CN112288728A - Photovoltaic cell bus corrosion condition analysis method based on image perception - Google Patents

Photovoltaic cell bus corrosion condition analysis method based on image perception Download PDF

Info

Publication number
CN112288728A
CN112288728A CN202011204282.9A CN202011204282A CN112288728A CN 112288728 A CN112288728 A CN 112288728A CN 202011204282 A CN202011204282 A CN 202011204282A CN 112288728 A CN112288728 A CN 112288728A
Authority
CN
China
Prior art keywords
photovoltaic cell
image
frame
area
delamination
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202011204282.9A
Other languages
Chinese (zh)
Inventor
郭燕
余波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to CN202011204282.9A priority Critical patent/CN112288728A/en
Publication of CN112288728A publication Critical patent/CN112288728A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Photovoltaic Devices (AREA)

Abstract

The invention discloses a photovoltaic cell bus corrosion condition analysis method based on image perception. The method comprises the following steps: detecting the collected photovoltaic cell assembly image to obtain a photovoltaic cell single-plate image; segmenting a frame region in a photovoltaic cell single-plate image to obtain a frame semantic perception map; obtaining a frame image of the photovoltaic cell panel according to the single plate image of the photovoltaic cell and the frame semantic perception map; and carrying out classification analysis on the frame images of the photovoltaic cell panel, judging whether the photovoltaic cell panel has a delaminating phenomenon, and if the photovoltaic cell panel has the delaminating phenomenon, carrying out measurement analysis on the bus corrosion condition of the photovoltaic cell panel according to the delaminating area. By using the invention, the utilization rate and the power generation power of the cell panel are improved, manpower and material resources are saved, and the operation and maintenance efficiency is improved.

Description

Photovoltaic cell bus corrosion condition analysis method based on image perception
Technical Field
The invention relates to the technical field of new energy photovoltaic power generation, computer vision and artificial intelligence, in particular to a photovoltaic cell bus corrosion condition analysis method based on image perception.
Background
The solar cell panel, which is exposed to a high-temperature and high-humidity environment for a long time, is subject to failure and is mainly characterized by two types in appearance, one is a partial swelling of the resin back plate portion and the other is a delamination phenomenon in which the vicinity of the outer periphery of the cell unit peels off from the protective glass. The delamination may cause the deterioration of the back sheet, and water vapor entering the solar cell panel may chemically react with the EVA, thereby generating acetic acid gas and corroding the periphery of the bus bar (thick wire electrode on the cell). At present, the method for detecting the corrosion of the photovoltaic cell panel mostly adopts the detection of the amount of acetic acid gas, and can not provide more accurate basis for the replacement of the photovoltaic cell panel.
Disclosure of Invention
The invention aims to provide a photovoltaic cell bus corrosion condition analysis method based on image perception aiming at the defects in the prior art.
A photovoltaic cell bus corrosion condition analysis method based on image perception comprises the following steps:
step 1, detecting an acquired photovoltaic cell assembly image to obtain a photovoltaic cell single-plate image;
step 2, segmenting a frame region in the photovoltaic cell single-plate image to obtain a frame semantic perception map;
step 3, obtaining a frame image of the photovoltaic cell panel according to the single plate image of the photovoltaic cell and the frame semantic perception map;
step 4, carry out classification analysis to photovoltaic cell board frame image, judge whether there is the delaminating phenomenon in photovoltaic cell board, if there is the delaminating phenomenon, then measure the analysis according to the delaminating area to the photovoltaic cell board bus-bar corrosion condition:
g(t)＝g(0)+∫f(h(t),T(t),s(t))*vdt
wherein g (0) is the bus corrosion measurement at the initial moment, h (t) represents the humidity sensor data at the t moment, T (t) represents the temperature sensor data at the t moment, if T (t)<T1Or T (t)>T2If f (h), (T), T (T), s (T) ═ 0, [ T [, [1,T2]Representing the ambient temperature at which the corrosion reaction conditions are met, s (t) is the delamination area, v represents the reference rate of reaction at the reference humidity, reference temperature, f (h (t), T (t), s (t)) is a function of the influence of the humidity, temperature, delamination area on the reference rate, satisfying:
Figure BDA0002756496020000011
h is the reference humidity, q (T (t), s (t)) is a function of the temperature, the delamination area and the reference rate.
The function of the influence of the humidity, temperature, delamination area on the reference rate is:
Figure BDA0002756496020000021
wherein H represents the reference humidity at which the reaction occurs.
According to the photovoltaic cell single-plate image and the frame semantic perception map, obtaining a photovoltaic cell panel frame image comprises the following steps:
binarizing the frame semantic perception map to obtain a frame perception binary mask;
and multiplying the frame sensing binary mask and the photovoltaic cell single-plate image point-to-point to obtain a photovoltaic cell panel frame image.
The step 1 specifically comprises the following steps:
and detecting the collected photovoltaic cell assembly image by using a photovoltaic single-board detection neural network to obtain a photovoltaic cell single-board image.
The step 2 specifically comprises the following steps:
and segmenting the frame area in the photovoltaic cell single-plate image by using a frame semantic segmentation neural network to obtain a frame semantic perception map.
Carry out classification analysis to photovoltaic cell board frame image, judge whether photovoltaic cell board has the delaminating phenomenon and include:
segmenting a frame image of the photovoltaic cell panel by utilizing a delaminating segmentation neural network, and outputting a delaminating semantic perception map, wherein the delaminating semantic perception map is used for distinguishing the semantics of a delaminating region, a normal region and other irrelevant regions;
and counting the number of the pixels in the delamination area in the delamination semantic perception map, and if the number is larger than a set threshold value, judging that the delamination phenomenon exists.
The method further comprises the following steps:
and carrying out opening operation on the frame perception binary mask, and multiplying the result after the opening operation by the point-to-point of the image of the photovoltaic cell single plate to obtain the frame image of the photovoltaic cell panel.
The delamination area was calculated as follows:
Figure BDA0002756496020000022
wherein N is1Number of pixels in the normal region, N2The number of pixels in the delamination area.
The delaminating semantic awareness map is also used to distinguish the semantics of a bus bar region from other types of regions.
The measurement analysis is carried out on the bus corrosion condition of the photovoltaic cell panel according to the delamination area:
g(t)＝g(0)+∫f(h(t),T(t),s(t),a(t))*vdt
wherein g (0) is the bus corrosion measurement at the initial moment, h (t) represents the humidity sensor data at the t moment, T (t) represents the temperature sensor data at the t moment, if T (t)<T1Or T (t)>T2F (h), (T), T (T), s (T), a (T)) 0, [ T [, (T) ], and1,T2]representing the ambient temperature meeting the corrosion reaction conditions, s (t) is the delamination area, a (t) is the area of the bus area, v represents the reference rate of reaction at the reference humidity and the reference temperature, f (h) (t), T (t), s (t), a (t) is the influence function of the humidity, the temperature, the delamination area and the area of the bus area on the reference rate, and the conditions are met:
Figure BDA0002756496020000031
q (T (t), s (t), a (t)) is a function of temperature, delamination area, bus bar area effect on reference rate.
Compared with the prior art, the invention has the following beneficial effects:
need not to carry out quantitative analysis to the adnexed acetic acid gas of backplate, only need image information, sensor data can realize the measurement analysis to the luminous photovoltaic cell generating line corrosion condition, for the panel change opportunity provides the quantization basis, has not only improved panel utilization ratio and generated power, and the material resources of using manpower sparingly moreover improve fortune dimension efficiency. The image data is processed through the deep neural network, so that the robustness is good, the generalization capability is strong, and the method is applicable to various scenes. Through multiple image operations, irrelevant working conditions are isolated, and the final classification result and the area integral analysis result are more accurate. The bus corrosion condition of the photovoltaic cell panel is analyzed through a mathematical model, and data which are difficult to observe are analyzed through data which are easy to obtain.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The delamination phenomenon directly corrodes the bus, resulting in a reduction in the generated power. If delamination is found, the panel is replaced immediately, and the good balance between the economic benefit and the power generation benefit cannot be achieved. Therefore, in addition to detecting the delamination phenomenon causing the bus bar corrosion, the bus bar corrosion condition should be analyzed, so as to improve the utilization rate of the photovoltaic cell panel and ensure higher power generation efficiency. The invention provides a photovoltaic cell bus corrosion condition analysis method based on image perception. The bus corrosion condition of the solar cell panel in the photovoltaic power station is analyzed through image processing and multi-dimensional information perception. FIG. 1 is a flow chart of the present invention. The following description will be made by way of specific examples.
Example 1:
the photovoltaic cell bus corrosion condition analysis method based on image perception comprises the following steps:
step 1, detecting the collected photovoltaic cell assembly image to obtain a photovoltaic cell single-plate image. Specifically, the collected photovoltaic cell module image is detected by using a photovoltaic single-board detection neural network to obtain a photovoltaic cell single-board image.
The unmanned aerial vehicle collects RGB images and sends the RGB images into a photovoltaic single-board detection neural network, the detection target is a photovoltaic cell panel single board, and the output result is an enclosure frame of the photovoltaic cell panel single board.
The method comprises the steps that a photovoltaic single board detection neural network is used, the input of the network is an RGB image collected by an unmanned aerial vehicle, features are extracted through a photovoltaic single board detection Encoder Encoder1, a first feature diagram Featuremap1 is output, the first feature diagram Featuremap1 is used as the input after being flattened, the input is sent into a first full-connection network FC1, and a surrounding frame of a photovoltaic cell single board is output. The specific training method comprises the following steps: the training set selects a plurality of RGB images collected by the unmanned aerial vehicle, the RGB images are marked as the coordinates of the upper left corner point and the lower right corner point of the bounding box, and the loss function adopts a mean square error loss function. And cutting the original image based on the surrounding frame, and sampling to obtain an equal-size image to obtain the photovoltaic cell single-plate image. The purpose of this step is to extract only one photovoltaic cell panel image to avoid being disturbed by adjacent photovoltaic cell panels when subsequent images are classified. The sampling to an equal size image is intended to facilitate input to a subsequent network.
And 2, segmenting a frame region in the photovoltaic cell single-plate image to obtain a frame semantic perception map. Specifically, the photovoltaic cell single-plate image is sent to a frame semantic segmentation neural network, pixel points are classified, and a frame semantic perception map is output.
The input of the frame semantic segmentation neural network is a photovoltaic cell single-plate image, the characteristics are extracted through a first semantic segmentation Encoder Encoder2, a second characteristic diagram Featuremap2 is output, the Featuremap2 is sent to a first semantic segmentation Decoder Decoder2 for sampling, and a frame semantic perception diagram is output. The specific training method of the frame semantic segmentation neural network comprises the following steps: the training set selects the cut photovoltaic cell single-board images with the same size, wherein the photovoltaic cell single-board images under the delaminating condition are included, the marks are marked by marking pixel points by using marking tools such as labelme, and the pixel points are frames, battery panels and other irrelevant items. The loss function is a cross-entropy loss function.
And 3, obtaining a frame image of the photovoltaic cell panel according to the single plate image of the photovoltaic cell and the frame semantic perception map. Specifically, binarization is carried out on the frame semantic perception map to obtain a frame perception binary mask; and multiplying the frame sensing binary mask and the photovoltaic cell single-plate image point-to-point to obtain a photovoltaic cell panel frame image.
First, an initial bounding box mask is generated based on a bounding box semantic perception map. And setting the pixel value of the pixel point belonging to the frame category in the frame semantic perception graph as 1, and setting the other pixel values as 0 to obtain an initial frame mask binary graph. And performing image morphological operation, specifically opening operation, namely performing expansion operation after image corrosion on the initial frame mask, so as to eliminate noise caused by image false detection and obtain a frame perception binary mask.
And (4) performing point-by-point multiplication operation on the frame perception binary mask and the photovoltaic cell single-plate image to obtain an image only containing information in a frame range. And multiplying the binary image and the photovoltaic cell single-plate image point by point, namely multiplying the pixel point with the pixel value of 0 in the binary image by 0, and multiplying the pixel point with the pixel value of 1 in the binary image by reserving the original pixel value of the cutting image. The method comprises the following steps that when the delamination condition of the photovoltaic cell panel occurs, the delamination phenomenon is firstly generated in a frame area, the step aims to only pay attention to image information of the frame area, and irrelevant working condition interference is isolated.
Step 4, carry out classification analysis to photovoltaic cell board frame image, judge whether there is the delaminating phenomenon in photovoltaic cell board, if there is the delaminating phenomenon, then measure the analysis according to the delaminating area to the photovoltaic cell board bus-bar corrosion condition:
g(t)＝g(0)+∫f(h(t),T(t),s(t))*vdt
wherein g (0) is the bus corrosion measurement at the initial moment, h (t) represents the humidity sensor data at the t moment, T (t) represents the temperature sensor data at the t moment, if T (t)<T1Or T (t)>T2If f (h), (T), T (T), s (T) ═ 0, [ T [, [1,T2]Representing the ambient temperature at which the corrosion reaction conditions are met, s (t) is the delamination area, v represents the reference rate of reaction at the reference humidity, reference temperature, f (h (t), T (t), s (t)) is a function of the influence of the humidity, temperature, delamination area on the reference rate, satisfying:
Figure BDA0002756496020000051
h is the reference humidity, q (T (t), s (t)) is a function of the temperature, the delamination area and the reference rate.
Carry out classification analysis to photovoltaic cell board frame image, judge whether photovoltaic cell board has the delaminating phenomenon and include: segmenting a frame image of the photovoltaic cell panel by utilizing a delaminating segmentation neural network, and outputting a delaminating semantic perception map, wherein the delaminating semantic perception map is used for distinguishing the semantics of a delaminating region, a normal region and other irrelevant regions; and counting the number of the pixels in the delamination area in the delamination semantic perception map, and if the number is larger than a set threshold value, judging that the delamination phenomenon exists.
And sending the frame image of the photovoltaic cell panel as an input into a delamination segmentation neural network, extracting features through a second semantic segmentation Encoder Encoder3, outputting a third feature map Featuremap3, sampling the third feature map Featuremap3 through a second semantic segmentation Decoder Decoder3, and outputting a delamination semantic perception map. The classification of the pixel points in the delaminating and dividing neural network is three, namely delaminating condition, normal condition and irrelevant item, and the irrelevant item in the invention is the pixel point belonging to the outside of the photovoltaic cell panel single plate. The specific training method of the delaminating and segmenting neural network comprises the following steps: the training set selects a plurality of photovoltaic cell panel frame images, the images are marked as pixel level marks, the types of pixel points are two types, namely, the type of an irrelevant item corresponds to an index 0, the type of a normal pixel point corresponds to an index 1, the type of a delamination pixel point corresponds to an index 2, the marks need to be subjected to one-hot coding, and a cross entropy loss function is adopted as a loss function.
Counting the number of pixel points with the pixel value of 1 and the pixel value of 2 in the delaminating semantic perception image, if the pixel point with the pixel value of 2 exists, and the pixel value is larger than a set threshold value m1Judging that there is delamination, preferably, m1Set to 10 in this embodiment, the threshold is set for the purpose of preventing false detection and ignoring the case where the delamination area is particularly small; when the delamination phenomenon is judged, counting the number of the pixels with the pixel value of 1 as N1The number of pixels with a pixel value of 2 is N2The area of delamination
Figure BDA0002756496020000052
The delamination detection results are obtained, and then analysis needs to be carried out by combining the data detected by the sensor. Data are collected through a sensor, a corrosion condition measurement and analysis mathematical model is constructed, and when a delaminating phenomenon exists in a frame region, the mathematical model is adopted to analyze the bus corrosion condition of a solar cell panel in a photovoltaic power station. The triggering condition is that the frame region has a delamination phenomenon. And collecting sensor data in real time, wherein the sensor comprises a temperature sensor and a humidity sensor. The temperature sensor only needs to be arranged on one surface of any photovoltaic cell panel, and the humidity sensor is a thin film sensor and arranged in the frame area of each photovoltaic cell panel. The reason is that the environmental temperature of the photovoltaic cell assemblies in the same range area can not change greatly, a temperature sensor is adopted to collect data for saving consumables, and the data is used as the temperature data of the photovoltaic cell panel in the whole area; humidity transducer because of the position difference can receive around the grass trend, the regional inequality factor influence of falling water, consequently need gather the humidity information of every panel, and dispose in the frame region, mainly gather the humidity in frame region, make follow-up calculation more accurate.
A bus corrosion condition measurement analysis mathematical model is constructed, and the bus corrosion is caused by the fact that water vapor enters the delaminating battery plate and acts with packaging material EVA (ethylene-vinyl acetate copolymer) filled between the protective glass and the back plate, and the water vapor is chemical, namely the temperature reaches a certain value to meet the reaction environment. Therefore, the influence function f (h (t), T (t), s (t)) of the humidity, the temperature and the delamination area on the reference rate is provided, and the following conditions are met:
Figure BDA0002756496020000061
wherein h (t) represents humidity sensor data at time t, T (t) represents temperature sensor data at time t, if T (t)<T1Or T (t)>T2If f (h), (T), T (T), s (T) ═ 0, [ T [, [1,T2]Representing the ambient temperature at which the corrosion reaction conditions are met, s (t) is the delamination area, H is the reference humidity, q (T) (t), s (t)) is a function of the temperature, delamination area effect on the reference rate. Thereby obtaining a bus corrosion condition measurement analysis mathematical model:
g(t)＝g(0)+∫f(h(t),T(t),s(t))*vdt
wherein g (0) is the bus corrosion measurement at the initial moment, can be set according to the corrosion condition, and starts immediately after installationWhen monitored, the initial value is 0 and v represents the reference rate of reaction at the reference temperature at the reference humidity. Since the catalytic temperature can be determined, the reference temperature is T1. When the measurement analysis value is larger than the set threshold value, the bus corrosion is serious, and the generated power can be ensured only by replacing the battery panel. Therefore, the measurement analysis model can provide a positive accurate basis for the battery plate replacement time.
Considering the amount of reactants, when H (t) is greater than H, namely the measured humidity is greater than the standard humidity, the reaction rate is increased because the EVA material amount is fixed and the water vapor amount and the humidity are in a linear and direct proportional relation; when H (t) is equal to H, namely the measured humidity is equal to the standard humidity, taking the standard reaction rate; when H (t) is less than H, i.e., the measured humidity is less than the standard humidity, a negative gain is made in the reaction rate. In addition, the greater the humidity, the greater the contact area of the reactants may be (water vapor may not be sufficient to completely cover the delamination area when the humidity is low, and all of the reactants may participate in the reaction upon entering the edge of the delamination area). Considering the influence of the catalytic temperature, T (T) epsilon [ T ] only when the catalytic temperature is satisfied1,T2]The reaction will take place. Within this interval, the reaction rate is faster at higher temperatures, and therefore
Figure BDA0002756496020000062
For normalization, the range is [0,1 ]]. Preferably, f (h (t), T (t), s (t)) is as follows:
Figure BDA0002756496020000063
wherein H represents the reference humidity at which the reaction occurs. The above functions integrate the effects of humidity, temperature, delamination area on corrosion.
Example 2:
the embodiment provides a photovoltaic cell bus corrosion condition analysis method based on image perception, and the method is different from the embodiment 1 in that the bus proportion in a delamination area is considered. If the delamination area includes a larger bus bar area, the bus bar will be corroded more severely. Thus, the delaminating semantic awareness map is also used to distinguish the semantics of the bus bar region from other types of regions. Carrying out measurement analysis on the bus corrosion condition of the photovoltaic cell panel according to the delamination area:
g(t)＝g(0)+∫f(h(t),T(t),s(t),a(t))*vdt
wherein g (0) is the bus corrosion measurement at the initial moment, h (t) represents the humidity sensor data at the t moment, T (t) represents the temperature sensor data at the t moment, if T (t)<T1Or T (t)>T2F (h), (T), T (T), s (T), a (T)) 0, [ T [, (T) ], and1,T2]representing the ambient temperature meeting the corrosion reaction conditions, s (t) is the delamination area, a (t) is the area of the bus area, v represents the reference rate of reaction at the reference humidity and the reference temperature, f (h) (t), T (t), s (t), a (t) is the influence function of the humidity, the temperature, the delamination area and the area of the bus area on the reference rate, and the conditions are met:
Figure BDA0002756496020000071
q (T (t), s (t), a (t)) is a function of temperature, delamination area, bus bar area effect on reference rate. Preferably, f (h (t), T (t), s (t), a (t)) is as follows:
Figure BDA0002756496020000072
wherein H represents the reference humidity of the reaction, b is a compensation coefficient, and b is greater than 1 due to the larger influence of the area of the bus area.
The above embodiments are merely preferred embodiments of the present invention, which should not be construed as limiting the present invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (10)

1. A photovoltaic cell bus corrosion condition analysis method based on image perception is characterized by comprising the following steps:
step 1, detecting an acquired photovoltaic cell assembly image to obtain a photovoltaic cell single-plate image;
step 2, segmenting a frame region in the photovoltaic cell single-plate image to obtain a frame semantic perception map;
step 3, obtaining a frame image of the photovoltaic cell panel according to the single plate image of the photovoltaic cell and the frame semantic perception map;
step 4, carry out classification analysis to photovoltaic cell board frame image, judge whether there is the delaminating phenomenon in photovoltaic cell board, if there is the delaminating phenomenon, then measure the analysis according to the delaminating area to the photovoltaic cell board bus-bar corrosion condition:
g(t)＝g(0)+∫f(h(t)，T(t)，s(t))*vdt
g (0) is bus corrosion measurement at initial time, h (T) represents humidity sensor data at T time, T (T) represents temperature sensor data at T time, and if T (T) < T1Or T (T) > T2If f (h), (T), T (T), s (T) ═ 0, [ T [, [1，T2]Representing the ambient temperature at which the corrosion reaction conditions are met, s (t) is the delamination area, v represents the reference rate of reaction at the reference humidity, reference temperature, f (h (t), T (t), s (t)) is a function of the influence of the humidity, temperature, delamination area on the reference rate, satisfying:
Figure FDA0002756496010000011
q (T (t), s (t)) is a function of temperature, delamination area, and reference rate.
2. The method of claim 1, wherein the function of the effect of humidity, temperature, delamination area on the reference rate is:
Figure FDA0002756496010000012
wherein H represents the reference humidity at which the reaction occurs.
3. The method of claim 1, wherein obtaining the border image of the photovoltaic cell panel according to the single-plate image of the photovoltaic cell and the semantic perception map of the border comprises:
binarizing the frame semantic perception map to obtain a frame perception binary mask;
and multiplying the frame sensing binary mask and the photovoltaic cell single-plate image point-to-point to obtain a photovoltaic cell panel frame image.
4. The method according to claim 1, wherein step 1 is specifically:
and detecting the collected photovoltaic cell assembly image by using a photovoltaic single-board detection neural network to obtain a photovoltaic cell single-board image.
5. The method according to claim 1, wherein the step 2 is specifically:
and segmenting the frame area in the photovoltaic cell single-plate image by using a frame semantic segmentation neural network to obtain a frame semantic perception map.
6. The method of claim 1, wherein the classifying and analyzing the images of the borders of the photovoltaic panel to determine whether the photovoltaic panel has delamination comprises:
segmenting a frame image of the photovoltaic cell panel by utilizing a delaminating segmentation neural network, and outputting a delaminating semantic perception map, wherein the delaminating semantic perception map is used for distinguishing the semantics of a delaminating region, a normal region and other irrelevant regions;
and counting the number of the pixels in the delamination area in the delamination semantic perception map, and if the number is larger than a set threshold value, judging that the delamination phenomenon exists.
7. The method of claim 1, further comprising:
and carrying out opening operation on the frame perception binary mask, and multiplying the result after the opening operation by the point-to-point of the image of the photovoltaic cell single plate to obtain the frame image of the photovoltaic cell panel.
8. The method of claim 1, wherein the delamination area is calculated as follows:
Figure FDA0002756496010000021
wherein N is1Number of pixels in the normal region, N2The number of pixels in the delamination area.
9. The method of claim 1, wherein the de-layered semantic awareness map is also used to distinguish the semantics of a bus bar region from other types of regions.
10. The method of claim 6, wherein the busbar corrosion of the photovoltaic panel is quantitatively analyzed in terms of delamination area:
g(t)＝g(0)+∫f(h(t)，T(t)，s(t)，a(t))*vdt
g (0) is bus corrosion measurement at initial time, h (T) represents humidity sensor data at T time, T (T) represents temperature sensor data at T time, and if T (T) < T1Or T (T) > T2F (h), (T), T (T), s (T), a (T)) 0, [ T [, (T) ], and1，T2]representing the ambient temperature meeting the corrosion reaction conditions, s (t) is the delamination area, a (t) is the area of the bus area, v represents the reference rate of reaction at the reference humidity and the reference temperature, f (h) (t), T (t), s (t), a (t) is the influence function of the humidity, the temperature, the delamination area and the area of the bus area on the reference rate, and the conditions are met:
Figure FDA0002756496010000022
q (T (t), s (t), a (t)) is a function of temperature, delamination area, bus bar area effect on reference rate.
CN202011204282.9A 2020-11-02 2020-11-02 Photovoltaic cell bus corrosion condition analysis method based on image perception Withdrawn CN112288728A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011204282.9A CN112288728A (en) 2020-11-02 2020-11-02 Photovoltaic cell bus corrosion condition analysis method based on image perception

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011204282.9A CN112288728A (en) 2020-11-02 2020-11-02 Photovoltaic cell bus corrosion condition analysis method based on image perception

Publications (1)

Publication Number Publication Date
CN112288728A true CN112288728A (en) 2021-01-29

Family

ID=74352895

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011204282.9A Withdrawn CN112288728A (en) 2020-11-02 2020-11-02 Photovoltaic cell bus corrosion condition analysis method based on image perception

Country Status (1)

Country Link
CN (1) CN112288728A (en)

Similar Documents

Publication Publication Date Title
CN111080691A (en) Infrared hot spot detection method and device for photovoltaic module
CN107490769B (en) A kind of fuel cell water distribution visualization on-line detecting system
CN114399719B (en) Transformer substation fire video monitoring method
CN113536440B (en) Data processing method based on BIM operation and maintenance management system
CN114743115A (en) Shelter recognition method and fault early warning system for photovoltaic electronic equipment
CN115326661A (en) Atmospheric environment-based monitoring method and system and readable storage medium
CN115239209B (en) Method, equipment and system for screening heavy metal-containing humus soil in refuse landfill
CN115601197B (en) Abnormal state detection method and device for photovoltaic power station
CN115457001A (en) Photovoltaic panel foreign matter detection method, system, device and medium based on VGG network
CN114399484A (en) Photovoltaic module defect detection method, device, equipment and medium
CN117576632B (en) Multi-mode AI large model-based power grid monitoring fire early warning system and method
CN116503354A (en) Method and device for detecting and evaluating hot spots of photovoltaic cells based on multi-mode fusion
CN117333776A (en) VOCs gas leakage detection method, device and storage medium
CN111339927A (en) Intelligent work state identification system for personnel in electric power business hall
Hwang et al. Intelligent cleanup scheme for soiled photovoltaic modules
CN117649123B (en) Wisdom garden management system
CN112288728A (en) Photovoltaic cell bus corrosion condition analysis method based on image perception
CN116580363A (en) Photovoltaic panel hot spot identification method, storage medium and electronic equipment
CN115186230B (en) Method and device for judging influence of typhoons on air quality
CN110610474A (en) Solar panel defect real-time detection method based on infrared image
CN116434258A (en) Automatic identification method for form data
CN114926774A (en) Method and device for identifying photovoltaic module fire based on video image
CN112129767A (en) Photovoltaic cell internal defect type detection method and system based on artificial intelligence
CN109583441A (en) Subtitle recognition method based on monitor supervision platform
CN118052356B (en) Industrial carbon emission remote sensing monitoring system based on big data

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20210129