WO2022059815A1

WO2022059815A1 - Artificial intelligence-based solar module diagnosis method and system using thermal image captured by drone

Info

Publication number: WO2022059815A1
Application number: PCT/KR2020/012504
Authority: WO
Inventors: 오원욱; 최훈주; 서근원
Original assignee: 주식회사 에스테코
Priority date: 2020-09-15
Filing date: 2020-09-16
Publication date: 2022-03-24
Also published as: KR102431935B1; KR20220036115A

Abstract

Provided is an artificial intelligence-based solar module diagnosis method and system using a thermal image captured by a drone. A solar module diagnosis method according to an embodiment of the present invention comprises: acquiring a thermal image of solar modules; and inputting the acquired thermal image to an artificial intelligence model, so as to analyze the failure and degradation of the solar modules, wherein the artificial intelligence model is an artificial intelligence model obtained through learning using the thermal image as an input and failure and degradation areas and modes of the solar modules as an output. Accordingly, it is possible to analyze the failure/degradation of solar modules on the basis of artificial intelligence, predict an output therefrom, and determine a grade thereof, by using a thermal image of the solar modules captured by a drone, so as to perform solar module diagnosis objectively, quickly, and easily without an expert.

Description

A method and system for diagnosing solar modules based on artificial intelligence using thermal images taken with a drone

The present invention relates to a photovoltaic module management technology, and more particularly, a photovoltaic module for analyzing failure/deterioration with a thermal image of a photovoltaic module taken with a drone in a photovoltaic power plant, predicting output, and judging a grade It relates to a diagnostic method and system.

The solar module of the solar power plant is a stable product that has been operating for more than 25 years. However, failure/deterioration of the solar module may occur due to incorrect installation by the operator, initial failure, natural disaster, negligent management, and the like. The failure/deterioration of a solar module appears in many different ways.

In order to understand the state of the photovoltaic module, it is possible to diagnose the failure/deterioration by photographing the photovoltaic module with a thermal imaging camera. However, there are problems such as the fact that the diagnosis has to be manually performed by an expert, and that it takes a lot of time and effort even if it is done.

In addition, even by an expert, it cannot be completely free from subjective diagnosis. In other words, there is a problem that the diagnosis results are different depending on who the expert is, so objectivity cannot be guaranteed.

The present invention has been devised to solve the above problems, and an object of the present invention is to analyze failure/deterioration based on artificial intelligence using thermal images of solar modules photographed by a drone, predict output, and grade To provide a solar module diagnostic method and system for determining

According to an embodiment of the present invention for achieving the above object, a solar module diagnosis method includes: acquiring thermal images of the solar modules; An analysis step of inputting the thermal image obtained in the acquisition step into the artificial intelligence model to analyze the failure and deterioration of the solar modules; It is an artificial intelligence model trained with the deterioration region and mode as output.

A method for diagnosing a solar module according to an embodiment of the present invention includes: predicting the output of the solar modules by using the thermal image acquired in the acquisition step and the analysis result in the analysis step; determining a rating for each of the solar modules based on the output predicted in the prediction step; and visualizing and outputting the analysis result in the analysis step, the prediction result in the prediction step, and the determination result in the determination step.

The artificial intelligence model is a Generative Adversarial Network (GAN), and GANs for Anomaly Detection (GAN-AD) may be applied.

The analysis step may include: a first analysis step of inputting the thermal image obtained in the acquisition step into the first artificial intelligence model, and analyzing the failure and deterioration of the solar modules; converting the thermal image acquired in the acquisition step by image processing based on the analysis result in the first analysis step; A second analysis step of inputting the thermal image converted in the conversion step into the second artificial intelligence model, and analyzing the failure and deterioration of the solar modules; may include.

The conversion step may perform image processing for emphasizing faulty and deteriorated areas.

The failure and degradation modes may include at least one of hot spots, bypass diode failure (BDF), shading, contamination, potential induced degradation (PID), short circuit, and disconnection.

In the conversion step, image processing is performed to detect and emphasize the point shape for the failure and degradation areas analyzed as hotspots, PIDs, or shorts in the failure and degradation modes in the first analysis step, and failure and degradation in the first analysis step Image processing is performed to detect and emphasize the bar shape for the failure and deterioration area analyzed by BDF mode, and the surface shape is detected for the failure and deterioration area analyzed as a single line in the failure and deterioration mode in the first analysis step. Thus, image processing for emphasizing is performed, and image processing for emphasizing by detecting irregular shapes can be performed on the failure and deterioration areas analyzed as shadows or contamination in the first analysis step.

The conversion step may further perform image processing for smoothing the area in which the failure and deterioration appear.

The analysis step may further include; a third analysis step of collecting the analysis results in the first analysis step and the analysis results in the second analysis step, and analyzing the failure and deterioration of the solar modules.

On the other hand, according to another embodiment of the present invention, a solar module diagnosis system, the acquisition unit for acquiring thermal images of the solar modules; and an analysis unit that inputs the thermal image obtained in the acquisition unit to the artificial intelligence model and analyzes the failure and deterioration of the solar modules, wherein the artificial intelligence model receives the thermal image as an input, and It is an artificial intelligence model trained with the deterioration region and mode as output.

As described above, according to embodiments of the present invention, by analyzing failure/deterioration based on artificial intelligence using thermal images of solar modules photographed by a drone, predicting output, and judging grade, objectively without an expert At the same time, it is possible to diagnose the solar module quickly and easily.

1 is a view provided for conceptual explanation of a solar module diagnostic system according to an embodiment of the present invention;

2 is a block diagram of the solar module diagnostic system shown in FIG. 1;

3 is a detailed block diagram of the solar module diagnostic platform 200 shown in FIG. 2 ;

4 is a diagram provided for explanation of failure/degradation mode;

5 is a diagram illustrating visual information output as a diagnosis result;

6 is a block diagram of a solar module diagnostic platform according to another embodiment of the present invention, and

7 is a block diagram of a solar module diagnostic platform according to another embodiment of the present invention.

Hereinafter, the present invention will be described in more detail with reference to the drawings.

1 is a diagram provided to explain the concept of a solar module diagnosis system according to an embodiment of the present invention.

In the solar module diagnosis system according to an embodiment of the present invention, as shown, when thermal images of solar modules are taken with the drone 100, the solar module diagnosis platform 200 analyzes the captured thermal images, It is a system built to diagnose a solar module and provide a diagnostic result through the solar power plant manager terminal 300 .

FIG. 2 is a block diagram of the solar module diagnostic system shown in FIG. 1 . The solar module diagnosis system according to the embodiment of the present invention is configured to include a drone 100 , a solar module diagnosis platform 200 , and a solar power plant manager terminal 300 as shown.

The drone 100 creates thermal images of solar modules by photographing as if scanning the sky above the solar power plant. The thermal image taken by the drone 100 is transmitted to the solar module diagnosis platform 200 .

The solar module diagnosis platform 200 analyzes the failure/deterioration of the solar modules based on artificial intelligence using the thermal image received from the drone 100, predicts the output based on this, and determines the grade.

The photovoltaic power plant manager terminal 300 is a terminal possessed/carried by the photovoltaic power plant manager, and receives and displays a diagnosis result from the photovoltaic module diagnosis platform 200 . Through this, the photovoltaic power plant manager can understand the status of the photovoltaic modules, and establish an after-sales service and preventive maintenance plan.

Hereinafter, the detailed structure and operation of the solar module diagnostic platform 200 will be described in detail with reference to FIG. 3 . 3 is a detailed block diagram of the solar module diagnostic platform 200 shown in FIG. 2 .

The solar module diagnosis platform 200, as shown, includes a thermal image acquisition unit 210, a failure/degradation analysis unit 220, a module output prediction unit 230 and a module grade determination unit 240, and a diagnosis result It is configured to include an output unit 250 , a diagnosis result storage unit 260 , and a diagnosis result analysis unit 270 .

The thermal image acquisition unit 210 receives and acquires thermal images of solar modules photographed by the drone 100 , and uses the acquired thermal images to the failure/deterioration analysis unit 220 and the module output prediction unit 230 . Enter

The failure/deterioration analysis unit 220 inputs the thermal image input by the thermal image acquisition unit 210 into the artificial intelligence model for failure/deterioration analysis, and identifies the failure and deterioration of the solar modules.

The artificial intelligence model provided in the failure/deterioration analysis unit 220 is an artificial intelligence model learned by taking thermal images of solar modules as input and outputting failure/deterioration regions and modes of solar modules.

The failure/deterioration mode can be classified into seven types as shown in FIG. 4 .

1) Hot-spot

Cell cracks, snail trails, appearing as dots or multiple dots in a cell in a module

2) BDF (Bypass Diode Failure)

Appears as module substring due to bypass diode failure, uniform high temperature, 1/3, 2/3 uniform brightness of rod (rectangular)-shaped module

3) Shading

Appears in irregular shape with shadows caused by telephone poles, weeds, fallen leaves, snow, etc.

4) Contamination

Appears in irregular shape due to surface contamination by dust adsorption, water scale, bird droppings, insect carcass, etc.

5) PID (Potential Induced Degradation)

A gradual increase in the number of hot-spot modules in the string (-) region, represented by multiple dots of multiple connected modules.

6) Short circuit

Non-uniform high temperature of string, represented by multiple dots of multiple connected modules

7) Disconnection

A uniform high temperature of the string, manifested by the uniform brightness of a number of connected modules

The AI model can be implemented as a Generative Adversarial Network (GAN), which is the most suitable model for detecting failure/degradation areas, and GAN-AD (GANs for Anomaly Detection) can be applied. However, this is exemplary and does not exclude implementation in other ways.

Accordingly, when the thermal image is input, the failure/deterioration analysis unit 220 outputs the failure/deterioration region and mode of the solar modules, so that it is possible to analyze the failure/deterioration status and mode for each solar module.

The module output prediction unit 230 predicts the output of the solar modules using the following data.

1) Thermal images of solar modules input from the thermal image acquisition unit 210

2) Analysis result by the failure/deterioration analysis unit 220

3) Weather/environment data at the time of thermal imaging by drone 100

4) Photovoltaic module specification/specification data

The module output prediction unit 230 may be implemented as an artificial intelligence model learned using the above data as an input and the expected output of the solar modules as an output, or may be implemented as an algorithm using a decision engine.

The module grade determination unit 240 determines a grade for each of the solar modules based on the output predicted by the module output prediction unit 230 .

The module grade determination unit 240 may be implemented as an artificial intelligence model learned by inputting the prediction result of the module output prediction unit 230 and outputting the grades of solar modules, or may be implemented as an algorithm using a decision engine. .

The diagnosis result output unit 250 visualizes and outputs the analysis result of the failure/degradation analysis unit 220 , the prediction result of the module output prediction unit 230 , and the determination result of the module grade determination unit 240 .

Specifically, the diagnosis result output unit 250 displays the corresponding information on the photovoltaic power plant map showing the arrangement state of the photovoltaic modules. 5 exemplifies information output by the diagnosis result output unit 250 .

Specifically, in the upper part of FIG. 5, information on the failure/degradation status and mode of the solar module is displayed on a module-by-module basis, and in the center of FIG. In the lower part, information about the grade determined for each of the solar modules is shown as a visualization result.

The diagnosis result output unit 250 transmits the visualized information shown in FIG. 5 to the solar power plant manager terminal 300 .

The diagnosis result storage unit 260 receives the analysis result of the failure/degradation analysis unit 220 , the prediction result of the module output prediction unit 230 , and the determination result of the module grade determination unit 240 from the thermal image acquisition unit 210 . It is saved together with the input thermal image.

The diagnosis result analysis unit 270 analyzes the data stored in the diagnosis result storage unit 260 to manage preventive maintenance and A/S of the solar modules.

6 is a block diagram of a solar module diagnostic platform according to another embodiment of the present invention. The solar module diagnosis platform 200 according to an embodiment of the present invention, as shown, includes a thermal image acquisition unit 210 , a failure/degradation analysis unit-1 221 , an image processing unit 222 , and a failure/deterioration Including the analysis unit-2 223 , the module output prediction unit 230 , the module grade determination unit 240 , the diagnosis result output unit 250 , the diagnosis result storage unit 260 , and the diagnosis result analysis unit 270 , is composed

The solar module diagnosis platform shown in FIG. 6 includes the 'failure/deterioration analysis unit 220' of the solar module diagnosis platform shown in FIG. ) and failure/deterioration analysis unit-2(223)'.

Accordingly, in FIG. 6 , the rest of the configurations except for the 'failure/degradation analysis unit-1 (221), the image processing unit 222, and the failure/degradation analysis unit-2(223)' can be inferred from the description of FIG. 3 . In , only the 'failure/degradation analysis unit-1 (221), the image processing unit 222, and the failure/degradation analysis unit-2(223)' will be described, and detailed descriptions of the remaining components will be omitted.

The failure/deterioration analysis unit-1 221 inputs the thermal image input by the thermal image acquisition unit 210 into the artificial intelligence model for failure/deterioration analysis, and identifies the failure and deterioration of the solar modules.

The artificial intelligence model provided in the failure/degradation analysis unit-1 (221) is an artificial intelligence model learned by inputting thermal images of solar modules and outputting failure/deterioration regions and modes of the solar modules. The failure/deterioration mode is classified into seven types as described above.

The image processing unit 222 converts the thermal image input by the thermal image acquisition unit 210 by image processing, and transmits the converted thermal image to the failure/degradation analysis unit-2 223 . The image processing performed is different according to the analysis result by the failure/deterioration analysis unit-1 (221). Specifically,

1) For the photovoltaic module whose failure mode is analyzed as hotspot, PID, and short by the failure/degradation analysis unit-1 (221), image processing is performed to detect and emphasize (enhance) the shape of a point,

2) For the photovoltaic module whose failure mode is analyzed as BDF by the failure/degradation analysis unit-1 (221), image processing is performed to detect and emphasize the bar shape,

3) For the solar module whose failure mode is analyzed as disconnection by the failure/degradation analysis unit-1 (221), image processing is performed to detect and emphasize the shape of the surface,

4) Image processing for detecting and emphasizing irregular shapes is performed on the photovoltaic module whose failure mode is analyzed as shadow and pollution by the failure/degradation analysis unit-1 (221).

The failure/deterioration analysis unit-2 223 inputs the thermal image converted by the image processing unit 222 into the artificial intelligence model for failure/deterioration analysis, and identifies the failure and deterioration of the solar modules.

The artificial intelligence model provided in the failure/degradation analysis unit-2 (222) is an artificial intelligence model learned by inputting the thermal image converted by thermal image processing and outputting the failure/deterioration region and mode of the solar modules. . Failure/deterioration modes are classified into 7 categories.

The failure/degradation analysis unit-2 (223) can utilize an artificial intelligence model of the same structure as the failure/degradation analysis unit-1 (221), but the input image is different during learning (original thermal image vs. image processing converted to thermal image), the parameters of the artificial intelligence model are set differently.

The failure/deterioration region and mode of the photovoltaic modules output from the failure/deterioration analysis unit-2 223 are output to the module output prediction unit 230 as a final analysis result for the failure and deterioration of the photovoltaic modules.

In the above embodiment, it has been assumed that the image processing unit 222 performs image processing for emphasizing by detecting a point shape, a bar shape, a planar shape, and an irregular shape, but variations are possible. For instance,

1) For photovoltaic modules whose failure modes are analyzed as hotspots, PIDs, and shorts by the failure/degradation analysis unit-1 (221), point shapes are detected and smoothed (weakened) in addition to image processing to emphasize (enhance) image processing is further performed,

2) For the photovoltaic module whose failure mode is analyzed as BDF by the failure/degradation analysis unit-1 (221), image processing for smoothing is further performed in addition to image processing to detect and emphasize the bar shape,

3) For the solar module whose failure mode is analyzed as disconnection by the failure/degradation analysis unit-1 (221), image processing for smoothing is further performed in addition to image processing to detect and emphasize the surface shape,

4) An irregular shape is detected for the solar module whose failure mode is analyzed as shading and contamination by the failure/degradation analysis unit-1 (221), and image processing for smoothing is further performed in addition to image processing to emphasize .

In this case, the image processing unit 222 further generates a thermal image in which the corresponding shape is smoothed in addition to the thermal image in which the corresponding shape is emphasized, and the failure/deterioration analysis unit-2 222 uses both the thermal images to generate a solar module. Analyze their failure/deterioration areas and modes.

7 is a block diagram of a solar module diagnostic platform according to another embodiment of the present invention. The solar module diagnosis platform 200 according to an embodiment of the present invention, as shown, includes a thermal image acquisition unit 210 , a failure/degradation analysis unit-1 221 , an image processing unit 222 , and a failure/deterioration Analysis unit-2 (223), failure/degradation analysis unit-3 (224), module output prediction unit 230 and module grade determination unit 240, diagnosis result output unit 250, diagnosis result storage unit 260 and a diagnosis result analysis unit 270 .

The solar module diagnosis platform shown in FIG. 7 is a failure/deterioration analysis unit-3 224 added to the solar module diagnosis platform shown in FIG. 6 .

The failure/deterioration analysis unit-3 (224) collects the failure/deterioration analysis result in the failure/degradation analysis unit-1 (221) and the failure/deterioration analysis result in the failure/deterioration analysis unit-2 (223), The failure/deterioration region and mode of the solar modules are finally analyzed, and the result is output to the module output prediction unit 230 .

In the previous embodiment, there is a difference in that the failure/deterioration analysis result of the failure/degradation analysis unit-2 223 is output to the module output prediction unit 230 as a final result.

The failure/deterioration analysis unit-3 (224) takes the failure/deterioration analysis results of the previous failure/degradation analysis units (221,223) as input and outputs the failure/deterioration region and mode of the solar modules as an output, and an artificial intelligence model learned It can be implemented as an algorithm using Decision Engine.

So far, a preferred embodiment has been described in detail for a method and system for diagnosing an artificial intelligence-based solar module using a thermal image taken by a drone.

In the above embodiment, the failure/

deterioration analyzers

220 , 221 , and 223 received the thermal images of the Taeyoung Kwang modules as input and analyzed the failure/deterioration regions and modes of the solar modules, and input data may be added. For example, weather/environment data at the time of thermal image shooting by the drone 100 and/or standard/specification data of a solar module may be added to the input.

On the other hand, it goes without saying that the technical idea of the present invention can be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment. In addition, the technical ideas according to various embodiments of the present invention may be implemented in the form of computer-readable codes recorded on a computer-readable recording medium. The computer-readable recording medium may be any data storage device readable by the computer and capable of storing data. For example, the computer-readable recording medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, or the like. In addition, the computer-readable code or program stored in the computer-readable recording medium may be transmitted through a network connected between computers.

In addition, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above, and the technical field to which the present invention belongs without departing from the gist of the present invention as claimed in the claims In addition, various modifications are possible by those of ordinary skill in the art, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

Claims

obtaining thermal images of solar modules;

An analysis step of analyzing the failure and deterioration of solar modules by inputting the thermal image obtained in the acquisition step into the artificial intelligence model;

artificial intelligence model,

A solar module diagnosis method, characterized in that it is an artificial intelligence model learned by taking a thermal image as an input and outputting the failure and deterioration regions and modes of the solar modules.
The method according to claim 1,

predicting the output of the solar modules by using the thermal image acquired in the acquisition step and the analysis result in the analysis step;

determining a rating for each of the solar modules based on the output predicted in the predicting step; and

Visualizing and outputting the analysis result in the analysis step, the prediction result in the prediction step, and the determination result in the determination step; Solar module diagnosis method further comprising a.
The method according to claim 1,

artificial intelligence model,

A solar module diagnosis method, characterized in that it is a Generative Adversarial Network (GAN) and GAN-AD (GANs for Anomaly Detection) is applied.
The method according to claim 1,

The analysis step is

a first analysis step of inputting the thermal image obtained in the acquisition step into the first artificial intelligence model, and analyzing the failure and deterioration of the solar modules; and

converting the thermal image acquired in the acquisition step by image processing based on the analysis result in the first analysis step;

A method for diagnosing a solar module comprising: inputting the thermal image converted in the conversion step into a second artificial intelligence model, and analyzing the failure and deterioration of the solar modules.
5. The method according to claim 4,

The conversion step is

A method for diagnosing a solar module, characterized in that image processing is performed to highlight areas of failure and deterioration.
6. The method of claim 5,

Failure and deterioration modes are

A method for diagnosing a solar module comprising at least one of hotspots, BDF (Bypass Diode Failure), shading, contamination, PID (Potential Induced Degradation), short circuit and disconnection.
7. The method of claim 6,

The conversion step is

In the first analysis step, image processing is performed to detect and emphasize point shapes for the failure and deterioration areas analyzed as hotspots, PIDs, or short circuits in failure and deterioration modes;

In the first analysis step, image processing is performed to detect and emphasize the bar shape for the failure and deterioration areas analyzed by BDF in the failure and deterioration modes,

In the first analysis step, image processing is performed to detect and emphasize the shape of the failure and deterioration areas analyzed as disconnection in the failure and deterioration modes;

A method for diagnosing a photovoltaic module, characterized in that, in the first analysis step, image processing is performed to detect and emphasize irregular shapes for the failure and deterioration areas analyzed as shadows or contamination in the failure and deterioration modes.
6. The method of claim 5,

The conversion step is

A method for diagnosing a photovoltaic module, characterized in that further image processing is performed for smoothing the area in which the failure and deterioration appear.
5. The method according to claim 4,

The analysis step is

A photovoltaic module diagnosis method further comprising; a third analysis step of collecting the analysis results in the first analysis step and the analysis results in the second analysis step to analyze the failure and deterioration of the solar modules.
an acquisition unit for acquiring thermal images of solar modules;

An analysis unit that inputs the thermal image obtained by the acquisition unit into the artificial intelligence model and analyzes the failure and deterioration of the solar modules;

artificial intelligence model,

A solar module diagnosis system, characterized in that it is an artificial intelligence model learned by taking thermal images as input and outputting failure and deterioration regions and modes of solar modules.