Photovoltaic panel dust cleaning robot based on infrared correlation sensor and cleaning method

Abstract

The invention provides a photovoltaic panel dust cleaning robot and method based on infrared correlation sensors, and the method comprises the steps: receiving monitoring signals of a plurality of infrared correlation sensors of the cleaning robot in real time, and judging the position of the cleaning robot on a cleaned photovoltaic panel; the electric quantity of a storage battery in the cleaning robot is monitored in real time, and then the photovoltaic panel surface cleanliness is evaluated by selecting to return to a cabin or a center corrosion method based on an OTSU method and K-means improvement; further controlling the cleaning robot to perform conventional cleaning or stubborn dirt back cleaning on the photovoltaic panel; during conventional cleaning or stubborn dirt cleaning, the air spraying and sucking conditions of the suction cups of the positioning mechanism and the moving mechanism are controlled according to the moving direction, moving positioning and re-adsorption are completed, and the suction cups are controlled to adsorb the photovoltaic panel with the optimal real-time adsorption pressure. Through an advanced image processing algorithm and an intelligent cleaning strategy, the automation level and the overall efficiency of photovoltaic panel cleaning are remarkably improved, and important technical support is provided for wide application of clean energy.

Description

Technical Field

[0001] The invention belongs to the technical field of photovoltaic panel cleaning robots, and in particular relates to a photovoltaic panel dust cleaning robot and a cleaning method based on an infrared counter-radiation sensor. Background Art

[0002] With the growth of global energy demand and the enhancement of environmental awareness, photovoltaic power generation has been widely used as a clean and renewable energy form. However, in the actual operation of photovoltaic power stations, pollutants such as dust, dirt, and bird droppings on the surface of photovoltaic panels will significantly affect the photoelectric conversion efficiency of photovoltaic panels, resulting in a decrease in power generation. Therefore, regular cleaning and maintenance of photovoltaic panels has become an important part of ensuring the efficient operation of photovoltaic power stations.

[0003] At present, the cleaning methods of photovoltaic panels mainly include manual cleaning, mechanical cleaning and automatic cleaning robots. Among them, manual cleaning is inefficient and costly, and it is difficult to meet the actual needs in large-scale photovoltaic power stations; although traditional mechanical cleaning equipment has improved cleaning efficiency, its mobile mechanism (such as wheeled or crawler type) is easy to cause wear and scratches on the photovoltaic panel when running on the surface of the photovoltaic panel, affecting the service life of the photovoltaic panel.

[0004] To solve these problems, automated cleaning robots have gradually become the mainstream of photovoltaic panel cleaning. However, existing automated cleaning robots still have the following technical difficulties:

[0005] 1) Insufficient mobility stability: On high-angle, wet or smooth photovoltaic panel surfaces, traditional wheeled, tracked or sliding mobile mechanisms (such as the Chinese patent application with application number 202111191376.1) are difficult to provide sufficient friction, which can easily cause the robot to slip or fall, affecting the continuity and safety of the cleaning task. 2) Potential damage to the surface of photovoltaic panels: During the movement and cleaning process of traditional cleaning robots, the mechanical contact parts may cause a certain degree of wear or scratches on the surface of photovoltaic panels, thereby reducing the service life and power generation efficiency of photovoltaic panels.

[0006] In addition, existing photovoltaic panel cleaning robots such as application number 202310400352.5 cannot judge the cleanliness based on the condition of stains when using machine vision, and then choose conventional cleaning methods or perform targeted repeated cleaning back and forth on a certain photovoltaic panel. Summary of the invention

[0007] In view of the above-mentioned defects, the present invention provides a photovoltaic panel dust cleaning robot and cleaning method based on an infrared counter-radiation sensor. The cleaning robot provided by the present invention adopts a positioning mechanism and a moving mechanism that can alternately perform vacuum adsorption to realize the movement of the cleaning robot, which provides the possibility for flexible adjustment of the cleaning path. The surface cleanliness of the photovoltaic panel is evaluated by the center corrosion method based on the OTSU method and the improved K-means, so as to judge whether conventional cleaning or stubborn dirt return cleaning should be adopted. The stubborn bones are accurately located during the stubborn dirt return cleaning process, and then the cleaning robot is controlled to accurately move to the vicinity of the stubborn stains and repeat left and right movements n times. The robot can automatically adjust the cleaning route according to the actual dirt distribution, avoid redundant cleaning, improve the cleaning efficiency, and reduce the operating energy consumption.

[0008] The present invention provides the following technical solution: a cleaning method of a photovoltaic panel dust cleaning robot based on an infrared counter-radiation sensor, wherein the method evaluates the surface cleanliness of the photovoltaic panel based on the OTSU method and the K-means improved central corrosion method, and then plans the cleaning path of the cleaning robot, comprising the following steps:

[0009] S1: receiving monitoring signals from the first infrared radiation sensor to the fourth infrared radiation sensor of the cleaning robot in real time, and determining the position of the cleaning robot on the photovoltaic panel to be cleaned;

[0010] S2: monitor the power of the battery inside the cleaning robot in real time. If the remaining power is insufficient or the entire photovoltaic panel has been cleaned, control the cleaning robot to return to the cabin; if the remaining power is sufficient and the entire photovoltaic panel has not been cleaned, further evaluate the surface cleanliness of the photovoltaic panel based on the OTSU method and the K-means improved central corrosion method;

[0011] S3: According to the cleanliness of the photovoltaic panel surface, if it is higher than the cleanliness threshold, the cleaning robot is controlled to perform regular cleaning on the photovoltaic panel row by row; if it is lower than the cleanliness threshold, the cleaning robot is controlled to perform back cleaning of stubborn dirt on the photovoltaic panel;

[0012] S4: When the cleaning robot performs routine cleaning or stubborn cleaning of the photovoltaic panel in step S2 or step S3, the spraying and suction of the suction cups included in the positioning mechanism and the moving mechanism are controlled respectively according to the moving directions of the two to complete the moving positioning and re-adsorption of the cleaning robot, and the optimal real-time adsorption pressure of the four positioning suction cups of the positioning mechanism and the four moving suction cups of the moving mechanism of the cleaning robot on the photovoltaic panel during the moving positioning and re-adsorption process is calculated, and the suction cups are controlled to adsorb the photovoltaic panel with the optimal real-time adsorption pressure.

[0013] Furthermore, the standard for judging in step S2 that the cleaning robot has cleaned the entire photovoltaic panel is: when the fourth infrared counter-radiation sensor located on the right side of the cleaning robot frame and the second infrared counter-radiation sensor at its rear end both detect signals, it proves that the cleaning robot has completed the cleaning of the photovoltaic panel.

[0014] Furthermore, the step S2 includes the following steps:

[0015] S21: Use median filtering to remove image noise and isolated points;

[0016] S22: Convert the image from RGB color space to Lab color space;

[0017] S23: extract ab components;

[0018] S24: Calculate the optimal threshold for binary segmentation of the OTSU image, and cluster the ab components using the improved K-means algorithm according to the optimal threshold;

[0019] S25: Evaluate the surface cleanliness of photovoltaic panels based on clustering results;

[0020] The step S22 of converting the image from the RGB color space to the Lab color space comprises the following steps:

[0021] S221: Convert RGB image to XYZ color space:

[0022]

[0023] Among them, X, Y and Z are the X component value, Y component value and Z component value of the XYZ color space respectively; R, G and B are the red channel value, green channel value and blue channel value of the image in the RGB color space respectively;

[0024] S222: Further convert each component of the XYZ color space to the Lab color space:

[0025] L * =116×f(Y)-16;

[0026] a * =500×(f(X)-f(Y));

[0027] b * =200×(f(Y)-f(Z));

[0028] Among them, L * Represents the color change from pure black to pure white, L * ∈; a * represents the color change from red to green, a *∈[127, -128]; b * The amount represents the color change from yellow to blue, b * ∈[127, -128]; f(t) is the color conversion nonlinear coefficient function about the independent variable t, and the independent variable t = X, Y or Z;

[0029] The calculation formula of the color conversion nonlinear coefficient function f(t) about the independent variable t is as follows:

[0030]

[0031] Furthermore, the step S24 includes the following steps:

[0032] S241: Setting an initial threshold k, k∈[0, 255], where k is the grayscale level of the kth image pixel in the range of [0, 255];

[0033] S242: When the initial threshold k is used to classify the image pixels, the probability of classifying them into class A is set to p A (k), the probability of classification into class B is p B (k), the average grayscale of the pixels assigned to A is m A (k), the average grayscale of the pixels assigned to B is m B (k):

[0034] p i is the probability of the pixel grayscale being i in the image, n i is the number of pixels with grayscale i in the image;

[0035]

[0036] S243: Construct the probability average grayscale coupling value of class A pixels and class B pixels and the grayscale value m of the entire image G The equilibrium equation is:

[0037] p A (k)*m A (k)+p B (k)*m B (k) = m G ;

[0038] Among them, p A (k)+p B (k) = 1;

[0039] S244: Construct the average gray level m of class A pixels A (k) and the average gray level m of class B pixels B(k) is the coupling variance σ with the gray value of the entire image 2 Calculation formula:

[0040] σ 2 =p A (k)(m A (k)-m G ) 2 +p B (k)(m B (k)-m G ) 2 ;

[0041] S245: Substitute the equilibrium equation constructed in step S243 into the coupling variance σ constructed in step S244 2 In the calculation formula, we get σ 2 =p A (k)p B (m A (k)-m B (k)) 2 ;

[0042] S246: Further calculate the cumulative mean m of k:

[0043] The simplified coupling variance σ is obtained 2 :

[0044]

[0045] S247: Let k traverse 0-255, and solve the coupling variance σ in the grayscale value of 0-255 2 The maximum gray value is the optimal threshold for OTSU image binary segmentation.

[0046] Furthermore, the conventional cleaning in step S3 includes the following steps:

[0047] S301: Control the cleaning robot to move longitudinally;

[0048] S302: judging whether the infrared signal at the front end is abnormal according to the infrared signals at the front end, the rear end, the left end and the right end obtained by real-time monitoring, and if not, controlling the cleaning robot to continue to move longitudinally; if yes, controlling the cleaning robot to move laterally;

[0049] S303: During the lateral movement of the cleaning robot, further determining whether the infrared signal at the left end or the rear end is abnormal, if so, controlling the cleaning robot to move downward by one vehicle body; if not, continuing to control the cleaning robot to move lateral;

[0050] S304: After controlling the cleaning robot to move one vehicle body downward, further determine whether the infrared signal at the rear end is abnormal. If not, control the cleaning robot to continue cleaning horizontally; if so, control the cleaning robot to return to the starting point to complete the routine cleaning of a photovoltaic panel.

[0051] Furthermore, the stubborn dirt recovery in step S3 includes the following steps:

[0052] S311: Identify whether there is dirt in the area where the cleaning robot is located according to the image collected by the camera; if so, control the cleaning robot to move one body longitudinally; if not, control the cleaning robot to clean the photovoltaic panel in a conventional cleaning mode;

[0053] S312: After the cleaning robot moves one car body longitudinally, the cleaning robot is further controlled to try to move rightward, and it is determined whether each infrared signal is normal. If so, the cleaning robot is controlled to move one car body; if not, the cleaning robot is controlled to return to the recognition area and move one car body to the left, and then controlled to move one car body to the right;

[0054] S313: Further determine whether it is manually set to control the cleaning robot to repeatedly move left and right n times and then return to its current position. If not, execute the S312 step; if so, further identify whether there is still dirt at the location of the cleaning robot; if not, control the cleaning robot to clean the photovoltaic panel in a conventional cleaning mode; if so, upload the special details of the area to the host computer.

[0055] Furthermore, in the step S4, the suction cups of the positioning mechanism and the moving mechanism are both adsorbed on the photovoltaic panel, and when the cleaning robot needs to be moved, the following steps are included:

[0056] S401: Controlling the four positioning suction cups of the positioning mechanism to eject air, so that the four positioning suction cups of the positioning mechanism are lifted upward;

[0057] S402: After controlling the positioning mechanism to move to the target point, controlling the four positioning suction cups of the positioning mechanism to fall downward at the first target point and absorb air, so that the four positioning suction cups of the positioning mechanism are adsorbed on the photovoltaic panel again;

[0058] S403: Control the four movable suction cups of the movable mechanism to eject air, so that the four movable suction cups of the movable mechanism are lifted upward;

[0059] S404: After controlling the mobile mechanism to move to the target point, the four mobile suction cups of the mobile mechanism are controlled to fall down and absorb air at the second target point position, so that the four mobile suction cups of the mobile mechanism are adsorbed on the photovoltaic panel again; thereby completing the moving positioning and re-adsorption of the cleaning robot.

[0060] Furthermore, the step S4 calculates the optimal real-time adsorption pressure of the four positioning suction cups of the positioning mechanism of the cleaning robot and the four moving suction cups of the moving mechanism on the photovoltaic panel during the moving positioning and re-adsorption process, including the following steps:

[0061] S411: Construct the relationship equation group between the photovoltaic panel inclination angle θ, the mass m of the cleaning robot and the real-time adsorption pressure P of the suction cup:

[0062]

[0063] Where g is the acceleration of gravity, μ is the friction coefficient of the suction cup, N is the support force of the photovoltaic panel on the suction cup, N * is the adsorption force of the suction cup on the photovoltaic panel, P is the adsorption pressure of the suction cup on the photovoltaic panel, and S is the contact area between the suction cup and the inclined surface;

[0064] S412: According to the relationship equation group of step S411, it can be obtained:

[0065] mg sinθ=μmg cosθ+μPS;

[0066] Among them, P(m, θ) is the adsorption force function according to the mass m of the cleaning robot and the inclination angle θ of the photovoltaic panel;

[0067] S413: Let P(m, θ) = P max , where P max is the maximum adsorption force; the maximum cleaning robot mass m that the suction cups of the positioning mechanism and the moving mechanism can bear is obtained max :

[0068]

[0069] S414: Based on the maximum cleaning robot mass m max , substituted into the adsorption force function P(m, θ), the optimal real-time adsorption pressure p of the cleaning robot is obtained:

[0070]

[0071] The present invention also provides a photovoltaic panel dust cleaning robot based on an infrared counter-radiation sensor for executing the method as described above, the robot comprising a square bottom frame, the robot further comprising a first infrared counter-radiation sensor, a second infrared counter-radiation sensor, a third infrared counter-radiation sensor and a fourth infrared counter-radiation sensor respectively arranged on the front end, rear end, left end and right end frames of the square bottom frame; four first gas pipelines for providing gas to four moving suction cups of the moving mechanism are arranged on the inner edge of the square bottom frame, the moving mechanism comprises four positioning blocks fixed at four vertices on the inner side of the square bottom frame, and a lead screw is arranged between two of the positioning blocks; a moving lead screw that moves left and right is arranged on the upper part of the square bottom frame, and a positioning mechanism that can move forward and backward is arranged on the moving lead screw; the front end and the rear end of the moving lead screw move on the front lead screw and the rear lead screw respectively through the front moving block and the rear moving block in turn;

[0072] Four positioning suction cups are arranged at the four vertices below the positioning mechanism, and four second gas pipelines corresponding to the four positioning suction cups are arranged at the four vertices above the dormitory positioning mechanism;

[0073] A cleaning roller brush and a roller brush motor are also provided at the lower part of the frame where the moving mechanism is located away from the center point; the cleaning robot also includes a host computer for executing any method as claimed in claim 1.

[0074] Furthermore, a square top frame is provided on the upper part of the square bottom frame, and a first electromagnetic valve, a second electromagnetic valve, a third electromagnetic valve and a fourth electromagnetic valve are provided on the top frame, wherein the first electromagnetic valve and the third electromagnetic valve are both connected to the first electric air pump in communication; the second electromagnetic valve and the fourth electromagnetic valve are both connected to the second electric air pump in communication;

[0075] The square top frame is also provided with a camera for real-time monitoring of the image information of the surface of the cleaned photovoltaic panel, and the camera is a V7670 camera.

[0076] The beneficial effects of the present invention are:

[0077] 1. The present invention uses the OV7670 camera as the core of machine vision acquisition, which can obtain the surface image of the photovoltaic panel in real time, realize the adaptive binary segmentation of the image through the OTSU method, accurately calculate the optimal threshold, quickly distinguish the dirt and background on the surface of the photovoltaic panel, and greatly improve the accuracy and efficiency of dirt detection. Combined with repeated central corrosion, the target area is further processed to ensure that the dirt and stubborn stains on the surface of the photovoltaic panel can be accurately identified. The algorithm gradually reduces the area of ​​interest, eliminates noise interference, improves the robustness of target recognition, and provides a reliable basis for the planning of the cleaning path.

[0078] 2. The cleaning robot provided by the present invention adopts a vacuum alternating adsorption mechanism composed of a positioning mechanism with four suction cups and a moving mechanism with four suction cups combined with an intelligent control system, which can realize highly automated photovoltaic panel cleaning operations. The robot can efficiently complete complex cleaning tasks without human intervention, thereby reducing labor costs and improving the overall operation and maintenance efficiency of photovoltaic power stations.

[0079] 3. The present invention adopts a vacuum alternating adsorption positioning and moving mechanism, which makes the photovoltaic panel cleaning robot excellent in stability, flexibility and cleaning efficiency. This innovative design not only improves the cleaning effect of photovoltaic panels, but also effectively extends the service life of the equipment, providing strong technical support for the intelligent operation and maintenance of photovoltaic power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] The present invention will be described in more detail below based on embodiments and with reference to the accompanying drawings, wherein:

[0081] Figure 1 It is a schematic diagram of the overall structure of a photovoltaic power plant composed of a cleaning robot of the present invention and photovoltaic panels to be cleaned;

[0082] Figure 2 Schematic diagram of the process of step S2 of the present invention;

[0083] Figure 3 This is a comparison chart of the recognition effects of the stubborn stains using the OTSU method and the K-means improved center corrosion method of the present invention;

[0084] Figure 4 This is a demonstration diagram of the dirt recognition effect after the OV7670 camera is used for binarization in the present invention;

[0085] Figure 5 A route map for controlling the cleaning robot to perform conventional cleaning;

[0086] Figure 6 This is a flow chart of a control method for controlling a cleaning robot to perform conventional cleaning according to the present invention;

[0087] Figure 7 A route map for controlling the cleaning robot to clean stubborn dirt;

[0088] Figure 8 This is a flow chart of a control method for controlling a cleaning robot to perform stubborn dirt cleaning according to the present invention;

[0089] Fig. 9 This is a force analysis diagram of the cleaning robot of the present invention when the four positioning suction cups are ejecting air and the moving suction cup is sucking air;

[0090] Fig.10This is a force analysis diagram of the cleaning robot of the present invention when the four positioning suction cups suck air and the moving suction cup blows air;

[0091] Fig.11 This is a side view of the overall structure of the cleaning robot of the present invention;

[0092] Fig.12 This is a top side view of the bottom frame structure of the cleaning robot of the present invention;

[0093] Fig.13 This is a bottom view of the bottom frame structure of the cleaning robot of the present invention. DETAILED DESCRIPTION

[0094] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0095] The present invention provides a cleaning method of a photovoltaic panel dust cleaning robot based on an infrared radiation sensor, such as Figure 1 As shown, the cleaning robot provided by the present invention moves and cleans between different fixed tilted frames of multiple photovoltaic panels by a mobile trolley, and the cleaning robot cleans multiple photovoltaic panels in one frame. The method evaluates the surface cleanliness of each cleaned photovoltaic panel on a fixed tilted frame based on the OTSU method and the K-means improved center corrosion method, and then plans the cleaning path of the cleaning robot, including the following steps:

[0096] S1: receiving monitoring signals from the first infrared radiation sensor to the fourth infrared radiation sensor of the cleaning robot in real time, and determining the position of the cleaning robot on the photovoltaic panel to be cleaned;

[0097] S2: Real-time monitoring of the power of the battery inside the cleaning robot. If the remaining power is insufficient or the entire photovoltaic panel has been cleaned, the cleaning robot is controlled to return to the cabin; if the remaining power is sufficient and the entire photovoltaic panel has not been cleaned, the surface cleanliness of the photovoltaic panel is further evaluated based on the OTSU method and the K-means improved center corrosion method;

[0098] S3: According to the surface cleanliness of the photovoltaic panel, if it is higher than the cleanliness threshold, the cleaning robot is controlled to perform regular cleaning on the photovoltaic panel row by row; if it is lower than the cleanliness threshold, the cleaning robot is controlled to perform back cleaning of stubborn dirt on the photovoltaic panel;

[0099] S4: When the cleaning robot performs routine cleaning or stubborn cleaning of the photovoltaic panel in step S2 or step S3, the spraying and suction of the suction cups included in the positioning mechanism and the moving mechanism are controlled respectively according to the moving directions of the two to complete the moving positioning and re-adsorption of the cleaning robot, and the optimal real-time adsorption pressure of the four positioning suction cups 34 of the positioning mechanism 3 and the four moving suction cups 20 of the moving mechanism 2 of the cleaning robot on the photovoltaic panel during the moving positioning and re-adsorption process is calculated, and the suction cups are controlled to adsorb the photovoltaic panel with the optimal real-time adsorption pressure.

[0100] The cleaning robot provided by the present invention has three modes: conventional cleaning, stubborn dirt cleaning, and insufficient remaining power returning to the cabin. It can cope with different situations and ensure efficient cleaning and autonomous charging. Through the vacuum alternating adsorption design, the positioning mechanism suction cup and the moving mechanism suction cup adsorb the surface of the photovoltaic panel in an alternating manner. When one side adsorbs, the other side is lifted to move, ensuring that the robot always runs stably on the photovoltaic panel. This method effectively prevents sliding, overturning or falling off that may be caused by traditional movement methods, especially when the photovoltaic panel has a large inclination angle or a smooth surface, and has higher reliability.

[0101] In addition, the cleaning robot provided by the present invention has a flexible and efficient moving mechanism and an alternating adsorption method, which enables the robot to achieve smooth and continuous movement on the surface of the photovoltaic panel. Compared with the traditional crawler or wheel mechanism, this design reduces the mechanical friction on the surface, while avoiding the interference of complex terrain on the moving path, and improving the cleaning efficiency.

[0102] The standard for judging whether the cleaning robot has cleaned the entire photovoltaic panel in step S2 is: when the fourth infrared counter-radiation sensor 14 located on the right side of the cleaning robot frame and the second infrared counter-radiation sensor 12 at its rear end both detect signals, it proves that the cleaning robot has completed the cleaning of the photovoltaic panel, proving that the right side of the cleaning robot is in contact with the right side frame of the photovoltaic panel, and at the same time, the bottom end of the cleaning robot is in contact with the bottom side frame of the photovoltaic panel, and then the cleaning robot reaches the lower right corner of the photovoltaic panel, proving that the cleaning robot has completed the cleaning of the photovoltaic panel.

[0103] When the A-end sensor (i.e., the first infrared sensor 11 at the front end) has a signal, it indicates that it has reached the top of the panel. Then the cleaning robot is controlled to translate laterally. If the C-end sensor (i.e., the third infrared sensor 13 at the left end) or the D-end sensor (i.e., the fourth infrared sensor 14 at the right end) has a signal, it means that a row of cleaning is completed, and the operation is repeated. When the B-end sensor (i.e., the second infrared sensor 12 at the rear end) and the C-end (i.e., the third infrared sensor 13 at the left end) have signals at the same time, it proves that the cleaning robot has cleaned all the photovoltaic panels on a fixed tilt frame, and the cleaning robot is controlled to return to the cabin.

[0104] As another preferred embodiment of the present invention, Figure 2 As shown, step S2 includes the following steps:

[0105] S21: Use median filtering to remove image noise and isolated points;

[0106] S22: Convert the image from RGB color space to Lab color space;

[0107] S23: extract ab components;

[0108] S24: Calculate the optimal threshold for binary segmentation of the OTSU image, and cluster the ab components using the improved K-means algorithm according to the optimal threshold;

[0109] S25: Evaluate the surface cleanliness of photovoltaic panels based on clustering results;

[0110] The conversion of the image from the RGB color space to the Lab color space in step S22 includes the following steps:

[0111] S221: Convert RGB image to XYZ color space:

[0112]

[0113] Among them, X, Y and Z are the X component value, Y component value and Z component value of the XYZ color space respectively; R, G and B are the red channel value, green channel value and blue channel value of the image in the RGB color space respectively;

[0114] X component: represents the mixture of red (Red) and green (Green), which reflects the sensitivity of the human eye to red and green. The X component is the result of the mixture of red and green light, and it contains the information of red and green light. Y component: represents luminance, which reflects the human eye's perception of brightness. The Y component is the brightness information of the image, which contains the brightness and darkness information of the image, but does not contain color information. In image processing and display devices, the Y component is usually used to control the brightness of the image. Z component: represents the mixture of blue (Blue) and green (Green), which reflects the human eye's sensitivity to blue and green;

[0115] S222: Further convert each component of the XYZ color space to the Lab color space:

[0116] L * = 116 × f(Y) - 16;

[0117] a * = 500 × (f(X) - f(Y));

[0118] b * =200×(f(X)-f(Z));

[0119] Among them, L * Represents the color change from pure black to pure white, L * ∈[0, 100]; a * represents the color change from red to green, a * ∈[127, -128]; b * The amount represents the color change from yellow to blue, b * ∈[127, -128]; In OpenCV, you can use the cv2.cvtColor() function to convert an RGB image to the Lab color space, and then directly access the L, a, and b components; f(t) is the color conversion nonlinear coefficient function with respect to the independent variable t, and the independent variable t = X, Y, or Z.

[0120] The calculation formula of the color conversion nonlinear coefficient function f(t) with respect to the independent variable t is as follows:

[0121]

[0122] As another preferred embodiment of the present invention, step S24 includes the following steps:

[0123] S241: Setting an initial threshold k, k∈[0, 255], where k is the grayscale level of the kth image pixel in the range of [0, 255];

[0124] S242: When the initial threshold k is used to classify the image pixels, the probability of classifying them into class A is set to p A (k), the probability of classification into class B is p B (k), the average grayscale of the pixels assigned to A is m A (k), the average grayscale of the pixels assigned to B is m B (k):

[0125] p i is the probability of the pixel grayscale being i in the image, n i is the number of pixels with grayscale i in the image;

[0126]

[0127] S243: Construct the probability average grayscale coupling value of class A pixels and class B pixels and the grayscale value m of the entire image G The equilibrium equation is:

[0128] p A (k)*m A (k)+p B (k)*m B (k) = mG ;

[0129] Among them, p A (k)+p B (k) = 1;

[0130] S244: Construct the average gray level m of class A pixels A (k) and the average gray level m of class B pixels B (k) is the coupling variance σ with the gray value of the entire image 2 Calculation formula:

[0131] σ 2 =p A (k)(m A (k)-m G ) 2 +p B (k)(m B (k)-m G ) 2 ;

[0132] S245: Substitute the equilibrium equation constructed in S243 into the coupling variance σ constructed in step S244 2 In the calculation formula, we get σ 2 =p A (k)p B (k)(m A (k)-m B (k)) 2 ;

[0133] S246: Further calculate the cumulative mean m of k:

[0134] The simplified coupling variance σ is obtained 2 :

[0135]

[0136] S247: Let k traverse 0-255, and solve the coupling variance σ in the grayscale value of 0-255 2 The maximum gray value is the optimal threshold for OTSU image binary segmentation. Figure 3

[0137] like Figure 3 (a) The different colored tapes represent the stains on the surface of the photovoltaic panel with different colors. Figure 3(b) is the optimal threshold of OTSU image binarization segmentation adopted in the present invention. The image recognition result is obtained by clustering the ab components using the improved K-means algorithm according to the optimal threshold. It can be seen that the method of steps S241-S247 provided in the present invention can effectively identify stubborn stains on the surface of photovoltaic panels, and then give an accurate cleanliness evaluation.

[0138] Figure 4 This is a demonstration of the dirt recognition effect after the OV7670 camera is used for binarization. Figure 4 It can be seen that the center erosion method can make the target smaller during the image processing morphological operation. The OTSU image binarization segmentation is used to obtain the optimal threshold of K-means clustering, and the image is divided into background and foreground according to the grayscale characteristics, so that the inter-class variance is maximized and the probability of misclassification is minimized. After analysis, OTSU is suitable for photovoltaic panel surface contamination assessment. As a camera image segmentation threshold selection standard, we incorporate this improved k-means clustering algorithm into the center erosion method to enhance its boundaries.

[0139] The present invention uses a camera as the machine vision acquisition center. When the camera detects a dirt block, the system automatically switches to the stubborn dirt return path mode to deeply clean the stubborn stains. This automated path planning avoids redundant cleaning, improves cleaning efficiency, and reduces energy consumption. Secondly, it can perform real-time cleanliness assessment, and quantitatively evaluate the cleanliness of the photovoltaic panel surface through the combined application of OTSU binarization and center corrosion methods. This function can not only judge the cleaning effect, but also dynamically adjust the cleaning strategy to ensure that the photovoltaic panel surface is always in an efficient power generation state.

[0140] like Figure 5 , Figure 6 As shown, Figure 6 The A data, B data, C data and D data in the figure represent the front infrared signal data, rear infrared signal data, left infrared signal data and rear infrared signal data of the cleaning robot respectively. The conventional cleaning in step S3 includes the following steps:

[0141] S301: Control the cleaning robot to move longitudinally;

[0142] S302: judging whether the infrared signal at the front end is abnormal according to the infrared signals at the front end, the rear end, the left end and the right end obtained by real-time monitoring, and if not, controlling the cleaning robot to continue to move longitudinally; if yes, controlling the cleaning robot to move laterally;

[0143] S303: During the lateral movement of the cleaning robot, further determine whether the infrared signal at the left end or the rear end is abnormal. If so, control the cleaning robot to move downward by one vehicle body; if not, continue to control the cleaning robot to move lateral;

[0144] S304: After the cleaning robot is controlled to move one vehicle body downward, it is further determined whether the infrared signal at the rear end is abnormal. If not, the cleaning robot is controlled to continue cleaning horizontally; if so, the cleaning robot is controlled to return to the starting point to complete the routine cleaning of a photovoltaic panel.

[0145] Through routine cleaning, photovoltaic panels can be cleaned regularly and efficiently, effectively preventing corrosion and damage to the surface of photovoltaic panels caused by dirt accumulation, extending the service life of photovoltaic panels, reducing equipment maintenance costs, and providing more reliable operation guarantees for photovoltaic systems.

[0146] like Figure 7 , Figure 8 As shown, the stubborn dirt cleaning in step S3 includes the following steps:

[0147] S311: Identify whether there is dirt in the area where the cleaning robot is located based on the image collected by the camera; if so, control the cleaning robot to move one body longitudinally; if not, control the cleaning robot to clean the photovoltaic panel in a conventional cleaning mode;

[0148] S312: After the cleaning robot moves one car body longitudinally, the cleaning robot is further controlled to try to move rightward, and it is determined whether each infrared signal is normal. If so, the cleaning robot is controlled to move one car body; if not, the cleaning robot is controlled to return to the recognition area and move one car body to the left, and then controlled to move one car body to the right;

[0149] S313: Further determine whether the cleaning robot is manually set to repeat n left and right movements before returning to its current position. If not, execute step S312; if so, further identify whether there are still dirt blocks at the location of the cleaning robot; if not, control the cleaning robot to clean the photovoltaic panels in a conventional cleaning mode; if so, upload the special details of the area to the host computer.

[0150] As another preferred embodiment of the present invention, in step S4, the suction cups of the positioning mechanism and the moving mechanism are both adsorbed on the photovoltaic panel, and when the cleaning robot needs to be moved, the following steps are included:

[0151] S401: Fig. 9 As shown, the four positioning suction cups 34 of the positioning mechanism are controlled to eject air, so that the four positioning suction cups 34 of the positioning mechanism 3 are lifted upward;

[0152] S402: After the positioning mechanism 3 is controlled to move to the target point, Fig.10 As shown, the four positioning suction cups 34 of the positioning mechanism 3 are controlled to fall downward at the first target point and absorb air, so that the four positioning suction cups 34 of the positioning mechanism 3 are adsorbed on the photovoltaic panel again;

[0153] S403: Control the four moving suction cups 20 of the moving mechanism to spray air, so that the four moving suction cups 20 of the moving mechanism 2 are lifted upward;

[0154] S404: After controlling the mobile mechanism 2 to move to the target point, the four mobile suction cups 20 of the mobile mechanism are controlled to fall down and absorb air at the second target point position, so that the four mobile suction cups 20 of the mobile mechanism 2 are adsorbed on the photovoltaic panel again; thereby completing the moving positioning and re-adsorption of the cleaning robot.

[0155] In step S4, the optimal real-time adsorption pressure of the four positioning suction cups 34 of the positioning mechanism 3 of the cleaning robot and the four moving suction cups 20 of the moving mechanism 2 on the photovoltaic panel during the moving positioning and re-adsorption process is calculated, including the following steps:

[0156] S411: Construct the relationship equation group between the photovoltaic panel inclination angle θ, the mass m of the cleaning robot and the real-time adsorption pressure P of the suction cup:

[0157]

[0158] Where g is the acceleration of gravity, μ is the friction coefficient of the suction cup, N is the support force of the photovoltaic panel on the suction cup, N * is the adsorption force of the suction cup on the photovoltaic panel, P is the adsorption pressure of the suction cup on the photovoltaic panel, and S is the contact area between the suction cup and the inclined surface;

[0159] S412: According to the relationship equations in step S411, we can get:

[0160] mg sinθ=μmg cosθ+μPS;

[0161] Among them, P(m, θ) is the adsorption force function according to the mass m of the cleaning robot and the inclination angle θ of the photovoltaic panel;

[0162] S413: Let P(m, θ) = P max , where P max is the maximum adsorption force; the maximum cleaning robot mass m that the suction cups of the positioning mechanism and the moving mechanism can bear is obtained max :

[0163]

[0164] S414: Based on the maximum cleaning robot mass m max , substituted into the adsorption force function P(m, θ), the optimal real-time adsorption pressure p of the cleaning robot is obtained:

[0165]

[0166] The adsorption mechanism achieves positioning and movement through vacuum suction, avoiding direct contact between mechanical parts such as wheels and tracks on the surface of photovoltaic panels, reducing scratches or wear on photovoltaic panels during cleaning, and thus extending the service life of photovoltaic panels. Through the calculation of the real-time optimal adsorption force of the four suction cups of the positioning mechanism and the four suction cups of the moving mechanism, the robot can adapt to the complex application scenarios of various photovoltaic power stations, including high-angle, dusty, humid or windy environments, ensuring that during the adsorption and desorption process, the adsorption force on the photovoltaic panel is too large to cause loopholes or the previously attached part is driven away from the inclined surface of the fixed tilt frame during the desorption process, resulting in the tearing of the photovoltaic panel.

[0167] like Figure 11-13 As shown, the present invention also provides a photovoltaic panel dust cleaning robot based on infrared counter-radiation sensors for executing the above method, the robot comprises a square bottom frame 1, and the robot further comprises a first infrared counter-radiation sensor 11, a second infrared counter-radiation sensor 12, a third infrared counter-radiation sensor 13 and a fourth infrared counter-radiation sensor 14 respectively arranged on the front end, rear end, left end and right end frames of the square bottom frame 1; four first gas pipelines 21 for providing gas to four moving suction cups 20 of the moving mechanism 2 are arranged on the inner edge of the square bottom frame 1, The moving mechanism 2 includes four positioning blocks 22 fixed at four vertices of the inner side of the square bottom frame 1, and a lead screw 23 is arranged between two positioning blocks 22; a moving lead screw 31 that can move left and right is arranged on the upper part of the square bottom frame 1, and a positioning mechanism 3 that can move forward and backward is arranged on the moving lead screw 31; the front end and the rear end of the moving lead screw 31 are respectively moved on the inner side of the square bottom frame 1 at positions corresponding to the front and rear of the front lead screw through the front moving block 32 and the rear moving block 33 (not marked in the figure) respectively.

[0168] Four positioning suction cups 34 are arranged at the four vertices below the positioning mechanism 3, and four second gas pipelines 35 corresponding to the four positioning suction cups 34 are arranged at the four vertices above the dormitory positioning mechanism 3;

[0169] A cleaning roller brush 24 and a roller brush motor 25 are also provided at the lower part of the frame where the moving mechanism 2 is located away from the center point; the cleaning robot also includes a host computer for executing any method as claimed in claims 1-8.

[0170] A square top frame 4 is also provided on the upper part of the square bottom frame 1, and a first electromagnetic gas valve 51, a second electromagnetic gas valve 52, a third electromagnetic gas valve 53 and a fourth electromagnetic gas valve 54 are provided on the top frame 4. The first electromagnetic gas valve 51 and the third electromagnetic gas valve 53 are both connected to the first electric air pump 61 for communication and are used to control the first electric air pump 61 to pump gas into part of the first gas pipeline and part of the branch of the second gas pipeline. The first electromagnetic gas valve 51 is used to control the first electric air pump 61 to pump gas into the two second gas pipelines 35 on the left side of the positioning mechanism 3, and the third electromagnetic gas valve 53 is used to control the first electric air pump 61 to pump gas into the two second gas pipelines 35 on the left side of the positioning mechanism 3. The left branch pipe and the lower branch pipe of the square first gas pipeline 21 are used to pump gas; the second electromagnetic gas valve 52 and the fourth electromagnetic gas valve 54 are both connected to the second electric air pump 62 for communication, and are used to control the second electric air pump 62 to pump gas into part of the first gas pipeline and part of the branch of the second gas pipeline. The second electromagnetic gas valve 52 is used to control the second electric air pump 62 to pump gas into the two second gas pipelines 35 on the right side of the positioning mechanism 3, and the fourth electromagnetic gas valve 54 is used to control the second electric air pump 62 to pump gas into the right branch pipe and the upper branch pipe of the square first gas pipeline 21;

[0171] The square top frame 4 is also provided with a camera 7 for real-time monitoring of the image information of the surface of the cleaned photovoltaic panel. The camera is a V7670 camera, which serves as a core device for machine vision acquisition.

[0172] Compared with the traditional mechanical adsorption mechanism, the cleaning robot provided by the present invention reduces the energy consumption required for continuous adsorption through alternating adsorption. In a single cleaning task, the robot can clearly determine whether to adopt conventional cleaning or stubborn dirt return cleaning cleaning strategies, and achieve the same or even more efficient cleaning effect at a lower energy cost. In addition, compared with traditional manual inspection and cleaning methods, this design significantly reduces the need for manual intervention. Visual recognition and automated cleaning based on the OTSU method and the improved central corrosion method of K-means to evaluate the surface cleanliness of photovoltaic panels can operate stably in various complex environments with high cleaning accuracy, ensuring that photovoltaic panels can continue to work efficiently.

[0173] Although the present invention has been described with reference to preferred embodiments, various modifications may be made thereto and parts thereof may be replaced by equivalents without departing from the scope of the present invention. In particular, the various technical features mentioned in the various embodiments may be combined in any manner as long as there are no structural conflicts. The present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A cleaning method for a photovoltaic panel dust cleaning robot based on an infrared counter-radiation sensor, wherein the method evaluates the surface cleanliness of the photovoltaic panel based on the OTSU method and the K-means improved central corrosion method, and then plans the cleaning path of the cleaning robot, characterized in that: The following steps are involved: S1: receiving monitoring signals from the first infrared radiation sensor to the fourth infrared radiation sensor of the cleaning robot in real time, and determining the position of the cleaning robot on the photovoltaic panel to be cleaned; S2: monitor the power of the battery inside the cleaning robot in real time. If the remaining power is insufficient or the entire photovoltaic panel has been cleaned, control the cleaning robot to return to the cabin; if the remaining power is sufficient and the entire photovoltaic panel has not been cleaned, further evaluate the surface cleanliness of the photovoltaic panel based on the OTSU method and the K-means improved central corrosion method; S3: According to the cleanliness of the photovoltaic panel surface, if it is higher than the cleanliness threshold, the cleaning robot is controlled to perform regular cleaning on the photovoltaic panel row by row; if it is lower than the cleanliness threshold, the cleaning robot is controlled to perform back cleaning of stubborn dirt on the photovoltaic panel; S4: When the cleaning robot performs routine cleaning or stubborn cleaning of the photovoltaic panel in step S2 or step S3, the spraying and suction of the suction cups included in the positioning mechanism (3) and the moving mechanism (2) are controlled according to the moving directions of the positioning mechanism (3) and the moving mechanism (2), respectively, to complete the moving positioning and re-adsorption of the cleaning robot, and the optimal real-time adsorption pressure of the four positioning suction cups (34) of the positioning mechanism (3) and the four moving suction cups (20) of the moving mechanism (2) on the photovoltaic panel during the moving positioning and re-adsorption process is calculated, and the suction cups are controlled to adsorb the photovoltaic panel with the optimal real-time adsorption pressure.

2. The cleaning method of the photovoltaic panel dust cleaning robot based on the infrared radiation sensor according to claim 1 is characterized in that: The standard for judging whether the cleaning robot has finished cleaning the entire photovoltaic panel in step S2 is: when the fourth infrared counter-radiation sensor (14) located on the right side of the cleaning robot frame and the second infrared counter-radiation sensor (12) at its rear end both detect signals, it proves that the cleaning robot has completed the cleaning of the photovoltaic panel.

3. The cleaning method of the photovoltaic panel dust cleaning robot based on the infrared radiation sensor according to claim 1 is characterized in that: The S2 step includes the following steps: S21: Use median filtering to remove image noise and isolated points; S22: Convert the image from RGB color space to Lab color space; S23: extract ab components; S24: Calculate the optimal threshold for binary segmentation of the OTSU image, and cluster the ab components using the improved K-means algorithm according to the optimal threshold; S25: Evaluate the cleanliness of the photovoltaic panel surface based on the clustering results; The step S22 of converting the image from the RGB color space to the Lab color space comprises the following steps: S221: Convert RGB image to XYZ color space: Among them, X, Y and Z are the X component value, Y component value and Z component value of the XYZ color space respectively; R, G and B are the red channel value, green channel value and blue channel value of the image in the RGB color space respectively; S222: Further convert each component of the XYZ color space to the Lab color space: L*=116×f(Y)-16; a*=500×(f(X)-f(Y)); b*=200×(f(Y)-f(Z)); Wherein, L* represents the color change from pure black to pure white, L*∈[0,100]; a* represents the color change from red to green, a*∈[127,-128]; b* represents the color change from yellow to blue, b*∈[127,-128]; f(t) is the color conversion nonlinear coefficient function with respect to the independent variable t, the independent variable t=X,Y or Z; The calculation formula of the color conversion nonlinear coefficient function f(t) about the independent variable t is as follows:

4. The cleaning method of the photovoltaic panel dust cleaning robot based on the infrared radiation sensor according to claim 3 is characterized in that: The step S24 comprises the following steps: S241: Setting an initial threshold k, k∈[0, 255], where k is the grayscale level of the kth image pixel in the range of [0, 255]; S242: When the initial threshold k is used to classify the image pixels, the probability of classifying them into class A is set to p A (k), the probability of classification into class B is p B (k), the average grayscale of the pixels assigned to A is m A (k), the average grayscale of the pixels assigned to B is m B (k): p i is the probability of the pixel grayscale being i in the image, n i is the number of pixels with grayscale i in the image; S243: Construct the probability average grayscale coupling value of class A pixels and class B pixels and the grayscale value m of the entire image G The equilibrium equation is: p A (k)*m A (k)+p B (k)*m B (k)=m G ; Among them, p A (k)+p B (k) = 1; S244: Construct the average gray level m of class A pixels A (k) and the average gray level m of class B pixels B (k) is the coupling variance σ with the gray value of the entire image 2 Calculation formula: σ 2 =p A (k)(m A (k)-m G ) 2 +p B (k)(m B (k)-m G ) 2 ; S245: Substitute the equilibrium equation constructed in step S243 into the coupling variance σ constructed in step S244 2 In the calculation formula, we get σ 2 =p A (k)p B (k)(m A (k)-m B (k)) 2 ; S246: Further calculate the cumulative mean m of k: The simplified coupling variance σ is obtained 2 : S247: Let k traverse 0-255, and solve the coupling variance σ in the grayscale value of 0-255 2 The maximum gray value is the optimal threshold for OTSU image binary segmentation.

5. The cleaning method of the photovoltaic panel dust cleaning robot based on infrared radiation sensor according to claim 1 is characterized in that: The conventional cleaning in step S3 includes the following steps: S301: Control the cleaning robot to move longitudinally; S302: judging whether the infrared signal at the front end is abnormal according to the infrared signals at the front end, the rear end, the left end and the right end obtained by real-time monitoring, and if not, controlling the cleaning robot to continue to move longitudinally; if yes, controlling the cleaning robot to move laterally; S303: During the lateral movement of the cleaning robot, further determining whether the infrared signal at the left end or the rear end is abnormal, if so, controlling the cleaning robot to move downward by one vehicle body; if not, continuing to control the cleaning robot to move lateral; S304: After controlling the cleaning robot to move one vehicle body downward, further determine whether the infrared signal at the rear end is abnormal. If not, control the cleaning robot to continue cleaning horizontally; if so, control the cleaning robot to return to the starting point to complete the routine cleaning of a photovoltaic panel.

6. The cleaning method of the photovoltaic panel dust cleaning robot based on infrared radiation sensor according to claim 1 is characterized in that: The stubborn dirt cleaning in step S3 includes the following steps: S311: Identify whether there is dirt in the area where the cleaning robot is located according to the image collected by the camera; if so, control the cleaning robot to move one body longitudinally; if not, control the cleaning robot to clean the photovoltaic panel in a conventional cleaning mode; S312: After the cleaning robot moves one car body longitudinally, the cleaning robot is further controlled to try to move rightward, and it is determined whether each infrared signal is normal. If so, the cleaning robot is controlled to move one car body; if not, the cleaning robot is controlled to return to the recognition area and move one car body to the left, and then controlled to move one car body to the right; S313: Further determine whether it is manually set to control the cleaning robot to repeatedly move left and right n times and then return to its current position. If not, execute the S312 step; if so, further identify whether there is still dirt at the location of the cleaning robot; if not, control the cleaning robot to clean the photovoltaic panel in a conventional cleaning mode; if so, upload the special details of the area to the host computer.

7. The cleaning method of a photovoltaic panel dust cleaning robot based on an infrared radiation sensor according to claim 1, characterized in that: In the step S4, the suction cups of the positioning mechanism and the moving mechanism are both adsorbed on the photovoltaic panel. When the cleaning robot needs to be moved, the following steps are included: S401: Controlling the four positioning suction cups (34) of the positioning mechanism to eject air, so that the four positioning suction cups (34) of the positioning mechanism (3) are lifted upwards; S402: After controlling the positioning mechanism (3) to move to the target point, controlling the four positioning suction cups (34) of the positioning mechanism (3) to fall downward at the first target point and absorb air, so that the four positioning suction cups (34) of the positioning mechanism (3) are adsorbed on the photovoltaic panel again; S403: controlling the four movable suction cups (20) of the movable mechanism to eject air, so that the four movable suction cups (20) of the movable mechanism (2) are lifted upwards; S404: After controlling the mobile mechanism (2) to move to the target point, the four mobile suction cups (20) of the mobile mechanism are controlled to fall downward and absorb air at the second target point position, so that the four mobile suction cups (20) of the mobile mechanism (2) are adsorbed on the photovoltaic panel again; thereby completing the moving positioning and re-adsorption of the cleaning robot.

8. The cleaning method of photovoltaic panel dust cleaning robot based on infrared radiation sensor according to claim 1 is characterized in that: The step S4 calculates the optimal real-time adsorption pressure of the four positioning suction cups (34) of the positioning mechanism (3) of the cleaning robot and the four moving suction cups (20) of the moving mechanism (2) on the photovoltaic panel during the moving positioning and re-adsorption process, including the following steps: S411: Construct the relationship equation group between the photovoltaic panel inclination angle θ, the mass m of the cleaning robot and the real-time adsorption pressure P of the suction cup: Where g is the acceleration of gravity, μ is the friction coefficient of the suction cup, N is the support force of the photovoltaic panel on the suction cup, N * is the adsorption force of the suction cup on the photovoltaic panel, P is the adsorption pressure of the suction cup on the photovoltaic panel, and S is the contact area between the suction cup and the inclined surface; S412: According to the relationship equation group of step S411, it can be obtained: mg sinθ=μmgcosθ+μPS; Among them, P(m, θ) is the adsorption force function according to the mass m of the cleaning robot and the inclination angle θ of the photovoltaic panel; S413: Let P(m, θ) = P max , where P max is the maximum adsorption force; the maximum cleaning robot mass m that the suction cups of the positioning mechanism and the moving mechanism can bear is obtained max : S414: Based on the maximum cleaning robot mass m max , substituted into the adsorption force function P(m, θ), the optimal real-time adsorption pressure p of the cleaning robot is obtained:

9. A photovoltaic panel dust cleaning robot based on an infrared radiation sensor that performs the method according to any one of claims 1 to 8, the robot comprising a square bottom frame (1), characterized in that: The robot further comprises a first infrared radiation sensor (11), a second infrared radiation sensor (12), a third infrared radiation sensor (13) and a fourth infrared radiation sensor (14) respectively arranged at the front end, rear end, left end and right end of the frame of the square bottom frame (1); four first gas pipelines (21) are arranged on the inner edge of the square bottom frame (1) to supply gas to four moving suction cups (20) of the moving mechanism (2); the moving mechanism (2) comprises four positioning blocks (22) fixed at four vertices of the inner side of the frame of the square bottom frame (1), and a lead screw (23) is arranged between two positioning blocks (22); a moving lead screw (31) that can move left and right is arranged on the upper part of the square bottom frame (1), and a positioning mechanism (3) that can move forward and backward is arranged on the moving lead screw (31); the front end and the rear end of the moving lead screw (31) are moved on the front lead screw (231) and the rear lead screw (232) respectively through the front moving block (32) and the rear moving block (33); Four positioning suction cups (34) are arranged at the four vertices below the positioning mechanism (3), and four second gas pipelines (35) corresponding to the four positioning suction cups (34) are arranged at the four vertices above the dormitory positioning mechanism (3); A cleaning roller brush (24) and a roller brush motor (25) are also provided at the lower part of the frame where the moving mechanism (2) is located away from the center point; the cleaning robot also includes a host computer for executing any method as claimed in claims 1-8.

10. The photovoltaic panel dust cleaning robot based on infrared radiation sensor according to claim 9, characterized in that: A square top frame (4) is also provided on the upper part of the square bottom frame (1); a first electromagnetic valve (51), a second electromagnetic valve (52), a third electromagnetic valve (53) and a fourth electromagnetic valve (54) are provided on the top frame (4); the first electromagnetic valve (51) and the third electromagnetic valve (53) are both connected to the first electric air pump (61); the second electromagnetic valve (52) and the fourth electromagnetic valve (54) are both connected to the second electric air pump (62); The square top frame (4) is also provided with a camera (7) for real-time monitoring of image information on the surface of the cleaned photovoltaic panel, and the camera is a V7670 camera.

Images