A detection information processing method for wind power and photovoltaic power generation equipment
By acquiring the grayscale image and tilt direction of the photovoltaic panel surface, using segmentation neural network and region growth algorithm, accurately segmenting the pollutant areas on the photovoltaic panel, solving the problem of inaccurate identification caused by reflection and shape diversity in photovoltaic panel pollutant detection. , improves the accuracy of detection.
Patent Information
- Application Number
- CN202411273279.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-12
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-12
AI Technical Summary
The existing method of photovoltaic panel pollutant detection due to the reflection of the photovoltaic panel surface and the diversity of pollutant shapes, resulting in inaccurate identification of pollutant characteristics, reducing the accuracy of pollutant area division.
The grayscale image and tilt direction of the photovoltaic panel surface are obtained, and the photovoltaic cell area is divided using the trained segmentation neural network, and the connecting domain is divided through area growth and gradient analysis, the pollution possibility of each connecting domain is calculated, and the reflective area is reduced. Influence, accurately divide contaminated areas.
It improves the segmentation accuracy of pollutant areas on the surface of photovoltaic panels, enhances the accuracy of detection information processing, reduces false detection, and improves the accuracy of pollutant detection.
Smart Images

Figure CN118799314B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of image data processing, and in particular to a detection information processing method for wind power and photovoltaic power generation equipment. Background Art
[0002] Wind power generation and photovoltaic power generation are important components of the green energy generation system. When there is insufficient sunshine, wind power cannot meet the needs of the power grid. Photovoltaic panels, as important energy supply equipment in photovoltaic power generation, can supplement wind energy and improve the energy output and stability of the entire power generation system. Therefore, photovoltaic panels need to maintain independent and stable power supply capabilities.
[0003] During the photovoltaic panel power generation process, the accumulation of pollutants, such as water stains, oil stains, bird droppings, etc., will cause the photovoltaic panel power generation efficiency to decrease, thereby affecting energy production and environmental sustainability. Therefore, it is crucial to detect and clean the pollutants on the photovoltaic panels in a timely and accurate manner. In order to solve this problem, the detection method through image processing came into being. It collects images of the surface of the photovoltaic panel and uses computer vision technology to process and analyze the images to distinguish the pollutant areas. Among them, the detection processing using the deep learning neural network model requires a large number of high-quality sample sets that are manually annotated; and the pollutants accumulated on the surface of the photovoltaic panel, such as oil stains, water stains, etc., are not fixed in shape, and it is difficult to form a moving number of samples for each pollutant and train a high-quality neural network model.
[0004] At present, in order to adapt to the diversity of shapes and types of pollutants on photovoltaic panels, digital image processing technology is used to detect photovoltaic panel pollutants. When detecting photovoltaic panel pollutants, the reflection on the photovoltaic panels will affect the regional detection of pollutants, resulting in inaccurate identification of pollutant features on the surface of the photovoltaic panels, making it impossible to effectively segment the pollutant area, reducing the accuracy of the photovoltaic panel pollutant image area division. Summary of the invention
[0005] The present application provides a detection information processing method for wind power and photovoltaic power generation equipment to solve the existing problems.
[0006] A detection information processing method for wind power and photovoltaic power generation equipment in the present application adopts the following technical solution:
[0007] An embodiment of the present application provides a detection information processing method for wind power and photovoltaic power generation equipment, the method comprising the following steps:
[0008] Obtain the grayscale image of the photovoltaic panel surface and the tilt direction of the photovoltaic panel; use the trained segmentation neural network to segment the photovoltaic cell area in the grayscale image of the photovoltaic panel surface;
[0009] Perform regional growth on the photovoltaic cell area and divide the photovoltaic cell area into several connected domains; according to the grayscale value of the pixels in each connected domain in different directions and the tilt direction of the photovoltaic panel, obtain the possibility that each connected domain belongs to the pollutant area under the influence of grayscale;
[0010] According to the gradient of the pixels in each connected domain and the tilt direction of the photovoltaic panel, the possibility of each connected domain belonging to the pollutant area under the gradient feature is obtained; according to the possibility of each connected domain belonging to the pollutant area under the influence of grayscale and the possibility of belonging to the pollutant area under the gradient feature, the pollution possibility of each connected domain is obtained;
[0011] According to the pollution possibility of each connected domain, the polluted area is segmented from all connected domains divided by the photovoltaic cell area.
[0012] Preferably, the method of obtaining the possibility that each connected domain belongs to a pollutant under the influence of grayscale according to the grayscale values of the pixels of each connected domain in different directions and the tilt direction of the photovoltaic panel includes the following specific steps:
[0013] The first The pixel point at the center of the smallest circumscribed circle of the connected domain is recorded as The initial point of a connected domain;
[0014] Take horizontal right as the initial direction, rotate clockwise for one circle, select a direction every 45 degrees, and get 8 directions;
[0015] According to the Starting from the initial point of the connected domain, follow the previous The gray value of the pixel is obtained The connected domain is a fitted straight line in each direction; is a preset quantity threshold;
[0016] According to The slope of the fitted straight line of each connected domain in each direction is obtained. The gray level decrease degree of each connected domain in each direction;
[0017] According to The gray level decrease degree of the connected domain in each direction and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to the pollutant area under the influence of grayscale.
[0018] Preferably, the method according to Starting from the initial point of the connected domain, follow the previous The gray value of the pixel is obtained The specific steps of fitting straight lines in each direction of a connected domain are as follows:
[0019] From Starting from the initial point of the connected domain, along the In each direction, the previous The sequence of gray values of pixels is recorded as The connected domain is Grayscale sequence in each direction;
[0020] Use the least squares method to The connected domain is The grayscale sequence in each direction is fitted linearly to obtain The connected domain is The fitted straight line in each direction.
[0021] Preferably, the The slope of the fitted straight line of each connected domain in each direction is obtained. The grayscale reduction degree of a connected domain in each direction includes the following specific steps:
[0022] When The connected domain is When the slope of the fitted line in the direction is negative, The connected domain is The absolute value of the slope of the fitted straight line in the direction is recorded as The connected domain is The degree of grayscale reduction in each direction;
[0023] When The connected domain is When the slope of the fitted line in the direction is non-negative, The connected domain is The grayscale reduction degree in each direction is set to 0.
[0024] Preferably, the The gray level decrease degree of the connected domain in each direction and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to the pollutant area under the influence of grayscale includes the following specific steps:
[0025] In the Among the grayscale decrease degrees in all directions of the connected domain, the direction corresponding to the maximum grayscale decrease degree is recorded as the target direction, and the The other directions that are not the target directions among all the directions corresponding to the connected domain are recorded as reference directions;
[0026] The first The gray level decrease degree of the connected domain in the target direction minus the gray level decrease degree of the first ... The connected domain is The difference in the grayscale decrease degree in the reference direction is recorded as The connected domain is The first difference in the reference direction;
[0027] Calculate the The mean of the first differences of the connected domains in all reference directions, and then calculate the The normalized value of the inverse proportion of the minimum angle between the target direction of the connected domain and the tilt direction of the photovoltaic panel is obtained, and the product of the mean value and the normalized value is recorded as The possibility that a connected domain belongs to the pollutant area under the influence of grayscale.
[0028] Preferably, the method of obtaining the possibility that each connected domain belongs to a pollutant region under the gradient feature according to the gradient of the pixel points in each connected domain and the tilt direction of the photovoltaic panel includes the following specific steps:
[0029] Use the Sobel operator to We can calculate the connected domains and obtain The gradient of each pixel in a connected domain;
[0030] The first The sum of the gradients of all pixels in a connected domain is denoted as The gradient of a connected domain;
[0031] According to The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, The target direction of the connected domain and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to the pollutant area under the gradient characteristics.
[0032] Preferably, the The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, The target direction of the connected domain and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to a pollutant region under the gradient characteristics includes the following specific steps:
[0033] According to The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, and the The target direction of the connected domain is obtained The first eigenvalue of a connected domain;
[0034] Calculate the The normalized value of the inverse proportion between the direction of the gradient of the first connected domain and the minimum angle value of the tilt direction of the photovoltaic panel is The magnitude of the gradient of the first connected domain, the normalized value of the inverse proportion of the minimum angle value, and the The product of the first eigenvalues of the connected domains is recorded as The possibility that a connected domain belongs to the pollutant area under the gradient characteristics.
[0035] Preferably, the The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, and the The target direction of the connected domain is obtained The first eigenvalue of a connected domain includes the following specific steps:
[0036] Calculate the The first Pixels to The reciprocal of the shortest distance between the boundaries of the connected domains, and then calculate and The sum of the values of the inverse proportion of the sum is multiplied by the reciprocal, and is recorded as The first The possibility of contamination of pixels; among them, For the The first The direction of the gradient of the pixel is The minimum angle value of the gradient direction of a connected domain, For the The first The direction of the gradient of the pixel is The minimum angle value of the target direction of the connected domain;
[0037] The first The average value of the contamination probability of all pixels in the connected domain is recorded as The first eigenvalue of a connected component.
[0038] Preferably, the contamination possibility of each connected domain is obtained according to the possibility that each connected domain belongs to the pollutant region under the influence of grayscale and the possibility that each connected domain belongs to the pollutant region under the gradient feature, and the specific steps include the following:
[0039] Calculate the The probability that a connected domain belongs to a pollutant under the influence of grayscale is The product of the probability that the connected domains belong to pollutants under the gradient characteristics is the normalized value of the product, and is recorded as The contamination possibility of a connected domain.
[0040] Preferably, the method of segmenting the polluted area from all the connected areas divided from the photovoltaic cell area according to the pollution possibility of each connected area includes the following specific steps:
[0041] The contamination probability of all connected domains is used as the input of the Otsu threshold algorithm to obtain the segmentation threshold;
[0042] In the photovoltaic cell area, the area composed of all connected domains whose pollution possibility is greater than the segmentation threshold is recorded as the polluted area.
[0043] The beneficial effects of the technical solution of this application are:
[0044] In the embodiment of the present application, the grayscale image of the photovoltaic panel surface and the tilt direction of the photovoltaic panel are obtained, and the photovoltaic cell area in the grayscale image of the photovoltaic panel surface is segmented, and the photovoltaic cell area is divided into several connected domains. According to the grayscale value of the pixel points in different directions of each connected domain and the tilt direction of the photovoltaic panel, the possibility of each connected domain belonging to a pollutant under the influence of grayscale is obtained. According to the analysis of the grayscale value of the pixel point, the possibility of belonging to a pollutant under the influence of grayscale is determined to calculate the possibility of pollution and ensure the accuracy of distinguishing the polluted area. According to the gradient of the pixel points in each connected domain and the tilt direction of the photovoltaic panel, the possibility of each connected domain belonging to a pollutant under the gradient feature is obtained, thereby obtaining the possibility of pollution of each connected domain. According to the analysis of the gradient of the pixel points, the possibility of belonging to a pollutant under the gradient feature is used to calculate the possibility of pollution, and further ensure the accuracy of distinguishing the polluted area. Thus, the polluted area is screened out from all the connected domains divided by the photovoltaic cell area. So far, the present application divides the grayscale image of the photovoltaic panel surface, calculates the possibility of pollution of each connected domain divided, accurately distinguishes the polluted area, improves the accuracy of the segmentation of the pollutant area in the grayscale image of the photovoltaic panel surface, and then improves the accuracy of the processing of the photovoltaic panel detection information. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0046] Figure 1A flowchart of a method for processing detection information of wind power and photovoltaic power generation equipment provided in an embodiment of the present application;
[0047] Figure 2 A flow chart for obtaining a contaminated area provided in an embodiment of the present application. DETAILED DESCRIPTION
[0048] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the detection information processing method for wind power and photovoltaic power generation equipment proposed in the present application, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0049] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
[0050] The following is a detailed description of a specific scheme of a detection information processing method for wind power and photovoltaic power generation equipment provided by the present application in conjunction with the accompanying drawings.
[0051] See also Figure 1 , which shows a flowchart of a method for processing detection information of wind power and photovoltaic power generation equipment provided by an embodiment of the present application, the method comprising the following steps:
[0052] Step S001: Acquire a grayscale image of the photovoltaic panel surface and the tilt direction of the photovoltaic panel; and use a trained segmentation neural network to segment the photovoltaic cell area in the grayscale image of the photovoltaic panel surface.
[0053] Since the photovoltaic panel is composed of multiple photovoltaic cells, during the detection of the pollutant area on the photovoltaic panel, the grid lines and reflective areas in the photovoltaic panel image will affect the detection of the pollutant area, causing false detection. Therefore, this embodiment first excludes the grid line area, and then analyzes the suspected area after threshold segmentation. By analyzing the gray value and gradient of the suspected area, the influence of the reflective area on the detection of the pollutant area is eliminated. Therefore, this embodiment reduces the influence of the reflective area on the detection of the pollutant area by distinguishing the reflective area of the photovoltaic panel from the pollutant area, thereby achieving accurate detection of the pollutant area of the photovoltaic panel.
[0054] Specifically, a camera is used to capture a surface image of a photovoltaic panel, which is used as a detection image in the process of detecting pollutant areas. The surface image is an RGB image. The surface image of the photovoltaic panel is grayed to obtain a gray image of the photovoltaic panel surface, and then the tilt direction of the photovoltaic panel is obtained.
[0055] It should be noted that: this embodiment is photographed in a direction perpendicular to the surface of the photovoltaic panel, and a frontal top view of the photovoltaic panel surface is obtained, and in this embodiment, the tilt direction of the photovoltaic panel is the vertical downward direction in the photographed image. Since the surface of the photovoltaic panel is usually divided into a grid line and a photovoltaic cell area, in which the grid line area will affect the detection of photovoltaic panel contaminants, it is necessary to remove the grid line area first.
[0056] The embodiment of the present application uses a segmentation neural network to segment the grid line area and the photovoltaic cell area in the grayscale image of the photovoltaic panel surface.
[0057] The relevant content of the segmentation neural network is as follows:
[0058] The segmentation neural network used in this embodiment is the Mask R-CNN neural network; the data set used is a photovoltaic panel surface grayscale image data set, in which M photovoltaic panel surface grayscale images are collected. In this embodiment, the size of M is set to 1000. Among them, Mask R-CNN is a well-known technology, and the specific method is not introduced here. The full name of Mask R-CNN in Chinese is "Mask Region-based Convolutional Neural Network", and the full name in English is "Mask Region-based Convolutional Neural Network".
[0059] The pixels that need to be segmented are divided into two categories, that is, the labeling process of the training set is: single-channel semantic label, the corresponding position pixel belongs to the grid line area is labeled as 0, and the corresponding position pixel belongs to the photovoltaic cell area is labeled as 1.
[0060] The task of the network is classification, so the loss function used is the cross entropy loss function.
[0061] The training set after label encoding is input into the neural network for training to obtain a segmentation model. The training of the neural network is a well-known technology and the specific process will not be repeated here.
[0062] The photovoltaic panel surface grayscale image to be detected is input into the trained segmentation model to obtain the grid line area and photovoltaic cell area in the photovoltaic panel surface grayscale image. This process is a well-known technology and the specific method is not introduced here.
[0063] Step S002: Performing regional growth on the photovoltaic cell area to divide the photovoltaic cell area into several connected domains; according to the grayscale value of the pixels in each connected domain in different directions and the tilt direction of the photovoltaic panel, obtaining the possibility that each connected domain belongs to a pollutant under the influence of grayscale.
[0064] Thus, the suspected areas on the photovoltaic panel are found through regional growth. These suspected areas may be caused by pollutants and reflective areas. On the photovoltaic panel, pollutants such as oil and bird droppings will accumulate downward along the tilt direction of the photovoltaic panel due to the influence of rain erosion. Therefore, the pollutant area shows that the gray value gradually decreases along the tilt direction of the photovoltaic panel. The gradient direction in the pollutant area is mostly consistent, but it is perpendicular to the edge at the edge. As the distance to the edge approaches, the gradient direction deviation also increases. The reflective area shows the characteristics of gray value gradually decreasing from the center to the surroundings, and the gradient direction gradually decreasing from the center to the edge.
[0065] Therefore, the region growing algorithm is used to grow the photovoltaic cell region to obtain several connected domains, which may be divided into pollutant regions and reflective regions.
[0066] It should be noted that the region growing algorithm is a well-known technology and the specific method is not introduced here. The growth criterion of the algorithm is that the difference between the grayscale values of the pixels is less than 30. This example is used for description. Other values can be set in other implementations and this application does not limit this.
[0067] After rainwater washes away pollutants, the grayscale value decreases along the tilt direction of the photovoltaic panel, while the grayscale value of the reflective area decreases from the middle to the edge. By calculating the degree of change of the grayscale value in each direction within the connected domain and finding the direction with the largest grayscale value drop, it can be inferred whether there is pollution. In particular, the connected domain close to the tilt direction of the photovoltaic panel and with a sharp drop in grayscale value is more likely to belong to the polluted area.
[0068] Among the several connected domains divided by the photovoltaic cell area, For example, the first connected domain The pixel point at the center of the smallest circumscribed circle of the connected domain is recorded as The initial point of a connected domain.
[0069] The quantity threshold preset in this embodiment , take horizontal right as the initial direction, rotate clockwise for one circle, select a direction every 45 degrees, and get 8 directions. This is used as an example for description. Other implementations can be set to other values, and this application does not limit this.
[0070] From Starting from the initial point of the connected domain, along the In each direction, the previous The sequence of gray values of pixels is recorded as The connected domain is The grayscale sequence in each direction.
[0071] What needs to be explained is: Starting from the initial point of the connected domain, when the number of pixels along any direction does not meet 20, the corresponding direction The value is the number of pixels in that direction.
[0072] Use the least squares method to The connected domain is The grayscale sequence in each direction is fitted linearly to obtain The connected domain is The least square method is a well-known technique, and the specific method will not be introduced here.
[0073] When The connected domain is When the slope of the fitted line in the direction is negative, The connected domain is The absolute value of the slope of the fitted straight line in the direction is recorded as The connected domain is The grayscale reduction degree in each direction.
[0074] When The connected domain is When the slope of the fitted line in the direction is non-negative, The connected domain is The grayscale reduction degree in each direction is set to 0.
[0075] What needs to be explained is: The connected domain is When the slope of the fitted line in the direction is non-negative, it means that Starting from the initial point of the connected domain, along the In each direction, the grayscale value of the pixel gradually increases or remains unchanged, that is, there is no grayscale decrease. Therefore, in this embodiment, the grayscale decrease degree at this time is set to 0. The slope calculation process of the least squares model is a well-known technology. The specific method is as follows: The connected domain is The slope of the fitted line in the direction, that is, The connected domain is Grayscale reduction in each direction The calculation formula is:
[0076]
[0077] in, For Starting from the initial point of the connected domain, along the The first direction The gray value of a pixel, For Starting from the initial point of the connected domain, along the The front in the direction The mean gray value of pixels. is the preset quantity threshold, Reflects a direction forward The average of the sequence numbers corresponding to the pixels. When the slope is negative, and the smaller the negative number, that is The smaller the time, the smaller the Starting from the initial point of the connected domain, along the The faster the gray value of the pixel in each direction decreases, the greater the degree of decrease.
[0078] In the Among the grayscale decrease degrees in the eight directions of a connected domain, the direction corresponding to the largest grayscale decrease degree is recorded as the target direction, and the other directions that are not the target directions are recorded as reference directions.
[0079] It should be noted that: if there are multiple maximum grayscale reduction degrees, the direction corresponding to any maximum grayscale reduction degree is the target direction, that is, each connected domain has only one target direction and seven reference directions.
[0080] Since the grayscale value of the pollutant area increases along the tilt direction of the photovoltaic panel, and the reflective area increases from the middle to the surrounding areas, and the degree of increase from the middle to the surrounding areas is similar, by comparing the maximum grayscale decrease with the grayscale decrease in other directions, the greater the difference in the comprehensive calculation, the greater the possibility that the connected domain contains a pollutant area, and the smaller the difference between the direction with the maximum decrease and the tilt direction of the photovoltaic panel, the greater the possibility that the connected domain contains a pollutant area.
[0081] From this we can know that The calculation formula for the possibility that a connected domain belongs to the pollutant area under the influence of grayscale is:
[0082]
[0083] in, For the The possibility that a connected domain belongs to the pollutant area under the influence of grayscale, For the The minimum angle between the target direction of the connected domain and the tilt direction of the photovoltaic panel, For the The gray level decrease degree of the connected domain in the target direction, For the The connected domain is The grayscale reduction degree of the reference direction, Y is the number of reference directions, is the maximum value of the minimum angle between the two directions, that is for The normalized angle value of for The normalized angle value that is inversely proportional to . For the The connected domain is The first difference in the reference direction.
[0084] What needs to be explained is: The smaller it is, the more likely it is that there is a polluted area in the xth connected domain. The smaller it is, the more likely there is a reflective area in the x-th connected domain, and the larger it is, the more likely there is a polluted area. and The product of The possibility of pollutants in a connected domain under the influence of grayscale. The bigger, the The more connected domains there are, the more likely there are pollutants.
[0085] In another embodiment, obtaining The normalized value of the inverse proportion of ,in is an exponential function with a natural constant as the base, To present the inverse proportional relationship and normalization processing, the implementer can set the inverse proportional function and normalization function according to the actual situation.
[0086] Step S003: According to the gradient of the pixel points in each connected domain and the tilt direction of the photovoltaic panel, the possibility of each connected domain being a pollutant under the gradient feature is obtained; according to the possibility of each connected domain being a pollutant under the influence of grayscale and the possibility of being a pollutant under the gradient feature, the possibility of pollution of each connected domain is obtained.
[0087] The first Take the first connected domain as an example and use the Sobel operator to We can calculate the connected domains and obtain The gradient of each pixel in a connected domain. It should be noted that the gradient of a pixel is a vector with magnitude and direction. The larger the gradient, the greater the possibility that the pixel belongs to the pollutant region in the direction with the largest grayscale drop. The smaller the difference between the gradient direction and the direction with the largest grayscale drop, the greater the possibility that the pixel belongs to the pollutant region. Among them, the Sobel operator is a well-known technology, and the specific method will not be introduced here.
[0088] The first The sum of the gradients of all pixels in a connected domain is denoted as The gradient of a connected domain. Therefore, the gradient is a vector, that is, the sum of the gradients is the sum of the vectors.
[0089] The gradient of the xth connected domain is determined based on the gradient of all pixels in the xth connected domain. The specific process is: statistically calculate the distribution of all kinds of gradients in the xth connected domain, calculate the probability of each gradient appearing, use the probability as a weight, take all kinds of gradients and the probability corresponding to each gradient as input, and use linear weighted fusion to obtain the gradient of the xth connected domain. Linear weighted fusion is a well-known technology, and the specific process is not repeated here. Next, the gradient direction of all pixels in the connected domain is compared with the direction with the largest grayscale drop, and the xth connected domain is obtained. The calculation formula for the possibility that a connected domain belongs to the pollutant area under the gradient characteristics is:
[0090]
[0091]
[0092] in, For the The first eigenvalue of a connected domain; For the The minimum angle between the gradient direction of the connected domain and the tilt direction of the photovoltaic panel, For the The number of pixels in a connected domain, For the The first Pixels to The shortest Euclidean distance between the boundaries of connected domains, For the The first The direction of the gradient of the pixel is The minimum angle value of the gradient direction of a connected domain, For the The first The direction of the gradient of the pixel is The minimum angle value of the target direction of the connected domain;
[0093] For the The possibility that a connected domain belongs to the pollutant region under the gradient characteristics, For the The size of the gradient of a connected domain, is the maximum value of the minimum angle between the two directions, that is for The normalized value of for The normalized value of the inverse proportion of for The normalized value of for The normalized value of the inverse proportion of .
[0094] It should be noted that the grayscale in the reflective connected domain decreases from the center to the surroundings, so when the gradients of the pixels in the opposite direction are added, they will cancel each other out. Therefore, the sum of the gradients of all the pixels in the reflective connected domain is small, while the normal photovoltaic panel area is flat and the gradient is small. Only the gradient of the contaminated area is along the tilt direction of the photovoltaic panel, that is, The bigger, the The more connected the domain is, the more likely it is that there is a polluted area. The first The pixel is When the connected domain boundary is The value is 1 to prevent the denominator from being 0. This is used as an example for description. Other implementations may be set to other values, and this application does not limit this. The closer the pixel point is to the boundary, the more likely it is to be the dividing point between the normal area and the polluted area, that is, the more important the pixel point is. The smaller it is, the more the pixel can reflect the gradient direction characteristics of the connected domain in which it is located. The smaller it is, the more the gradient direction of the pixel point conforms to the target direction of the connected domain, so Indicates The first The greater the probability that a pixel point belongs to the pixel point in the polluted area, the more likely it is a pixel point in the polluted area. , as well as The product of The possibility of pollutants in a connected domain under gradient characteristics, The bigger, the The more connected domains there are, the more likely there is a polluted area.
[0095] From this we can know that The calculation formula for the pollution possibility of a connected domain is:
[0096]
[0097] in, For the The contamination possibility of a connected domain, For the The possibility of pollutants in a connected domain under the influence of grayscale, For the The possibility of pollutants in a connected domain under gradient characteristics, is a linear normalization function that normalizes the data value to Within the range.
[0098] According to the above method, the pollution possibility of each connected domain divided by the photovoltaic cell area in the photovoltaic panel surface grayscale image is obtained.
[0099] Step S004: According to the pollution possibility of each connected domain, the polluted area is screened out from all the connected domains divided by the photovoltaic cell area.
[0100] In this embodiment, the contamination possibilities of all connected domains are used as inputs of the Otsu threshold algorithm to obtain a segmentation threshold. The Otsu threshold algorithm is a well-known technology, and the specific process will not be described in detail.
[0101] In the photovoltaic cell area, the area formed by all connected domains with a probability of contamination greater than the segmentation threshold is recorded as the contaminated area. This completes the segmentation of the contaminated area on the photovoltaic panel surface. The process of obtaining the contaminated area is as follows: Figure 2 shown.
[0102] This application is now complete.
[0103] In summary, in the embodiment of the present application, the grayscale image of the photovoltaic panel surface and the tilt direction of the photovoltaic panel are obtained, the photovoltaic cell area in the grayscale image of the photovoltaic panel surface is segmented, and the photovoltaic cell area is divided into several connected domains. According to the grayscale value of the pixel points in different directions of each connected domain and the tilt direction of the photovoltaic panel, the possibility of each connected domain belonging to a pollutant under the influence of grayscale is obtained. According to the gradient of the pixel points in each connected domain and the tilt direction of the photovoltaic panel, the possibility of each connected domain belonging to a pollutant under the gradient feature is obtained, thereby obtaining the possibility of pollution of each connected domain, thereby screening out the polluted area from all the connected domains divided from the photovoltaic cell area. The present application divides the grayscale image of the photovoltaic panel surface, calculates the possibility of pollution of each divided connected domain, accurately distinguishes the polluted area, and improves the accuracy of the regional division of the photovoltaic panel pollutant image based on computer vision.
[0104] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.
Claims
1. A detection information processing method for wind power and photovoltaic power generation equipment, characterized in that: The method comprises the following steps: Obtain the grayscale image of the photovoltaic panel surface and the tilt direction of the photovoltaic panel; use the trained segmentation neural network to segment the photovoltaic cell area in the grayscale image of the photovoltaic panel surface; Perform regional growth on the photovoltaic cell area and divide the photovoltaic cell area into several connected domains; according to the grayscale value of the pixels in each connected domain in different directions and the tilt direction of the photovoltaic panel, obtain the possibility that each connected domain belongs to the pollutant area under the influence of grayscale; The direction of the maximum grayscale drop in the connected domain is recorded as the target direction; the pollution possibility of each pixel is obtained according to the angle between the gradient direction of each pixel in the connected domain and the gradient direction of the connected domain, the angle between the gradient direction of the pixel and the target direction of the connected domain, and the inverse of the shortest distance between the pixel and the boundary of the connected domain; the possibility of each connected domain belonging to the pollutant area under the gradient feature is obtained according to the pollution possibility of the pixel in each connected domain and the minimum angle between the gradient direction of the connected domain and the tilt direction of the photovoltaic panel; the pollution possibility of each connected domain is obtained according to the possibility of each connected domain belonging to the pollutant area under the influence of grayscale and the possibility of belonging to the pollutant area under the gradient feature; According to the pollution possibility of each connected domain, the polluted area is segmented from all connected domains divided by the photovoltaic cell area.
2. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 1, characterized in that: The method of obtaining the possibility that each connected domain belongs to a pollutant under the influence of grayscale according to the grayscale values of the pixels in different directions of each connected domain and the tilt direction of the photovoltaic panel includes the following specific steps: The first The pixel point at the center of the smallest circumscribed circle of the connected domain is recorded as The initial point of a connected domain; Take horizontal right as the initial direction, rotate clockwise for one circle, select a direction every 45 degrees, and get 8 directions; According to the Starting from the initial point of the connected domain, follow the previous The gray value of the pixel is obtained The connected domain is a fitted straight line in each direction; is a preset quantity threshold; According to The slope of the fitted straight line of each connected domain in each direction is obtained. The gray level decrease degree of each connected domain in each direction; According to The gray level decrease degree of the connected domain in each direction and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to the pollutant area under the influence of grayscale.
3. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 2, characterized in that: The basis from Starting from the initial point of the connected domain, follow the previous The gray value of the pixel is obtained The specific steps of fitting straight lines in each direction of a connected domain are as follows: From Starting from the initial point of the connected domain, along the In each direction, the previous The sequence of gray values of pixels is recorded as The connected domain is Grayscale sequence in each direction; Use the least squares method to The connected domain is The grayscale sequence in each direction is fitted linearly to obtain The connected domain is The fitted straight line in each direction.
4. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 2, characterized in that: According to the The slope of the fitted straight line of each connected domain in each direction is obtained. The grayscale reduction degree of a connected domain in each direction includes the following specific steps: When The connected domain is When the slope of the fitted line in the direction is negative, The connected domain is The absolute value of the slope of the fitted straight line in the direction is recorded as The connected domain is The degree of grayscale reduction in each direction; When The connected domain is When the slope of the fitted line in the direction is non-negative, The connected domain is The grayscale reduction degree in each direction is set to 0.
5. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 2, characterized in that: According to the The gray level decrease degree of the connected domain in each direction and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to the pollutant area under the influence of grayscale includes the following specific steps: In the Among the grayscale decrease degrees in all directions of the connected domain, the direction corresponding to the maximum grayscale decrease degree is recorded as the target direction, and the The other directions that are not the target directions among all the directions corresponding to the connected domain are recorded as reference directions; The first The gray level decrease degree of the connected domain in the target direction minus the gray level decrease degree of the first ... The connected domain is The difference in the grayscale decrease degree in the reference direction is recorded as The connected domain is The first difference in the reference direction; Calculate the The mean of the first differences of the connected domains in all reference directions, and then calculate the The normalized value of the inverse proportion of the minimum angle between the target direction of the connected domain and the tilt direction of the photovoltaic panel is obtained, and the product of the mean value and the normalized value is recorded as The possibility that a connected domain belongs to the pollutant area under the influence of grayscale.
6. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 5, characterized in that: The method of obtaining the possibility that each connected domain belongs to a pollutant region under the gradient feature according to the possibility of pollution of the pixel points in each connected domain and the minimum angle between the gradient direction of the connected domain and the tilt direction of the photovoltaic panel includes the following specific steps: Use the Sobel operator to We can calculate the connected domains and obtain The gradient of each pixel in a connected domain; The first The sum of the gradients of all pixels in a connected domain is denoted as The gradient of a connected domain; According to The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, The target direction of the connected domain and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to the pollutant area under the gradient characteristics.
7. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 6, characterized in that: According to the The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, The target direction of the connected domain and the tilt direction of the photovoltaic panel are obtained. The possibility that a connected domain belongs to a pollutant region under the gradient characteristics includes the following specific steps: According to The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, and the The target direction of the connected domain is obtained The first eigenvalue of a connected domain; Calculate the The normalized value of the inverse proportion between the direction of the gradient of the first connected domain and the minimum angle value of the tilt direction of the photovoltaic panel is The magnitude of the gradient of the first connected domain, the normalized value of the inverse proportion of the minimum angle value, and the The product of the first eigenvalues of the connected domains is recorded as The possibility that a connected domain belongs to the pollutant area under the gradient characteristics.
8. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 7, characterized in that: According to the The gradient of the connected domain, The gradient of each pixel in the connected domain, the distance between pixels, and the The target direction of the connected domain is obtained The first eigenvalue of a connected domain includes the following specific steps: Calculate the The first Pixels to The reciprocal of the shortest distance between the boundaries of the connected domains, and then calculate and The sum of the values of the inverse proportion of the sum is multiplied by the reciprocal, and is recorded as The first The possibility of contamination of pixels; among them, For the The first The direction of the gradient of the pixel is The minimum angle value of the gradient direction of a connected domain, For the The first The direction of the gradient of the pixel is The minimum angle value of the target direction of the connected domain; The first The average value of the contamination probability of all pixels in the connected domain is recorded as The first eigenvalue of a connected component.
9. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 1, characterized in that: The method of obtaining the pollution possibility of each connected domain according to the possibility that each connected domain belongs to the pollutant region under the influence of grayscale and the possibility that each connected domain belongs to the pollutant region under the gradient feature includes the following specific steps: Calculate the The probability that a connected domain belongs to a pollutant under the influence of grayscale is The product of the probability that the connected domains belong to pollutants under the gradient characteristics is the normalized value of the product, and is recorded as The contamination possibility of a connected domain.
10. A detection information processing method for wind power and photovoltaic power generation equipment according to claim 1, characterized in that: The method of segmenting the polluted area from all the connected areas divided by the photovoltaic cell area according to the pollution possibility of each connected area includes the following specific steps: The contamination probability of all connected domains is used as the input of the Otsu threshold algorithm to obtain the segmentation threshold; In the photovoltaic cell area, the area composed of all connected domains whose pollution possibility is greater than the segmentation threshold is recorded as the polluted area.
Citation Information
Patent Citations
Method for detecting pollution discharge in river based on computer vision
CN116229276A
Flexible circuit board defect detection method and system
CN117808812A
Methods for detection of contaminants on optical fiber connectors
US20190339456A1
Hough transform-based license plate tilt correction method
WO2023097931A1
