Unmanned aerial vehicle inspection method and device, computer device, readable storage medium and program product

Abstract

The application relates to an unmanned aerial vehicle inspection method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring infrared thermal imaging data, discharge spectrum data and visible light imaging data of a target power device based on a carrying device of an unmanned aerial vehicle, determining the contour of a target area, locking the target power device based on the contour and a defocus detection assembly, and collecting high-resolution spectrum data; analyzing the high-resolution spectrum data, establishing a quantitative model between the energy level difference and the redshift wavelength offset, determining the correlation mapping relationship between the discharge defect type and the redshift spectrum, correcting the correlation mapping relationship, obtaining the final correlation mapping relationship, distributing a preset weight, fusing to obtain a fault image, carrying out noise reduction processing, and outputting the discharge defect detection result of the target power device and the processed fault image. The method can improve the accuracy of power equipment defect detection.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) inspection technology, and in particular to a UAV inspection method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Technology

[0002] With the continuous expansion of the power system, the inspection and maintenance of power equipment faces challenges such as low efficiency, high risk, and incomplete coverage. Traditional manual inspection methods are limited by factors such as terrain and environment, making it difficult to efficiently inspect power equipment in high-altitude and remote areas, and their ability to identify early potential defects in equipment is also limited. Therefore, drone inspection technology has been introduced.

[0003] In related technologies, existing drone inspection technologies mostly use a single type of optical sensor, which can only collect imaging data in a single band and cannot comprehensively capture the operating status information of power equipment. For example, relying solely on visible light imaging makes it difficult to detect thermal and discharge defects inside the equipment, and relying solely on infrared thermal imaging makes it difficult to accurately locate the specific physical form of defects, resulting in inaccurate final detection results. Summary of the Invention

[0004] Therefore, it is necessary to provide a drone inspection method, device, computer equipment, computer-readable storage medium, and computer program product that can provide accurate power equipment fault detection results to address the aforementioned technical problems.

[0005] Firstly, this application provides a method for unmanned aerial vehicle (UAV) inspection, including:

[0006] The equipment carried by the drone acquires infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment;

[0007] Based on the visible light imaging data, the outline of the target area is determined, the target power equipment is locked based on the outline and the defocus detection component, and high-resolution spectral data of the target power equipment is acquired based on the focus detection component.

[0008] The high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift, and based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined.

[0009] Based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship;

[0010] Based on the final association mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectrum data and visible light imaging data, and fused to obtain the fault image of the target power equipment;

[0011] The fault image is denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output.

[0012] In one embodiment, before acquiring high-resolution spectral data of the target power equipment based on the positive focus detection component, the method further includes:

[0013] The real-time measurement parameters of the positive focus detection component are determined, and a first model of the light field propagation path is constructed based on the real-time measurement parameters;

[0014] A priori constraint on the distribution of surface temperature of the target power equipment is determined, the priori constraint is mapped to an optical refractive index correction matrix, and a second model is constructed based on the differentiated optical attenuation coefficient weights configured for different temperature gradient intervals.

[0015] Align and fuse the first model and the second model to obtain the light field spatial geometric model;

[0016] Based on the optical field spatial geometric model, the adaptation deviation between the optical parameters and the temperature constraint parameters is calculated, and the real-time measurement parameters are corrected based on the adaptation deviation to obtain the final measurement parameters.

[0017] Based on the final measurement parameters, the focus detection component is adjusted.

[0018] In one embodiment, the step of analyzing the high-resolution spectral data to establish a quantitative model between the energy level difference and the redshift wavelength shift includes:

[0019] The high-resolution spectral data is divided into bands to obtain the divided spectral data;

[0020] Feature peak positions are extracted from the split spectral data, and energy level differences are calculated based on the feature peak positions;

[0021] Obtain the real-time temperature and pressure of the target area;

[0022] Based on the energy level difference, the real-time temperature, and the real-time pressure, a quantitative model is established between the energy level difference and the redshift wavelength shift.

[0023] In one embodiment, the step of correcting the correlation mapping relationship based on the energy flux density attenuation coefficient to obtain the final correlation mapping relationship includes:

[0024] Obtain the spatial distance between the drone and the target area, as well as the ambient atmospheric humidity;

[0025] Calculate the energy flux density attenuation coefficient based on the spatial distance and the ambient atmospheric humidity;

[0026] When the energy flux density attenuation coefficient is greater than the preset reference coefficient, the severity level of the discharge defect in the correlation mapping relationship is adjusted upward to obtain the final correlation mapping relationship;

[0027] If the energy flux density attenuation coefficient is less than the preset reference coefficient, the severity level of the discharge defect in the correlation mapping relationship is downgraded to obtain the final correlation mapping relationship.

[0028] In one embodiment, the step of assigning preset weights to infrared thermal imaging data, discharge spectral data, and visible light imaging data based on the final association mapping relationship, and fusing them to obtain a fault image of the target power equipment, includes:

[0029] Based on the final association mapping relationship, the dominant characteristics of the discharge defects of the target power equipment are determined;

[0030] The current inspection scenario is determined. Based on the dominant features and the current inspection scenario, preset weights are assigned to infrared thermal imaging data, discharge spectral data and visible light imaging data, and weighted fusion is performed to obtain the fault image of the target power equipment.

[0031] In one embodiment, the step of denoising the fault image to obtain a processed fault image includes:

[0032] Collect real-time data on the illumination interference dimension in the current inspection scenario, and generate an illumination intensity interference vector based on the real-time data;

[0033] The light intensity interference vector is compared with the standard light intensity interference vector to determine the light difference vector;

[0034] The illumination difference vector is converted into a three-dimensional blurred form, and the correlation coefficients of the fault image in different dimensions are determined respectively;

[0035] Based on the correlation coefficient, the fault image is subjected to noise reduction processing to obtain the processed fault image.

[0036] Secondly, this application also provides a drone inspection device, comprising:

[0037] The acquisition module is used to acquire infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment based on the equipment carried by the UAV;

[0038] The acquisition module is used to determine the outline of the target area based on the visible light imaging data, lock the target power equipment based on the outline and the defocus detection component, and acquire high-resolution spectral data of the target power equipment based on the focus detection component.

[0039] The determination module is used to analyze the high-resolution spectral data, establish a quantitative model between energy level difference and redshift wavelength shift, and determine the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum based on the quantitative model.

[0040] The correction module is used to correct the correlation mapping relationship based on the energy flux density attenuation coefficient to obtain the final correlation mapping relationship;

[0041] The fusion module is used to assign preset weights to infrared thermal imaging data, discharge spectrum data and visible light imaging data based on the final correlation mapping relationship, and fuse them to obtain a fault image of the target power equipment.

[0042] The output module is used to perform noise reduction processing on the fault image to obtain the processed fault image, and output the discharge defect detection result and the processed fault image of the target power equipment.

[0043] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0044] The equipment carried by the drone acquires infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment;

[0045] Based on the visible light imaging data, the outline of the target area is determined, the target power equipment is locked based on the outline and the defocus detection component, and high-resolution spectral data of the target power equipment is acquired based on the focus detection component.

[0046] The high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift, and based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined.

[0047] Based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship;

[0048] Based on the final association mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectrum data and visible light imaging data, and fused to obtain the fault image of the target power equipment;

[0049] The fault image is denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output.

[0050] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0051] The equipment carried by the drone acquires infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment;

[0052] Based on the visible light imaging data, the outline of the target area is determined, the target power equipment is locked based on the outline and the defocus detection component, and high-resolution spectral data of the target power equipment is acquired based on the focus detection component.

[0053] The high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift, and based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined.

[0054] Based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship;

[0055] Based on the final association mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectrum data and visible light imaging data, and fused to obtain the fault image of the target power equipment;

[0056] The fault image is denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output.

[0057] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0058] The equipment carried by the drone acquires infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment;

[0059] Based on the visible light imaging data, the outline of the target area is determined, the target power equipment is locked based on the outline and the defocus detection component, and high-resolution spectral data of the target power equipment is acquired based on the focus detection component.

[0060] The high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift, and based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined.

[0061] Based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship;

[0062] Based on the final association mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectrum data and visible light imaging data, and fused to obtain the fault image of the target power equipment;

[0063] The fault image is denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output.

[0064] The aforementioned UAV inspection method, device, computer equipment, computer-readable storage medium, and computer program product first acquire infrared thermal imaging data, discharge spectrum data, and visible light imaging data of the target power equipment using the UAV's onboard equipment; based on the visible light imaging data, the outline of the target area is determined; based on the outline and a defocus detection component, the target power equipment is locked, and high-resolution spectral data of the target power equipment is acquired based on a focus detection component; the high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift, and based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined; based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship; based on the final correlation mapping relationship, preset weights are assigned to the infrared thermal imaging data, discharge spectrum data, and visible light imaging data, and they are fused to obtain a fault image of the target power equipment; the fault image is denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output. In this way, by using multi-source heterogeneous optical sensors to achieve comprehensive multi-band data acquisition, the collaborative detection module can accurately lock onto the target power equipment, optimize the data acquisition accuracy in complex environments, the quantitative spectral redshift analysis module can achieve quantitative identification of defects, and the multi-modal data fusion diagnostic module can improve the reliability of diagnostic results, thus significantly improving the comprehensiveness, accuracy and efficiency of power equipment defect detection. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a diagram illustrating the application environment of the UAV inspection method in one embodiment.

[0067] Figure 2 This is a flowchart illustrating a drone inspection method in one embodiment;

[0068] Figure 3 This is a structural block diagram of a drone inspection device in one embodiment;

[0069] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0071] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0072] The drone inspection method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located on the cloud or other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0073] In one exemplary embodiment, such as Figure 2 As shown, a method for drone inspection is provided, which can be applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps 202 to 212. Wherein:

[0074] Step 202: Acquire infrared thermal imaging data, discharge spectrum data, and visible light imaging data of the target power equipment based on the equipment mounted on the UAV.

[0075] For example, the incident light of the target power equipment is collected by the sensors on the UAV, and the incident light of the target power equipment is split into infrared band, ultraviolet band and visible light band to determine the infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment.

[0076] Incident light refers to the light rays that enter the optical path of the system, generated during the reflection, radiation, or discharge process from the surface of the target power equipment. For example, infrared light radiated from the surface of transmission line conductors, ultraviolet light generated by insulator discharge, and sunlight reflected from tower structures are all incident light.

[0077] Infrared thermal imaging data refers to two-dimensional image data that reflects the surface temperature distribution of the target power equipment, with each pixel corresponding to a temperature value. Discharge spectral data refers to ultraviolet spectral signal data generated during the discharge process of the target power equipment, including information such as discharge intensity and wavelength distribution. Visible light imaging data refers to two-dimensional image data that reflects the shape and outline of the target power equipment.

[0078] Step 204: Based on visible light imaging data, determine the outline of the target area, lock the target power equipment based on the outline and the defocus detection component, and acquire high-resolution spectral data of the target power equipment based on the focus detection component.

[0079] Optionally, the defocus detection component extracts the contour feature signal of the target area using large field-of-view and visible light imaging data, transmits it to the focus detection component to lock onto the target power equipment, and acquires high-resolution spectral data of the target power equipment.

[0080] The defocus detection component uses a wide-angle lens with a field of view of 60° to quickly capture the outline of the target area and extract the outline feature signals of the target area, such as the linear features of power transmission lines and the rectangular features of towers. The focus detection component uses a telephoto lens with a field of view of 5°, which can acquire the spectral data of the target equipment with a small field of view and high resolution.

[0081] Step 206: Analyze the high-resolution spectral data, establish a quantitative model between the energy level difference and the redshift wavelength shift, and based on the quantitative model, determine the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum.

[0082] High-resolution spectral data refers to spectral signal data with a spectral resolution ≥ 0.1 nm that can clearly present the characteristic peak positions of molecular and particle energy level transitions.

[0083] For example, high-resolution spectral data is analyzed, and a quantitative model of quantum transition energy level difference-redshift wavelength shift is established based on the spectral characteristics of molecules and particles during the discharge process of the target power equipment. That is, a quantitative model between energy level difference and redshift wavelength shift is established. Based on the historical discharge defect data of the target power equipment and the division of redshift wavelength shift intervals, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined.

[0084] Among them, the types of discharge defects include corona discharge, arc discharge, and partial discharge, which are different in terms of the quantum transition characteristics of molecules / particles during the discharge process.

[0085] Among them, the redshift wavelength shift is the value by which the peak position of the spectral characteristic generated by the target power equipment during the discharge process shifts towards a longer wavelength relative to the peak position under standard conditions. For example, the characteristic peak position of the nitrogen discharge spectrum under standard conditions is 260nm, while the peak position in actual detection is 260.5nm, and the redshift wavelength shift is 0.5nm.

[0086] Among them, the quantitative model refers to the model that establishes the mathematical correspondence between the quantum transition energy level difference and the redshift wavelength shift, which can calculate the redshift wavelength shift through the energy level difference, or inversely deduce the energy level difference through the redshift wavelength shift.

[0087] Step 208: Based on the energy flux density attenuation coefficient, correct the correlation mapping relationship to obtain the final correlation mapping relationship.

[0088] Optionally, the correlation mapping relationship can be corrected by incorporating the energy flux density attenuation coefficient to obtain the final correlation mapping relationship.

[0089] The correlation mapping relationship is the correspondence between the type and severity of discharge defects and the redshift spectral characteristics (redshift wavelength shift, characteristic peak shape).

[0090] Step 210: Based on the final association mapping relationship, assign preset weights to infrared thermal imaging data, discharge spectrum data and visible light imaging data, and fuse them to obtain the fault image of the target power equipment.

[0091] For example, based on the final association mapping relationship, the importance of the data and its correlation with the defect features are determined, and preset weights are assigned to infrared thermal imaging data, discharge spectrum data and visible light imaging data respectively. For example, when the dominant feature of the discharge defect is temperature anomaly, the weight of infrared thermal imaging data is assigned as 0.6, the weight of discharge spectrum data is 0.3, and the weight of visible light imaging data is 0.1. The imaging data, discharge spectrum data and visible light imaging data after weight assignment are fused to obtain the fault image of the target power equipment.

[0092] Step 212: Denoise the fault image to obtain the processed fault image, and output the discharge defect detection result of the target power equipment and the processed fault image.

[0093] Optionally, an ambient light intensity interference factor is introduced to denoise the fault image, resulting in a processed fault image. The discharge defect detection results of the target power equipment and the processed fault image are then output.

[0094] The discharge defect detection results include the type, location, and severity of discharge defects in the target power equipment.

[0095] The severity level refers to the degree of impact of the discharge defect on the safe operation of the target power equipment. For example, it can be divided into three levels: mild, moderate, and severe. For example, a mild defect has a corona discharge intensity ≤ [missing information - likely a value]. No obvious equipment damage; moderate defect: arc discharge duration ≤1s, slight corrosion on equipment surface; severe defect: partial discharge amplitude ≥1000pC, aging of internal insulation of equipment.

[0096] Among them, ambient light intensity interference factors are ambient light-related factors that affect image quality, such as light intensity, light angle, and light uniformity.

[0097] In the aforementioned UAV inspection method, infrared thermal imaging data, discharge spectrum data, and visible light imaging data of the target power equipment are acquired based on the equipment mounted on the UAV. Based on the visible light imaging data, the outline of the target area is determined. The target power equipment is then locked based on the outline and a defocus detection component, and high-resolution spectral data of the target power equipment is acquired based on a focus detection component. The high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift. Based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined. Based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship. Based on the final correlation mapping relationship, preset weights are assigned to the infrared thermal imaging data, discharge spectrum data, and visible light imaging data, and they are fused to obtain a fault image of the target power equipment. The fault image is then denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output. In this way, by using multi-source heterogeneous optical sensors to achieve comprehensive multi-band data acquisition, the collaborative detection module can accurately lock onto the target power equipment, optimize the data acquisition accuracy in complex environments, the quantitative spectral redshift analysis module can achieve quantitative identification of defects, and the multi-modal data fusion diagnostic module can improve the reliability of diagnostic results, thus significantly improving the comprehensiveness, accuracy and efficiency of power equipment defect detection.

[0098] In an exemplary embodiment, before acquiring high-resolution spectral data of the target power equipment based on the positive-focus detection component, the method further includes: determining the real-time measurement parameters of the positive-focus detection component, and constructing a first model of the light field propagation path based on the real-time measurement parameters; determining the prior constraints of the surface temperature distribution of the target power equipment, mapping the prior constraints to a light field refractive index correction matrix, and constructing a second model by configuring differentiated light field attenuation coefficient weights based on different temperature gradient intervals; aligning and fusing the first model and the second model to obtain a light field spatial geometric model; calculating the adaptation deviation between the optical parameters and the temperature constraint parameters based on the light field spatial geometric model, and correcting the real-time measurement parameters based on the adaptation deviation to obtain the final measurement parameters; and adjusting the positive-focus detection component based on the final measurement parameters.

[0099] In practical implementation, before the positive focus detection component operates, the detection field of view of the target power equipment is anchored, and real-time measurement parameters of aperture, field of view, and focal length are collected to generate an optical parameter set. Prior constraints on the surface temperature distribution of the target power equipment are determined, generating a temperature constraint parameter set. Based on the optical parameter set, an initial geometric model of the light field propagation path is constructed, locking the light field coverage boundary of the detection field of view to obtain the first model. The temperature constraint parameter set is mapped to a light field refractive index correction matrix, and differentiated light field attenuation coefficient weights are configured based on different temperature gradient intervals to obtain the second model. Using the optical center of a typical hot spot of the target power equipment as a reference, the spatial coordinate centers of the first and second models are aligned and fused to obtain a light field spatial geometric model. The adaptation deviation between the optical parameter set and the temperature constraint parameter set in the light field spatial geometric model is calculated, and the optical parameter set is corrected according to the adaptation deviation to obtain the final measurement parameters. Based on the final measurement parameters, the positive focus detection component is adjusted.

[0100] The final measurement parameters include the focal length compensation range, the field of view convergence threshold, and the optical field adaptation parameters of the data aperture adaptation value. The detection field of view refers to the spatial area that the optical system can effectively detect, i.e., the area occupied by the target electrical equipment in optical imaging, such as a circular area with a diameter of 5m. The optical parameter set refers to the set of parameters composed of real-time measured parameters of data aperture, field of view, and focal length. For example, the optical parameter set for a certain detection scenario is: {data aperture is 20mm, horizontal field of view is 5°, vertical field of view is 3.75°, focal length is 100mm}. The prior constraints on surface temperature distribution are retrieved from a pre-stored mapping table of device type and temperature constraints. This mapping table contains fields such as device type, temperature range, and hotspot temperature threshold. The temperature constraint parameter set refers to the set of parameters composed of key parameters in the prior constraints on surface temperature distribution.

[0101] In this embodiment, the data aperture refers to the effective aperture that allows light to pass through in the optical system, which determines the amount of light entering and the resolution; the field of view refers to the angle of the spatial range that the optical system can observe, which is divided into the horizontal field of view and the vertical field of view; the focal length parameter refers to the distance from the focal point of the optical lens to the center of the lens, which determines the magnification and the range of sharpness of the image.

[0102] The focal length compensation range refers to the range of focal length adjustment determined to ensure that the actual focal length of the positive focus detection component matches the light field adaptation requirements. Specifically, the upper and lower limits of the compensation range are determined by linear calculation based on the magnitude of the adaptation deviation. The width of the range is determined according to the equipment type and detection scenario. For example, if the focal length required for light field adaptation is 100mm, the focal length compensation range is [98, 102]mm, meaning the actual focal length is adjusted within the range of 98-102mm. The field of view convergence threshold refers to the maximum allowable deviation value during the field of view adjustment process. When the deviation between the actual value of the field of view and the adaptation requirement value is less than this threshold, it is considered that the adjustment is in place.

[0103] In one embodiment, the initial geometric model of the light field propagation path refers to a geometric model that describes the propagation path of the light field from the light path inlet to the surface of the target device, considering only optical parameters, namely data aperture, field of view, and focal length, and ignoring the effects of temperature. The light field coverage boundary refers to the boundary of the maximum spatial range that the light field can reach within the detection field of view, and is determined by the optical parameters.

[0104] The optical field refractive index correction matrix reflects the influence of different temperature regions on the optical field refractive index. The elements of the matrix are the refractive index correction coefficients for the corresponding regions. Specifically, the matrix elements are calculated through linear interpolation based on the relationship between temperature gradient and refractive index. The matrix dimension is determined according to the division of the detection field of view; for example, a 3×3 matrix corresponds to 9 regions. In this case, the 3×3 optical field refractive index correction matrix... .

[0105] The differentiated light field attenuation coefficient weights are weighted values ​​configured for different temperature gradient ranges, reflecting the degree of influence of temperature on light field attenuation within that range. Typical hotspots refer to areas of abnormal surface temperature, concentrated discharge signals, or obvious morphological characteristics on the tested power equipment; these are the reference areas for aligning the light field model. For example, in an area where the insulator surface temperature is 38°C, temperatures exceeding the normal upper limit of 35°C and the discharge signal intensity on the conductor surface are... These regions have obvious anomalous features and are suitable as alignment references.

[0106] The optical center refers to the geometric center of a typical hotspot area, that is, the center coordinate point of the boundary contour of the hotspot area. In other words, the geometric center is calculated by extracting the boundary contour of the hotspot area and using the minimum bounding rectangle method.

[0107] Spatial coordinate center alignment refers to adjusting the origin of the coordinate system of the first and second models to the optical center of the typical hotspot, so that the spatial coordinate systems of the two models are consistent. For example, if the original origin of the first model is (0,0,0), the original origin of the second model is (1,1,1), and the coordinates of the optical center of the typical hotspot are (10,5,20), then the origin of the coordinate system of both models will be translated to (10,5,20) to achieve coordinate system unification.

[0108] In another embodiment, the temperature anomaly features of infrared thermal imaging data, the morphological contour features of visible light imaging data, and the redshift anomaly features of discharge spectrum data are correlated and matched to identify typical hotspot areas of the target power equipment. The optical center coordinates of the typical hotspot areas are extracted as alignment reference points and denoted as the reference origin. Load the light field propagation stability adaptation factor into the first model Simultaneously, a temperature-optical field coupling adaptation factor is loaded into the second model. Using the origin as the core, the spatial coordinate system of the models after loading the adaptation factor is transformed to the reference coordinate system. Gradient fitting is performed on the overlapping region of the light field coverage boundary of the two models to obtain the initial fused model. The deviation value of the light field parameters of the initial fused model is then calculated. ,like Then the initial fusion model is regarded as a light field space geometric model, where, Set the parameter deviation threshold; otherwise, adjust. , until Thus, the spatial geometric model of the light field is obtained.

[0109] Among them, the optical field parameter deviation value The calculation formula is shown in formula (1).

[0110] (1)

[0111] in, The mean refractive index of the medium through which the light field propagates; The optical path length for a typical hotspot area; To detect the wavelength of light; The light field diffusion angle; For the adaptation coefficients of the light field parameters; Real-time light intensity values ​​for typical hotspots; The average energy density of the light field; Micro-angle of the light field focusing surface Deviation; The rated light intensity value for a typical hotspot; This is the rated value of the light field energy density.

[0112] In this embodiment, temperature anomaly features refer to characteristics in infrared thermal imaging data where the temperature value exceeds the prior constraint range of surface temperature distribution, such as hot spot temperature and abnormal temperature gradient. For example, if the temperature of a certain area on the surface of an insulator is 42°C, exceeding the prior constraint hot spot temperature threshold of 40°C, the temperature anomaly feature of this area is described as a hot spot temperature exceeding the standard by 2°C and a temperature gradient of 4°C / cm. That is, the temperature anomaly feature is extracted by the threshold comparison method.

[0113] Morphological contour features refer to characteristics in visible light imaging data that reflect the structure or defect morphology of equipment, such as the edges, corners, and shapes of defects. Redshift anomaly features refer to characteristics in discharge spectral data where the redshift wavelength shift exceeds the redshift range of normally operating equipment. Correlation matching refers to spatially corresponding temperature anomaly regions, morphological contour feature regions, and redshift anomaly regions to find overlapping areas that simultaneously satisfy all three types of anomaly features.

[0114] In the above embodiments, by collecting key parameters, constructing an optical field model, and optimizing the model parameters, the optical field parameters are accurately matched with the temperature distribution of the device under test, which significantly improves the stability and accuracy of spectral data acquisition in complex environments such as low light, and provides reliable optical field parameter support for subsequent quantitative analysis of defects.

[0115] In an exemplary embodiment, parsing high-resolution spectral data and establishing a quantitative model between energy level difference and redshift wavelength shift includes: splitting the high-resolution spectral data into bands to obtain split spectral data; extracting characteristic peak positions from the split spectral data and calculating the energy level difference based on the characteristic peak positions; obtaining the real-time temperature and real-time pressure of the target region; and establishing a quantitative model between energy level difference and redshift wavelength shift based on the energy level difference, real-time temperature, and real-time pressure.

[0116] In practice, high-resolution spectral data is divided into multiple sub-bands based on the energy level transition spectral ranges of different molecules and particles. Each sub-band corresponds to the spectral signal of a specific molecule or particle. The vibrational-rotational characteristic peaks of the molecule are extracted, and the vibrational-rotational energy level difference of the corresponding transition process is calculated. Simultaneously, the characteristic peak positions of the electronic energy level transitions of the particles are extracted, and the electronic energy level differences corresponding to the transition processes are calculated. Based on the real-time temperature T and real-time pressure P of the discharge region as environmental correction factors, a quantitative model of quantum transition energy level difference-redshift wavelength shift is constructed for different types.

[0117] in, ,in, The wavelengths of the characteristic peaks of the corresponding molecules are denoted as λ. ,in, The wavelength of the characteristic peak position of the corresponding particle.

[0118] Among them, the quantitative model corresponding to the molecule As shown in formula (2), the quantitative model corresponding to the particle As shown in formula (3).

[0119] (2)

[0120] (3)

[0121] in, Where is Planck's constant, c is the speed of light, T0 is the temperature under standard conditions, and P0 is the pressure under standard conditions. Calibration coefficients for the redshift characteristics of molecules; The calibration coefficients for the redshift characteristics of the particles.

[0122] In the above embodiments, by accurately extracting the energy level transition characteristics of molecules and particles and calculating the energy level difference, an environmental correction factor is introduced to construct a classification and quantitative model, thereby realizing the quantitative calculation of redshift. This provides a scientific quantitative basis for the accurate classification and severity assessment of discharge defects, significantly improving the accuracy and reliability of defect analysis.

[0123] In an exemplary embodiment, the correlation mapping relationship is corrected based on the energy flux density attenuation coefficient to obtain the final correlation mapping relationship, including: obtaining the spatial distance between the UAV and the target area and the ambient atmospheric humidity; calculating the energy flux density attenuation coefficient based on the spatial distance and ambient atmospheric humidity; if the energy flux density attenuation coefficient is greater than a preset reference coefficient, adjusting the severity level of the discharge defect in the correlation mapping relationship upwards to obtain the final correlation mapping relationship; if the energy flux density attenuation coefficient is less than the preset reference coefficient, adjusting the severity level of the discharge defect in the correlation mapping relationship downwards to obtain the final correlation mapping relationship.

[0124] The formula for calculating the energy flux density attenuation coefficient is shown in formula (4).

[0125] (4)

[0126] in, This is the distance attenuation coefficient. The humidity attenuation coefficient is... This is the normalized value of the spatial distance; Here, is the normalized value of atmospheric humidity, where, The value ranges from 0.02 to 0.05, and is fitted according to the inspection distance (0.02 for distance ≤ 50m, 0.03 for distance < 100m, and 0.05 for distance > 100m). The value ranges from 0.01 to 0.03, and is fitted according to humidity (0.01 for RH≤50%, 0.02 for 50%<RH≤70%, and 0.03 for RH>70%).

[0127] In practical implementation, the energy flux density attenuation coefficient under the current discharge scenario is determined by combining the spatial distance between the target area and the UAV inspection terminal, as well as the ambient atmospheric humidity. The corresponding correlation logic between the energy flux density attenuation coefficient and the redshift and discharge defect severity is then established. Based on the energy flux density attenuation coefficient, the corresponding correlation logic is adjusted to obtain the correlation mapping relationship. Specifically, if the energy flux density attenuation coefficient is higher than a preset reference coefficient, the defect type severity level is increased by R1 level, and the correlation range of the corresponding defect type's redshift spectral characteristics is expanded. for ,in, For the adjustment value of H1, This is the adjustment value for H2.

[0128] If the flux density attenuation coefficient is lower than the preset reference coefficient, the severity level of the defect type matched in the corresponding redshift range will be downgraded by R2 levels. At the same time, the correlation range of the redshift spectral characteristics of the corresponding defect type will be narrowed. for ,in, 3 is the adjustment value for H1. This is the adjustment value for H2.

[0129] In one embodiment, , , , , , The calculation formulas are shown in formulas (5)-(10).

[0130] (5)

[0131] (6)

[0132] (7)

[0133] (8)

[0134] (9)

[0135] (10)

[0136] in, To adjust the level value, the value should be a positive integer; The severity level of the defect type; The energy flux density attenuation coefficient; These are preset coefficients; The ambient atmospheric humidity; Standard humidity; The rounding up symbol; The floor symbol is used for rounding down. The value ranges from 0.3 to 0.7, depending on the type of equipment (insulators). =0.5, wire =0.6), R0 is the severity level of the defect type matched for the corresponding redshift range. When it is mild, R0 takes the value of 1, when it is moderate, it takes the value of 2, and when it is severe, it takes the value of 3.

[0137] In the above embodiments, preliminary defect classification is achieved by setting differentiated redshift ranges for different devices, energy flux density attenuation coefficient is determined by combining spatial distance and humidity, and then the severity level of the defect and the redshift correlation range are dynamically adjusted based on the attenuation coefficient. This makes the classification and severity assessment of discharge defects more in line with actual inspection scenarios, significantly improving the accuracy of quantitative analysis and environmental adaptability, and providing accurate defect feature basis for subsequent multimodal data fusion diagnosis.

[0138] In an exemplary embodiment, based on the final correlation mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectral data, and visible light imaging data, and fused to obtain a fault image of the target power equipment. This includes: determining the dominant features of the discharge defect of the target power equipment based on the final correlation mapping relationship; determining the current inspection scenario; assigning preset weights to the infrared thermal imaging data, discharge spectral data, and visible light imaging data based on the dominant features and the current inspection scenario, and performing weighted fusion to obtain a fault image of the target power equipment.

[0139] In practice, based on the final correlation mapping relationship, the dominant features of the discharge defect pair of the target power equipment are determined, the current inspection scenario is determined, and based on the dominant features and the current inspection scenario, preset weights are assigned to infrared thermal imaging data, discharge spectral data and visible light imaging data, and weighted fusion is performed to obtain the fault image of the target power equipment.

[0140] In one embodiment, the defect feature fields (redshift, temperature, morphology) in the final association mapping relationship are traversed, and the correlation degree between each feature and the discharge defect type is calculated as: feature conformity / sum of all feature conformities. The feature with the highest correlation degree is the dominant feature.

[0141] The current inspection scenario refers to the environmental conditions and equipment status during drone inspections, mainly including light intensity (strong light / normal / low light), weather (sunny / cloudy / rainy), and equipment type (insulator / conductor / transformer). For example, in a low light (light intensity 5 lux) + rainy (humidity 90%RH) + transformer scenario, visible light data has high noise, resulting in blurred images; infrared data is affected by moisture, such as temperature measurement deviation ±3℃; and discharge spectrum data is attenuated by humidity, such as redshift measurement deviation ±0.03nm.

[0142] In the above embodiments, by accurately anchoring the core features of defect diagnosis and using a multimodal attention mechanism to combine the scene and dominant features to allocate data weights, the data weight allocation is more in line with the nature of the defect and the inspection environment, which significantly improves the pertinence and effectiveness of multimodal data fusion and lays a high-quality data fusion foundation for the output of subsequent comprehensive diagnostic results.

[0143] In an exemplary embodiment, denoising a fault image to obtain a processed fault image includes: acquiring real-time data of the illumination interference dimension in the current inspection scenario; generating an illumination intensity interference vector based on the real-time data; comparing the illumination intensity interference vector with a standard illumination intensity interference vector to determine an illumination difference vector; converting the illumination difference vector into a three-dimensional fuzzy form and determining the correlation coefficients of the fault image in different dimensions; and denoising the fault image based on the correlation coefficients to obtain the processed fault image.

[0144] In practice, the ambient light intensity interference factor of a predefined interference dimension is collected to obtain the light intensity interference vector. and the standard light intensity interference vector Comparative analysis was performed to obtain the illumination difference vector. ,in, is the ambient light intensity interference factor for the i1th predefined interference dimension; n is the total number of predefined interference dimensions; The standard illumination intensity interference factor for the i1th predefined interference dimension; For the i1th predefined interference dimension, the difference interference factor is determined based on the correlation characteristics of the influence of each illumination interference dimension on the imaging channel. The dimensions in the data are clustered into at least two classes, and the clustering identifier is denoted as . Where m is the number of clusters, and the dimensions within each cluster are divided into the first influence domain, the second influence domain, and the third influence domain by combining the standard illumination interference historical characteristics of the corresponding cluster. At the same time, the correlation coefficient of each dimension within its respective cluster is calculated.

[0145] Based on the correlation coefficient, calculate the correlation coefficient within each cluster. The contribution weights in different influence domains, and the corresponding Convert to 3D fuzzy form ,in, , , These correspond to the influence levels of the first, second, and third influence domains, respectively. , , Contribution weights are assigned based on the first, second, and third influence domains, respectively. For the corresponding The correlation coefficient.

[0146] Based on three-dimensional fuzzy form And combined with the corresponding The historical influence standard deviations of infrared thermal imaging data, discharge spectral data, and visible light imaging data, respectively, and the weights assigned to these historical influence standard deviations, are used to determine the corresponding... Prioritize, and apply priority to all The data is sorted, and the imaging noise reduction intensity of each cluster and each dimension is configured based on the sorting results. Data noise reduction preprocessing is also performed.

[0147] In one embodiment, the predefined interference dimensions n1=5, namely light intensity (i1=1), light angle (i1=2), light uniformity (i1=3), spectral distribution (i1=4), and glare level (i1=5), are used to collect the interference factor G1={ for a certain inspection scene. =0.8 (Actual illumination 800 lux / Standard 1000 lux) =0.9 (actual angle 30° / standard 0°) =0.7 (actual uniformity 70% / standard 100%) =0.95 (actual spectrum and standard deviation of 5%) =0.6 (actual glare level 6 / standard 10)}; Standard vector G0={1.0,1.0,1.0,1.0,1.0}, its interference factor under standard operating conditions is 1; Calculate the difference vector The negative sign indicates that the actual interference is lower than the standard. Of the five dimensions, illumination intensity (i1=1), illumination uniformity (i1=3), and glare intensity (i1=5) primarily affect the visible light imaging channel. Their correlation characteristic is that they affect morphological recognition, clustering into one class C1 with m=2. Another class C2 consists of illumination angle (i1=2) and spectral distribution (i1=4), primarily affecting the infrared and spectral channels. For class C1, considering historical characteristics, such as illumination intensity deviation > ±0.3 having a significant impact on imaging, the influence domain is divided into: the first influence domain L1 (… Absolute value > 0.3, such as =-0.3、 =-0.4), second influence domain L2 ( Absolute value 0.1-0.3, such as =-0.2), third influence domain L3 ( (Absolute value < 0.1, no corresponding dimension); calculate the correlation coefficient of each dimension within class C1. It is the Pearson correlation coefficient based on dimensionality and image quality: =0.8 (Light intensity is highly correlated with visible light signal-to-noise ratio) =0.9 (uniformity has a more significant impact) =0.7 (glare effect is secondary). In specific implementation, clustering is achieved using the K-means algorithm (k=2-3), where the boundary values ​​of the influence domain are determined based on the analysis of historical interference data (≥500 groups).

[0148] The formula for calculating the contribution weight is shown in formula (11).

[0149] (11)

[0150] The total impact coefficient is generally set to 1, and ,and This is the sum of the influence coefficients corresponding to the L1, L2, and L3 influence domains.

[0151] In the above embodiments, by quantifying illumination interference deviation, clustering and dividing the influence domain based on interference dimension, constructing three-dimensional fuzzy interference features, determining interference processing priority, and configuring differentiated noise reduction intensity, the noise reduction of imaging data is made more targeted, significantly improving the quality and purity of multimodal data, and providing high-quality data assurance for the accuracy of the final comprehensive diagnostic results.

[0152] To illustrate the drone inspection method in this application in detail, an embodiment is described below. For example, this application describes a drone inspection method in a specific scenario.

[0153] First, the incident light from the target power equipment is collected by the sensors on the UAV. The incident light from the target power equipment is then split into infrared, ultraviolet and visible light bands to determine the infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment.

[0154] The defocus detection component uses wide field of view and visible light imaging data to extract the contour feature signal of the target area, transmits it to the focus detection component to lock onto the target power equipment, and acquires high-resolution spectral data of the target power equipment.

[0155] By analyzing high-resolution spectral data and based on the spectral characteristics of molecules and particles during the discharge process of the target power equipment, a quantitative model of the quantum transition energy level difference versus redshift wavelength shift is established. This model establishes a quantitative relationship between the energy level difference and the redshift wavelength shift. Based on historical discharge defect data and the redshift wavelength shift intervals of the target power equipment, the correlation mapping between the discharge defect types and the redshift spectrum is determined. The correlation mapping is then corrected using the energy flux density attenuation coefficient to obtain the final correlation mapping.

[0156] Based on the final correlation mapping relationship, the importance of the data and its correlation with defect features are determined. Preset weights are assigned to infrared thermal imaging data, discharge spectral data, and visible light imaging data, respectively. The weighted imaging data, discharge spectral data, and visible light imaging data are then fused to obtain a fault image of the target power equipment. An ambient light intensity interference factor is introduced to denoise the fault image, resulting in a processed fault image. The discharge defect detection results and the processed fault image of the target power equipment are then output.

[0157] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0158] Based on the same inventive concept, this application also provides a drone inspection device for implementing the drone inspection method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more drone inspection device embodiments provided below can be found in the limitations of the drone inspection method described above, and will not be repeated here.

[0159] In one exemplary embodiment, such as Figure 3 As shown, a drone inspection device is provided, including: an acquisition module 301, a data collection module 302, a determination module 303, a correction module 304, a fusion module 305, and an output module 306, wherein:

[0160] The acquisition module is used to acquire infrared thermal imaging data, discharge spectrum data, and visible light imaging data of the target power equipment based on the equipment mounted on the UAV.

[0161] The acquisition module is used to determine the outline of the target area based on the visible light imaging data, lock the target power equipment based on the outline and the defocus detection component, and acquire high-resolution spectral data of the target power equipment based on the focus detection component.

[0162] The determination module is used to analyze the high-resolution spectral data, establish a quantitative model between energy level difference and redshift wavelength shift, and determine the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum based on the quantitative model.

[0163] The correction module is used to correct the correlation mapping relationship based on the energy flux density attenuation coefficient to obtain the final correlation mapping relationship.

[0164] The fusion module is used to assign preset weights to infrared thermal imaging data, discharge spectrum data and visible light imaging data based on the final association mapping relationship, and fuse them to obtain the fault image of the target power equipment.

[0165] The output module is used to perform noise reduction processing on the fault image to obtain the processed fault image, and output the discharge defect detection result and the processed fault image of the target power equipment.

[0166] In one exemplary embodiment, the above-described apparatus further includes an adjustment module for:

[0167] The real-time measurement parameters of the positive focus detection component are determined, and a first model of the light field propagation path is constructed based on the real-time measurement parameters;

[0168] The prior constraints on the distribution of surface temperature of the target power equipment are determined, the prior constraints are mapped to the optical field refractive index correction matrix, and the weights of the optical field attenuation coefficients are configured based on different temperature gradient intervals to construct a second model.

[0169] The first and second models are aligned and fused to obtain the light field spatial geometric model.

[0170] Based on the optical field spatial geometric model, the adaptation deviation between optical parameters and temperature constraint parameters is calculated, and the real-time measurement parameters are corrected based on the adaptation deviation to obtain the final measurement parameters.

[0171] Based on the final measurement parameters, the focus detection component is adjusted.

[0172] In one exemplary embodiment, the above-described apparatus further includes a building module for:

[0173] High-resolution spectral data is band-splitting to obtain the split spectral data;

[0174] Extract the characteristic peak positions from the split spectral data, and calculate the energy level difference based on the characteristic peak positions;

[0175] Obtain the real-time temperature and pressure of the target area;

[0176] A quantitative model is established based on the energy level difference, real-time temperature, and real-time pressure to determine the relationship between the energy level difference and the redshift wavelength shift.

[0177] In one exemplary embodiment, the above-described correction module is further configured to:

[0178] Obtain the spatial distance between the drone and the target area, as well as the ambient atmospheric humidity;

[0179] Calculate the energy flux density attenuation coefficient based on spatial distance and ambient atmospheric humidity;

[0180] When the energy flux density attenuation coefficient is greater than the preset reference coefficient, the severity level of the discharge defect in the correlation mapping relationship is adjusted upward to obtain the final correlation mapping relationship.

[0181] When the energy flux density attenuation coefficient is less than the preset reference coefficient, the severity level of the discharge defect in the correlation mapping relationship is downgraded to obtain the final correlation mapping relationship.

[0182] In one exemplary embodiment, the fusion module is further configured to:

[0183] Based on the final correlation mapping relationship, the dominant characteristics of the discharge defects of the target power equipment are determined;

[0184] The current inspection scenario is determined. Based on the dominant features and the current inspection scenario, preset weights are assigned to infrared thermal imaging data, discharge spectral data and visible light imaging data, and weighted fusion is performed to obtain the fault image of the target power equipment.

[0185] In one exemplary embodiment, the above-described apparatus further includes a processing module for:

[0186] Collect real-time data on the light interference dimension in the current inspection scenario, and generate a light intensity interference vector based on the real-time data;

[0187] The light intensity interference vector is compared with the standard light intensity interference vector to determine the light difference vector;

[0188] The illumination difference vector is converted into a three-dimensional blurred form, and the correlation coefficients of the fault image in different dimensions are determined respectively;

[0189] Based on the correlation coefficient, noise reduction processing is performed on the fault image to obtain the processed fault image.

[0190] Each module in the aforementioned UAV inspection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0191] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. 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 the environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a drone inspection method.

[0192] The display unit of this computer device is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of this computer device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad set on the casing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0193] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0194] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0195] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0196] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0197] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0198] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0199] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0200] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for unmanned aerial vehicle (UAV) inspection, characterized in that, The method includes: The equipment carried by the drone acquires infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment; Based on the visible light imaging data, the outline of the target area is determined, the target power equipment is locked based on the outline and the defocus detection component, and high-resolution spectral data of the target power equipment is acquired based on the focus detection component. The high-resolution spectral data is analyzed to establish a quantitative model between energy level difference and redshift wavelength shift, and based on the quantitative model, the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum is determined. Based on the energy flux density attenuation coefficient, the correlation mapping relationship is corrected to obtain the final correlation mapping relationship; Based on the final association mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectrum data and visible light imaging data, and fused to obtain the fault image of the target power equipment; The fault image is denoised to obtain a processed fault image, and the discharge defect detection result and the processed fault image of the target power equipment are output.

2. The method according to claim 1, characterized in that, Before acquiring high-resolution spectral data of the target power equipment based on the positive focus detection component, the following steps are also included: The real-time measurement parameters of the positive focus detection component are determined, and a first model of the light field propagation path is constructed based on the real-time measurement parameters; A priori constraint on the distribution of surface temperature of the target power equipment is determined, the priori constraint is mapped to an optical refractive index correction matrix, and a second model is constructed based on the differentiated optical attenuation coefficient weights configured for different temperature gradient intervals. Align and fuse the first model and the second model to obtain the light field spatial geometric model; Based on the optical field spatial geometric model, the adaptation deviation between the optical parameters and the temperature constraint parameters is calculated, and the real-time measurement parameters are corrected based on the adaptation deviation to obtain the final measurement parameters. Based on the final measurement parameters, the focus detection component is adjusted.

3. The method according to claim 1, characterized in that, The process of analyzing the high-resolution spectral data and establishing a quantitative model between energy level differences and redshift wavelength shift includes: The high-resolution spectral data is divided into bands to obtain the divided spectral data; Feature peak positions are extracted from the split spectral data, and energy level differences are calculated based on the feature peak positions; Obtain the real-time temperature and pressure of the target area; Based on the energy level difference, the real-time temperature, and the real-time pressure, a quantitative model is established between the energy level difference and the redshift wavelength shift.

4. The method according to claim 1, characterized in that, The step of correcting the correlation mapping relationship based on the energy flux density attenuation coefficient to obtain the final correlation mapping relationship includes: Obtain the spatial distance between the drone and the target area, as well as the ambient atmospheric humidity; Calculate the energy flux density attenuation coefficient based on the spatial distance and the ambient atmospheric humidity; When the energy flux density attenuation coefficient is greater than the preset reference coefficient, the severity level of the discharge defect in the correlation mapping relationship is adjusted upward to obtain the final correlation mapping relationship; If the energy flux density attenuation coefficient is less than the preset reference coefficient, the severity level of the discharge defect in the correlation mapping relationship is downgraded to obtain the final correlation mapping relationship.

5. The method according to claim 1, characterized in that, Based on the final correlation mapping relationship, preset weights are assigned to infrared thermal imaging data, discharge spectral data, and visible light imaging data, and fused to obtain a fault image of the target power equipment, including: Based on the final association mapping relationship, the dominant characteristics of the discharge defects of the target power equipment are determined; The current inspection scenario is determined. Based on the dominant features and the current inspection scenario, preset weights are assigned to infrared thermal imaging data, discharge spectral data and visible light imaging data, and weighted fusion is performed to obtain the fault image of the target power equipment.

6. The method according to claim 5, characterized in that, The noise reduction process for the fault image to obtain the processed fault image includes: Collect real-time data on the illumination interference dimension in the current inspection scenario, and generate an illumination intensity interference vector based on the real-time data; The light intensity interference vector is compared with the standard light intensity interference vector to determine the light difference vector; The illumination difference vector is converted into a three-dimensional blurred form, and the correlation coefficients of the fault image in different dimensions are determined respectively; Based on the correlation coefficient, the fault image is subjected to noise reduction processing to obtain the processed fault image.

7. A drone inspection device, characterized in that, The device includes: The acquisition module is used to acquire infrared thermal imaging data, discharge spectrum data and visible light imaging data of the target power equipment based on the equipment carried by the UAV; The acquisition module is used to determine the outline of the target area based on the visible light imaging data, lock the target power equipment based on the outline and the defocus detection component, and acquire high-resolution spectral data of the target power equipment based on the focus detection component. The determination module is used to analyze the high-resolution spectral data, establish a quantitative model between energy level difference and redshift wavelength shift, and determine the correlation mapping relationship between the discharge defect type of the target power equipment and the redshift spectrum based on the quantitative model. The correction module is used to correct the correlation mapping relationship based on the energy flux density attenuation coefficient to obtain the final correlation mapping relationship; The fusion module is used to assign preset weights to infrared thermal imaging data, discharge spectrum data and visible light imaging data based on the final correlation mapping relationship, and fuse them to obtain a fault image of the target power equipment. The output module is used to perform noise reduction processing on the fault image to obtain the processed fault image, and output the discharge defect detection result and the processed fault image of the target power equipment.

8. 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 method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images