A cleaning robot control method and system for a photovoltaic power station

Abstract

The present application relates to the technical field of cleaning robot control, and particularly relates to a cleaning robot control method and system for a photovoltaic power station, the method comprising: acquiring an initial background image of a to-be-detected area on a surface of the photovoltaic power station under a target shooting angle when a near-infrared light source is turned off and an initial target image when the near-infrared light source is turned on; determining a light transmission index corresponding to each pixel point in the initial target image; segmenting out a target stain area and screening out a target environmental reference area corresponding to each target stain area from the initial background image; screening out an environmental area matching each target stain area; determining a stain characteristic index corresponding to each target stain area, so as to adjust and control a cleaning mode of the cleaning robot. The present application quantifies the stain characteristic index corresponding to the target stain area, thereby realizing the differentiation of the stain type represented by the target stain area and improving the rationality of the adjustment and control of the cleaning mode of the cleaning robot.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cleaning robot control, in particular to a cleaning robot control method and system for a photovoltaic power station. BACKGROUND

[0002] A photovoltaic power station usually occupies a wide area, which can reach thousands of mu, and can be built in complex terrain areas such as rooftops, deserts, and slopes. Photovoltaic panels convert solar energy into electricity by absorbing sunlight, and dust and other stains on the surface often block sunlight, thereby reducing the photoelectric conversion efficiency. Therefore, a cleaning robot is often used to clean the photovoltaic panels to maintain high light transmittance and power generation efficiency of the photovoltaic panels.

[0003] The cleaning robot usually has two working modes, namely dry brushing and wet brushing. Dry brushing is usually suitable for loose dust, dry powder, and other stains that can be easily removed by mechanical force, and does not require the use of cleaning liquid, which can save consumables and avoid potential effects of cleaning liquid residue on photovoltaic panels. Wet brushing is usually suitable for stubborn stains such as oil stains, hardened dust, dried mud stains, bird droppings, and mold, and removes stains by spraying cleaning liquid combined with mechanical brushing. Therefore, during the cleaning process controlled by the cleaning robot, the cleaning mode of the cleaning robot often needs to be adjusted by identifying the type of stain. Currently, when identifying the type of stain, the method commonly used is to identify the type of stain according to the difference in gray value.

[0004] However, when the cleaning mode of the cleaning robot is adjusted and controlled by identifying the type of stain according to the difference in gray value, the following technical problems often occur:

[0005] The gray values corresponding to some different types of stains that require different cleaning modes can be similar, such as the gray difference between dried mud stains and loose dust, which is often not obvious. Therefore, when identifying the type of stain based only on the difference in gray value, it can be difficult to accurately distinguish between different types of stains that require different cleaning modes, resulting in poor accuracy of stain type identification, and further resulting in poor rationality of cleaning robot cleaning mode adjustment and control. SUMMARY

[0006] To solve the technical problem of poor rationality of cleaning robot cleaning mode adjustment and control due to poor accuracy of stain type identification, the present application provides a cleaning robot control method and system for a photovoltaic power station.

[0007] In a first aspect, the present application provides a cleaning robot control method for a photovoltaic power station, which comprises:

[0008] obtaining an initial background image of a to-be-detected area on the surface of the photovoltaic power station at a target shooting angle when a near-infrared light source is turned off and an initial target image when the near-infrared light source is turned on;

[0009] According to the gray difference between the pixel points at the same position in the initial background image and the initial target image, the light transmission index corresponding to each pixel point in the initial target image is determined;

[0010] According to the light transmission index corresponding to all pixel points in the initial target image, the target stain area is segmented from the initial target image, and the target environment reference area corresponding to each target stain area is selected from the initial background image;

[0011] Obtain the reference background image of the to-be-detected area under the preset shooting angle different from the target shooting angle when the near-infrared light source is turned off, and select the environment area matched with each target stain area from all reference background images as the matched environment reference area corresponding to each target stain area;

[0012] According to the gray difference between the target environment reference area and the matched environment reference area corresponding to each target stain area, and the light transmission index corresponding to the pixel points in each target stain area, the stain characteristic index corresponding to each target stain area is determined;

[0013] According to the stain characteristic index corresponding to each target stain area, the cleaning mode of the cleaning robot is adjusted.

[0014] In combination with the above first aspect, in a possible implementation manner, after the stain represented by the target stain area in the to-be-detected area is cleaned by the cleaning robot each time, the method comprises the following steps.

[0015] Any one target stain area in the initial target image is determined as a marker stain area, and any cleaning of the stain represented by the marker stain area is determined as a marker cleaning.

[0016] Obtain the marker cleaning background image and the marker cleaning target image of the to-be-detected area under the target shooting angle when the near-infrared light source is turned off and when the near-infrared light source is turned on after the marker cleaning of the stain represented by the marker stain area.

[0017] According to the gray difference between the pixel points at the same position in the marker cleaning background image and the marker cleaning target image, the light transmission index corresponding to each pixel point in the marker cleaning target image is determined.

[0018] According to the light transmission index corresponding to all pixel points in the marker cleaning target image, the marker cleaning stain area is segmented from the marker cleaning target image.

[0019] From all marker cleaning stain areas, select the marker cleaning stain area with a non-empty intersection with the marker stain area as a marker matching area, and determine the stain characteristic index corresponding to the marker matching area.

[0020] determine the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning;

[0021] if the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning is less than or equal to a preset cleaning effectiveness threshold, change the cleaning method to implement the next cleaning of the stain represented by the marked stain area after the marked cleaning;

[0022] if the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning is greater than the preset cleaning effectiveness threshold, use the same cleaning method as the marked cleaning to implement the next cleaning of the stain represented by the marked stain area after the marked cleaning.

[0023] In combination with the first aspect, in a possible implementation manner, if the same cleaning method as the marked cleaning is used to implement the next cleaning of the stain represented by the marked stain area after the marked cleaning, the method for obtaining the cleaning intensity of the next cleaning includes:

[0024] determine the cleaning effectiveness of the stain represented by the marked stain area after the previous cleaning of the marked cleaning as a reference previous effectiveness;

[0025] determine the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning as a reference marked effectiveness;

[0026] determine an intensity correction coefficient according to the area of the marked matching area, the average of the light transmission indexes of all pixel points in the marked matching area, and the difference between the reference marked effectiveness and the reference previous effectiveness;

[0027] determine a target cleaning intensity corresponding to the next cleaning of the marked cleaning according to the intensity correction coefficient and the cleaning intensity in the marked cleaning process;

[0028] when the next cleaning of the stain represented by the marked stain area after the marked cleaning is implemented, adjust the cleaning intensity to the target cleaning intensity.

[0029] In combination with the first aspect, in a possible implementation manner, the method for determining the light transmission index corresponding to each pixel point in the initial target image according to the gray difference between the pixel points at the same position in the initial background image and the initial target image includes:

[0030] determining any one pixel point in the initial target image as a marker pixel point, and screening a pixel point with the same position as the marker pixel point from the initial background image as a matching environmental point corresponding to the marker pixel point;

[0031] determining a difference between the gray value of the marker pixel point and the gray value of the matching environmental point corresponding to the marker pixel point as a light transmission index corresponding to the marker pixel point.

[0032] In combination with the first aspect, in a possible implementation manner, the step of segmenting the target stain region from the initial target image according to the light transmission indexes corresponding to all pixel points in the initial target image comprises:

[0033] obtaining a standard background image when a near-infrared light source is turned off and a standard target image when the near-infrared light source is turned on of a standard photovoltaic panel under a target shooting angle;

[0034] determining a light transmission index corresponding to each pixel point in the standard target image according to a gray difference between pixel points at the same position in the standard background image and the standard target image;

[0035] determining a mean value of the light transmission indexes corresponding to all pixel points in the standard target image as a no-stain representative index;

[0036] determining an absolute value of a difference between the light transmission index corresponding to each pixel point in the initial target image and the no-stain representative index as an index deviation degree corresponding to each pixel point in the initial target image;

[0037] if the index deviation degree corresponding to the pixel point in the initial target image is greater than a preset deviation abnormal threshold, determining the pixel point in the initial target image as an abnormal light transmission point;

[0038] performing connected domain extraction on a region composed of all abnormal light transmission points in the initial target image, and determining the extracted connected domain as a target stain region.

[0039] In combination with the first aspect, in a possible implementation manner, the step of screening a target environmental reference region corresponding to each target stain region from the initial background image comprises:

[0040] determining any one target stain region in the initial target image as a marker stain region, and screening a region with the same position as the marker stain region from the initial background image as a target environmental reference region corresponding to the marker stain region.

[0041] With reference to the first aspect, in a possible implementation manner, the filtering, from all the reference background images, of an environmental region matching each target stain region as a matching environmental reference region corresponding to each target stain region comprises:

[0042] obtaining a reference target image of the to-be-detected region under a preset shooting angle different from the target shooting angle and when the near-infrared light source is turned on;

[0043] determining any one target stain region in the initial target image as a marker stain region;

[0044] filtering, from each reference target image, a region representing the same stain as the marker stain region as a reference matching region corresponding to the marker stain region by using the SIFT algorithm;

[0045] determining, as a matching background image corresponding to each reference matching region, a reference background image belonging to the same preset shooting angle as the reference matching region;

[0046] filtering, from the matching background image corresponding to each reference matching region, a region having the same position as the reference matching region as a matching environmental reference region corresponding to the marker stain region.

[0047] With reference to the first aspect, in a possible implementation manner, the determining of the stain feature index corresponding to each target stain region according to the gray difference between the target environmental reference region and the matching environmental reference region corresponding to each target stain region and the light transmission index corresponding to each pixel point in each target stain region comprises:

[0048] constructing, as a whole matching region set corresponding to each target stain region, the target environmental reference region corresponding to each target stain region and all the matching environmental reference regions;

[0049] determining, as a gray representative index, a mean value of the gray values corresponding to all the pixel points in each region in the whole matching region set corresponding to each target stain region, to obtain a gray representative index set corresponding to each target stain region;

[0050] determining, as a target gray variation difference corresponding to each target stain region, a difference between different gray representative indexes in the gray representative index set corresponding to each target stain region;

[0051] determining, as the stain feature index corresponding to each target stain region, the target gray variation difference corresponding to each target stain region and a mean value of the light transmission indexes corresponding to all the pixel points in each target stain region.

[0052] In a possible implementation manner of the first aspect, the target gray variation difference of each target stain area is determined according to a difference between different gray representative indexes in the set of gray representative indexes corresponding to each target stain area.

[0053] determining any one target stain area in the initial target image as a marked stain area;

[0054] determining an absolute value of a difference between each two gray representative indexes in the set of gray representative indexes corresponding to the marked stain area as a local gray variation, to obtain a set of local gray variations corresponding to the marked stain area;

[0055] determining a mean value of all local gray variations in the set of local gray variations corresponding to the marked stain area as the target gray variation difference corresponding to the marked stain area.

[0056] In a second aspect, the present application provides a cleaning robot control system for a photovoltaic power station, the system comprising:

[0057] an image acquisition module configured to acquire an initial background image of a to-be-detected area on a surface of the photovoltaic power station under a target shooting angle when a near-infrared light source is turned off and an initial target image when the near-infrared light source is turned on;

[0058] a light transmission index determination module configured to determine a light transmission index corresponding to each pixel in the initial target image according to a gray variation between the pixels at the same position in the initial background image and the initial target image;

[0059] a segmentation and screening module configured to segment target stain areas from the initial target image according to the light transmission index corresponding to all pixels in the initial target image, and screen a target environment reference area corresponding to each target stain area from the initial background image;

[0060] an acquisition and screening module configured to acquire a reference background image of the to-be-detected area under a preset shooting angle different from the target shooting angle when the near-infrared light source is turned off, and screen an environment area matching each target stain area from all the reference background images as a matching environment reference area corresponding to each target stain area;

[0061] a stain feature index determination module configured to determine a stain feature index corresponding to each target stain area according to a gray variation between the target environment reference area and the matching environment reference area corresponding to each target stain area, and the light transmission index corresponding to the pixels in each target stain area;

[0062] an adjustment control module configured to adjust and control a cleaning mode of the cleaning robot according to the stain feature index corresponding to each target stain area.

[0063] In a third aspect, a server is provided, comprising a memory and a processor. The memory is configured to store executable program code, and the processor is configured to invoke and run the executable program code from the memory, so that the device executes the method in the first aspect or any possible implementation manner of the first aspect.

[0064] In a fourth aspect, a computer program product is provided, comprising computer program code, which, when executed on a computer, causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0065] In a fifth aspect, a computer-readable storage medium is provided, which stores computer program code, which, when executed on a computer, causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0066] The present application has the following beneficial effects:

[0067] The control method of the cleaning robot for the photovoltaic power station quantifies the stain feature index corresponding to the target stain area, thereby realizing the differentiation of the stain type represented by the target stain area, solving the technical problem of poor rationality of the cleaning mode adjustment control of the cleaning robot due to poor accuracy of stain type identification, and improving the rationality of the cleaning mode adjustment control of the cleaning robot. Specifically, the present application analyzes a plurality of factors related to stain type differentiation based on the initial background image, the initial target image and the reference background image, such as the gray difference between the target environment reference area and the matching environment reference area, and the light transmission index, thereby quantifying the stain feature index representing the stain type, and adjusting and controlling the cleaning mode of the cleaning robot based on the stain feature index corresponding to the target stain area, thereby improving the rationality of the cleaning mode adjustment control of the cleaning robot. BRIEF DESCRIPTION OF DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0069] Figure 1 The flowchart of the control method of the cleaning robot for the photovoltaic power station of the present application;

[0070] Figure 2A schematic structural diagram of a cleaning robot control system for a photovoltaic power station according to the present application;

[0071] Figure 3 A schematic structural diagram of a computer device according to the present application. DETAILED DESCRIPTION

[0072] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined object of the application, the specific implementation, structure, features and effects of the technical solutions proposed according to the present application are described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0073] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0074] Reference Figure 1 , shows the flow of some embodiments of a cleaning robot control method for a photovoltaic power station according to the present application. The cleaning robot control method for a photovoltaic power station includes the following steps:

[0075] Step S1, obtaining the initial background image when the near-infrared light source is turned off and the initial target image when the near-infrared light source is turned on of the to-be-detected area on the surface of the photovoltaic power station at the target shooting angle.

[0076] Wherein, the to-be-detected area can be a region divided on the surface of the photovoltaic panel of the photovoltaic power station. In actual situations, the photovoltaic power station usually occupies a wide area, which can reach thousands of mu, so it is difficult to achieve full coverage of the surface of the photovoltaic power station by one shooting when obtaining the surface image of the photovoltaic power station. Therefore, the present application realizes the identification of the camera coverable area in the surface of the photovoltaic power station, i.e. the to-be-detected area, by means of block identification of the surface of the photovoltaic power station. It should be noted that in order to improve the identification efficiency, the to-be-detected area can be as large as possible.

[0077] The target shooting angle can be a front angle, in which case the camera can be directly aimed at the front of the region to be detected. The initial background image when the near-infrared light source is off can be an image captured when the near-infrared light source is off, which captures reflected light generated by the scene being illuminated by ambient light (such as sunlight, etc.). The initial target image when the near-infrared light source is on can be an image captured when the near-infrared light source is on, which captures reflected light generated by the scene being illuminated by ambient light and actively projected near-infrared light. The infrared light source is often turned on and off every 1 second. Therefore, the time length between the capture time corresponding to the initial background image and the capture time corresponding to the initial target image can be 1 second.

[0078] As an example, the shooting angle can be adjusted to a front angle by a near-infrared sensitive camera, and an image of the region to be detected captured when the near-infrared light source is off is recorded as an initial background image, and an image of the region to be detected captured when the near-infrared light source is on is recorded as an initial target image. The near-infrared sensitive camera is an imaging device specially designed to detect near-infrared band (usually about 780 nm to 1100 nm) light.

[0079] Step S2, determining the light transmission index corresponding to each pixel point in the initial target image according to the gray difference between the pixel points at the same position in the initial background image and the initial target image.

[0080] It should be noted that due to the high reflection area of the photovoltaic panel structure itself, and the fact that it is in a natural environment, the ambient light is complex, such as the movement of clouds, changes in the angle of the sun, etc., which affects the intensity of the light on the surface of the photovoltaic panel, so the gray difference between the images obtained before and after the near-infrared light source is turned on can to some extent exclude the interference of the fluctuation of the ambient light, and obtain an accurate light transmission index.

[0081] As an example, this step can include the following steps:

[0082] First, any one pixel point in the initial target image is determined as a marker pixel point, and a pixel point with the same position as the marker pixel point is selected from the initial background image as a matching ambient point corresponding to the marker pixel point.

[0083] Second, the difference between the gray value of the marker pixel point and the gray value of the matching ambient point corresponding to the marker pixel point is determined as the light transmission index corresponding to the marker pixel point.

[0084] It should be noted that the greater the light transmission index corresponding to the marker pixel point, the greater the increase in the corresponding brightness after the near-infrared light source is turned on.

[0085] Step S3, according to the light transmission index corresponding to all pixel points in the initial target image, segmenting the target stain area from the initial target image, and screening the target environment reference area corresponding to each target stain area from the initial background image.

[0086] As an example, the present step can include the following steps:

[0087] First, obtain the standard background image of the standard photovoltaic panel under the target shooting angle when the near-infrared light source is turned off, and the standard target image when the near-infrared light source is turned on.

[0088] Among them, the standard photovoltaic panel can be a clean photovoltaic panel with the same material model as the material model of the photovoltaic panel to be detected. The standard background image can be an image of the standard photovoltaic panel collected when the near-infrared light source is turned off. The standard target image can be an image of the standard photovoltaic panel collected when the near-infrared light source is turned on.

[0089] Second, according to the gray difference between the pixel points at the same position in the standard background image and the standard target image, determine the light transmission index corresponding to each pixel point in the standard target image.

[0090] It should be noted that the method for obtaining the light transmission index corresponding to the pixel points in the standard target image can be the same as the method for obtaining the light transmission index corresponding to the pixel points in the initial target image, which will not be repeated here.

[0091] Third, the average of the light transmission index corresponding to all pixel points in the above standard target image is determined as the no-stain representative index.

[0092] Fourth, the absolute value of the difference between the light transmission index corresponding to each pixel point in the above initial target image and the above no-stain representative index is determined as the index deviation degree corresponding to each pixel point in the above initial target image.

[0093] Fifth, if the index deviation degree corresponding to the pixel points in the above initial target image is greater than the preset deviation threshold, the pixel points in the above initial target image are determined as abnormal light transmission points.

[0094] Among them, the preset deviation threshold can be a pre-set threshold, which can be 0.6.

[0095] It should be noted that the stain interference often makes the light transmission degree of the corresponding area deviate from the light transmission degree when there is no stain, so the abnormal light transmission point often represents the position point affected by the stain interference.

[0096] Sixth, the region formed by all abnormal light transmission points in the above initial target image is subjected to connected component extraction, and the extracted connected component is determined as the target stain area.

[0097] It should be noted that the target stain area can represent a stain on the to-be-detected area.

[0098] Step 7, any one of the target stain areas in the initial target image is determined as a marked stain area, and an area in the initial background image with the same position as the marked stain area is selected as a target environment reference area corresponding to the marked stain area.

[0099] It should be noted that the target stain area can represent a stain on the to-be-detected area.

[0100] Step S4, obtaining a reference background image of the to-be-detected area under a preset shooting angle different from the target shooting angle when the near-infrared light source is turned off, and selecting an environment area matching each target stain area from all reference background images as a matching environment reference area corresponding to each target stain area.

[0101] The preset shooting angle can be a shooting angle different from the target shooting angle. The number of preset shooting angles can be preset, which can be 6. The preset shooting angle can be a different shooting angle obtained by tilting the camera. The reference background image can be an image of the to-be-detected area under different preset shooting angles when the near-infrared light source is turned off.

[0102] As an example, the step can include the following steps:

[0103] Step 1, obtaining a reference background image of the to-be-detected area under a preset shooting angle different from the target shooting angle when the near-infrared light source is turned off by a near-infrared sensitive camera.

[0104] Step 2, obtaining a reference target image of the to-be-detected area under a preset shooting angle different from the target shooting angle when the near-infrared light source is turned on by a near-infrared sensitive camera.

[0105] The reference target image can be an image of the to-be-detected area under different preset shooting angles when the near-infrared light source is turned on.

[0106] Step 3, any one of the target stain areas in the initial target image is determined as a marked stain area.

[0107] Step 4, selecting an area in each reference target image with the same stain as the stain represented by the marked stain area as a reference matching area corresponding to the marked stain area by a SIFT (Scale-Invariant Feature Transform) algorithm.

[0108] Fifthly, the reference background image with the same shooting angle as the preset shooting angle of each reference matching area is determined as the matching background image corresponding to each reference matching area.

[0109] The matching background image corresponding to the reference matching area can be an image representing the ambient light condition with the same shooting angle as the reference matching area.

[0110] Sixthly, the region with the same position as each reference matching area is screened from the matching background image corresponding to each reference matching area as the matching ambient reference region corresponding to the marked stain region, and a plurality of matching ambient reference regions corresponding to the marked stain region are obtained.

[0111] The number of matching ambient reference regions can be the same as the number of reference matching areas. The matching ambient reference region corresponding to the marked stain region can represent the ambient light condition at different shooting angles at the marked stain region.

[0112] Step S5, determining the stain feature index corresponding to each target stain region according to the gray difference between the target ambient reference region corresponding to each target stain region and the matching ambient reference region, and the light transmission index corresponding to the pixel point in each target stain region.

[0113] It should be noted that there are usually various stains on the photovoltaic panel, and different cleaning methods need to be taken for different stains. Common stains usually include mirror stains, diffuse reflection stains and shielding stains. Mirror stains, such as bird droppings and dried mud stains, have smooth surfaces and usually appear as strong mirror reflection points, i.e. high light points, in images. Diffuse reflection stains, such as dust, have rough surfaces and no obvious high light points in images due to diffuse reflection effect. Shielding stains, such as fallen leaves and snow, block light and often cause a sharp decrease in light transmission.

[0114] As an example, the present step can include the following steps:

[0115] Firstly, the target ambient reference region corresponding to each target stain region and all matching ambient reference regions constitute a whole matching region set corresponding to each target stain region.

[0116] Secondly, the mean value of the gray values of all pixel points in each region in the whole matching region set corresponding to each target stain region is determined as the gray representative index, and a gray representative index set corresponding to each target stain region is obtained.

[0117] Thirdly, the target gray variation difference corresponding to each target stain region is determined according to the difference between different gray representative indexes in the gray representative index set corresponding to each target stain region. The target gray variation difference can include the following sub-steps:

[0118] The first sub-step involves identifying any one of the target stain areas in the initial target image as the marked stain area.

[0119] The second sub-step involves determining the absolute value of the difference between every two gray-level representative indicators in the set of gray-level representative indicators corresponding to the marked stain area as the local gray-level difference, thereby obtaining the set of local gray-level differences corresponding to the marked stain area.

[0120] The third sub-step is to determine the mean of all local grayscale differences in the set of local grayscale differences corresponding to the marked stain area as the target grayscale change difference corresponding to the marked stain area.

[0121] The fourth step is to determine the stain feature index corresponding to each target stain area based on the difference in target grayscale change corresponding to each target stain area and the average light transmittance of all pixels in each target stain area.

[0122] For example, the formula for determining the stain characteristic index corresponding to the target stain area can be:

[0123] ;

[0124] in, It is the first i The stain characteristic index corresponding to each target stained area. i It is the sequence number of the target stain region in the initial target image. It is a normalization function. It is the first i The difference in target grayscale variation corresponding to each target stain area. It is the first i The average transmittance of all pixels within a target stain area.

[0125] It should be noted that in practice, if a target stain area does not show significant grayscale changes before and after the near-infrared light source is turned on, that is, if the light transmittance is low and the grayscale changes between images at different shooting angles are small, then it is more likely to meet the characteristics of a masking stain. It can characterize the first i The grayscale changes of a target stain area between images taken from different shooting angles. It can characterize the first i The grayscale changes of a target stain area before and after the near-infrared light source is turned on. Therefore, when The smaller the size, the more likely it is to indicate the first i The more likely a target stained area is to be a masking stain. The larger the value, the more likely it is to indicate the first iThe more likely the stain represented by the target stain area is a mirror surface stain. When The more likely the stain represented by the target stain area is a diffuse reflection stain. i The more likely the stain represented by the target stain area is a diffuse reflection stain.

[0126] Step S6, adjusting the cleaning mode of the cleaning robot according to the stain characteristic index corresponding to each target stain area.

[0127] As an example, the present step can include the following steps:

[0128] First, if the stain characteristic index corresponding to the target stain area is less than or equal to a first preset stain threshold, the stain type represented by the target stain area is determined to be a concealing stain.

[0129] The first preset stain threshold can be a preset threshold, which can be 0.3.

[0130] Second, if the stain characteristic index corresponding to the target stain area is greater than the first preset stain threshold and less than or equal to a second preset stain threshold, the stain type represented by the target stain area is determined to be a diffuse reflection stain.

[0131] The second preset stain threshold can be a preset threshold, which can be greater than the first preset stain threshold. For example, the second preset stain threshold can be 0.6.

[0132] Third, if the stain characteristic index corresponding to the target stain area is greater than the second preset stain threshold, the stain type represented by the target stain area is determined to be a mirror surface stain.

[0133] Fourth, if the stain type represented by the target stain area is a concealing stain or a diffuse reflection stain, the cleaning mode when cleaning the stain represented by the target stain area can be set to wet brushing.

[0134] Fifth, if the stain type represented by the target stain area is a mirror surface stain, the cleaning mode when cleaning the stain represented by the target stain area can be set to dry brushing.

[0135] Optionally, in actual situations, the same stain area may need to be cleaned multiple times before it can be cleaned completely, and after each cleaning of the stain represented by the target stain area in the above to-be-detected area by the cleaning robot, the following steps can be included:

[0136] First, any one of the target stain areas in the initial target image is determined to be a marked stain area, and any one cleaning of the stain represented by the marked stain area is determined to be a marked cleaning.

[0137] Secondly, obtaining a mark cleaning background image of the to-be-detected area under the target shooting angle when the near-infrared light source is turned off and a mark cleaning target image when the near-infrared light source is turned on after the mark cleaning of the stains represented by the mark stain area.

[0138] The mark cleaning background image can be an image when the near-infrared light source is turned off. The mark cleaning target image can be an image when the near-infrared light source is turned on.

[0139] Thirdly, determining the light transmission index corresponding to each pixel point in the mark cleaning target image according to the gray scale difference between the pixel points at the same position in the mark cleaning background image and the mark cleaning target image.

[0140] It should be noted that the method for obtaining the light transmission index corresponding to the pixel point in the mark cleaning target image can be the same as the method for obtaining the light transmission index corresponding to the pixel point in the initial target image, which will not be repeated here.

[0141] Fourthly, segmenting the mark cleaning stain area from the mark cleaning target image according to the light transmission index corresponding to all the pixel points in the mark cleaning target image.

[0142] The mark cleaning stain area can represent the stains after the mark cleaning.

[0143] It should be noted that the method for obtaining the mark cleaning stain area can be the same as the method for obtaining the target stain area, which will not be repeated here.

[0144] Fifthly, screening the mark cleaning stain area with a non-empty intersection with the mark stain area from all the mark cleaning stain areas as a mark matching area, and determining the stain feature index corresponding to the mark matching area.

[0145] It should be noted that if there is no mark cleaning stain area with a non-empty intersection with the mark stain area in all the mark cleaning stain areas, it often means that the stains represented by the mark stain area have been cleaned after the mark cleaning and do not need to be cleaned again. The method for obtaining the stain feature index corresponding to the mark matching area can be the same as the method for obtaining the stain feature index corresponding to the target stain area, which will not be repeated here.

[0146] Sixthly, determining the cleaning effectiveness of the stains represented by the mark stain area after the mark cleaning according to the light transmission index change between the mark matching area and the mark stain area and the stain feature index change between the mark matching area and the mark stain area.

[0147] For example, the formula for determining the cleaning effectiveness of the stains represented by the mark stain area after the mark cleaning can be:

[0148] ;

[0149] wherein, v is the cleaning effectiveness of the marked stain after the marked cleaning. is a normalization function. is the mean of the light transmission index of all pixel points in the marked matching area. is the mean of the light transmission index of all pixel points in the marked stain area. is the stain characteristic index corresponding to the marked stain area. is the stain characteristic index corresponding to the marked matching area.

[0150] It should be noted that in actual situations, when cleaning the stain, if the cleaning method used matches the required cleaning method, the light transmission rate of the corresponding stain area will often recover significantly and tend to be consistent, resulting in a significant reduction in the corresponding stain characteristic index. can represent the light transmission rate recovery after the marked cleaning. can represent the stain characteristic index reduction. Therefore, when v is greater, it often means that the cleaning effectiveness of the marked cleaning is relatively better.

[0151] Step 7, if the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning is less than or equal to the preset cleaning effectiveness threshold, change the cleaning method to implement the next cleaning of the stain represented by the marked stain area after the marked cleaning.

[0152] wherein, the preset cleaning effectiveness threshold can be a pre-set threshold, which can be 0.4.

[0153] Step 8, if the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning is greater than the preset cleaning effectiveness threshold, use the same cleaning method as the marked cleaning to implement the next cleaning of the stain represented by the marked stain area after the marked cleaning.

[0154] It should be noted that if the stain represented by the target stain area is not completely cleaned after more than a preset number of cleaning times, the stain represented by the target stain area may not be a real surface stain, but a defect such as a scratch. At this time, cleaning can be stopped, and the worker is reminded to check the area. Wherein, the preset number can be a pre-set number, which can be 10.

[0155] Optionally, when the next cleaning of the stain represented by the marked stain area after the marked cleaning is performed using the same cleaning method as the marked cleaning, the method for obtaining the cleaning intensity can include the following steps:

[0156] The cleaning effectiveness after the previous cleaning before the marked cleaning of the stain represented by the marked stain area is determined as a reference pre-effectiveness.

[0157] It should be noted that the method for obtaining the cleaning effectiveness after the previous cleaning before the marked cleaning can be the same as the method for obtaining the cleaning effectiveness after the marked cleaning, which will not be described here.

[0158] The cleaning effectiveness after the marked cleaning of the stain represented by the marked stain area is determined as a reference marked effectiveness.

[0159] The intensity correction coefficient is determined according to the area of the marked matching area, the mean value of the light transmission index corresponding to all pixel points in the marked matching area, the difference between the reference marked effectiveness and the reference pre-effectiveness.

[0160] For example, the formula corresponding to the determination of the intensity correction coefficient can be:

[0161] ;

[0162] wherein, E is the intensity correction coefficient. is a normalization function. M is the area of the marked matching area. is an exponential function with a natural constant as the base. is the mean value of the light transmission index corresponding to all pixel points in the marked matching area. is the reference marked effectiveness. is the reference pre-effectiveness.

[0163] It should be noted that in actual situations, when the stain area is large and its light transmission rate interference to the photovoltaic panel is large, the cleaning effectiveness is poor, and the cleaning intensity should be increased. When M is larger, it means that the stain area after the marked cleaning is larger, which means that the cleaning intensity should be increased. When is smaller, it means that the stain after the marked cleaning has a greater interference to the light transmission rate of the photovoltaic panel, which means that the cleaning intensity should be increased. When is smaller, it means that the cleaning effectiveness after the marked cleaning is poor, which means that the cleaning intensity should be increased. Therefore, when E is larger, it means that the cleaning intensity should be increased.

[0164] The target cleaning intensity corresponding to the next cleaning of the marked cleaning is determined according to the intensity correction coefficient and the cleaning intensity in the marked cleaning process.

[0165] The cleaning intensity can be represented by the cleaning brush pressure during cleaning.

[0166] For example, the formula for determining the target cleaning intensity corresponding to the next cleaning of the marked cleaning can be:

[0167]

[0168] The target cleaning intensity corresponding to the next cleaning of the marked cleaning is represented by p The cleaning intensity during the marked cleaning process is represented by The maximum cleaning intensity allowed by the same cleaning mode as the marked cleaning is represented by The intensity correction coefficient is represented by E

[0169] In the fifth step, when the next cleaning of the marked stain area is performed, the cleaning intensity is adjusted to the target cleaning intensity.

[0170] It should be noted that when cleaning the stain represented by the target stain area, if the cleaning mode is frequently changed, it often indicates that the stain represented by the target stain area has been re-contaminated during the cleaning process, such as new dust or bird droppings falling, or the stain represented by the target stain area may not be a real surface stain, but a defect such as a scratch. At this time, the cleaning can be stopped and the staff can be reminded to check the area. If it is re-contamination, the cleaning mode can be re-determined to clean the stain represented by the target stain area. If it is a scratch or other defect, maintenance can be performed by the staff.

[0171] Optionally, the abnormality possibility of the marked stain area after the marked cleaning can include the following steps:

[0172] In the first step, the cleaning effectiveness of the marked stain area after the marked cleaning and all previous cleanings is constructed into a cleaning effectiveness sequence of the marked stain area after the marked cleaning.

[0173] The cleaning effectiveness sequence can be a time sequence.

[0174] In the second step, the difference between adjacent cleaning effectiveness in the cleaning effectiveness sequence is determined as the effectiveness difference, and an effectiveness difference sequence is obtained.

[0175] In the third step, the consecutive negative effectiveness differences in the effectiveness difference sequence are constructed into an initial difference sub-section.

[0176] In the fourth step, the longest initial difference sub-section is determined as the target difference sub-section.

[0177] ​​The fifth step is to determine the formula of the abnormal possibility of the marked stain area after the stain is marked and cleaned, which can be:

[0178] ;

[0179] wherein, w is the abnormal possibility of the marked stain area after the stain is marked and cleaned, and the greater the value, the more likely secondary pollution or multiple stains are superimposed in the cleaning process, and the more the marked stain area needs to be re-judged and cleaned. is a normalization function. is the number of elements in the target difference sub-section. is the number of elements in the validity difference sequence. is the cleaning validity of the marked stain area after the stain is cleaned for the first time. v is the cleaning validity of the marked stain area after the stain is marked and cleaned. d is the mean value of all elements in the target difference sub-section, and the greater the value, the greater the decline in cleaning effect. When is greater, the smaller the growth value of cleaning effectiveness.

[0180] Referring to Figure 2 , based on the same inventive concept as the above method embodiment, the present application provides a cleaning robot control system for a photovoltaic power station, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the above computer program is executed by the processor to realize the steps of a cleaning robot control method for a photovoltaic power station, which can specifically include:

[0181] The image acquisition module 201 is configured to acquire an initial background image of a to-be-detected area on the surface of the photovoltaic power station under a target shooting angle when a near-infrared light source is turned off and an initial target image when the near-infrared light source is turned on.

[0182] The light transmission index determination module 202 is configured to determine a light transmission index corresponding to each pixel point in the initial target image according to the gray difference between the pixel points at the same position in the initial background image and the initial target image.

[0183] The segmentation and screening module 203 is configured to segment a target stain area from the initial target image according to the light transmission index corresponding to all pixel points in the initial target image, and screen a target environment reference area corresponding to each target stain area from the initial background image.

[0184] The acquisition and screening module 204 is configured to acquire reference background images of the detection area under a preset shooting angle different from the target shooting angle when the near-infrared light source is turned off, and screen out, from all the reference background images, an environmental area matched with each target stain area as a matched environmental reference area corresponding to each target stain area.

[0185] The stain feature index determination module 205 is configured to determine a stain feature index corresponding to each target stain area according to the gray difference between the target environmental reference area corresponding to each target stain area and the matched environmental reference area, and the light transmission index corresponding to the pixel points in each target stain area.

[0186] The adjustment control module 206 is configured to adjust and control the cleaning mode of the cleaning robot according to the stain feature index corresponding to each target stain area.

[0187] Figure 3 is a structural schematic diagram of a computer device provided by an embodiment of the present application. As shown in the example, Figure 3 the computer device 300 includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302, wherein the processor 302 executes the computer program 303, so that the computer device can execute any one of the above-mentioned cleaning robot control methods for photovoltaic power stations.

[0188] Based on the same inventive concept as the above method embodiments, the present application provides a server including a memory and a processor. The memory is configured to store executable program code, and the processor is configured to call and run the executable program code from the memory, so that the device executes any one of the above-mentioned cleaning robot control methods for photovoltaic power stations.

[0189] Based on the same inventive concept as the above method embodiments, the present application provides a computer program product, which includes computer program code. When the computer program code runs on a computer, it makes the computer execute any one of the above-mentioned cleaning robot control methods for photovoltaic power stations.

[0190] Based on the same inventive concept as the above method embodiments, the present application provides a computer readable storage medium, which stores computer program code. When the computer program code runs on a computer, it makes the computer execute any one of the above-mentioned cleaning robot control methods for photovoltaic power stations.

[0191] To sum up, based on the initial background image, the initial target image and the reference background image, the application analyzes multiple factors related to the stain type differentiation, such as the gray difference between the target environment reference area and the matching environment reference area, and the light transmission index, thereby quantifying the stain characteristic index representing the stain type, and adjusting and controlling the cleaning mode of the cleaning robot based on the stain characteristic index corresponding to the target stain area, thereby improving the rationality of the adjustment and control of the cleaning mode of the cleaning robot.

[0192] The above examples are only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A cleaning robot control method for a photovoltaic power plant, characterized by, The method comprises the following steps: acquiring an initial background image of a to-be-detected area on a surface of a photovoltaic power station under a target shooting angle when a near-infrared light source is turned off and an initial target image when the near-infrared light source is turned on; determining a light transmission index corresponding to each pixel point in the initial target image according to the gray difference between the pixel points at the same position in the initial background image and the initial target image; segmenting a target stain area from the initial target image according to the light transmission index corresponding to all the pixel points in the initial target image, and screening a target environment reference area corresponding to each target stain area from the initial background image; acquiring a reference background image of the to-be-detected area under a preset shooting angle different from the target shooting angle when the near-infrared light source is turned off, and screening an environment area matching each target stain area from all the reference background images as a matching environment reference area corresponding to each target stain area; determining a stain characteristic index corresponding to each target stain area according to the gray difference between the target environment reference area and the matching environment reference area corresponding to each target stain area and the light transmission index corresponding to the pixel points in each target stain area; adjusting and controlling the cleaning mode of the cleaning robot according to the stain characteristic index corresponding to each target stain area.

2. The cleaning robot control method for a photovoltaic power station according to claim 1, characterized in that, After the stain represented by the target stain area in the to-be-detected area is cleaned by the cleaning robot each time, the method comprises the following steps: determining any one target stain area in the initial target image as a marked stain area, and determining any one cleaning of the stain represented by the marked stain area as a marked cleaning; acquiring a marked cleaning background image of the to-be-detected area under the target shooting angle when the near-infrared light source is turned off and a marked cleaning target image when the near-infrared light source is turned on after the stain represented by the marked stain area is cleaned by the marked cleaning; determining a light transmission index corresponding to each pixel point in the marked cleaning target image according to the gray difference between the pixel points at the same position in the marked cleaning background image and the marked cleaning target image; segmenting a marked cleaning stain area from the marked cleaning target image according to the light transmission index corresponding to all the pixel points in the marked cleaning target image; screening a marked matching area from all the marked cleaning stain areas, as an intersection between the marked matching area and the marked stain area is not empty, and determining a stain characteristic index corresponding to the marked matching area; determining the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning according to the change of the light transmission index between the marked matching area and the marked stain area and the change of the stain characteristic index between the marked matching area and the marked stain area; if the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning is less than or equal to a preset cleaning effectiveness threshold, changing the cleaning mode to implement the next cleaning of the stain represented by the marked stain area after the marked cleaning. If the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning is greater than a preset cleaning effectiveness threshold, the next cleaning of the stain represented by the marked stain area is performed by using the same cleaning mode as the marked cleaning.

3. The cleaning robot control method for a photovoltaic power station according to claim 2, wherein, If the next cleaning of the stain represented by the marked stain area is performed by using the same cleaning mode as the marked cleaning, a method for obtaining the cleaning intensity of the next cleaning includes: determining the cleaning effectiveness of the stain represented by the marked stain area after the previous cleaning of the marked cleaning as a reference previous effectiveness; determining the cleaning effectiveness of the stain represented by the marked stain area after the marked cleaning as a reference marked effectiveness; determining an intensity correction coefficient according to the area of the marked matching area, the mean value of the light transmission index corresponding to all pixel points in the marked matching area, and the difference between the reference marked effectiveness and the reference previous effectiveness; determining a target cleaning intensity corresponding to the next cleaning of the marked cleaning according to the intensity correction coefficient and the cleaning intensity in the marked cleaning process; adjusting the cleaning intensity to the target cleaning intensity when the next cleaning of the stain represented by the marked stain area is performed.

4. The cleaning robot control method for a photovoltaic power station according to claim 1, wherein, The method for determining the light transmission index corresponding to each pixel point in the initial target image according to the gray difference between the pixel points at the same position in the initial background image and the initial target image includes: determining any one pixel point in the initial target image as a marked pixel point, and screening a pixel point at the same position as the marked pixel point from the initial background image as a matching environment point corresponding to the marked pixel point; determining the difference between the gray value of the marked pixel point and the gray value of the matching environment point corresponding to the marked pixel point as the light transmission index corresponding to the marked pixel point.

5. The cleaning robot control method for a photovoltaic power station according to claim 1, wherein, The method for segmenting the target stain area from the initial target image according to the light transmission index corresponding to all pixel points in the initial target image includes: obtaining a standard background image when a standard photovoltaic panel is under a target shooting angle and a near-infrared light source is turned off, and a standard target image when the near-infrared light source is turned on; determining the light transmission index corresponding to each pixel point in the standard target image according to the gray difference between the pixel points at the same position in the standard background image and the standard target image; determining the mean value of the light transmission index corresponding to all pixel points in the standard target image as a no-stain representative index; determining the absolute value of the difference between the light transmission index corresponding to each pixel point in the initial target image and the no-stain representative index as the index deviation degree corresponding to each pixel point in the initial target image; if the index deviation degree corresponding to the pixel point in the initial target image is greater than a preset deviation anomaly threshold, determining the pixel point in the initial target image as an abnormal light transmission point; performing connected domain extraction on the area formed by all abnormal light transmission points in the initial target image, and determining the extracted connected domain as a target stain area.

6. The cleaning robot control method for a photovoltaic power station according to claim 1, wherein, The method for screening a target environment reference area corresponding to each target stain area from the initial background image includes: Determine any one target stain area in the initial target image as a marked stain area, and screen an area in the initial background image with the same position as the marked stain area as a target environment reference area corresponding to the marked stain area.

7. The cleaning robot control method for a photovoltaic power station according to claim 1, wherein, The screening of the environment area matching each target stain area from all the reference background images as the matching environment reference area corresponding to each target stain area comprises: Obtain a reference target image of the to-be-detected area under a preset shooting angle different from the target shooting angle and when the near-infrared light source is turned on; Determine any one target stain area in the initial target image as a marked stain area; Screen an area in each reference target image with the same stain as the stain represented by the marked stain area as a reference matching area corresponding to the marked stain area by using the SIFT algorithm; Determine a reference background image with the same preset shooting angle as each reference matching area as a matching background image corresponding to each reference matching area; Screen an area in the matching background image corresponding to each reference matching area with the same position as the reference matching area as the matching environment reference area corresponding to the marked stain area.

8. The cleaning robot control method for a photovoltaic power station according to claim 1, wherein, The determination of the stain feature index corresponding to each target stain area according to the gray difference between the target environment reference area and the matching environment reference area corresponding to each target stain area and the light transmission index corresponding to the pixel points in each target stain area comprises: Construct a whole matching area set corresponding to each target stain area by using the target environment reference area corresponding to each target stain area and all the matching environment reference areas; Determine the mean value of the gray values of all the pixel points in each area in the whole matching area set corresponding to each target stain area as a gray representative index to obtain a gray representative index set corresponding to each target stain area; Determine a target gray change difference corresponding to each target stain area according to the difference between different gray representative indexes in the gray representative index set corresponding to each target stain area; Determine the stain feature index corresponding to each target stain area according to the target gray change difference corresponding to each target stain area and the mean value of the light transmission indexes of all the pixel points in each target stain area.

9. The cleaning robot control method for a photovoltaic power station according to claim 8, wherein, The determination of the target gray change difference corresponding to each target stain area according to the difference between different gray representative indexes in the gray representative index set corresponding to each target stain area comprises: Determine any one target stain area in the initial target image as a marked stain area; Determine the absolute value of the difference between each two gray representative indexes in the gray representative index set corresponding to the marked stain area as a local gray difference to obtain a local gray difference set corresponding to the marked stain area; Determine the mean value of all the local gray differences in the local gray difference set corresponding to the marked stain area as the target gray change difference corresponding to the marked stain area.

10. A cleaning robot control system for a photovoltaic power plant, characterized in that A control method for a cleaning robot of a photovoltaic power station is provided, including a processor and a memory, the processor being configured to process instructions stored in the memory to implement the control method.

Images