Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image

Abstract

The invention discloses an unmanned aerial vehicle image-based photovoltaic power station intelligent inspection method and system, and relates to the technical field of unmanned aerial vehicle intelligent inspection, and the system comprises an inspection data acquisition module, a photovoltaic module image module, an intelligent inspection processing module, an image defect determination module and an intelligent evaluation control module, the inspection data acquisition module synchronously acquires unmanned aerial vehicle flight data and photovoltaic module image data through acquisition equipment, data entry and NTP technologies, and the intelligent inspection processing module is used for acquiring the influence degree of the unmanned aerial vehicle flight data on the photovoltaic module image. According to the method, the image resolution, the image view angle and the color rendition degree of the photovoltaic module image are obtained, and an image defect judgment model is constructed through a neural network algorithm, so that the problem that the flight state and the flight path cannot be adjusted according to the influence of the flight condition on the photovoltaic module image in the intelligent inspection process of the unmanned aerial vehicle is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of drone intelligent inspection technology, and in particular to a method and system for intelligent inspection of photovoltaic power stations based on drone images. Background Art

[0002] With the rapid development of the photovoltaic power generation industry, the scale of photovoltaic power stations continues to expand and the number is increasing. The importance of their inspection work has become increasingly prominent. The traditional inspection method of photovoltaic power stations mainly relies on manual labor. Manual inspection is not only inefficient, but also greatly restricted by factors such as terrain and weather. It is difficult to fully and timely discover various problems of photovoltaic components. In addition, manual inspection has problems such as strong subjectivity and difficulty in ensuring accuracy, which easily leads to some potential fault hazards being ignored. With the continuous advancement of drone technology and image processing technology, new solutions have been provided for the inspection of photovoltaic power stations. Drones can quickly and flexibly reach every corner of the photovoltaic power station, overcome terrain obstacles, and efficiently collect image data. At the same time, combined with advanced image processing algorithms and intelligent analysis technology, the images obtained by drones can be accurately analyzed and abnormal conditions of photovoltaic components can be accurately identified. Therefore, the development of a photovoltaic power station intelligent inspection method and system based on drone images has important practical significance.

[0003] Although the existing intelligent inspection method and system for photovoltaic power stations based on drone images have made great progress, there are still some problems that need to be optimized. During the intelligent inspection of photovoltaic power stations, drones are unable to adjust the flight status and flight path of the drone according to the impact of flight conditions on the images of photovoltaic modules. Moreover, during the drone flight path planning process, although the intelligent optimization algorithms studied at home and abroad can plan the drone flight path according to the impact of drone flight conditions on collected images, they have problems such as premature maturation, slow convergence, and low precision, resulting in inaccurate images of photovoltaic modules collected, affecting the inspection results of photovoltaic power stations. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for intelligent inspection of photovoltaic power stations based on drone images to solve the problems raised in the above background technology.

[0005] To solve the above technical problems, the technical solutions adopted by the present invention are as follows: In the first aspect, an intelligent inspection system for photovoltaic power stations based on drone images includes an inspection data acquisition module, a photovoltaic module image module, an intelligent inspection processing module, an image defect determination module, and an intelligent evaluation and control module, wherein the modules are communicatively connected;

[0006] The inspection data acquisition module synchronously acquires drone flight data and photovoltaic module image data through acquisition equipment, data entry and NTP technology. The acquisition equipment includes ultrasonic height sensors, light sensors, accelerometers, visible light cameras and laser rangefinders, providing data support for the subsequent intelligent inspection process.

[0007] The photovoltaic module image module uses photovoltaic module image data to obtain the image resolution, image viewing angle, and color reproduction of the photovoltaic module image through calculation and AI image color difference comparison technology, laying the foundation for subsequent acquisition of the impact of drone flight data on photovoltaic module images;

[0008] The intelligent inspection processing module combines calculation and regression analysis to obtain the degree of influence of drone flight data on photovoltaic module images, providing technical support for drone intelligent inspection of photovoltaic power stations;

[0009] The image defect determination module uses a neural network algorithm to construct an image defect determination model;

[0010] The intelligent evaluation and control module analyzes the drone data in combination with the image defect determination model, evaluates the degree of impact of the drone data on the photovoltaic module image data, and takes corresponding measures to deal with it.

[0011] A further improvement of the technical solution of the present invention is that the inspection data acquisition module obtains the UAV flight data through the acquisition device, including the following steps:

[0012] The UAV flight data includes the UAV's flight altitude, flight angle, and light intensity of the flight environment;

[0013] An ultrasonic height sensor is installed on the bottom of the drone. The ultrasonic height sensor transmits ultrasonic pulses and receives the reflected ultrasonic pulses. The time interval from the transmission of the ultrasonic pulse to the reception of the ultrasonic pulse is measured. The flight altitude of the drone is calculated based on the formula of ultrasonic propagation speed and displacement in the air.

[0014] Establish a three-dimensional space coordinate system, which includes the x-axis, y-axis and z-axis. Use the accelerometer to measure the acceleration of the drone in the x-axis, y-axis and z-axis directions. Under the action of gravity, the posture of the drone changes, and the acceleration measured by the accelerometer changes accordingly. Formula, calculate the flight angle of the drone, where is the tilt angle relative to the direction of gravity, is the acceleration collected by the accelerometer, is the acceleration due to gravity;

[0015] Light sensors are installed around the visible light camera equipped with the drone to collect the light intensity of the drone's flight environment.

[0016] A further improvement of the technical solution of the present invention is that the inspection data acquisition module obtains photovoltaic module image data through acquisition equipment and data entry, and uses NTP technology to synchronize the UAV flight data and photovoltaic module image data, including the following process:

[0017] The photovoltaic module image data includes a real-time photovoltaic module image, a photovoltaic module length, a photovoltaic module image pixel length, and a visible light camera focal length;

[0018] The drone is equipped with a visible light camera, and the visible light camera shooting parameters are initialized and set. When the drone operates according to the flight trajectory, the visible light camera takes real-time images of the photovoltaic modules to obtain real-time images of the photovoltaic modules;

[0019] Install a laser rangefinder on the belly of the drone, start the laser rangefinder, and adjust the measurement angle using the drone's gimbal. The laser rangefinder calculates the distance based on the delay between the emitted laser beam and the reflected beam to obtain the length of the photovoltaic module.

[0020] Input the real-time photovoltaic module image into Photoshop software and measure the pixel length of the photovoltaic module image; consult the visible light camera parameter manual and obtain the visible light camera focal length through data entry;

[0021] A system of drones, NTP servers, and ground control stations is constructed. The drone and ground control station are started. The drone and ground control station send synchronization requests to NTP respectively, adding timestamps to the collected drone flight data and photovoltaic module image data. The drone transmits the timestamps of drone flight data and photovoltaic module image data to the ground control station via a wireless communication link. The drone flight data and photovoltaic module image data are matched and synchronized according to the timestamps.

[0022] A further improvement of the technical solution of the present invention is that the process of obtaining the image resolution, image viewing angle and color reproduction of the photovoltaic module image by the photovoltaic module image module includes:

[0023] A1. Calculate the PV module image resolution using the PV module length and PV module image pixel length:

[0024]

[0025] Where R is the PV module image resolution, l is the PV module image pixel length, and L is the PV module length;

[0026] A2. Based on the principles of geometric optics and the relationship between similar triangles, the process of calculating the viewing angle of the photovoltaic module image using the photovoltaic module length and the focal length of the visible light camera is as follows:

[0027]

[0028] in, is the viewing angle of the photovoltaic module image, L is the length of the photovoltaic module, is the focal length of the visible light camera;

[0029] A3. Establish a standard color library and use Gaussian filtering to denoise the photovoltaic module image. Use the AI ​​image analysis algorithm to extract the color data from the denoised photovoltaic module image. Compare the extracted color data with the standard color library. Use the AI ​​image analysis algorithm to calculate the color difference data of various colors in the photovoltaic module image. Define the color difference index of various colors in the photovoltaic module image. Calculate the color difference restoration degree by calculating the ratio of the color difference data of the same color to the corresponding color difference index.

[0030] A further improvement of the technical solution of the present invention is that the intelligent inspection processing module combines calculation and regression analysis methods to obtain the degree of influence of drone flight data on photovoltaic module images, including:

[0031] The degree of influence of the drone flight data on the photovoltaic module image includes the degree of influence of the drone flight altitude on the resolution of the photovoltaic module image, the degree of influence of the drone flight angle on the viewing angle of the photovoltaic module image, and the degree of influence of the light intensity of the drone flight environment on the color reproduction of the photovoltaic module image.

[0032] A further improvement of the technical solution of the present invention is that the process of the intelligent inspection processing module obtaining the degree of influence of the drone's flight altitude and flight angle on the photovoltaic module image resolution and photovoltaic module image viewing angle respectively includes:

[0033] B1. Select and As a data reference time point, obtain and The UAV flight altitude, UAV flight angle, PV module image resolution, and PV module image viewing angle at the moment;

[0034] B2. The process of calculating the impact of the drone's flight altitude on the PV panel image resolution is as follows:

[0035]

[0036]

[0037] in, The impact of the UAV's flight altitude on the resolution of photovoltaic module images; and for and The drone's flight altitude at the moment; and for and PV module image resolution at the moment;

[0038] B3. The process of calculating the impact of the drone's flight angle on the viewing angle of the photovoltaic module image is as follows:

[0039]

[0040]

[0041]

[0042]

[0043] in, The degree of influence of the UAV flight angle on the viewing angle of the photovoltaic module image; and for and The drone's flight angle at the moment; and for and The viewing angle of the photovoltaic module image at the moment; and They are and The difference in the UAV's flight angle within a certain time period and and The difference in viewing angle of the photovoltaic module image within a certain time period; is the proportional coefficient between the viewing angle of the PV module image and the flight angle of the UAV.

[0044] A further improvement of the technical solution of the present invention is that the process of the intelligent inspection processing module obtaining the degree of influence of the light intensity of the UAV flight environment on the color reproduction of the photovoltaic module image includes:

[0045] C1. Use regression analysis to construct a linear regression model CR=aE+b, where CR is the color reproduction, E is the light intensity of the UAV flight environment, and a and b are the linear regression model parameters;

[0046] The light intensity of the drone's flight environment and the color reproduction of the photovoltaic module images at the same time are used as the data set, which is divided into a training set and a test set in a ratio of 7:3.

[0047] Use the training set data to train the linear regression model, calculate the linear regression model parameters through the least squares method, and obtain a preliminary linear regression model;

[0048] The performance of the trained linear regression model was evaluated using test data. The linear regression model was optimized by adjusting model parameters to obtain the linear relationship between the light intensity of the UAV flight environment and the color reproduction of the photovoltaic module image.

[0049] C2. Select and As a data reference time point, obtain The light intensity of the drone's flight environment at the moment and and Color restoration of photovoltaic module images at the moment;

[0050] C3. Using the linear relationship between the light intensity of the UAV flight environment and the color reproduction of the photovoltaic module image, the process of calculating the degree of influence of the light intensity of the UAV flight environment on the color reproduction of the photovoltaic module image is as follows:

[0051]

[0052]

[0053] in, The degree of influence of the light intensity of the UAV flight environment on the color reproduction of photovoltaic module images; for The light intensity of the drone’s flight environment at the moment; and for and The color restoration degree of the photovoltaic module image at the moment, a and b are the parameters of the linear regression model.

[0054] A further improvement of the technical solution of the present invention is that the image defect determination module adopts a neural network algorithm, and the process of constructing an image defect determination model includes:

[0055] A neural network algorithm was used to construct a neural network model. The data sets included the drone's flight altitude and its impact on the resolution of photovoltaic module images, the drone's flight angle and its impact on the viewing angle of photovoltaic module images, and the drone's flight environment's light intensity and its impact on the color reproduction of photovoltaic module images. The model was divided into a training set and a validation set in a 7:3 ratio.

[0056] The architecture of the neural network model includes an input layer, a hidden layer, and an output layer. The input layer has three input nodes, corresponding to the drone's flight altitude, flight angle, and light intensity of the flight environment, respectively. Three hidden layers are set, each with 32 neurons. The output layer has three output nodes, which output the degree of influence of the drone's flight altitude on the resolution of the photovoltaic module image, the degree of influence of the drone's flight angle on the viewing angle of the photovoltaic module image, and the degree of influence of the drone's flight environment light intensity on the color reproduction of the photovoltaic module image.

[0057] Input the training set data into the neural network model, perform weighted summation on the training set data through forward propagation, obtain the prediction result of the output layer, perform backpropagation on the error between the prediction result of the output layer and the real data, update the weight and configuration of the network, and obtain the trained neural network model;

[0058] The validation set data is input into the trained neural network model to obtain the output result of the output layer in the neural network model. Based on the difference between the output result and the true result, the performance of the trained neural network model is evaluated, the model parameters are adjusted and the performance of the neural network model is optimized. The neural network model is deployed in a photovoltaic power station intelligent inspection system based on drone images to obtain an image defect judgment model.

[0059] A further improvement to the technical solution of the present invention is that the intelligent assessment control module analyzes drone data in conjunction with the image defect determination model, assesses the degree of impact of the drone data on the photovoltaic module image data, and takes corresponding measures to address the impact, including the following steps:

[0060] The drone data is input into the image defect judgment model to obtain the degree of influence of the drone data on the photovoltaic module image data. When the degree of influence of the drone data on the photovoltaic module image data is between 0% and 30%, it is classified as low impact; when the degree of influence of the drone data on the photovoltaic module image data is between 30% and 60%, it is classified as medium impact; when the degree of influence of the drone data on the photovoltaic module image data is between 60% and 100%, it is classified as high impact;

[0061] When the drone's flight altitude has a low impact on the resolution of PV panel images, the drone will conduct normal inspections and maintain monitoring of the drone's flight altitude. When the drone's flight altitude has a medium impact on the resolution of PV panel images, the drone's altitude will be adjusted to improve the image resolution. When the drone's flight altitude has a high impact on the resolution of PV panel images, the drone's current inspection mission will be suspended, the flight altitude will be replanned, and a high-impact warning signal will be issued.

[0062] When the drone's flight angle has a low impact on the viewing angle of the photovoltaic module image, the drone will perform normal inspection operations and maintain its flight attitude. When the drone's flight angle has a medium impact on the viewing angle of the photovoltaic module image, an angle compensation strategy will be formulated to adjust the drone's attitude and correct the drone's flight angle. When the drone's flight angle has a high impact on the viewing angle of the photovoltaic module image, the drone's current inspection mission will be suspended, the drone's attitude will be recalibrated, and a high-impact warning signal will be issued.

[0063] When the light intensity of the drone's flight environment has a low impact on the color reproduction of photovoltaic module images, the drone performs normal inspection work; when the light intensity of the drone's flight environment has a medium impact on the color reproduction of photovoltaic module images, the visible light camera parameters on the drone are adjusted to control the exposure time, sensitivity and white balance; when the light intensity of the drone's flight environment has a high impact on the color reproduction of photovoltaic module images, the current inspection task of the drone is suspended, the drone inspection area is changed, the photovoltaic module images that are severely affected are marked, and an early warning signal is issued.

[0064] In a second aspect, a method for intelligent inspection of photovoltaic power stations based on drone images is provided, which is used to implement the above-mentioned intelligent inspection system for photovoltaic power stations based on drone images, and comprises the following steps:

[0065] Step 1: Use an ultrasonic height sensor, accelerometer, and light sensor to collect the drone's flight altitude, flight angle, and light intensity in the flight environment. Use a visible light camera, laser rangefinder, and Photoshop software to collect real-time PV module images, PV module length, PV module image pixel length, and visible light camera focal length. Use NTP technology to synchronize the drone data with the PV module image data.

[0066] Step 2: Using the PV module length, the PV module image pixel length, and the visible light camera focal length, calculate the image resolution and image viewing angle of the PV module image. Using AI image color difference comparison technology, obtain the color reproduction degree of the real-time PV module image.

[0067] Step 3: Calculate the impact of the drone's flight altitude and flight angle on the resolution and viewing angle of the photovoltaic module image, and combine calculation and regression analysis to determine the impact of the light intensity of the drone's flight environment on the color reproduction of the photovoltaic module image.

[0068] Step 4: Use neural network algorithm to build an image defect judgment model;

[0069] Step 5: Analyze the drone data using the image defect determination model, categorize the impact of the drone data on the PV module image data, and take appropriate measures.

[0070] Beneficial effects of the invention: The invention provides an intelligent inspection method and system for photovoltaic power stations based on drone images, which has made significant technological progress in many aspects compared with traditional methods. In terms of data acquisition, the invention closely combines sensors, accelerometers, visible light cameras and laser rangefinders with NTP technology to accurately capture drone data and photovoltaic module data, and obtain the image resolution, image viewing angle and color restoration of photovoltaic module images through calculation and AI image color difference comparison technology. Combined with calculation and regression analysis, the degree of influence of drone flight data on photovoltaic module images is obtained, and then an image defect judgment model is constructed through a neural network algorithm, achieving real-time and comprehensive monitoring of drone photovoltaic power station inspections. During the inspection process, the drone data is analyzed according to the image defect judgment model, and the degree of influence of drone flight height on photovoltaic module image resolution and drone The authors formulate corresponding response strategies based on the degree of influence of flight angle on image viewing angle and the degree of influence of light intensity in the UAV flight environment on color reproduction, which solves the problem that UAVs cannot adjust the flight status and flight path of UAVs according to the impact of flight conditions on the images of photovoltaic modules during intelligent inspection of photovoltaic power stations, and the problem that the intelligent optimization algorithms studied at home and abroad are not perfect enough in the process of UAV flight path planning, resulting in inaccurate images of collected photovoltaic modules and affecting the inspection results of photovoltaic power stations. The methods of the present invention are ensured to refine a dynamic monitoring standard of a photovoltaic power station intelligent inspection method and system based on UAV images in a more precise range, so that the monitored data becomes a more accurate indicator under the same conditions. The research and development and application of this method significantly enhance the intelligence level of UAVs in the inspection process of photovoltaic power stations and greatly improve the inspection efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0072] Figure 1 This is a block diagram of an intelligent inspection system for photovoltaic power stations based on drone images according to the present invention;

[0073] Figure 2 This is a flow chart of a method for intelligent inspection of photovoltaic power stations based on drone images according to the present invention. DETAILED DESCRIPTION

[0074] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0075] Example 1, as Figure 1 As shown, the present invention provides a photovoltaic power station intelligent inspection system based on drone images, including an inspection data acquisition module, a photovoltaic component image module, an intelligent inspection processing module, an image defect determination module and an intelligent evaluation control module, wherein each module is communicatively connected;

[0076] The inspection data acquisition module uses acquisition equipment, data entry, and NTP technology to synchronously acquire drone flight data and PV panel image data. The acquisition equipment includes ultrasonic height sensors, light sensors, accelerometers, visible light cameras, and laser rangefinders, providing data support for subsequent intelligent inspection processes.

[0077] The photovoltaic module imaging module uses photovoltaic module image data to obtain the image resolution, image viewing angle, and color reproduction of photovoltaic module images through calculation and AI image color difference comparison technology, laying the foundation for subsequent determination of the impact of drone flight data on photovoltaic module images;

[0078] The intelligent inspection processing module combines calculation and regression analysis to determine the impact of drone flight data on PV module images, providing technical support for drone intelligent inspections of PV power stations.

[0079] The image defect judgment module uses a neural network algorithm to build an image defect judgment model;

[0080] The intelligent assessment and control module analyzes drone data in combination with the image defect judgment model, assesses the degree of impact of drone data on photovoltaic module image data, and takes corresponding measures to deal with it.

[0081] The inspection data collection module obtains drone flight data through collection equipment, including the following process:

[0082] Among them, the UAV flight data includes the UAV’s flight altitude, flight angle and the light intensity of the flight environment;

[0083] An ultrasonic height sensor is installed on the bottom of the drone. The ultrasonic height sensor transmits ultrasonic pulses and receives the reflected ultrasonic pulses. The time interval from the transmission of the ultrasonic pulse to the reception of the ultrasonic pulse is measured. The flight altitude of the drone is calculated based on the formula of ultrasonic propagation speed and displacement in the air.

[0084] Establish a three-dimensional space coordinate system, which includes the x-axis, y-axis and z-axis. Use the accelerometer to measure the acceleration of the drone in the x-axis, y-axis and z-axis directions. Under the action of gravity, the posture of the drone changes, and the acceleration measured by the accelerometer changes accordingly. Formula, calculate the flight angle of the drone, where is the tilt angle relative to the direction of gravity, is the acceleration collected by the accelerometer, is the acceleration due to gravity;

[0085] Light sensors are installed around the visible light camera equipped with the drone to collect the light intensity of the drone's flight environment.

[0086] The inspection data acquisition module acquires PV module image data through acquisition equipment and data entry. The process of synchronizing UAV flight data and PV module image data using NTP technology includes the following:

[0087] The photovoltaic module image data includes real-time photovoltaic module images, photovoltaic module lengths, photovoltaic module image pixel lengths, and visible light camera focal lengths;

[0088] The drone is equipped with a visible light camera, and the visible light camera shooting parameters are initialized and set. When the drone operates according to the flight trajectory, the visible light camera takes real-time images of the photovoltaic modules to obtain real-time images of the photovoltaic modules;

[0089] Install a laser rangefinder on the belly of the drone, start the laser rangefinder, and adjust the measurement angle using the drone's gimbal. The laser rangefinder calculates the distance based on the delay between the emitted laser beam and the reflected beam to obtain the length of the photovoltaic module.

[0090] Input the real-time photovoltaic module image into Photoshop software and measure the pixel length of the photovoltaic module image; consult the visible light camera parameter manual and obtain the visible light camera focal length through data entry;

[0091] A system of drones, NTP servers, and ground control stations is constructed. The drone and ground control station are started. The drone and ground control station send synchronization requests to NTP respectively, adding timestamps to the collected drone flight data and photovoltaic module image data. The drone transmits the timestamps of drone flight data and photovoltaic module image data to the ground control station via a wireless communication link. The drone flight data and photovoltaic module image data are matched and synchronized according to the timestamps.

[0092] The photovoltaic module image module obtains the image resolution, image viewing angle, and color reproduction of the photovoltaic module image through the following process:

[0093] A1. Calculate the PV module image resolution using the PV module length and PV module image pixel length:

[0094]

[0095] Where R is the PV module image resolution, l is the PV module image pixel length, and L is the PV module length;

[0096] A2. Based on the principles of geometric optics and the relationship between similar triangles, the process of calculating the viewing angle of the photovoltaic module image using the photovoltaic module length and the focal length of the visible light camera is as follows:

[0097]

[0098] in, is the viewing angle of the photovoltaic module image, L is the length of the photovoltaic module, is the focal length of the visible light camera;

[0099] A3. Establish a standard color library and use Gaussian filtering to denoise the photovoltaic module image. Use the AI ​​image analysis algorithm to extract the color data from the denoised photovoltaic module image. Compare the extracted color data with the standard color library. Use the AI ​​image analysis algorithm to calculate the color difference data of various colors in the photovoltaic module image. Define the color difference index of various colors in the photovoltaic module image. Calculate the color difference restoration degree by calculating the ratio of the color difference data of the same color to the corresponding color difference index.

[0100] The intelligent inspection processing module combines calculation and regression analysis to obtain the degree of impact of drone flight data on PV panel images. The process includes:

[0101] Among them, the impact of drone flight data on photovoltaic module images includes the impact of drone flight altitude on photovoltaic module image resolution, the impact of drone flight angle on photovoltaic module image viewing angle, and the impact of drone flight environment light intensity on photovoltaic module image color reproduction.

[0102] The intelligent inspection processing module obtains the degree of influence of the drone's flight altitude and flight angle on the PV module image resolution and PV module image viewing angle, including the following process:

[0103] B1. Select and As a data reference time point, obtain and The UAV flight altitude, UAV flight angle, PV module image resolution, and PV module image viewing angle at the moment;

[0104] B2. The process of calculating the impact of the drone's flight altitude on the PV panel image resolution is as follows:

[0105]

[0106]

[0107] in, The impact of the UAV's flight altitude on the resolution of photovoltaic module images; and for and The drone's flight altitude at the moment; and for and PV module image resolution at the moment;

[0108] B3. The process of calculating the impact of the drone's flight angle on the viewing angle of the photovoltaic module image is as follows:

[0109]

[0110]

[0111]

[0112]

[0113] in, The degree of influence of the UAV flight angle on the viewing angle of the photovoltaic module image; and for and The drone's flight angle at the moment; and for and The viewing angle of the photovoltaic module image at the moment; and They are and The difference in the UAV's flight angle within a certain time period and and The difference in viewing angle of the photovoltaic module image within a certain time period; is the proportional coefficient between the photovoltaic module image viewing angle and the UAV flight angle.

[0114] The intelligent inspection processing module obtains the degree of influence of the light intensity of the UAV flight environment on the color reproduction of the photovoltaic module image, including the following process:

[0115] C1. Use regression analysis to construct a linear regression model CR=aE+b, where CR is the color reproduction, E is the light intensity of the UAV flight environment, and a and b are the linear regression model parameters;

[0116] The light intensity of the drone's flight environment and the color reproduction of the photovoltaic module images at the same time are used as the data set, which is divided into a training set and a test set in a ratio of 7:3.

[0117] Use the training set data to train the linear regression model, calculate the linear regression model parameters through the least squares method, and obtain a preliminary linear regression model;

[0118] The performance of the trained linear regression model was evaluated using test data. The linear regression model was optimized by adjusting model parameters to obtain the linear relationship between the light intensity of the UAV flight environment and the color reproduction of the photovoltaic module image.

[0119] C2. Select and As a data reference time point, obtain The light intensity of the drone's flight environment at the moment and and Color restoration of photovoltaic module images at the moment;

[0120] C3. Using the linear relationship between the light intensity of the UAV flight environment and the color reproduction of the photovoltaic module image, the process of calculating the degree of influence of the light intensity of the UAV flight environment on the color reproduction of the photovoltaic module image is as follows:

[0121]

[0122]

[0123] in, The degree of influence of the light intensity of the UAV flight environment on the color reproduction of photovoltaic module images; for The light intensity of the drone’s flight environment at the moment; and for and The color restoration degree of the photovoltaic module image at the moment, a and b are the parameters of the linear regression model.

[0124] The image defect determination module uses a neural network algorithm. The process of building an image defect determination model includes:

[0125] A neural network algorithm was used to construct a neural network model. The data sets included the drone's flight altitude and its impact on the resolution of photovoltaic module images, the drone's flight angle and its impact on the viewing angle of photovoltaic module images, and the drone's flight environment's light intensity and its impact on the color reproduction of photovoltaic module images. The model was divided into a training set and a validation set in a 7:3 ratio.

[0126] The architecture of the neural network model includes an input layer, a hidden layer, and an output layer. The input layer has three input nodes, corresponding to the drone's flight altitude, flight angle, and light intensity of the flight environment, respectively. Three hidden layers are set, each with 32 neurons. The output layer has three output nodes, which output the degree of influence of the drone's flight altitude on the resolution of the photovoltaic module image, the degree of influence of the drone's flight angle on the viewing angle of the photovoltaic module image, and the degree of influence of the drone's flight environment light intensity on the color reproduction of the photovoltaic module image.

[0127] Input the training set data into the neural network model, perform weighted summation on the training set data through forward propagation, obtain the prediction result of the output layer, perform backpropagation on the error between the prediction result of the output layer and the real data, update the weight and configuration of the network, and obtain the trained neural network model;

[0128] The validation set data is input into the trained neural network model to obtain the output result of the output layer in the neural network model. Based on the difference between the output result and the true result, the performance of the trained neural network model is evaluated, the model parameters are adjusted and the performance of the neural network model is optimized. The neural network model is deployed in a photovoltaic power station intelligent inspection system based on drone images to obtain an image defect judgment model.

[0129] The intelligent assessment and control module analyzes drone data using an image defect determination model, assesses the impact of drone data on PV module image data, and takes appropriate measures. The process includes:

[0130] The drone data is input into the image defect judgment model to obtain the degree of influence of the drone data on the photovoltaic module image data. When the degree of influence of the drone data on the photovoltaic module image data is between 0% and 30%, it is classified as low impact; when the degree of influence of the drone data on the photovoltaic module image data is between 30% and 60%, it is classified as medium impact; when the degree of influence of the drone data on the photovoltaic module image data is between 60% and 100%, it is classified as high impact;

[0131] When the drone's flight altitude has a low impact on the resolution of PV panel images, the drone will conduct normal inspections and maintain monitoring of the drone's flight altitude. When the drone's flight altitude has a medium impact on the resolution of PV panel images, the drone's altitude will be adjusted to improve the image resolution. When the drone's flight altitude has a high impact on the resolution of PV panel images, the drone's current inspection mission will be suspended, the flight altitude will be replanned, and a high-impact warning signal will be issued.

[0132] When the drone's flight angle has a low impact on the viewing angle of the photovoltaic module image, the drone will perform normal inspection operations and maintain its flight attitude. When the drone's flight angle has a medium impact on the viewing angle of the photovoltaic module image, an angle compensation strategy will be formulated to adjust the drone's attitude and correct the drone's flight angle. When the drone's flight angle has a high impact on the viewing angle of the photovoltaic module image, the drone's current inspection mission will be suspended, the drone's attitude will be recalibrated, and a high-impact warning signal will be issued.

[0133] When the light intensity of the drone's flight environment has a low impact on the color reproduction of photovoltaic module images, the drone performs normal inspection work; when the light intensity of the drone's flight environment has a medium impact on the color reproduction of photovoltaic module images, the visible light camera parameters on the drone are adjusted to control the exposure time, sensitivity and white balance; when the light intensity of the drone's flight environment has a high impact on the color reproduction of photovoltaic module images, the current inspection task of the drone is suspended, the drone inspection area is changed, the photovoltaic module images that are severely affected are marked, and an early warning signal is issued.

[0134] Example 2, as Figure 2 As shown, based on Example 1, the present invention provides a technical solution: a method for intelligent inspection of photovoltaic power stations based on drone images, which is used to implement the above-mentioned intelligent inspection system for photovoltaic power stations based on drone images, and is composed of the following steps:

[0135] Step 1: Use an ultrasonic height sensor, accelerometer, and light sensor to collect the drone's flight altitude, flight angle, and light intensity in the flight environment. Use a visible light camera, laser rangefinder, and Photoshop software to collect real-time PV module images, PV module length, PV module image pixel length, and visible light camera focal length. Use NTP technology to synchronize the drone data with the PV module image data.

[0136] Step 2: Using the PV module length, the PV module image pixel length, and the visible light camera focal length, calculate the image resolution and image viewing angle of the PV module image. Using AI image color difference comparison technology, obtain the color reproduction degree of the real-time PV module image.

[0137] Step 3: Calculate the impact of the drone's flight altitude and flight angle on the resolution and viewing angle of the photovoltaic module image, and combine calculation and regression analysis to determine the impact of the light intensity of the drone's flight environment on the color reproduction of the photovoltaic module image.

[0138] Step 4: Use neural network algorithm to build an image defect judgment model;

[0139] Step 5: Analyze the drone data using the image defect determination model, categorize the impact of the drone data on the PV module image data, and take appropriate measures.

[0140] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A photovoltaic power station intelligent inspection system based on drone images, including an inspection data acquisition module, a photovoltaic module image module, an intelligent inspection processing module, an image defect determination module and an intelligent evaluation control module, wherein: The communication connection of each module is characterized by: The inspection data acquisition module synchronously acquires UAV flight data and photovoltaic module image data through acquisition equipment, data entry and NTP technology, wherein the acquisition equipment includes an ultrasonic height sensor, a light sensor, an accelerometer, a visible light camera and a laser rangefinder; The photovoltaic module image module uses the photovoltaic module image data to obtain the image resolution, image viewing angle and color reproduction of the photovoltaic module image through calculation and AI image color difference comparison technology; The intelligent inspection processing module combines calculation and regression analysis to obtain the degree of influence of drone flight data on photovoltaic module images; The image defect determination module uses a neural network algorithm to construct an image defect determination model; The intelligent evaluation and control module analyzes the drone data in combination with the image defect determination model, evaluates the degree of impact of the drone data on the photovoltaic module image data, and takes corresponding measures to deal with it.

2. The intelligent inspection system for photovoltaic power plants based on drone images according to claim 1 is characterized by: The inspection data acquisition module acquires the UAV flight data through the acquisition device, including the following steps: The UAV flight data includes the UAV's flight altitude, flight angle, and light intensity of the flight environment; An ultrasonic height sensor is installed on the bottom of the drone. The ultrasonic height sensor transmits ultrasonic pulses and receives the reflected ultrasonic pulses. The time interval from the transmission of the ultrasonic pulse to the reception of the ultrasonic pulse is measured. The flight altitude of the drone is calculated based on the formula of ultrasonic propagation speed and displacement in the air. Establish a three-dimensional space coordinate system, which includes the x-axis, y-axis and z-axis. Use the accelerometer to measure the acceleration of the drone in the x-axis, y-axis and z-axis directions. Under the action of gravity, the posture of the drone changes, and the acceleration measured by the accelerometer changes accordingly. Formula, calculate the flight angle of the drone, where is the tilt angle relative to the direction of gravity, is the acceleration collected by the accelerometer, is the acceleration due to gravity; Light sensors are installed around the visible light camera equipped with the drone to collect the light intensity of the drone's flight environment.

3. The intelligent inspection system for photovoltaic power plants based on drone images according to claim 2 is characterized by: The inspection data acquisition module acquires photovoltaic module image data through acquisition equipment and data entry, and uses NTP technology to synchronize UAV flight data with photovoltaic module image data. The process includes: The photovoltaic module image data includes a real-time photovoltaic module image, a photovoltaic module length, a photovoltaic module image pixel length, and a visible light camera focal length; The drone is equipped with a visible light camera, and the visible light camera shooting parameters are initialized and set. When the drone operates according to the flight trajectory, the visible light camera takes real-time images of the photovoltaic modules to obtain real-time images of the photovoltaic modules; Install a laser rangefinder on the belly of the drone, start the laser rangefinder, and adjust the measurement angle using the drone's gimbal. The laser rangefinder calculates the distance based on the delay between the emitted laser beam and the reflected beam to obtain the length of the photovoltaic module. Input the real-time photovoltaic module image into Photoshop software and measure the pixel length of the photovoltaic module image; consult the visible light camera parameter manual and obtain the visible light camera focal length through data entry; A system of drones, NTP servers, and ground control stations is constructed. The drone and ground control station are started. The drone and ground control station send synchronization requests to NTP respectively, adding timestamps to the collected drone flight data and photovoltaic module image data. The drone transmits the timestamps of drone flight data and photovoltaic module image data to the ground control station via a wireless communication link. The drone flight data and photovoltaic module image data are matched and synchronized according to the timestamps.

4. The intelligent inspection system for photovoltaic power plants based on drone images according to claim 3 is characterized by: The process of obtaining the image resolution, image viewing angle and color reproduction of the photovoltaic module image by the photovoltaic module image module includes: A1. Calculate the PV module image resolution using the PV module length and PV module image pixel length: ； Where R is the PV module image resolution, l is the PV module image pixel length, and L is the PV module length; A2. Based on the principles of geometric optics and the relationship between similar triangles, the process of calculating the viewing angle of the photovoltaic module image using the photovoltaic module length and the focal length of the visible light camera is as follows: ； in, is the viewing angle of the photovoltaic module image, L is the length of the photovoltaic module, is the focal length of the visible light camera; A3. Establish a standard color library and use Gaussian filtering to denoise the photovoltaic module image. Use the AI ​​image analysis algorithm to extract the color data from the denoised photovoltaic module image. Compare the extracted color data with the standard color library. Use the AI ​​image analysis algorithm to calculate the color difference data of various colors in the photovoltaic module image. Define the color difference index of various colors in the photovoltaic module image. Calculate the color difference restoration degree by calculating the ratio of the color difference data of the same color to the corresponding color difference index.

5. The intelligent inspection system for photovoltaic power stations based on drone images according to claim 4 is characterized by: The intelligent inspection processing module combines calculation and regression analysis methods to obtain the degree of influence of drone flight data on photovoltaic module images, including the following process: The degree of influence of the drone flight data on the photovoltaic module image includes the degree of influence of the drone flight altitude on the resolution of the photovoltaic module image, the degree of influence of the drone flight angle on the viewing angle of the photovoltaic module image, and the degree of influence of the light intensity of the drone flight environment on the color reproduction of the photovoltaic module image.

6. The photovoltaic power station intelligent inspection system based on drone images according to claim 5 is characterized by: The process of the intelligent inspection processing module obtaining the degree of influence of the drone's flight altitude and flight angle on the photovoltaic module image resolution and photovoltaic module image viewing angle respectively includes: B1. Select and As a data reference time point, obtain and The UAV flight altitude, UAV flight angle, PV module image resolution, and PV module image viewing angle at the moment; B2. The process of calculating the impact of the drone's flight altitude on the PV panel image resolution is as follows: ； ； in, The impact of the UAV's flight altitude on the resolution of photovoltaic module images; and for and The drone's flight altitude at the moment; and for and PV module image resolution at the moment; B3. The process of calculating the impact of the drone's flight angle on the viewing angle of the photovoltaic module image is as follows: ； ； ； ； in, The degree of influence of the UAV flight angle on the viewing angle of the photovoltaic module image; and for and The drone's flight angle at the moment; and for and The viewing angle of the photovoltaic module image at the moment; and They are and The difference in the UAV's flight angle within a certain time period and and The difference in viewing angle of the photovoltaic module image within a certain time period; is the proportional coefficient between the photovoltaic module image viewing angle and the UAV flight angle.

7. The photovoltaic power station intelligent inspection system based on drone images according to claim 6 is characterized by: The process of the intelligent inspection processing module obtaining the degree of influence of the light intensity of the UAV flight environment on the color reproduction of the photovoltaic module image includes: C1. Use regression analysis to construct a linear regression model CR=aE+b, where CR is the color reproduction, E is the light intensity of the UAV flight environment, and a and b are the linear regression model parameters; The light intensity of the drone's flight environment and the color reproduction of the photovoltaic module images at the same time are used as the data set, which is divided into a training set and a test set in a ratio of 7:

3. Use the training set data to train the linear regression model, calculate the linear regression model parameters through the least squares method, and obtain a preliminary linear regression model; The performance of the trained linear regression model was evaluated using test data. The linear regression model was optimized by adjusting model parameters to obtain the linear relationship between the light intensity of the UAV flight environment and the color reproduction of the photovoltaic module image. C2. Select and As a data reference time point, obtain The light intensity of the drone's flight environment at the moment and and Color restoration of photovoltaic module images at the moment; C3. Using the linear relationship between the light intensity of the UAV flight environment and the color reproduction of the photovoltaic module image, the process of calculating the degree of influence of the light intensity of the UAV flight environment on the color reproduction of the photovoltaic module image is as follows: ； ； in, The degree of influence of the light intensity of the UAV flight environment on the color reproduction of photovoltaic module images; for The light intensity of the drone’s flight environment at the moment; and for and The color restoration degree of the photovoltaic module image at the moment, a and b are the parameters of the linear regression model.

8. The photovoltaic power station intelligent inspection system based on drone images according to claim 7 is characterized by: The image defect determination module uses a neural network algorithm to construct an image defect determination model, including the following steps: A neural network algorithm was used to construct a neural network model. The data sets included the drone's flight altitude and its impact on the resolution of photovoltaic module images, the drone's flight angle and its impact on the viewing angle of photovoltaic module images, and the drone's flight environment's light intensity and its impact on the color reproduction of photovoltaic module images. The model was divided into a training set and a validation set in a 7:3 ratio. The architecture of the neural network model includes an input layer, a hidden layer, and an output layer. The input layer has three input nodes, corresponding to the drone's flight altitude, flight angle, and light intensity of the flight environment, respectively. Three hidden layers are set, each with 32 neurons. The output layer has three output nodes, which output the degree of influence of the drone's flight altitude on the resolution of the photovoltaic module image, the degree of influence of the drone's flight angle on the viewing angle of the photovoltaic module image, and the degree of influence of the drone's flight environment light intensity on the color reproduction of the photovoltaic module image. Input the training set data into the neural network model, perform weighted summation on the training set data through forward propagation, obtain the prediction result of the output layer, perform backpropagation on the error between the prediction result of the output layer and the real data, update the weight and configuration of the network, and obtain the trained neural network model; The validation set data is input into the trained neural network model to obtain the output result of the output layer in the neural network model. Based on the difference between the output result and the true result, the performance of the trained neural network model is evaluated, the model parameters are adjusted and the performance of the neural network model is optimized. The neural network model is deployed in a photovoltaic power station intelligent inspection system based on drone images to obtain an image defect judgment model.

9. The photovoltaic power station intelligent inspection system based on drone images according to claim 8, characterized in that: The intelligent assessment control module analyzes drone data in combination with the image defect determination model, assesses the degree of impact of the drone data on the photovoltaic module image data, and takes corresponding measures to address the impact, including the following process: The drone data is input into the image defect judgment model to obtain the degree of influence of the drone data on the photovoltaic module image data. When the degree of influence of the drone data on the photovoltaic module image data is between 0% and 30%, it is classified as low impact; when the degree of influence of the drone data on the photovoltaic module image data is between 30% and 60%, it is classified as medium impact; when the degree of influence of the drone data on the photovoltaic module image data is between 60% and 100%, it is classified as high impact; When the drone's flight altitude has a low impact on the resolution of PV panel images, the drone will conduct normal inspections and maintain monitoring of the drone's flight altitude. When the drone's flight altitude has a medium impact on the resolution of PV panel images, the drone's altitude will be adjusted to improve the image resolution. When the drone's flight altitude has a high impact on the resolution of PV panel images, the drone's current inspection mission will be suspended, the flight altitude will be replanned, and a high-impact warning signal will be issued. When the drone's flight angle has a low impact on the viewing angle of the photovoltaic module image, the drone will perform normal inspection operations and maintain its flight attitude. When the drone's flight angle has a medium impact on the viewing angle of the photovoltaic module image, an angle compensation strategy will be formulated to adjust the drone's attitude and correct the drone's flight angle. When the drone's flight angle has a high impact on the viewing angle of the photovoltaic module image, the drone's current inspection mission will be suspended, the drone's attitude will be recalibrated, and a high-impact warning signal will be issued. When the light intensity of the drone's flight environment has a low impact on the color reproduction of photovoltaic module images, the drone performs normal inspection work; when the light intensity of the drone's flight environment has a medium impact on the color reproduction of photovoltaic module images, the visible light camera parameters on the drone are adjusted to control the exposure time, sensitivity and white balance; when the light intensity of the drone's flight environment has a high impact on the color reproduction of photovoltaic module images, the current inspection task of the drone is suspended, the drone inspection area is changed, the photovoltaic module images that are severely affected are marked, and an early warning signal is issued.

10. A method for intelligent inspection of photovoltaic power stations based on drone images, implemented based on the intelligent inspection system for photovoltaic power stations based on drone images according to any one of claims 1 to 9, characterized in that: It consists of the following steps: Step 1: Use an ultrasonic height sensor, accelerometer, and light sensor to collect the drone's flight altitude, flight angle, and light intensity in the flight environment. Use a visible light camera, laser rangefinder, and Photoshop software to collect real-time PV module images, PV module length, PV module image pixel length, and visible light camera focal length. Use NTP technology to synchronize the drone data with the PV module image data. Step 2: Using the PV module length, the PV module image pixel length, and the visible light camera focal length, calculate the image resolution and image viewing angle of the PV module image. Using AI image color difference comparison technology, obtain the color reproduction degree of the real-time PV module image. Step 3: Calculate the impact of the drone's flight altitude and flight angle on the resolution and viewing angle of the photovoltaic module image, and combine calculation and regression analysis to determine the impact of the light intensity of the drone's flight environment on the color reproduction of the photovoltaic module image. Step 4: Use neural network algorithm to build an image defect judgment model; Step 5: Analyze the drone data using the image defect determination model, categorize the impact of the drone data on the PV module image data, and take appropriate measures.

Images