing strategy. Update the defect determination results, equipment health score, and this test data to the equipment health record.

8. A method for a power distribution network inspection robot, based on the power distribution network inspection robot system according to any one of claims 1 to 7, characterized in that: include: Read the 3D model and historical inspection data of the power distribution equipment, control the robot to move to the detection point, collect multimodal data from multiple sensors and record spatiotemporal positioning information; Based on the 3D model and the spatiotemporal positioning information, the multimodal data is registered to a unified device coordinate system; Anomaly features are extracted from the registered modal data and anomaly judgment is performed to obtain the anomaly judgment results and confidence levels for each modality. Spatial cross-validation is performed on the anomaly determination results based on the physical rule base. Spatial consistency analysis and physical logic consistency analysis are performed on the anomaly features of different modalities. Anomaly level evaluation is performed on the validation results to obtain spatial validation results. The feature parameters extracted in this inspection are compared with the execution time sequence of historical inspection data to calculate the changing trend and rate of change of the feature parameters and assess the evolution status of the defects. Defects are determined by combining the spatial verification results and the evolutionary state, and corresponding operations are performed based on the determination results.

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 power distribution network inspection robot system according to 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 power distribution network inspection robot system according to any one of claims 1 to 7.

Images