Full-autonomous unmanned aerial vehicle hyperspectral inspection method for anomaly detection of photovoltaic power station

Abstract

The invention relates to an unmanned aerial vehicle hyperspectral inspection method, in particular to a photovoltaic power station anomaly detection-oriented full-autonomous unmanned aerial vehicle hyperspectral inspection method. The method comprises the following steps: 1, performing task scheduling on the unmanned aerial vehicle through an unmanned aerial vehicle field, and planning a route and an inspection period of the unmanned aerial vehicle; 2, the unmanned aerial vehicle flies to a to-be-detected target area according to a planned route and an inspection cycle, and spectral data of the target area are collected; 3, carrying out radiation correction and normalization processing on the collected spectral data; 4, detecting accumulated dust, pollutants and vegetation on the surface of the photovoltaic panel by utilizing a spectral feature analysis and machine learning method based on the spectral data subjected to radiation correction and normalization processing; and 5, based on a detection result, evaluating the influence degree of accumulated dust, pollutants and vegetation on the photoelectric conversion efficiency of the photovoltaic panel. According to the invention, a scientific basis can be provided for efficient operation of the photovoltaic power station and health management of the ecological environment.

Description

Technical Field

[0001] The present invention relates to a hyperspectral inspection method for unmanned aerial vehicles (UAVs), and in particular to a fully autonomous hyperspectral inspection method for unmanned aerial vehicles (UAVs) for photovoltaic power station anomaly detection. Background Art

[0002] With the large-scale application of photovoltaic power stations in special environments such as deserts, power station components are easily affected by environmental factors such as internal vegetation growth, dust accumulation, bird activities, etc. These factors will affect the power generation efficiency of photovoltaic panels and the ecological and environmental stability of power stations. The accumulation of dust and pollutants on photovoltaic panels will not only reduce the photoelectric conversion efficiency, but also accelerate the aging of components and increase maintenance costs. Vegetation growth may cause shading, further reducing the output power of photovoltaic panels, and also accompanied by fire hazards.

[0003] Traditional manual inspection methods are inefficient, labor-intensive, and difficult to achieve large-scale, refined monitoring. Inspections are especially challenging in complex environments such as deserts and mountainous areas.

[0004] Therefore, there is an urgent need for a fully autonomous, efficient and accurate inspection system to ensure the long-term efficient operation of photovoltaic power stations and the healthy management of the ecological environment. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problems that the manual inspection method used in the prior art is inefficient, labor-intensive, and difficult to achieve large-scale and refined monitoring, especially in complex environments such as deserts and mountainous areas, and to provide a fully autonomous UAV hyperspectral inspection method for photovoltaic power station anomaly detection.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is:

[0007] A fully autonomous drone hyperspectral inspection method for photovoltaic power station anomaly detection includes the following steps:

[0008] Step 1: Use the drone airport to dispatch drone tasks and plan the drone’s routes and inspection cycles;

[0009] Step 2: The drone flies to the target area where the photovoltaic panels in the photovoltaic power station to be inspected are located according to the planned route and inspection cycle, and uses the hyperspectral imaging sensor carried by the drone to collect spectral data of the photovoltaic panels and their surrounding environment in the target area;

[0010] Step 3: Perform radiation correction and normalization on the collected spectral data to ensure the accuracy and consistency of the data;

[0011] Step 4: Based on the spectral data after radiation correction and normalization, the dust accumulation, pollutants and vegetation on the surface of the photovoltaic panel are detected by using spectral feature analysis and machine learning methods to obtain the detection results;

[0012] Step 5: Based on the detection results, evaluate the impact of dust accumulation, pollutants and vegetation on the photovoltaic panel photoelectric conversion efficiency, and issue a maintenance alarm when the impact of dust accumulation and / or pollutants and / or vegetation exceeds a preset threshold, thereby completing the abnormal detection of the photovoltaic power station.

[0013] Furthermore, step 3 is specifically as follows:

[0014] Step 3.1: Use the following polynomial regression model to perform radiometric correction on the collected spectral data to compensate for spectral changes under different lighting conditions and ensure the accuracy and consistency of the data:

[0015] R cor =R raw -P(I)

[0016] Where: R cor is the corrected spectral reflectance, R raw is the original collected spectral reflectance, P(I) is the polynomial regression model of the light intensity I;

[0017] Step 3.2: Perform minimum-maximum normalization on the spectral data after radiation correction to eliminate the deviation under different acquisition conditions. The mathematical expression of normalization is:

[0018]

[0019] Where: R norm represents the normalized spectral reflectance, R min and R max are the minimum and maximum reflectance values ​​in the spectral data, respectively.

[0020] Furthermore, step 4 is specifically as follows:

[0021] Step 4.1: Based on the spectral data after radiation correction and normalization, the spectral characteristics of vegetation and photovoltaic panels are analyzed to extract the spectral characteristic differences. Based on the spectral characteristic differences, a mapping index is used to enhance the spectral contrast between vegetation and photovoltaic panels. The formula of the mapping index is as follows:

[0022]

[0023] Where: ESVI is the mapping index, R nir is the reflectivity in the near infrared band, R red is the red band reflectivity, R swiris the short-wave infrared reflectivity, R green is the reflectance of green band, α and β are adjustment factors;

[0024] Step 4.2: Determine the optimal segmentation threshold of the mapping index ESVI by Otsu method to achieve effective segmentation of vegetation areas:

[0025]

[0026]

[0027] Where: T * is the threshold value, is the between-class variance, w1(T) and w2(T) are the thresholds T * The weights of the background and foreground, μ1(T) and μ2(T) are the average gray values ​​of the background and foreground respectively;

[0028] Step 4.3: Use machine learning methods to detect the spectral data after vegetation segmentation, so as to detect dust, pollutants and vegetation occlusion on the surface of photovoltaic panels, obtain test results, and complete abnormal detection of photovoltaic power stations.

[0029] Furthermore, in step 4.3, the machine learning method calculates the dust accumulation and pollutant shielding area on the surface of the photovoltaic panel by the following formula:

[0030]

[0031] At the same time, the degree of vegetation occlusion is calculated by the following formula:

[0032]

[0033] Where: A is the dust accumulation and pollutant shielding area on the photovoltaic panel surface, p i is the abnormal coverage of the i-th pixel, n is the total number of pixels, and R is the actual ground area represented by each pixel in the spectral data; C veg is the percentage of vegetation coverage, A veg is the area covered by vegetation, A total The total area of ​​PV panels.

[0034] Furthermore, in step 4.3, the abnormal coverage of the i-th pixel is obtained by:

[0035] A1. By building a spectral change detection model based on time series, the current spectral characteristics are compared with historical data to identify spectral changes caused by dust accumulation and pollutants. The spectral change detection model uses a dynamic time warping method, and the formula is as follows:

[0036]

[0037] Where: x i and (i) They are all time series signals, and π(i) is the alignment path;

[0038] A2. Based on the spectral changes caused by dust accumulation and pollutants, determine whether there are pollutants. If so, compare the extracted spectral features with the existing spectral library and calculate the spectral similarity through the spectral correlation coefficient. The expression is as follows:

[0039]

[0040] Where: r is the correlation coefficient between the currently acquired spectral feature and the reference spectrum, x i and i are the values ​​of the current spectral feature and the reference spectrum at the i-th pixel, and x i and i The mean of

[0041] Identify the pollutant type according to the value of r; if not, proceed to step A3;

[0042] A3. Use the following support vector regression calculation formula to perform regression calculation to determine the coverage of dust and pollutants, that is, the abnormal coverage of the i-th pixel:

[0043]

[0044] Where: α i is the coefficient of the support vector, K(x i , x) is the kernel function, b is the bias term, and x is the input spectral feature.

[0045] Furthermore, in step 5, the photoelectric conversion efficiency loss percentage is calculated by the following formula:

[0046]

[0047] Where: η is the percentage of photoelectric conversion efficiency loss, P a is the photoelectric conversion efficiency in the current state, P i is the photoelectric conversion efficiency under ideal conditions.

[0048] The beneficial effects of the present invention are:

[0049] Through the coordinated work of fully autonomous drones and drone fields, comprehensive autonomous inspections inside and outside the photovoltaic power station are achieved, reducing labor costs. The use of hyperspectral imaging technology, combined with spectral variation analysis and machine learning algorithms, enables accurate monitoring of photovoltaic power stations, including dust accumulation on photovoltaic panels, pollutant accumulation, and vegetation coverage. The system can provide a scientific basis for the efficient operation of photovoltaic power stations and the healthy management of the ecological environment, which helps to improve the power generation efficiency of photovoltaic panels and reduce maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a schematic diagram of a fully autonomous drone hyperspectral inspection system for photovoltaic power station anomaly detection;

[0051] Figure 2 It is a flow chart of a fully autonomous drone hyperspectral inspection system for photovoltaic power station anomaly detection. DETAILED DESCRIPTION

[0052] In order to make the purpose, advantages and features of the present invention clearer, the following is a further detailed description of a fully autonomous drone hyperspectral inspection method for photovoltaic power station anomaly detection proposed by the present invention in combination with the accompanying drawings and specific embodiments. According to the following specific implementation methods, the advantages and features of the present invention will be clearer.

[0053] See also Figure 2 In this embodiment, a fully autonomous drone hyperspectral inspection method for photovoltaic power station anomaly detection specifically includes the following steps:

[0054] (1) See Figure 1 , and manage and dispatch fully autonomous drones through drone farms (tower or ground-based). The drone farm has a task planning function, which can automatically plan routes according to preset cycles or detection requirements, and assign drones to perform inspection tasks. Fully autonomous drones do not require human control. Through the automatic route planning function, they can fly autonomously to the area where the photovoltaic panels are located to carry out inspection tasks. Drones use hyperspectral imaging sensors to obtain spectral data of photovoltaic panels and the surrounding environment, which is used to detect abnormal conditions including dust accumulation on the surface of photovoltaic panels, accumulation of pollutants, and vegetation obstruction.

[0055] (2) The UAV flies to the target area where the photovoltaic panels in the photovoltaic power station to be inspected are located according to the planned route and inspection cycle, and uses the hyperspectral imaging sensor carried by the UAV to collect spectral data of the photovoltaic panels and their surrounding environment in the target area.

[0056] (3) The collected data are first radiometrically corrected to compensate for spectral variations under different lighting conditions and ensure data consistency and accuracy.

[0057] Radiation correction uses a polynomial regression model to correct the spectral deviation under different light intensities. The specific formula is as follows:

[0058] R cor =R raw -P(I)

[0059] Where: R cor is the corrected spectral reflectance, R raw is the original collected spectral reflectance, P(I) is the polynomial regression model of the light intensity I;

[0060] Next, the data is normalized using the minimum-maximum method to eliminate the deviations under different acquisition conditions and improve the accuracy of subsequent analysis. The mathematical representation of normalization is as follows:

[0061]

[0062] Where: R norm represents the normalized spectral reflectance, R min and R max are the minimum and maximum reflectance values ​​in the spectral data, respectively.

[0063] The preprocessed data can be used for subsequent anomaly detection and analysis.

[0064] (4) Based on remote sensing image data, the spectral reflectance characteristics of desert vegetation and photovoltaic panels are deeply analyzed to extract typical spectral feature differences. In view of the high reflectance of vegetation and the low reflectance of photovoltaic panels, a new mapping index is designed to enhance the spectral contrast between vegetation signals and photovoltaic panels and suppress the interference of photovoltaic panels.

[0065] The formula for the cartographic index is as follows:

[0066]

[0067] Where: ESVI is the mapping index, R nir is the reflectivity in the near infrared band, R red is the red band reflectivity, R swir is the short-wave infrared reflectivity, R green is the green band reflectivity, α and β are adjustment factors, α and β are used to control the inhibitory effect of the shortwave infrared band on the photovoltaic panel and the normalization of the index to prevent the denominator from approaching zero.

[0068] Through statistical analysis of typical vegetation and background samples, the optimal segmentation threshold of the mapping index is determined using the Otsu method, thereby achieving effective segmentation of the vegetation area.

[0069] The optimal threshold calculation formula of Otsu method is as follows:

[0070]

[0071]

[0072] Where: T * is the threshold value, is the between-class variance, w1(T) and w2(T) are the thresholds T * The weights of the background and foreground are as follows, μ1(T) and μ2(T) are the average grayscale values ​​of the background and foreground, respectively.

[0073] (5) Through machine learning methods, accurate detection of dust accumulation and pollutant identification in photovoltaic power stations can be achieved.

[0074] Dust accumulation monitoring evaluates the degree of dust accumulation by analyzing the spectral characteristics of the photovoltaic panel surface. Changes in the spectral characteristics reflect the coverage of dust on the photovoltaic panel. The impact of dust accumulation is quantified by calculating the characteristic spectral reflectance of the photovoltaic panel surface and comparing it with the baseline reflectance of a clean photovoltaic panel. By building a spectral change detection model based on time series, the current spectral characteristics are compared with historical data to identify spectral changes caused by dust accumulation and pollutants. Change detection uses the dynamic time warping (DTW) method, and the formula is as follows:

[0075]

[0076] Where: x i and (i) They are all time series signals, π(i) is the alignment path, and the difference in dust accumulation or pollutants is identified by calculating the minimum distance of spectral changes.

[0077] If there are pollutants, the extracted spectral features are compared with the existing spectral library (the spectral data in it is the reference spectrum), and the spectral similarity is calculated by the spectral correlation coefficient. The expression is as follows:

[0078]

[0079] Where: r is the correlation coefficient between the currently acquired spectral feature and the reference spectrum, x i and i are the values ​​of the current spectral feature and the reference spectrum at the i-th pixel, and x i and i The mean of

[0080] A high correlation (r close to 1) indicates a match with a certain type in the pollutant spectral library.

[0081] For the quantification of dust accumulation and pollutant levels, support vector regression (SVR) can be used for regression estimation. The regression model is expressed as:

[0082]

[0083] Where: α i is the coefficient of the support vector, K(x i , x) is the kernel function, b is the bias term, and x is the input spectral feature; the model can accurately estimate the coverage of dust and pollutants, that is, the abnormal coverage of the i-th pixel.

[0084] As for the impact of vegetation, the vegetation growth inside the photovoltaic power station is analyzed to evaluate its shading degree to the photovoltaic panels and its impact on the power generation efficiency of the power station. Combined with multi-temporal remote sensing data, the dynamic changes of vegetation are analyzed, and the vegetation coverage is calculated by the vegetation mapping index (ESVI in the above steps). The calculation formula for the shading degree of vegetation is as follows:

[0085]

[0086] Where: C veg is the percentage of vegetation coverage, A veg is the area covered by vegetation, A total The total area of ​​PV panels.

[0087] The trained model is applied to the remote sensing images of the photovoltaic power station area to accurately extract abnormal areas, such as dust accumulation, pollutants and vegetation areas, and evaluate their impact on power generation efficiency and safety. Based on the spatial resolution of the remote sensing image, the coverage area of ​​the abnormal area can be calculated by the following formula:

[0088]

[0089] Where: A is the dust accumulation, pollutants and vegetation blocking area on the surface of the photovoltaic panel, p i is the abnormal coverage of the ith pixel, n is the total number of pixels, and R is the size of the actual ground area represented by each pixel in the spectral data.

[0090] (6) The overall impact assessment of dust accumulation, pollutants and vegetation can be carried out by the change in photoelectric conversion efficiency, which is defined as follows:

[0091]

[0092] Where: η is the percentage of photoelectric conversion efficiency loss, P a is the photoelectric conversion efficiency in the current state, P i is the photoelectric conversion efficiency under ideal conditions.

Claims

1. A fully autonomous drone hyperspectral inspection method for photovoltaic power station anomaly detection, characterized in that: The following steps are involved: Step 1: Use the drone airport to dispatch drone tasks and plan the drone’s routes and inspection cycles; Step 2: The drone flies to the target area where the photovoltaic panels in the photovoltaic power station to be inspected are located according to the planned route and inspection cycle, and uses the hyperspectral imaging sensor carried by the drone to collect spectral data of the photovoltaic panels and their surrounding environment in the target area; Step 3: Perform radiation correction and normalization on the collected spectral data to ensure the accuracy and consistency of the data; Step 4: Based on the spectral data after radiation correction and normalization, the dust accumulation, pollutants and vegetation on the surface of the photovoltaic panel are detected by using spectral feature analysis and machine learning methods to obtain the detection results; Step 5: Based on the detection results, evaluate the impact of dust accumulation, pollutants and vegetation on the photovoltaic panel photoelectric conversion efficiency, and issue a maintenance alarm when the impact of dust accumulation and / or pollutants and / or vegetation exceeds a preset threshold, thereby completing the abnormal detection of the photovoltaic power station.

2. The fully autonomous UAV hyperspectral inspection method for photovoltaic power station anomaly detection according to claim 1 is characterized in that: Step 3 is as follows: Step 3.1: Use the following polynomial regression model to perform radiometric correction on the collected spectral data to compensate for spectral changes under different lighting conditions and ensure the accuracy and consistency of the data: R cor =R raw -P(I) Where: R cor is the corrected spectral reflectance, R raw is the original collected spectral reflectance, P(I) is the polynomial regression model of the light intensity I; Step 3.2: Perform minimum-maximum normalization on the spectral data after radiation correction to eliminate the deviation under different acquisition conditions. The mathematical expression of normalization is: Where: R norm represents the normalized spectral reflectance, R min and R max are the minimum and maximum reflectance values ​​in the spectral data, respectively.

3. The fully autonomous UAV hyperspectral inspection method for photovoltaic power station anomaly detection according to claim 1 or 2 is characterized in that: Step 4 is as follows: Step 4.1: Based on the spectral data after radiation correction and normalization, the spectral characteristics of vegetation and photovoltaic panels are analyzed to extract the spectral characteristic differences. Based on the spectral characteristic differences, a mapping index is used to enhance the spectral contrast between vegetation and photovoltaic panels. The formula of the mapping index is as follows: Where: ESVI is the mapping index, R nir is the reflectivity in the near infrared band, R red is the red band reflectivity, R swir is the short-wave infrared reflectivity, R green is the reflectance of green band, α and β are adjustment factors; Step 4.2: Determine the optimal segmentation threshold of the mapping index ESVI by Otsu method to achieve effective segmentation of vegetation areas: Where: T * is the threshold value, is the between-class variance, w1(T) and w2(T) are the thresholds T * The weights of the background and foreground, μ1(T) and μ2(T) are the average gray values ​​of the background and foreground respectively; Step 4.3: Use machine learning methods to detect the spectral data after vegetation segmentation, so as to detect dust, pollutants and vegetation occlusion on the surface of photovoltaic panels, obtain test results, and complete abnormal detection of photovoltaic power stations.

4. The fully autonomous UAV hyperspectral inspection method for photovoltaic power station anomaly detection according to claim 3 is characterized in that: In step 4.3, the machine learning method calculates the dust accumulation and pollutant shielding area on the surface of the photovoltaic panel by the following formula: At the same time, the degree of vegetation occlusion is calculated by the following formula: Where: A is the dust accumulation and pollutant shielding area on the photovoltaic panel surface, p i is the abnormal coverage of the i-th pixel, n is the total number of pixels, and R is the actual ground area represented by each pixel in the spectral data; C veg is the percentage of vegetation coverage, A veg is the area covered by vegetation, A total The total area of ​​PV panels.

5. The fully autonomous UAV hyperspectral inspection method for photovoltaic power station anomaly detection according to claim 4 is characterized in that: In step 4.3, the abnormal coverage of the i-th pixel p is obtained by the following method: i : A1. By building a spectral change detection model based on time series, the current spectral characteristics are compared with historical data to identify spectral changes caused by dust accumulation and pollutants. The spectral change detection model uses a dynamic time warping method, and the formula is as follows: Where: x i and (i) They are all time series signals, and π(i) is the alignment path; A2. Based on the spectral changes caused by dust accumulation and pollutants, determine whether there are pollutants. If so, compare the extracted spectral features with the existing spectral library and calculate the spectral similarity through the spectral correlation coefficient. The expression is as follows: Where: r is the correlation coefficient between the currently acquired spectral feature and the reference spectrum, x i and i are the values ​​of the current spectral feature and the reference spectrum at the i-th pixel, and x i and i The mean of Identify the pollutant type according to the value of r; if not, proceed to step A3; A3. Use the following support vector regression calculation formula to perform regression calculation to determine the coverage of dust and pollutants, that is, the abnormal coverage of the i-th pixel: Where: α i is the coefficient of the support vector, K(x i , x) is the kernel function, b is the bias term, and x is the input spectral feature.

6. The fully autonomous UAV hyperspectral inspection method for photovoltaic power station anomaly detection according to claim 1 is characterized by: In step 5, the photoelectric conversion efficiency loss percentage is calculated by the following formula: Where: η is the percentage of photoelectric conversion efficiency loss, P a is the photoelectric conversion efficiency in the current state, P i is the photoelectric conversion efficiency under ideal conditions.

Images