Photovoltaic power generation automatic cleaning control system

Abstract

The application belongs to the field of photovoltaics, and discloses a photovoltaic power generation automatic cleaning control system, which comprises a UAV and a computing server; the computing server is used for dividing the distribution area of photovoltaic components into multiple patrol areas, and for calculating the patrol cycle of each patrol area; the UAV is used for respectively patrolling each patrol area based on the patrol cycle, and acquiring the surface image of the photovoltaic components in the patrol area; the computing server is further used for judging whether the photovoltaic components need to be cleaned based on the surface image.The application divides the distribution area of photovoltaic components into multiple patrol areas, and then uses the UAV to patrol the patrol areas, thereby improving the efficiency of patrolling the photovoltaic components, and enabling the photovoltaic components that need to be cleaned to be cleaned in time.

Description

Technical Field

[0001] This invention relates to the field of photovoltaics, and more particularly to an automatic cleaning control system for photovoltaic power generation. Background Technology

[0002] Photovoltaic power stations are typically located in relatively barren and uninhabited areas. In these areas, due to factors such as wind and sand, the photovoltaic modules in the power stations need to be cleaned periodically to improve power generation efficiency. Currently, the method of manually inspecting the photovoltaic modules that need cleaning is generally used. However, since photovoltaic power stations are usually very large, manual inspections are inefficient and cannot ensure that the modules requiring cleaning are cleaned in a timely manner. Summary of the Invention

[0003] In view of this, the purpose of this invention is to disclose an automatic cleaning control system for photovoltaic power generation, which solves the problem of how to improve the efficiency of detecting photovoltaic modules that need cleaning during the daily maintenance of photovoltaic modules.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] An automatic cleaning control system for photovoltaic power generation includes a drone and a computing server;

[0006] The computing server is used to divide the distribution area of ​​photovoltaic modules into multiple inspection areas, and to calculate the inspection cycle for each inspection area;

[0007] The drones are used to patrol each patrol area according to the patrol cycle and acquire surface images of the photovoltaic modules in the patrol area;

[0008] The computing server is also used to determine whether photovoltaic modules need cleaning based on surface images.

[0009] Optionally, the computing server includes a partitioning module, an image recognition module, and a communication module;

[0010] The partitioning module is used to divide the distribution area of ​​photovoltaic modules into multiple inspection areas and calculate the inspection cycle for each inspection area;

[0011] The communication module is used to send the patrol schedule to the drone;

[0012] The image recognition module is used to determine whether photovoltaic modules need cleaning based on surface images.

[0013] Optionally, the distribution area of ​​photovoltaic modules can be divided into multiple inspection areas, including:

[0014] Calculate the effective duration of the patrol area division;

[0015] The distribution area of ​​photovoltaic modules is divided into multiple inspection areas based on the effective duration of the inspection area division.

[0016] Optionally, the patrol cycle for each patrol area is calculated, including:

[0017] Calculate the patrol coefficient for each patrol area;

[0018] The patrol cycle for each patrol area is calculated based on the patrol coefficient.

[0019] Optionally, the computing server may also include a path planning module;

[0020] The path planning module is used to calculate the shortest patrol path for the patrol area;

[0021] The communication module is also used to send the shortest patrol path to the drone.

[0022] Optionally, drones can also be used to patrol the area based on the shortest patrol path.

[0023] Optionally, the image recognition module includes a pixel value optimization unit and an image recognition unit;

[0024] The pixel value optimization unit is used to perform optimization calculations on the surface image using a set optimization algorithm to obtain an optimized image;

[0025] The image recognition unit is used to identify the optimized image and determine whether the photovoltaic module needs cleaning.

[0026] Optionally, the image recognition unit includes a storage subunit and a comparison subunit;

[0027] The storage sub-unit is used to store standard images of photovoltaic modules that do not require cleaning;

[0028] The comparison sub-unit is used to compare the standard surface image that does not require cleaning with the surface image, and to determine whether the photovoltaic module needs cleaning based on the comparison result.

[0029] Optionally, an automatic cleaning module is also included.

[0030] The automatic cleaning module is used to clean the photovoltaic modules that need cleaning after receiving cleaning instructions from the computing server.

[0031] Optionally, the computing server may also include an instruction generation module;

[0032] The instruction generation module is used to obtain the location information of the photovoltaic modules that need to be cleaned, and generate cleaning instructions based on the location information;

[0033] The communication module is also used to send cleaning instructions to the automatic cleaning module.

[0034] This invention improves the efficiency of photovoltaic module inspection by dividing the distribution area of ​​photovoltaic modules into multiple inspection zones and then using drones to inspect these zones, thus enabling photovoltaic modules that need cleaning to be cleaned in a timely manner. Attached Figure Description

[0035] This disclosure will be more fully understood from the detailed description and accompanying drawings given below, which are given by way of illustration only and are not intended to limit the disclosure, and wherein:

[0036] Figure 1 This is a schematic diagram of an automatic cleaning control system for photovoltaic power generation according to the present invention.

[0037] Figure 2 This is another schematic diagram of an automatic cleaning control system for photovoltaic power generation according to the present invention. Detailed Implementation

[0038] In the following detailed description, numerous specific details are set forth for illustrative purposes in order to provide a thorough understanding of the disclosed embodiments. The concepts and features of the invention will be readily apparent to those skilled in the art from the specification, claims, and drawings disclosed herein. The following embodiments further illustrate various aspects of the invention but are not intended to limit the scope of the invention.

[0039] like Figure 1 As shown in one embodiment, the present invention provides an automatic cleaning control system for photovoltaic power generation, including a drone and a computing server;

[0040] The computing server is used to divide the distribution area of ​​photovoltaic modules into multiple inspection areas, and to calculate the inspection cycle for each inspection area;

[0041] The drones are used to patrol each patrol area according to the patrol cycle and acquire surface images of the photovoltaic modules in the patrol area;

[0042] The computing server is also used to determine whether photovoltaic modules need cleaning based on surface images.

[0043] This invention improves the efficiency of photovoltaic module inspection by dividing the distribution area of ​​photovoltaic modules into multiple inspection zones and then using drones to inspect these zones, thus enabling photovoltaic modules that need cleaning to be cleaned in a timely manner.

[0044] Because drones fly much faster than pedestrians and can capture images of photovoltaic modules from a higher vantage point with better visibility, inspection efficiency is improved.

[0045] Optionally, the computing server includes a partitioning module, an image recognition module, and a communication module;

[0046] The partitioning module is used to divide the distribution area of ​​photovoltaic modules into multiple inspection areas and calculate the inspection cycle for each inspection area;

[0047] The communication module is used to send the patrol schedule to the drone;

[0048] The image recognition module is used to determine whether photovoltaic modules need cleaning based on surface images.

[0049] Dividing the distribution area allows photovoltaic modules with similar cleanliness to be grouped into the same inspection zone. This improves the efficiency of drones in identifying photovoltaic modules requiring cleaning when inspecting this area, and also extends the service life of the drones. Because the inspection cycle in this invention is related to the condition of the photovoltaic modules, if the photovoltaic modules have consistently been in good condition in recent inspections, the inspection cycle for that area can be adaptively extended. This reduces the need for drones to fly too frequently, as frequent flights accelerate the aging of drone components and shorten their service life.

[0050] Optionally, the distribution area of ​​photovoltaic modules can be divided into multiple inspection areas, including:

[0051] Calculate the effective duration of the patrol area division;

[0052] The distribution area of ​​photovoltaic modules is divided into multiple inspection areas based on the effective duration of the inspection area division.

[0053] Specifically, the effective duration of the patrol area division is calculated, including:

[0054] The effective duration of the first and second patrol area divisions is a set fixed value and does not need to be calculated;

[0055] The calculation method for the effective duration of the third and subsequent patrol area divisions is as follows:

[0056] Calculate the cleanliness index of the photovoltaic modules within the time periods corresponding to the effective duration of the inspection area division for the k-th and k+1-th inspections, respectively.

[0057] Calculate the effective duration of the (k+2)th inspection area division based on the cleanliness status index:

[0058]

[0059] efdur k+2 and efdur k+1 cleidx represents the effective duration of the (k+2)th and (k+1)th patrol area divisions, respectively. k+1 and cleidx k...

[0060] The calculation of the (k+2)th effective duration considers not only the (k+1)th effective duration but also the cleanliness index of the photovoltaic modules within the time periods corresponding to the kth and (k+1)th effective durations. The greater the change in the cleanliness index within these two time periods, the greater the change in the effective duration. This ensures that the effective duration changes with the actual cleanliness status, avoiding the problems of too few or too many divisions caused by pre-setting a fixed effective duration. Because pre-selected effective durations are difficult to accurately reflect the actual cleanliness of the photovoltaic modules, setting values ​​in advance can easily affect inspection efficiency or the service life of drones.

[0061] Dividing the patrol area too frequently can also prevent the division from being too infrequent. If the division is too frequent, the drone will have to wait for the results of another cleaning cycle after only a few patrols, impacting patrol efficiency. Conversely, if the division is too infrequent, the cleanliness of the photovoltaic modules within each patrol area will vary significantly, leading to excessively frequent drone patrols and reducing the drone's service life.

[0062] Specifically, the function for calculating the cleanliness index is:

[0063]

[0064] cleidx represents the cleanliness index, α is the proportion, phonoe is the set of all patrol areas within the effective time period corresponding to the patrol area division, Npne is the number of patrol areas in phonoe, and clenum is the number of patrol areas. a Within the effective time period defined for the patrol area, the average number of cleaning operations for patrol area 'a' is defined as follows: 'snpe' is the set variance of the average number of cleaning operations; 'simfc' is the surface image cleaning probability difference coefficient of the patrol area within the time period corresponding to the effective time period defined for the patrol area; 'scpe' represents the set reference value for the coefficient; 'simfc' = 'macle - micle', where 'macle' and 'micle' represent the maximum and minimum cleaning probability parameters of all patrol areas within the time period corresponding to the effective time period defined for the patrol area, respectively. The calculation function for the cleaning probability parameters is as follows:

[0065]

[0066] Among them, cle b The mxpro parameter represents the cleanliness probability parameter of the patrol area b. gThis represents the similarity between the surface image of photovoltaic module g in inspection area b and the standard surface image of photovoltaic modules that do not require cleaning, during the last inspection within the time period corresponding to the effective duration of the inspection area division. Nfb is the number of photovoltaic modules in inspection area b, and mdn is the number of photovoltaic modules in inspection area b. b The median similarity between the surface images of all photovoltaic modules in inspection area b and the standard surface images of photovoltaic modules that do not require cleaning, during the last inspection within the time period corresponding to the effective duration of the inspection area.

[0067] In addition to considering the differences in average cleaning frequency among different patrol areas, the cleaning status parameters also consider cleaning probability parameters. This allows the cleaning status parameters to comprehensively represent the cleaning status of patrol areas from different perspectives, making them more effective in representing the overall cleaning status of all patrol areas. The greater the difference in cleaning probability parameters, the greater the difference in average cleaning frequency, indicating a greater difference in cleaning status between patrol areas. This necessitates a shorter interval for dividing patrol areas, resulting in a negative right-hand side of the effective time calculation function, thus shortening the effective time for the next patrol area division. The cleaning probability parameter is primarily calculated based on similarity; a higher average similarity results in a higher cleaning probability parameter, indicating a greater need for cleaning.

[0068] Specifically, the distribution area of ​​photovoltaic modules is divided into multiple inspection zones based on the effective duration of the inspection area division, including:

[0069] After the effective time period corresponding to the previous patrol area division ends, the next patrol area division will be carried out:

[0070] S101, store all photovoltaic modules into the collection colsem;

[0071] S102, randomly select the starting photovoltaic module from the set colsem, remove the starting photovoltaic module from the set colsem and save it to the Qth inspection area set;

[0072] S103, save all photovoltaic modules in the set colsem whose distance from the starting photovoltaic module is less than the set distance threshold to a temporary set;

[0073] S104. If the number of elements in the temporary set is 0, proceed to S105; otherwise, proceed to S106.

[0074] S105, clear the temporary set, increment the value of Q by 1, and proceed to S102;

[0075] S106, calculate the correlation value between each element in the temporary set and the starting photovoltaic module, obtain the maximum correlation value. If the maximum correlation value is greater than the set correlation value standard value, remove the photovoltaic module corresponding to the maximum correlation value from the set colsem and store it in the Qth inspection area set, and proceed to S107; if the maximum correlation value is less than or equal to the set correlation value standard value, proceed to S105.

[0076] S107, take the photovoltaic module corresponding to the largest correlation value as the starting photovoltaic module. If the number of elements in the set colsem is greater than 0, then proceed to S103; if the number of elements in the set colsem is equal to 0, then assign the photovoltaic modules in each inspection area set to the same inspection area, and proceed to S108.

[0077] S108 If the number of photovoltaic modules in the inspection area is less than the set number threshold, then the inspection area is merged into another inspection area that is adjacent to it and has the largest number of photovoltaic modules.

[0078] Existing technologies also include methods that divide the distribution area of ​​photovoltaic modules into zones and then conduct inspections based on these zones. However, existing technologies often use an average zoning method, where each zone is the same size and generally has a regular shape, most commonly a rectangle. This zoning method does not consider the correlation between photovoltaic modules assigned to the same inspection area, easily grouping modules with significantly different cleanliness levels into the same distribution area. This obviously forces drones to increase the frequency of inspections of the same area, resulting in low inspection efficiency and reduced drone service life.

[0079] The inspection areas obtained by this invention may have irregular shapes, which differs from existing technologies. This invention employs a cyclical division method, grouping photovoltaic modules with similar cleanliness levels into the same area. If, during the division process, there are no photovoltaic modules meeting the set conditions near the starting photovoltaic module, the current area division ends, and the process moves to the next area division, until all elements in the set `colsem` are assigned to their corresponding inspection areas.

[0080] Specifically, the function for calculating the association value is:

[0081]

[0082] assvl represents associated values, simvl u,v This indicates the similarity between the surface images of the initial photovoltaic module u and the surface images of photovoltaic module v during the last inspection within the time period corresponding to the effective duration of the previous inspection area division.

[0083] Specifically, the more similar the surface images of the initial photovoltaic module u and the photovoltaic module v are, the greater the correlation value, indicating that the cleanliness of the initial photovoltaic module u and the photovoltaic module v are more similar.

[0084] Optionally, the patrol cycle for each patrol area is calculated, including:

[0085] Calculate the patrol coefficient for each patrol area;

[0086] The patrol cycle for each patrol area is calculated based on the patrol coefficient.

[0087] In addition to adaptively calculating the effective duration of the patrol area division, this invention also adaptively calculates the patrol cycle within the patrol area. This allows the patrol frequency of the drone to change with the cleaning status of the photovoltaic modules in the patrol area. On the one hand, it can avoid frequent patrols and extend the service life of the drone. On the other hand, it can avoid excessively long patrol cycles that may prevent timely detection of photovoltaic modules that need cleaning.

[0088] Specifically, the inspection coefficient is calculated using the following function:

[0089]

[0090] In the function, ptlcoe n The patrol coefficient corresponding to the nth patrol cycle; clmnu n slum represents the number of photovoltaic modules that have been cleaned in the nth inspection cycle; slum represents the total number of photovoltaic modules in the inspection area.

[0091] Specifically, the first inspection cycle is set to a fixed value.

[0092] The inspection coefficient is mainly calculated based on the number of photovoltaic modules that have been cleaned in the previous inspection cycle. The more photovoltaic modules that have been cleaned in the previous inspection cycle, the smaller the inspection coefficient will be.

[0093] Specifically, the patrol cycle for each patrol area is calculated based on the patrol coefficient, including:

[0094] Use the following function to calculate the (n+1)th inspection cycle:

[0095]

[0096] In the function, T n+1 and T n ...

[0097] The smaller the inspection coefficient, the greater the probability that the (n+1)th inspection cycle needs to be extended, which reduces the frequency of drone inspections compared to a fixed inspection cycle. Conversely, the larger the inspection coefficient, the greater the probability that the (n+1)th inspection cycle needs to be shortened, which allows for the timely detection of photovoltaic modules requiring cleaning compared to a fixed inspection cycle.

[0098] Optionally, the computing server may also include a path planning module;

[0099] The path planning module is used to calculate the shortest patrol path for the patrol area;

[0100] The communication module is also used to send the shortest patrol path to the drone.

[0101] Specifically, the shortest inspection path can be calculated using algorithms such as Dijkstra's algorithm, Bellman-Ford's algorithm, Floyd's algorithm, and SPFA algorithm.

[0102] Optionally, drones can also be used to patrol the area based on the shortest patrol path.

[0103] Specifically, the shortest patrol path varies for different patrol areas, and the drone patrols the patrol area according to the shortest patrol path corresponding to each patrol area.

[0104] Optionally, the image recognition module includes a pixel value optimization unit and an image recognition unit;

[0105] The pixel value optimization unit is used to perform optimization calculations on the surface image using a set optimization algorithm to obtain an optimized image;

[0106] The image recognition unit is used to identify the optimized image and determine whether the photovoltaic module needs cleaning.

[0107] Optimizing surface images can increase the amount of image information contained in the image, thereby improving the accuracy of subsequent similarity calculations.

[0108] Specifically, the surface image is optimized using a pre-defined optimization algorithm to obtain an optimized image, including:

[0109] Calculate the image feature coefficients of the optimized image:

[0110]

[0111] Where fetidx represents the image feature coefficient, bgotsu represents the number of pixels with gray values ​​greater than the threshold calculated by the otsu algorithm, smotsu represents the number of pixels with gray values ​​less than the threshold calculated by the otsu algorithm, and sall represents the total number of pixels in the optimized image.

[0112] If the image feature coefficients are greater than the corresponding image feature coefficient thresholds, the following function is used to optimize the surface image to obtain the optimized image:

[0113] cnfimg D = cns×log(cg+gray D )

[0114] Among them, cnfimg D This represents the grayscale value of pixel D in the optimized image cnfimg, where cns is the optimization coefficient, cg is a constant coefficient, and gray is the grayscale value. D Let be the gray value of pixel D in the grayscale image corresponding to the surface image;

[0115] If the image feature coefficients are less than or equal to the corresponding image feature coefficient thresholds, the surface image is optimized using the following method to obtain the optimized image:

[0116] If gray D If ∈(grystw, grysth], then

[0117]

[0118] If gray D If ∈(gryson, grystw], then

[0119]

[0120] If gray D If ∈[0, gryson], then

[0121]

[0122] Where gryson, grystw, and grysth represent the first grayscale threshold, the second grayscale threshold, and the third grayscale threshold, respectively. <grystw<grysth,hig one hig two hig thr These represent the first calculation parameter, the second calculation parameter, and the third calculation parameter, respectively. one Less than hig two hig two Less than hig thr .

[0123] gray D The valid value range for gray is 0 to 255. D If the value exceeds the range, then the gray value will be adjusted to the corresponding boundary value within the range.D The value is restricted.

[0124] Image feature coefficients are primarily calculated based on the difference in the number of pixels on either side of a threshold. A larger difference in the number of pixels results in a larger image feature coefficient, indicating a more concentrated distribution range of grayscale values ​​and a lower content of image information. Therefore, it is necessary to expand the distribution range to increase the image information content. Thus, when the image feature coefficient is greater than the corresponding image feature coefficient threshold, this invention uses a logarithmic transformation to expand the distribution range. When the image feature coefficient is less than or equal to the corresponding image feature coefficient threshold, this invention uses different optimization calculation functions based on the different pixel values ​​of pixel D to optimize the calculation, making the grayscale value distribution range more uniform and thus increasing the image information content. It is evident that this invention does not use a single optimization calculation function but rather employs different optimization calculation functions based on the degree of concentration of the distribution range, thereby obtaining more accurate optimization results.

[0125] In some embodiments, graymx is the maximum gray value in the grayscale image corresponding to the surface image.

[0126] In some embodiments, and

[0127] Optionally, the image recognition unit includes a storage subunit and a comparison subunit;

[0128] The storage sub-unit is used to store standard surface images of photovoltaic modules that do not require cleaning;

[0129] The comparison sub-unit is used to compare the standard surface image that does not require cleaning with the surface image, and to determine whether the photovoltaic module needs cleaning based on the comparison result.

[0130] Specifically, the similarity between two images can be compared. If the similarity is less than the set cleaning threshold, it means that the photovoltaic module needs to be cleaned.

[0131] Optional, such as Figure 2 As shown, it also includes an automatic cleaning module.

[0132] The automatic cleaning module is used to clean the photovoltaic modules that need cleaning after receiving cleaning instructions from the computing server.

[0133] Specifically, the automatic cleaning module may include unmanned cleaning robots, automatic cleaning vehicles, etc.

[0134] Optionally, the computing server may also include an instruction generation module;

[0135] The instruction generation module is used to obtain the location information of the photovoltaic modules that need to be cleaned, and generate cleaning instructions based on the location information;

[0136] The communication module is also used to send cleaning instructions to the automatic cleaning module.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An automatic cleaning control system for photovoltaic power generation, characterized in that, Including drones and computing servers; The computing server is used to divide the distribution area of ​​photovoltaic modules into multiple inspection areas, and to calculate the inspection cycle for each inspection area; The drones are used to patrol each patrol area according to the patrol cycle and acquire surface images of the photovoltaic modules in the patrol area; The computing server is also used to determine whether photovoltaic modules need cleaning based on surface images; The distribution area of ​​photovoltaic modules is divided into multiple inspection areas, including: Calculate the effective duration of the patrol area division; The distribution area of ​​photovoltaic modules is divided into multiple inspection areas based on the effective duration of the inspection area division. The effective duration for calculating the patrol area division includes: The effective duration of the first and second patrol area divisions is a set fixed value and does not need to be calculated; The calculation method for the effective duration of the third and subsequent patrol area divisions is as follows: Calculate the cleanliness index of the photovoltaic modules within the time periods corresponding to the effective duration of the inspection area division for the k-th and k+1-th inspections, respectively. Calculate the effective duration of the (k+2)th inspection area division based on the cleanliness status index: efdur k+2 and efdur k+1 cleidx represents the effective duration of the (k+2)th and (k+1)th patrol area divisions, respectively. k+1 and cleidx k ...

2. The photovoltaic power generation automatic cleaning control system according to claim 1, characterized in that, The computing server includes a partitioning module, an image recognition module, and a communication module; The partitioning module is used to divide the distribution area of ​​photovoltaic modules into multiple inspection areas and calculate the inspection cycle for each inspection area; The communication module is used to send the patrol schedule to the drone; The image recognition module is used to determine whether photovoltaic modules need cleaning based on surface images.

3. The photovoltaic power generation automatic cleaning control system according to claim 1, characterized in that, Calculate the patrol cycle for each patrol area, including: Calculate the patrol coefficient for each patrol area; The patrol cycle for each patrol area is calculated based on the patrol coefficient.

4. The photovoltaic power generation automatic cleaning control system according to claim 2, characterized in that, The computing server also includes a path planning module; The path planning module is used to calculate the shortest patrol path for the patrol area; The communication module is also used to send the shortest patrol path to the drone.

5. The photovoltaic power generation automatic cleaning control system according to claim 4, characterized in that, Drones are also used to patrol areas based on the shortest patrol path.

6. The photovoltaic power generation automatic cleaning control system according to claim 2, characterized in that, The image recognition module includes a pixel value optimization unit and an image recognition unit; The pixel value optimization unit is used to perform optimization calculations on the surface image using a set optimization algorithm to obtain an optimized image; The image recognition unit is used to identify the optimized image and determine whether the photovoltaic module needs cleaning.

7. The photovoltaic power generation automatic cleaning control system according to claim 6, characterized in that, The image recognition unit includes a storage subunit and a comparison subunit; The storage sub-unit is used to store standard images of photovoltaic modules that do not require cleaning; The comparison sub-unit is used to compare the standard surface image that does not require cleaning with the surface image, and to determine whether the photovoltaic module needs cleaning based on the comparison result.

8. The photovoltaic power generation automatic cleaning control system according to claim 2, characterized in that, It also includes an automatic cleaning module. The automatic cleaning module is used to clean the photovoltaic modules that need cleaning after receiving cleaning instructions from the computing server.

9. The photovoltaic power generation automatic cleaning control system according to claim 8, characterized in that, The computing server also includes an instruction generation module; The instruction generation module is used to obtain the location information of the photovoltaic modules that need to be cleaned, and generate cleaning instructions based on the location information; The communication module is also used to send cleaning instructions to the automatic cleaning module.

Images