WO2022246580A1 - Method and system for measuring power loss in photovoltaic solar modules - Google Patents

Method and system for measuring power loss in photovoltaic solar modules Download PDF

Info

Publication number
WO2022246580A1
WO2022246580A1 PCT/CL2021/050046 CL2021050046W WO2022246580A1 WO 2022246580 A1 WO2022246580 A1 WO 2022246580A1 CL 2021050046 W CL2021050046 W CL 2021050046W WO 2022246580 A1 WO2022246580 A1 WO 2022246580A1
Authority
WO
WIPO (PCT)
Prior art keywords
processor
photovoltaic solar
region
image
power loss
Prior art date
Application number
PCT/CL2021/050046
Other languages
Spanish (es)
French (fr)
Inventor
Rodrigo Sebastián BARRAZA VICENCIO
Patricio VALDIVIA LEFORT
Danilo Alejandro ESTAY BARRIENTOS
Robinson Daniel CAVIERES ABARCA
Original Assignee
Universidad Técnica Federico Santa María
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 Universidad Técnica Federico Santa María filed Critical Universidad Técnica Federico Santa María
Priority to PCT/CL2021/050046 priority Critical patent/WO2022246580A1/en
Publication of WO2022246580A1 publication Critical patent/WO2022246580A1/en

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T1/00—General purpose image data processing
    • G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02S—GENERATION OF ELECTRIC POWER BY CONVERSION OF INFRARED RADIATION, VISIBLE LIGHT OR ULTRAVIOLET LIGHT, e.g. USING PHOTOVOLTAIC [PV] MODULES
    • H02S50/00—Monitoring or testing of PV systems, e.g. load balancing or fault identification
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/50—Photovoltaic [PV] energy

Definitions

  • the present invention relates to the field of measurement and testing, more specifically with measurement by optical means, and in particular provides a method and system for measuring the power loss of photovoltaic solar modules.
  • WO 2014/154515 describes a method for determining contamination of a photovoltaic solar module, eg due to soiling, from image analysis. For this, the contrast in the image is measured and a contrast transfer function is determined which is a function of the measured contrast values. In this way, a calibration curve is used to determine the degree of soiling.
  • the method described in this document requires the addition of evaluation elements that have black and white bands to determine the contrast values, which makes it difficult to implement it on a large scale.
  • document JP 2016-205910 describes a system and method for monitoring a photovoltaic plant. For this, he makes use of photographs of the modules that make up said plant.
  • This document describes the use of a classifier to identify the presence of foreign material in the modules, where the classifier is trained with images of clean modules and modules with foreign material.
  • the system described in this document does not allow to determine the power loss due to the presence of the foreign material. Consequently, a system and method are required to overcome the deficiencies of the state of the art.
  • the present invention provides a method for measuring the power loss of photovoltaic solar modules that is characterized in that it comprises the steps of: acquiring, by means of a processor, a digital image; recognizing, in said image, at least one region of interest corresponding to a photovoltaic solar module, by means of said processor; acquiring, through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyzing said region of interest and said irradiance information by using, by said processor, a trained convolutional neural network, obtaining, after using said convolutional neural network, a plurality of probability values corresponding to a plurality of power loss percentage values, by said processor; and determining said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor.
  • the method is characterized in that it comprises the step of capturing said digital image by means of a camera, prior to the step of acquiring said digital image by means of said processor.
  • the method is characterized in that said step of acquiring said digital image by said processor comprises transmitting said digital image from said camera to said processor.
  • said step of transmitting said digital image from said camera to said processor is a wired transmission, a transmission wireless, as well as a combination between them.
  • the method is characterized in that it additionally comprises the step of storing said captured digital image in a storage memory and in that said step of acquiring said digital image comprises reading said digital image from said storage memory, by means of said processor.
  • the method is characterized in that said step of recognizing said region of interest comprises evaluating said image by means of a second trained convolutional neural network that is executed by said processor.
  • said second convolutional neural network results in a binary matrix, said binary matrix that classifies the pixels according to whether or not they correspond to a photovoltaic solar module.
  • said second convolutional neural network additionally delivers the number of photovoltaic solar modules detected as a result.
  • the method is characterized in that said irradiance information is obtained by means of a pyranometer that is in communication with said processor.
  • the method is characterized in that it further comprises obtaining a timestamp corresponding to said digital image and in that said irradiance information corresponds to the moment indicated by said timestamp.
  • the method is characterized in that it is performed substantially in real time.
  • the method is characterized in that, prior to the step of analyzing said region of interest and said irradiance information, it comprises correcting the perspective and dimension of each recognized solar module, by means of said processor.
  • the method is characterized in that it additionally comprises displaying said image on a screen operatively connected to said processor, wherein said display comprises indicating, for each photovoltaic solar module recognized in said image, said percent power loss.
  • said indication is made by means of a graduated color scale.
  • the present invention also provides a system for measuring the power loss of photovoltaic solar modules that is characterized in that it comprises a processor configured to: acquire a digital image; recognizing, in said image, at least one region of interest corresponding to a photovoltaic solar module; acquiring irradiance information of the region in which said photovoltaic solar module is located; analyzing said region of interest and said irradiance information by using a trained convolutional neural network, obtaining, after using said convolutional neural network, a plurality of probability values corresponding to a plurality of loss percentage values of power; and determining said power loss percentage as that which corresponds to a maximum probability value.
  • the system is characterized in that it additionally comprises a camera in communication with said processor.
  • system is characterized in that it additionally comprises a pyranometer in communication with said processor.
  • system is characterized in that it additionally comprises a display operatively connected to said processor.
  • Fig. 1 illustrates a flowchart of a first embodiment of the method that is the subject of the present invention.
  • Fig. 2 illustrates a schematic diagram of a second embodiment of the method that is the subject of the present invention.
  • Fig. 3 illustrates a schematic diagram of a first embodiment of the system that is the object of the present invention.
  • Fig. 4 shows a grayscale photograph in which two photovoltaic solar modules are included.
  • Fig. 5 schematically shows a photograph in which a first region of interest has been highlighted.
  • Fig. 6 schematically shows a photograph in which a second region of interest has been highlighted.
  • Fig. 7 shows a grayscale photograph in which a photovoltaic solar module is included.
  • Fig. 8 schematically illustrates the region of interest corresponding to the photovoltaic solar module of Figure 7.
  • Fig. 9 illustrates a graph in which the edges of the region of interest of Figure 8 are schematically illustrated.
  • Fig. 10 illustrates the image of Figure 7, after an affine transformation has been applied to it.
  • a method for measuring the power loss of photovoltaic solar modules which essentially comprises the steps of: acquiring (1), by means of a processor (11), a digital image; recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11); acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyze (4) said region of interest and said irradiance information through the use, by said processor (11), of a trained convolutional neural network (41), obtain (5), after the use of said convolutional neural network (41), a plurality of probability values that correspond to a plurality of power loss percentage values, by said processor (11); and determining (6) said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor (11).
  • said processor (11) can acquire (1) said image by any way known to a person normally skilled in the art.
  • said processor (11) can acquire said image from a camera (12).
  • said processor (11) can download said image from the Internet, read said image from a storage memory, or obtain said image from a remote server, as well as a combination of them.
  • said processor (11) can acquire a plurality of images.
  • a plurality shall be understood as two or more of the elements to which reference is made. The number of elements that form part of said plurality does not limit the scope of the present invention as long as it is greater than or equal to two.
  • the method comprises the steps of: recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11 ); acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyze (4) said region of interest and said irradiance information through the use, by said processor (11), of a trained convolutional neural network (41), obtain (5), after the use of said convolutional neural network (41), a plurality of probability values that correspond to a plurality of power loss percentage values, by said processor (11); and determining (6) said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor (11).
  • the digital image that is acquired by said processor (11) can be any image in the visible spectrum.
  • the visible spectrum shall be understood as that range of wavelengths in which light is perceptible by the human eye and which extends, without this limiting the scope of the present invention, between 380 nm and 780 nm.
  • said image is a color image. Said image can be described in any color space, such as, for example and without limitation, RGB, RGBA, HSV, HSL, CMY or CMYK. Additionally, the dimensions of said image do not limit the scope of the present invention.
  • said processor (11) acquires (1) said image from a camera (12)
  • said camera (12) can capture (7) said image prior to the acquisition (1) of said image by said processor (11).
  • Said camera (12) can transmit said digital image to said processor (11), store it in a storage memory (13), store it in a remote repository or in the cloud, as well as a combination of these options, without this limiting the scope of the present invention.
  • said camera (12) can transmit said image to said processor (11).
  • Said transmission can be carried out by wired or wired means, as well as a combination between them, without this limiting the scope of the present invention.
  • said camera (12) can be connected to said processor (11) by means of a cable selected from the group consisting of USB cable, UTP cable, STP cable, optical fiber , coaxial cable, as well as a combination thereof.
  • said camera (12) can be wirelessly connected to said processor through a connection selected from the group consisting of Wi-Fi connections, Bluetooth connections, Li-Fi, as well as a combination between them.
  • the transmission of said image from said camera (12) to said processor (11) can be done through a wide area network (WAN, for its acronym in English Wide Area NetWork), for example, Internet.
  • WAN wide area network
  • said camera (12) can store said image in a storage memory (13).
  • Said storage memory (13) can be an internal memory or a removable memory without this limiting the scope of the present invention. Additionally, the capacity of said memory does not limit the scope of the present invention.
  • said processor (11) can acquire (1) said image by reading said image from said storage memory.
  • the method that is the object of the present invention further comprises the step of recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11).
  • said region of interest may correspond to all or part of said photovoltaic solar module.
  • said processor (11) recognizes said region of interest (21) does not limit the scope of the present invention.
  • said processor (11) can recognize (2) said region of interest (21) by evaluating it with a second trained convolutional neural network.
  • Said second neural network can be trained with images of modules obtained with webscouting techniques so that it is capable of detecting solar panels.
  • said second convolutional neural network can be based on an instantaneous segmentation Mask RCNN convolutional neural network.
  • Said second convolutional neural network can deliver as a result a binary matrix that classifies the pixels of the image according to whether or not they correspond to a photovoltaic solar module.
  • said second convolutional neural network can individually identify and segment each module, isolating it from the whole scene.
  • Said second convolutional neural network will then analyze each pixel of the image individually and assign it a label, depending on whether said pixel corresponds to a module. Said analysis will generate one or more clouds of pixels that will each correspond to a region of interest (21), and each of these clouds will be considered as an individual module.
  • said second convolutional neural network can deliver as a result a matrix of dimensions (A, L, NI), where A and L represent the width and length in pixels of the image respectively, and NI the number of detected modules.
  • the content of the array may consist of a binary array, where each pixel is assigned a value of 1 if it belongs to the panel category, and 0 otherwise.
  • the Figure 4 shows a grayscale photograph that can be acquired by the processor (11).
  • Figures 5 illustrate the same image, but in which a first region of interest (21) has been highlighted, corresponding to a first photovoltaic solar module.
  • Figure 6 illustrates the same image as in Figure 4, but in which a second region of interest (21) has been highlighted, corresponding to a second solar module. photovoltaic.
  • Said processor (11) can, optionally and without this limiting the scope of the present invention, perform additional processing of said image.
  • said processor (11) can crop said image, rotate said image, center said region of interest (21), as well as a combination of all these processes.
  • said processor (11) can correct the perspective and dimension of each solar module recognized in said image.
  • said processor (11) can execute an affine transformation to correct perspective and an automated resizing taking as reference the corners of each solar module recognized in said image. .
  • figure 7 shows a photograph in which a photovoltaic solar module is included and figure 8 shows the region of interest (21), corresponding to the photovoltaic solar module of figure 7.
  • said processor can execute an ant algorithm that identifies the edges of said region of interest (21).
  • said processor (11) can define a quadrilateral from said region of interest (21). For this, for each identified edge, a linear regression is applied, in such a way that the lines that define said quadrilateral correspond to the edges of the photovoltaic solar module.
  • said processor (11) can find the intersections between them, which it will identify as the corners of the photovoltaic solar module.
  • Figure 9 illustrates a graph which includes the edges detected by said processor, the lines that best fit said edges and the intersections between said lines.
  • FIG. 10 shows a modified photograph that corresponds to an affine transformation applied to the photograph of Figure 7, using the vertex positions indicated in Figure 9.
  • the method additionally comprises acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located.
  • Said irradiance information allows, for example and without this limiting the scope of the present invention, to take into account the weather condition of the day on which the image is captured.
  • said irradiance information makes it possible to identify cloudy days, sunny days or a combination of both, and consider said information when determining the power loss of the photovoltaic solar module.
  • the region in which said photovoltaic solar module is located shall be understood as a geographical extension in which the climatic conditions are substantially the same as in the position in which said photovoltaic solar module is located.
  • said region may be defined in terms of a radius around said photovoltaic solar module.
  • Said radius can be, for example and without this limiting the scope of the present invention, 10 km, preferably 5 km and even more preferably 1 km.
  • said region may be defined by irregular shapes around said photovoltaic solar module that a specialist has determined to have substantially the same climatic conditions.
  • the means by which said processor (11) acquires said irradiance information does not limit the scope of the present invention.
  • said processor (1 1 ) can download said irradiance information from the internet, read said irradiance information from a storage memory, obtain said irradiance information from a remote server, as well as a combination of these options.
  • said processor (11) can obtain said irradiance information through a pyranometer (31) that is in communication with said processor (11) and that is positioned in said region in which said photovoltaic solar module is located.
  • Said communication between said pyranometer (31) and said processor (11) can be wired or wireless, as well as a combination of both, without this limiting the scope of the present invention.
  • said pyranometer (31) can be connected to said processor (11) by means of a cable selected from the group consisting of USB cable, UTP cable, STP cable, optical fiber , coaxial cable, as well as a combination thereof.
  • said pyranometer (31) can be wirelessly connected to said processor through a connection selected from the group consisting of Wi-Fi connections, Bluetooth connections, Li-Fi, as well as a combination between them.
  • the transmission of said irradiance information from said pyranometer (31) to said processor (11) can be done through a wide area network (WAN, for its Wide Area NetWork), for example, Internet.
  • WAN wide area network
  • said pyranometer (31) can store said irradiance information in a storage memory.
  • Said storage memory can be an internal memory or a removable memory without this limiting the scope of the present invention. Additionally, the capacity of said memory does not limit the scope of the present invention.
  • said processor (11) can acquire (3) said irradiance information by reading said irradiance information from said storage memory.
  • Said image and said irradiance information can comprise additional information, commonly known as metadata, which can, in a preferred embodiment, be read by said processor (11).
  • said digital image may comprise a timestamp indicating the moment in which it was captured.
  • said irradiance information may include information regarding the time at which said irradiance measurement was obtained.
  • said processor (11) can obtain said time stamp corresponding to said digital image and acquire (3) said irradiance information corresponding to the moment indicated by said time stamp. time of said digital image.
  • the method that is the subject of the present invention can be performed substantially in real time.
  • the method is carried out substantially in real time when the time difference between the capture of the digital image, the measurement of the irradiance information and the determination of the power loss of the photovoltaic solar module is less than a threshold value.
  • said threshold value can be 10 seconds, preferably 1 second and even more preferably 0.5 seconds.
  • the method that is the object of the present invention additionally comprises the step of analyzing (4) said region of interest and said irradiance information through the use, by said processor (11), of a trained convolutional neural network (41 ).
  • the training of such a convolutional neural network (41) typically involves two stages, namely training and validation. In the training stage, you are given a set of images labeled with their irradiance and their respective level of loss.
  • the validation stage provides the network with a set of images independent of the training set and verifies that the power loss percentages determined by the convolutional neural network for the image set agree with the corresponding power loss values. to each image.
  • any convolutional neural network architecture can be used to implement such a trained convolutional neural network (41).
  • said convolutional neural network can be chosen from the group formed by RESNET, INCEPTION, VGG, among others.
  • Said trained convolutional neural network (41) delivers as a result a plurality of probability values that correspond to a plurality of power loss percentage values.
  • Said plurality of probability values can be understood, without this limiting the scope of the present invention, as a vector of probability values.
  • the number of probability values that said vector forms does not limit the scope of the present invention and will depend, for example and without this limiting the scope of the present invention, on the required precision, as well as the uncertainty values that are assigned to it.
  • it may be required to determine if said power loss is within a range or if it is above a threshold value. In this sense, a person normally versed in the matter will understand how to configure said trained convolutional neural network (41) to satisfy said requirements.
  • said vector of probability values contains a number of probability values greater than 5, more preferably greater than 10 and even more preferably greater than 20, which may correspond, for example, and without this limiting the scope of the present invention, with power loss percentage intervals of 20%, 10% or 5% respectively.
  • said processor (11) can determine (6) the percentage of power loss of each photovoltaic solar module. For this, said processor (11) can recognize said power loss percentage as that which corresponds to a maximum value of probability, as schematically illustrated in Figure 2.
  • the method may comprise an additional stage of displaying (8) said image on a screen (81) operatively connected to said processor. (eleven ).
  • Said display also comprises indicating, for each photovoltaic solar module recognized in said image, its corresponding percentage of power loss.
  • Said indication can be made in any manner known to a person normally skilled in the art.
  • said indication can be made using a graduated color scale. For this, a graphic scale can be displayed indicating the value corresponding to each color and each region of interest (21) can be colored with the color corresponding to its percentage of power loss.
  • the present invention also provides a system for measuring the power loss of photovoltaic solar modules comprising a processor (11) configured to: acquire (1) a digital image; recognize (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module; acquiring (3) irradiance information of the region in which said photovoltaic solar module is located; analyzing (4) said region of interest and said irradiance information by using a trained convolutional neural network, obtaining (5), after using said convolutional neural network, a plurality of probability values corresponding to a plurality of power loss percentage values; Y determining (6) said power loss percentage as the one that corresponds to a maximum probability value.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The present invention relates to a method for measuring power loss in photovoltaic solar modules, characterised in that it comprises the steps of: acquiring a digital image; recognising in said image at least one region of interest that corresponds to a photovoltaic solar module; acquiring irradiance information about the region in which said photovoltaic solar module is located; analysing said region of interest and said irradiance information by using a trained convolutional neural network; obtaining a plurality of probability values that correspond to a plurality of power loss percentage values; and determining said power loss percentage as that which corresponds to a maximum probability value. The present invention further relates to a method for measuring power loss in photovoltaic solar modules.

Description

MÉTODO Y SISTEMA PARA LA MEDICIÓN DE LA PÉRDIDA DE POTENCIA DE MÓDULOS SOLARES FOTOVOLTAICOS METHOD AND SYSTEM FOR MEASURING THE POWER LOSS OF PHOTOVOLTAIC SOLAR MODULES
CAMPO TÉCNICO DE LA INVENCIÓN TECHNICAL FIELD OF THE INVENTION
La presente invención se relaciona con el campo de la medición y ensayo, más específicamente con la medición mediante medios ópticos y en particular proporciona un método y sistema para la medición de la pérdida de potencia de módulos solares fotovoltaicos. The present invention relates to the field of measurement and testing, more specifically with measurement by optical means, and in particular provides a method and system for measuring the power loss of photovoltaic solar modules.
ANTECEDENTES DE LA INVENCIÓN BACKGROUND OF THE INVENTION
Diversas tecnologías se han desarrollado para evaluar el impacto del ensuciamiento ( soiling ) en módulos solares fotovoltaicos, ya sea basándose en variables eléctricas, ambientales o en análisis de imágenes. Various technologies have been developed to assess the impact of soiling in photovoltaic solar modules, either based on electrical or environmental variables or image analysis.
Por ejemplo, el documento WO 2014/154515 describe un método para determinar la contaminación de un módulo solar fotovoltaico, por ejemplo, debido a soiling, a partir del análisis de imágenes. Para esto, se mide el contraste en la imagen y se determina una función de transferencia de contraste que es una función de los valores de contraste medidos. De esta forma, se utiliza una curva de calibración que permite determinar el grado de soiling. Sin embargo, el método descrito en este documento requiere la adición de elementos de evaluación que poseen bandas blancas y negras para determinar los valores de contraste, lo cual dificulta la implementación del mismo a gran escala. For example, WO 2014/154515 describes a method for determining contamination of a photovoltaic solar module, eg due to soiling, from image analysis. For this, the contrast in the image is measured and a contrast transfer function is determined which is a function of the measured contrast values. In this way, a calibration curve is used to determine the degree of soiling. However, the method described in this document requires the addition of evaluation elements that have black and white bands to determine the contrast values, which makes it difficult to implement it on a large scale.
Por otra parte, el documento JP 2016-205910 describe un sistema y método para el monitoreo de una planta fotovoltaica. Para esto, hace uso de fotografías de los módulos que forman dicha planta. Este documento describe el uso de un clasificador para identificar la presencia de material externo en los módulos, en donde el clasificador se entrena con imágenes de módulos limpios y de módulos con material externo. Sin embargo, el sistema descrito en este documento no permite determinar la pérdida de potencia debido a la presencia del material externo. En consecuencia, se requiere de un sistema y método que permita superar las deficiencias del estado de la técnica. On the other hand, document JP 2016-205910 describes a system and method for monitoring a photovoltaic plant. For this, he makes use of photographs of the modules that make up said plant. This document describes the use of a classifier to identify the presence of foreign material in the modules, where the classifier is trained with images of clean modules and modules with foreign material. However, the system described in this document does not allow to determine the power loss due to the presence of the foreign material. Consequently, a system and method are required to overcome the deficiencies of the state of the art.
SUMARIO DE LA INVENCIÓN SUMMARY OF THE INVENTION
La presente invención proporciona un método para la medición de la pérdida de potencia de módulos solares fotovoltaicos que se caracteriza porque comprende los pasos de: adquirir, mediante un procesador, una imagen digital; reconocer, en dicha imagen, al menos una región de interés correspondiente a un módulo solar fotovoltaico, mediante dicho procesador; adquirir, mediante dicho procesador, información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar dicha región de interés y dicha información de irradiancia mediante la utilización, por parte de dicho procesador, de una red neuronal convolucional entrenada, obtener, luego de la utilización de dicha red neuronal convolucional, una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia, mediante dicho procesador; y determinar dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad, mediante dicho procesador. The present invention provides a method for measuring the power loss of photovoltaic solar modules that is characterized in that it comprises the steps of: acquiring, by means of a processor, a digital image; recognizing, in said image, at least one region of interest corresponding to a photovoltaic solar module, by means of said processor; acquiring, through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyzing said region of interest and said irradiance information by using, by said processor, a trained convolutional neural network, obtaining, after using said convolutional neural network, a plurality of probability values corresponding to a plurality of power loss percentage values, by said processor; and determining said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor.
En una realización preferida, el método se caracteriza porque comprende el paso de capturar dicha imagen digital mediante una cámara, previo al paso de adquirir dicha imagen digital mediante dicho procesador. En una realización más preferida, el método se caracteriza porque dicho paso de adquirir dicha imagen digital mediante dicho procesador comprende transmitir dicha imagen digital desde dicha cámara a dicho procesador. En una realización aún más preferida, el método se caracteriza porque dicho paso de transmitir dicha imagen digital desde dicha cámara a dicho procesador es una transmisión cableada, una transmisión inalámbrica, así como una combinación entre ellas. En otra realización más preferida, el método se caracteriza porque comprende, adicionalmente, el paso de almacenar dicha imagen digital capturada en una memoria de almacenamiento y porque dicho paso de adquirir dicha imagen digital comprende leer dicha imagen digital desde dicha memoria de almacenamiento, mediante dicho procesador. In a preferred embodiment, the method is characterized in that it comprises the step of capturing said digital image by means of a camera, prior to the step of acquiring said digital image by means of said processor. In a more preferred embodiment, the method is characterized in that said step of acquiring said digital image by said processor comprises transmitting said digital image from said camera to said processor. In an even more preferred embodiment, the method is characterized in that said step of transmitting said digital image from said camera to said processor is a wired transmission, a transmission wireless, as well as a combination between them. In another more preferred embodiment, the method is characterized in that it additionally comprises the step of storing said captured digital image in a storage memory and in that said step of acquiring said digital image comprises reading said digital image from said storage memory, by means of said processor.
En otra realización preferida, el método se caracteriza porque dicho paso de reconocer dicha región de interés comprende evaluar dicha imagen mediante una segunda red neuronal convolucional entrenada que es ejecutada mediante dicho procesador. En una realización más preferida, el método se caracteriza porque dicha segunda red neuronal convolucional entrega como resultado una matriz binaria, dicha matriz binaria que clasifica los pixeles de acuerdo a si corresponden o no a un módulo solar fotovoltaico. En otra realización más preferida, el método se caracteriza porque dicha segunda red neuronal convolucional, adicionalmente, entrega como resultado el número de módulos solares fotovoltaicos detectados. In another preferred embodiment, the method is characterized in that said step of recognizing said region of interest comprises evaluating said image by means of a second trained convolutional neural network that is executed by said processor. In a more preferred embodiment, the method is characterized in that said second convolutional neural network results in a binary matrix, said binary matrix that classifies the pixels according to whether or not they correspond to a photovoltaic solar module. In another more preferred embodiment, the method is characterized in that said second convolutional neural network additionally delivers the number of photovoltaic solar modules detected as a result.
En una realización preferida adicional, el método se caracteriza porque dicha información de irradiancia se obtiene mediante un piranómetro que se encuentra en comunicación con dicho procesador. In a further preferred embodiment, the method is characterized in that said irradiance information is obtained by means of a pyranometer that is in communication with said processor.
En otra realización preferida, el método se caracteriza porque comprende, adicionalmente, obtener una marca de tiempo correspondiente con dicha imagen digital y porque dicha información de irradiancia corresponde al momento indicado por dicha marca de tiempo. In another preferred embodiment, the method is characterized in that it further comprises obtaining a timestamp corresponding to said digital image and in that said irradiance information corresponds to the moment indicated by said timestamp.
En una realización preferida, el método se caracteriza porque se realiza sustancialmente en tiempo real. In a preferred embodiment, the method is characterized in that it is performed substantially in real time.
En otra realización preferida, el método se caracteriza porque, previo al paso de analizar dicha región de interés y dicha información de irradiancia, comprende corregir la perspectiva y dimensión de cada módulo solar reconocido, mediante dicho procesador. In another preferred embodiment, the method is characterized in that, prior to the step of analyzing said region of interest and said irradiance information, it comprises correcting the perspective and dimension of each recognized solar module, by means of said processor.
En una realización preferida adicional, el método se caracteriza porque comprende, adicionalmente, desplegar dicha imagen en una pantalla conectada operativamente con dicho procesador, en donde dicho despliegue comprende indicar, para cada módulo solar fotovoltaico reconocido en dicha imagen, dicho porcentaje de pérdida de potencia. En una realización más preferida, el método se caracteriza porque dicha indicación se realiza mediante una escala de colores graduada. In a further preferred embodiment, the method is characterized in that it additionally comprises displaying said image on a screen operatively connected to said processor, wherein said display comprises indicating, for each photovoltaic solar module recognized in said image, said percent power loss. In a more preferred embodiment, the method is characterized in that said indication is made by means of a graduated color scale.
La presente invención proporciona, además, un sistema para la medición de la pérdida de potencia de módulos solares fotovoltaicos que se caracteriza porque comprende un procesador configurado para: adquirir una imagen digital; reconocer, en dicha imagen, al menos una región de interés correspondiente a un módulo solar fotovoltaico; adquirir información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar dicha región de interés y dicha información de irradiancia mediante la utilización de una red neuronal convolucional entrenada, obtener, luego de la utilización de dicha red neuronal convolucional, una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia; y determinar dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad. The present invention also provides a system for measuring the power loss of photovoltaic solar modules that is characterized in that it comprises a processor configured to: acquire a digital image; recognizing, in said image, at least one region of interest corresponding to a photovoltaic solar module; acquiring irradiance information of the region in which said photovoltaic solar module is located; analyzing said region of interest and said irradiance information by using a trained convolutional neural network, obtaining, after using said convolutional neural network, a plurality of probability values corresponding to a plurality of loss percentage values of power; and determining said power loss percentage as that which corresponds to a maximum probability value.
En una realización preferida, el sistema se caracteriza porque comprende, adicionalmente, una cámara en comunicación con dicho procesador. In a preferred embodiment, the system is characterized in that it additionally comprises a camera in communication with said processor.
En otra realización preferida, el sistema se caracteriza porque comprende, adicionalmente, un piranómetro en comunicación con dicho procesador. In another preferred embodiment, the system is characterized in that it additionally comprises a pyranometer in communication with said processor.
En una realización preferida adicional, el sistema se caracteriza porque comprende, adicionalmente, una pantalla conectada operativamente con dicho procesador. In a further preferred embodiment, the system is characterized in that it additionally comprises a display operatively connected to said processor.
BREVE DESCRIPCIÓN DE LAS FIGURAS BRIEF DESCRIPTION OF THE FIGURES
La Fig. 1 ilustra un diagrama de flujo de una primera realización del método que es objeto de la presente invención. La Fig. 2 ilustra un diagrama esquemático de una segunda realización del método que es objeto de la presente invención. Fig. 1 illustrates a flowchart of a first embodiment of the method that is the subject of the present invention. Fig. 2 illustrates a schematic diagram of a second embodiment of the method that is the subject of the present invention.
La Fig. 3 ilustra un diagrama esquemático de una primera realización del sistema que es objeto de la presente invención. La Fig. 4 muestra una fotografía en escala de grises en la cual se incluyen dos módulos solares fotovoltaicos. Fig. 3 illustrates a schematic diagram of a first embodiment of the system that is the object of the present invention. Fig. 4 shows a grayscale photograph in which two photovoltaic solar modules are included.
La Fig. 5 muestra, esquemáticamente, una fotografía en la cual se ha resaltado una primera región de interés. Fig. 5 schematically shows a photograph in which a first region of interest has been highlighted.
La Fig. 6 muestra, esquemáticamente, una fotografía en la cual se ha resaltado una segunda región de interés. Fig. 6 schematically shows a photograph in which a second region of interest has been highlighted.
La Fig. 7 muestra una fotografía en escala de grises en la cual se incluye un módulo solar fotovoltaico. Fig. 7 shows a grayscale photograph in which a photovoltaic solar module is included.
La Fig. 8 ilustra, esquemáticamente, la región de interés correspondiente con el módulo solar fotovoltaico de la Figura 7. La Fig. 9 ilustra un gráfico en el que se ilustran, esquemáticamente, los bordes de la región de interés de la Figura 8. Fig. 8 schematically illustrates the region of interest corresponding to the photovoltaic solar module of Figure 7. Fig. 9 illustrates a graph in which the edges of the region of interest of Figure 8 are schematically illustrated.
La Fig. 10 ilustra la imagen de la Figura 7, luego de que se le ha aplicado una transformación afín. DESCRIPCIÓN DETALLADA DE LA INVENCIÓN Fig. 10 illustrates the image of Figure 7, after an affine transformation has been applied to it. DETAILED DESCRIPTION OF THE INVENTION
A continuación, se describirá de manera detallada la presente invención, haciendo referencia para esto a las figuras que acompañan la presente invención. In the following, the present invention will be described in detail, referring to the figures that accompany the present invention.
En un primer objeto de la presente invención, se proporciona un método para la medición de la pérdida de potencia de módulos solares fotovoltaicos que comprende, esencialmente, los pasos de: adquirir (1 ), mediante un procesador (11 ), una imagen digital; reconocer (2), en dicha imagen, al menos una región de interés (21) correspondiente a un módulo solar fotovoltaico, mediante dicho procesador (11 ); adquirir (3), mediante dicho procesador, información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar (4) dicha región de interés y dicha información de irradiancia mediante la utilización, por parte de dicho procesador (11 ), de una red neuronal convolucional entrenada (41 ), obtener (5), luego de la utilización de dicha red neuronal convolucional (41 ), una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia, mediante dicho procesador (11 ); y determinar (6) dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad, mediante dicho procesador (11 ). In a first object of the present invention, a method for measuring the power loss of photovoltaic solar modules is provided, which essentially comprises the steps of: acquiring (1), by means of a processor (11), a digital image; recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11); acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyze (4) said region of interest and said irradiance information through the use, by said processor (11), of a trained convolutional neural network (41), obtain (5), after the use of said convolutional neural network (41), a plurality of probability values that correspond to a plurality of power loss percentage values, by said processor (11); and determining (6) said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor (11).
En el contexto de la presente invención, sin que esto limite el alcance de la presente invención, se entenderá que dicho procesador (11 ) puede adquirir (1) dicha imagen mediante cualquier modo conocido para una persona normalmente versada en la materia. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede adquirir dicha imagen desde una cámara (12). Sin embargo, en otras realizaciones preferidas, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede descargar dicha imagen desde internet, leer dicha imagen desde una memoria de almacenamiento, u obtener dicha imagen desde un servidor remoto, así como una combinación entre las mismas. In the context of the present invention, without this limiting the scope of the present invention, it will be understood that said processor (11) can acquire (1) said image by any way known to a person normally skilled in the art. For example, and without this limiting the scope of the present invention, said processor (11) can acquire said image from a camera (12). However, in other preferred embodiments, without this limiting the scope of the present invention, said processor (11) can download said image from the Internet, read said image from a storage memory, or obtain said image from a remote server, as well as a combination of them.
De manera opcional, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede adquirir una pluralidad de imágenes. En el contexto de la presente invención, sin que esto limite el alcance de la misma, se entenderá como pluralidad a dos o más de los elementos a los que se hace referencia. El número de elementos que forme parte de dicha pluralidad no limita el alcance de la presente invención en tanto sea mayor o igual que dos. Optionally, without this limiting the scope of the present invention, said processor (11) can acquire a plurality of images. In the context of the present invention, without this limiting its scope, a plurality shall be understood as two or more of the elements to which reference is made. The number of elements that form part of said plurality does not limit the scope of the present invention as long as it is greater than or equal to two.
En aquellas realizaciones preferida en las cuales el procesador (11 ) adquiere una pluralidad de imágenes, sin que esto limite el alcance de la presente invención, por cada una de dichas imágenes que forman parte de la pluralidad, el método comprende los pasos de: reconocer (2), en dicha imagen, al menos una región de interés (21 ) correspondiente a un módulo solar fotovoltaico, mediante dicho procesador (11 ); adquirir (3), mediante dicho procesador, información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar (4) dicha región de interés y dicha información de irradiancia mediante la utilización, por parte de dicho procesador (11 ), de una red neuronal convolucional entrenada (41 ), obtener (5), luego de la utilización de dicha red neuronal convolucional (41 ), una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia, mediante dicho procesador (11 ); y determinar (6) dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad, mediante dicho procesador (11 ). In those preferred embodiments in which the processor (11) acquires a plurality of images, without this limiting the scope of the present invention, for each of said images that are part of the plurality, the method comprises the steps of: recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11 ); acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyze (4) said region of interest and said irradiance information through the use, by said processor (11), of a trained convolutional neural network (41), obtain (5), after the use of said convolutional neural network (41), a plurality of probability values that correspond to a plurality of power loss percentage values, by said processor (11); and determining (6) said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor (11).
En lo sucesivo, se describirá la invención haciendo referencia a una única imagen. Sin embargo, debe entenderse que aquellas opciones descritas para dicha imagen de manera individual pueden ser aplicables a una pluralidad de imágenes sin que esto limite el alcance de la presente invención. Hereinafter, the invention will be described with reference to a single image. However, it should be understood that those options described for said image individually may be applicable to a plurality of images without this limiting the scope of the present invention.
La imagen digital que es adquirida por dicho procesador (11 ) puede ser cualquier imagen en el espectro visible. En el contexto de la presente invención, sin que esto limite el alcance de la misma, se entenderá como espectro visible a aquel rango de longitudes de onda en la cual la luz es perceptible por el ojo humano y que se extiende, sin que esto limite el alcance de la presente invención, entre los 380 nm y los 780 nm. En una realización preferida, sin que esto limite el alcance de la presente invención, dicha imagen es una imagen en colores. Dicha imagen puede estar descrita en cualquier espacio de colores, como puede ser, por ejemplo y sin limitarse a estos, RGB, RGBA, HSV, HSL, CMY o CMYK. Adicionalmente, las dimensiones de dicha imagen no limitan el alcance de la presente invención. En aquellas realizaciones preferidas en las cuales dicho procesador (11 ) adquiere (1 ) dicha imagen desde una cámara (12), dicha cámara (12) puede capturar (7) dicha imagen previo a la adquisición (1) de dicha imagen por parte de dicho procesador (11 ). Dicha cámara (12) puede transmitir dicha imagen digital a dicho procesador (11 ), almacenarla en una memoria de almacenamiento (13), almacenarla en un repositorio remoto o en la nube, así como una combinación de estas opciones, sin que esto limite el alcance de la presente invención. The digital image that is acquired by said processor (11) can be any image in the visible spectrum. In the context of the present invention, without this limiting its scope, the visible spectrum shall be understood as that range of wavelengths in which light is perceptible by the human eye and which extends, without this limiting the scope of the present invention, between 380 nm and 780 nm. In a preferred embodiment, without this limiting the scope of the present invention, said image is a color image. Said image can be described in any color space, such as, for example and without limitation, RGB, RGBA, HSV, HSL, CMY or CMYK. Additionally, the dimensions of said image do not limit the scope of the present invention. In those preferred embodiments in which said processor (11) acquires (1) said image from a camera (12), said camera (12) can capture (7) said image prior to the acquisition (1) of said image by said processor (11). Said camera (12) can transmit said digital image to said processor (11), store it in a storage memory (13), store it in a remote repository or in the cloud, as well as a combination of these options, without this limiting the scope of the present invention.
En una realización preferida, sin que esto limite el alcance de la presente invención, dicha cámara (12) puede transmitir dicha imagen a dicho procesador (11 ). Dicha transmisión puede realizarse mediante medios cableados o alámbricos, así como una combinación entre los mismos, sin que esto limite el alcance de la presente invención. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicha cámara (12) puede estar conectada a dicho procesador (11) mediante un cable que se selecciona del grupo formado por cable USB, cable UTP, cable STP, fibra óptica, cable coaxial, así como una combinación entre los mismos. En otro ejemplo, y sin que esto limite el alcance de la presente invención, dicha cámara (12) puede estar conectada de manera inalámbrica con dicho procesador mediante una conexión que se selecciona del grupo formado por conexiones Wi-Fi, conexiones por Bluetooth, conexiones Li-Fi, así como una combinación entre las mismas. En una realización preferida, sin que esto limite el alcance de la presente invención, la transmisión de dicha imagen desde dicha cámara (12) hacia dicho procesador (11 ) puede realizarse a través de una red de área amplia (WAN, por sus siglas en inglés Wide Area NetWork), por ejemplo, internet. In a preferred embodiment, without this limiting the scope of the present invention, said camera (12) can transmit said image to said processor (11). Said transmission can be carried out by wired or wired means, as well as a combination between them, without this limiting the scope of the present invention. For example, and without this limiting the scope of the present invention, said camera (12) can be connected to said processor (11) by means of a cable selected from the group consisting of USB cable, UTP cable, STP cable, optical fiber , coaxial cable, as well as a combination thereof. In another example, and without this limiting the scope of the present invention, said camera (12) can be wirelessly connected to said processor through a connection selected from the group consisting of Wi-Fi connections, Bluetooth connections, Li-Fi, as well as a combination between them. In a preferred embodiment, without this limiting the scope of the present invention, the transmission of said image from said camera (12) to said processor (11) can be done through a wide area network (WAN, for its acronym in English Wide Area NetWork), for example, Internet.
En otra realización preferida, sin que esto limite el alcance de la presente invención, dicha cámara (12) puede almacenar dicha imagen en una memoria de almacenamiento (13). Dicha memoria de almacenamiento (13) puede ser una memoria interna o una memoria extraíble sin que esto limite el alcance de la presente invención. Adicionalmente, la capacidad de dicha memoria no limita el alcance de la presente invención. En esta realización preferida, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede adquirir (1 ) dicha imagen mediante la lectura de dicha imagen desde dicha memoria de almacenamiento. El método que es objeto de la presente invención comprende, adicionalmente, el paso de reconocer (2), en dicha imagen, al menos una región de interés (21 ) correspondiente a un módulo solar fotovoltaico, mediante dicho procesador (11 ). En el contexto de la presente invención, sin que esto limite el alcance de la presente invención, dicha región de interés puede corresponder a la totalidad o a una parte de dicho módulo solar fotovoltaico. La forma en la cual dicho procesador (11 ) reconozca dicha región de interés (21 ) no limita el alcance de la presente invención. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede reconocer (2) dicha región de interés (21 ) mediante su evaluación con una segunda red neuronal convolucional entrenada. Dicha segunda red neuronal puede ser entrenada con imágenes de módulos obtenidas con técnicas de webscouting para que sea capaz de detectar paneles solares. Además, sin que esto limite el alcance de la presente invención, dicha segunda red neuronal convolucional puede estar basada en una red neuronal convolucional Mask RCNN de segmentación instancial. In another preferred embodiment, without this limiting the scope of the present invention, said camera (12) can store said image in a storage memory (13). Said storage memory (13) can be an internal memory or a removable memory without this limiting the scope of the present invention. Additionally, the capacity of said memory does not limit the scope of the present invention. In this preferred embodiment, without this limiting the scope of the present invention, said processor (11) can acquire (1) said image by reading said image from said storage memory. The method that is the object of the present invention further comprises the step of recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11). In the context of the present invention, without this limiting the scope of the present invention, said region of interest may correspond to all or part of said photovoltaic solar module. The way in which said processor (11) recognizes said region of interest (21) does not limit the scope of the present invention. For example, and without this limiting the scope of the present invention, said processor (11) can recognize (2) said region of interest (21) by evaluating it with a second trained convolutional neural network. Said second neural network can be trained with images of modules obtained with webscouting techniques so that it is capable of detecting solar panels. In addition, without this limiting the scope of the present invention, said second convolutional neural network can be based on an instantaneous segmentation Mask RCNN convolutional neural network.
Dicha segunda red neuronal convolucional puede entregar como resultado una matriz binaria que clasifica los pixeles de la imagen de acuerdo a si corresponden o no a un módulo solar fotovoltaico. Para esto, dicha segunda red neuronal convolucional puede identificar y segmentar individualmente cada módulo, aislándolo de la escena completa. Dicha segunda red neuronal convolucional, luego, analizará cada pixel de la imagen individualmente y le asignará una etiqueta, dependiendo si dicho pixel corresponde a un módulo. Dicho análisis generará una o más nubes de pixeles que corresponderán, cada una, a una región de interés (21 ), y cada una de estas nubes se considerará como un módulo individual. Said second convolutional neural network can deliver as a result a binary matrix that classifies the pixels of the image according to whether or not they correspond to a photovoltaic solar module. For this, said second convolutional neural network can individually identify and segment each module, isolating it from the whole scene. Said second convolutional neural network will then analyze each pixel of the image individually and assign it a label, depending on whether said pixel corresponds to a module. Said analysis will generate one or more clouds of pixels that will each correspond to a region of interest (21), and each of these clouds will be considered as an individual module.
En un ejemplo de realización, sin que esto limite el alcance de la presente invención, dicha segunda red neuronal convolucional puede entregar como resultado una matriz de dimensiones (A, L, NI), donde A y L representan el ancho y largo en pixeles de la imagen respectivamente, y NI el número de módulos detectados. En este ejemplo, sin que esto limite el alcance de la presente invención, el contenido de la matriz puede consistir en un conjunto binario, donde a cada pixel se le asigna un valor de 1 si pertenece a la categoría de panel, y 0 en cualquier otro caso. Por ejemplo, y sin que esto limite el alcance de la presente invención, la Figura 4 muestra una fotografía en escala de grises que puede ser adquirida por el procesador (11). A su vez, las Figuras 5 ilustra la misma imagen, pero en la cual se ha resaltado una primera región de interés (21 ) que corresponde a un primer módulo solar fotovoltaico. De manera equivalente, sin que esto limite el alcance de la presente invención, la Figura 6 ilustra la misma imagen que en la Figura 4, pero en la cual se ha resaltado una segunda región de interés (21 ) que corresponde a un segundo módulo solar fotovoltaico. In an embodiment, without this limiting the scope of the present invention, said second convolutional neural network can deliver as a result a matrix of dimensions (A, L, NI), where A and L represent the width and length in pixels of the image respectively, and NI the number of detected modules. In this example, without limiting the scope of the present invention, the content of the array may consist of a binary array, where each pixel is assigned a value of 1 if it belongs to the panel category, and 0 otherwise. another case. For example, and without limiting the scope of the present invention, the Figure 4 shows a grayscale photograph that can be acquired by the processor (11). In turn, Figures 5 illustrate the same image, but in which a first region of interest (21) has been highlighted, corresponding to a first photovoltaic solar module. Equivalently, without this limiting the scope of the present invention, Figure 6 illustrates the same image as in Figure 4, but in which a second region of interest (21) has been highlighted, corresponding to a second solar module. photovoltaic.
Dicho procesador (11 ) puede, de manera opcional y sin que esto limite el alcance de la presente invención, realizar un procesamiento adicional de dicha imagen. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede recortar dicha imagen, rotar dicha imagen, centrar dicha región de interés (21 ), así como una combinación de todos esos procesamientos. En una realización preferida, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede corregir la perspectiva y dimensión de cada módulo solar reconocido en dicha imagen. Para esto, en una realización preferida y sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede ejecutar una transformación afín para corregir la perspectiva y un redimensionamiento automatizado tomando como referencia las esquinas de cada módulo solar reconocido en dicha imagen. Por ejemplo, y sin que esto limite el alcance de la presente invención, la figura 7 muestra una fotografía en la cual se incluye un módulo solar fotovoltaico y la figura 8 muestra la región de interés (21 ), correspondiente con el módulo solar fotovoltaico de la figura 7. Said processor (11) can, optionally and without this limiting the scope of the present invention, perform additional processing of said image. For example, and without this limiting the scope of the present invention, said processor (11) can crop said image, rotate said image, center said region of interest (21), as well as a combination of all these processes. In a preferred embodiment, without this limiting the scope of the present invention, said processor (11) can correct the perspective and dimension of each solar module recognized in said image. For this, in a preferred embodiment and without this limiting the scope of the present invention, said processor (11) can execute an affine transformation to correct perspective and an automated resizing taking as reference the corners of each solar module recognized in said image. . For example, and without this limiting the scope of the present invention, figure 7 shows a photograph in which a photovoltaic solar module is included and figure 8 shows the region of interest (21), corresponding to the photovoltaic solar module of figure 7.
En una primera etapa, dicho procesador puede ejecutar un algoritmo hormiga que identifica los bordes de dicha región de interés (21 ). Con posterioridad a la detección de los bordes, dicho procesador (11 ) puede definir un cuadrilátero a partir de dicha región de interés (21 ). Para esto, por cada borde identificado, se aplica una regresión lineal, de forma tal que las rectas que definen dicho cuadrilátero se corresponden con los bordes del módulo solar fotovoltaico. Utilizando dichas rectas, dicho procesador (11 ) puede encontrar las intersecciones entre ellas, las cuales identificará como las esquinas del módulo solar fotovoltaico. Por ejemplo, y sin que esto limite el alcance de la presente invención, la Figura 9 ilustra un gráfico en el cual se incluyen los bordes detectados por dicho procesador, las rectas que mejor ajustan dichos bordes y las intersecciones entre dichas rectas. In a first stage, said processor can execute an ant algorithm that identifies the edges of said region of interest (21). After the detection of the edges, said processor (11) can define a quadrilateral from said region of interest (21). For this, for each identified edge, a linear regression is applied, in such a way that the lines that define said quadrilateral correspond to the edges of the photovoltaic solar module. Using said straight lines, said processor (11) can find the intersections between them, which it will identify as the corners of the photovoltaic solar module. For example, and without limiting the scope of the present invention, Figure 9 illustrates a graph which includes the edges detected by said processor, the lines that best fit said edges and the intersections between said lines.
Luego de encontrar dichas intersecciones, que se corresponden con las esquinas del módulo, dicho procesador (11 ) realiza una corrección de perspectiva y un redimensionamiento de dicha región de interés (21 ), aplicando para esto una transformación afín. Por ejemplo, y sin que esto limite el alcance de la presente invención, la Figura 10 muestra una fotografía modificada que se corresponde con una transformación afín aplicada a la fotografía de la Figura 7, utilizando las posiciones de los vértices indicados en la Figura 9. After finding said intersections, which correspond to the corners of the module, said processor (11) performs a perspective correction and a resizing of said region of interest (21), applying an affine transformation for this. For example, and without limiting the scope of the present invention, Figure 10 shows a modified photograph that corresponds to an affine transformation applied to the photograph of Figure 7, using the vertex positions indicated in Figure 9.
El método comprende, adicionalmente, adquirir (3), mediante dicho procesador, información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico. Dicha información de irradiancia permite, por ejemplo y sin que esto limite el alcance de la presente invención, tomar en consideración la condición climática del día en el cual se captura la imagen. Por ejemplo, dicha información de irradiancia permite identificar días nublados, días soleados o una combinación entre ambas, y considerar dicha información al momento de determinar la pérdida de potencia del módulo solar fotovoltaico. En el contexto de la presente invención, sin que esto limite el alcance de la misma, se entenderá como la región en la cual se encuentra dicho módulo solar fotovoltaico a una extensión geográfica en la cual las condiciones climáticas sean sustancialmente las mismas que en la posición en la cual se encuentra dicho módulo solar fotovoltaico. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicha región puede estar definida en términos de un radio alrededor de dicho módulo solar fotovoltaico. Dicho radio puede ser, por ejemplo y sin que esto limite el alcance de la presente invención, de 10 km, preferentemente de 5 km y aún más preferentemente, de 1 km. Sin embargo, en otras realizaciones preferidas y sin que esto limite el alcance de la presente invención, dicha región puede estar definida por formas irregulares en torno de dicho módulo solar fotovoltaico que un especialista ha determinado que poseen sustancialmente las mismas condiciones climáticas. The method additionally comprises acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located. Said irradiance information allows, for example and without this limiting the scope of the present invention, to take into account the weather condition of the day on which the image is captured. For example, said irradiance information makes it possible to identify cloudy days, sunny days or a combination of both, and consider said information when determining the power loss of the photovoltaic solar module. In the context of the present invention, without this limiting its scope, the region in which said photovoltaic solar module is located shall be understood as a geographical extension in which the climatic conditions are substantially the same as in the position in which said photovoltaic solar module is located. For example, and without this limiting the scope of the present invention, said region may be defined in terms of a radius around said photovoltaic solar module. Said radius can be, for example and without this limiting the scope of the present invention, 10 km, preferably 5 km and even more preferably 1 km. However, in other preferred embodiments and without this limiting the scope of the present invention, said region may be defined by irregular shapes around said photovoltaic solar module that a specialist has determined to have substantially the same climatic conditions.
Los medios mediante los cuales dicho procesador (11 ) adquiera dicha información de irradiancia no limita el alcance de la presente invención. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicho procesador (1 1 ) puede descargar dicha información de irradiancia desde internet, leer dicha información de irradiancia desde una memoria de almacenamiento, obtener dicha información de irradiancia desde un servidor remoto, así como una combinación entre dichas opciones. En una realización preferida, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede obtener dicha información de irradiancia mediante un piranómetro (31 ) que se encuentra en comunicación con dicho procesador (11 ) y que se posiciona en dicha región en la cual se encuentra dicho módulo solar fotovoltaico. The means by which said processor (11) acquires said irradiance information does not limit the scope of the present invention. For example, and without this limiting the scope of the present invention, said processor (1 1 ) can download said irradiance information from the internet, read said irradiance information from a storage memory, obtain said irradiance information from a remote server, as well as a combination of these options. In a preferred embodiment, without this limiting the scope of the present invention, said processor (11) can obtain said irradiance information through a pyranometer (31) that is in communication with said processor (11) and that is positioned in said region in which said photovoltaic solar module is located.
Dicha comunicación entre dicho piranómetro (31 ) y dicho procesador (11 ) puede ser cableada o inalámbrica, así como una combinación entre ambas, sin que esto limite el alcance de la presente invención. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicho piranómetro (31 ) puede estar conectado a dicho procesador (11 ) mediante un cable que se selecciona del grupo formado por cable USB, cable UTP, cable STP, fibra óptica, cable coaxial, así como una combinación entre los mismos. En otro ejemplo, y sin que esto limite el alcance de la presente invención, dicho piranómetro (31 ) puede estar conectado de manera inalámbrica con dicho procesador mediante una conexión que se selecciona del grupo formado por conexiones Wi-Fi, conexiones por Bluetooth, conexiones Li-Fi, así como una combinación entre las mismas. En una realización preferida, sin que esto limite el alcance de la presente invención, la transmisión de dicha información de irradiancia desde dicho piranómetro (31) hacia dicho procesador (11 ) puede realizarse a través de una red de área amplia (WAN, por sus siglas en inglés Wide Area NetWork), por ejemplo, internet. Said communication between said pyranometer (31) and said processor (11) can be wired or wireless, as well as a combination of both, without this limiting the scope of the present invention. For example, and without this limiting the scope of the present invention, said pyranometer (31) can be connected to said processor (11) by means of a cable selected from the group consisting of USB cable, UTP cable, STP cable, optical fiber , coaxial cable, as well as a combination thereof. In another example, and without this limiting the scope of the present invention, said pyranometer (31) can be wirelessly connected to said processor through a connection selected from the group consisting of Wi-Fi connections, Bluetooth connections, Li-Fi, as well as a combination between them. In a preferred embodiment, without this limiting the scope of the present invention, the transmission of said irradiance information from said pyranometer (31) to said processor (11) can be done through a wide area network (WAN, for its Wide Area NetWork), for example, Internet.
En otra realización preferida, sin que esto limite el alcance de la presente invención, dicho piranómetro (31 ) puede almacenar dicha información de irradiancia en una memoria de almacenamiento. Dicha memoria de almacenamiento puede ser una memoria interna o una memoria extraíble sin que esto limite el alcance de la presente invención. Adicionalmente, la capacidad de dicha memoria no limita el alcance de la presente invención. En esta realización preferida, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede adquirir (3) dicha información de irradiancia mediante la lectura de dicha información de irradiancia desde dicha memoria de almacenamiento. In another preferred embodiment, without this limiting the scope of the present invention, said pyranometer (31) can store said irradiance information in a storage memory. Said storage memory can be an internal memory or a removable memory without this limiting the scope of the present invention. Additionally, the capacity of said memory does not limit the scope of the present invention. In this preferred embodiment, without this limiting the scope of the present invention, said processor (11) can acquire (3) said irradiance information by reading said irradiance information from said storage memory.
Dicha imagen y dicha información de irradiancia pueden comprender información adicional, comúnmente conocida como metadatos, que pueden, en una realización preferida, ser leídas por dicho procesador (11 ). Por ejemplo, y sin que esto limite el alcance de la presente invención, dicha imagen digital puede comprender una marca de tiempo que indica el momento en el cual fue capturada. De manera equivalente, sin que esto limite el alcance de la presente invención, dicha información de irradiancia puede incluir información respecto de la hora en la cual se obtuvo dicha medición de irradiancia. En esta realización preferida, sin que esto limite el alcance de la presente invención, dicho procesador (11 ) puede obtener dicha marca de tiempo correspondiente con dicha imagen digital y adquirir (3) dicha información de irradiancia que corresponde al momento indicado por dicha marca de tiempo de dicha imagen digital. Said image and said irradiance information can comprise additional information, commonly known as metadata, which can, in a preferred embodiment, be read by said processor (11). For example, and without this limiting the scope of the present invention, said digital image may comprise a timestamp indicating the moment in which it was captured. Equivalently, without this limiting the scope of the present invention, said irradiance information may include information regarding the time at which said irradiance measurement was obtained. In this preferred embodiment, without this limiting the scope of the present invention, said processor (11) can obtain said time stamp corresponding to said digital image and acquire (3) said irradiance information corresponding to the moment indicated by said time stamp. time of said digital image.
Sin embargo, en otras realizaciones preferidas, el método que es objeto de la presente invención puede ser realizado sustancialmente en tiempo real. En el contexto de la presente invención, sin que esto limite el alcance de la misma, se entenderá que el método se realiza sustancialmente en tiempo real cuando la diferencia temporal entre la captura de la imagen digital, la medición de la información de irradiancia y la determinación de la pérdida de potencia del módulo solar fotovoltaico es menor que un valor umbral. Por ejemplo, y sin que esto limite el alcance de la presente invención, dicho valor umbral puede ser 10 segundos, preferentemente 1 segundo y aún más preferentemente 0,5 segundos. Una persona normalmente versada en la materia notará que, cuando el método que es objeto de la presente invención se realiza sustancialmente en tiempo real, dicha imagen digital y dicha información de irradiancia no requieren de información temporal. However, in other preferred embodiments, the method that is the subject of the present invention can be performed substantially in real time. In the context of the present invention, without this limiting the scope of the same, it will be understood that the method is carried out substantially in real time when the time difference between the capture of the digital image, the measurement of the irradiance information and the determination of the power loss of the photovoltaic solar module is less than a threshold value. For example, and without this limiting the scope of the present invention, said threshold value can be 10 seconds, preferably 1 second and even more preferably 0.5 seconds. A person normally skilled in the art will notice that, when the method that is the object of the present invention is carried out substantially in real time, said digital image and said irradiance information do not require temporal information.
El método que es objeto de la presente invención comprende, adicionalmente, el paso de analizar (4) dicha región de interés y dicha información de irradiancia mediante la utilización, por parte de dicho procesador (11 ), de una red neuronal convolucional entrenada (41 ). El entrenamiento de dicha red neuronal convolucional (41 ) típicamente conlleva dos etapas, a saber, entrenamiento y validación. En la etapa de entrenamiento, se le entrega un conjunto de imágenes etiquetadas con su irradiancia y su respectivo nivel de pérdida. La etapa de validación le entrega a la red un conjunto de imágenes independientes del conjunto de entrenamiento y verifica que los porcentajes de pérdida de potencia determinado por la red neuronal convolucional para el conjunto de imágenes esté en concordancia con los valores de pérdida de potencia que corresponde a cada imagen. The method that is the object of the present invention additionally comprises the step of analyzing (4) said region of interest and said irradiance information through the use, by said processor (11), of a trained convolutional neural network (41 ). The training of such a convolutional neural network (41) typically involves two stages, namely training and validation. In the training stage, you are given a set of images labeled with their irradiance and their respective level of loss. The validation stage provides the network with a set of images independent of the training set and verifies that the power loss percentages determined by the convolutional neural network for the image set agree with the corresponding power loss values. to each image.
Cualquier arquitectura de red neuronal convolucional puede ser utilizada para implementar dicha red neuronal convolucional entrenada (41 ). Por ejemplo, y sin que esto limite el alcance de la presente invención, dicha red neuronal convolucional puede escogerse del grupo formado por RESNET, INCEPTION, VGG, entre otras. Any convolutional neural network architecture can be used to implement such a trained convolutional neural network (41). For example, and without this limiting the scope of the present invention, said convolutional neural network can be chosen from the group formed by RESNET, INCEPTION, VGG, among others.
Dicha red neuronal convolucional entrenada (41 ) entrega como resultado una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia. Dicha pluralidad de valores de probabilidad puede entenderse, sin que esto limite el alcance de la presente invención, como un vector de valores de probabilidad. El número de valores de probabilidad que forme dicho vector no limita el alcance de la presente invención y dependerá, por ejemplo y sin que esto limite el alcance de la presente invención, de la precisión requerida, así como de los valores de incerteza que se le atribuyan a la red neuronal convolucional entrenada (41 ). Adicionalmente, en ciertas realizaciones y sin que esto limite el alcance de la presente invención, puede requerirse determinar si dicha pérdida de potencia se encuentra dentro de un rango o si se encuentra por sobre un valor umbral. En este sentido, una persona normalmente versada en la materia entenderá cómo configurar dicha red neuronal convolucional entrenada (41 ) para satisfacer dichos requerimientos. Said trained convolutional neural network (41) delivers as a result a plurality of probability values that correspond to a plurality of power loss percentage values. Said plurality of probability values can be understood, without this limiting the scope of the present invention, as a vector of probability values. The number of probability values that said vector forms does not limit the scope of the present invention and will depend, for example and without this limiting the scope of the present invention, on the required precision, as well as the uncertainty values that are assigned to it. attributed to the trained convolutional neural network (41). Additionally, in certain embodiments and without this limiting the scope of the present invention, it may be required to determine if said power loss is within a range or if it is above a threshold value. In this sense, a person normally versed in the matter will understand how to configure said trained convolutional neural network (41) to satisfy said requirements.
En una realización preferida, dicho vector de valores probabilidades contiene un número de valores de probabilidad mayor que 5, más preferentemente mayor que 10 y aún más preferentemente mayor que 20, lo cual se puede corresponder, por ejemplo y sin que esto limite el alcance de la presente invención, con intervalos de porcentaje pérdida de potencia de 20%, 10% o 5% respectivamente. Con posterioridad a la obtención del vector de valores de probabilidad, dicho procesador (11 ) puede determinar (6) el porcentaje de pérdida de potencia de cada módulo solar fotovoltaico. Para esto, dicho procesador (11 ) puede reconocer dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad, tal como se ilustra esquemáticamente en la Figura 2. In a preferred embodiment, said vector of probability values contains a number of probability values greater than 5, more preferably greater than 10 and even more preferably greater than 20, which may correspond, for example, and without this limiting the scope of the present invention, with power loss percentage intervals of 20%, 10% or 5% respectively. After obtaining the vector of probability values, said processor (11) can determine (6) the percentage of power loss of each photovoltaic solar module. For this, said processor (11) can recognize said power loss percentage as that which corresponds to a maximum value of probability, as schematically illustrated in Figure 2.
De manera opcional, sin que esto limite el alcance de la presente invención y tal como se ilustra esquemáticamente en la Figura 3, el método puede comprender una etapa adicional de desplegar (8) dicha imagen en una pantalla (81 ) conectada operativamente con dicho procesador (11 ). Dicho despliegue comprende, además, indicar, para cada módulo solar fotovoltaico reconocido en dicha imagen, su correspondiente porcentaje de pérdida de potencia. Dicha indicación se puede hacer de cualquier manera conocida por una persona normalmente versada en la materia. De manera ventajosa, sin que esto limite el alcance de la presente invención, dicha indicación puede realizarse realiza mediante una escala de colores graduada. Para esto, se puede desplegar una escala gráfica indicando el valor correspondiente a cada color y cada región de interés (21 ) se puede colorear con el color correspondiente a su porcentaje de pérdida de potencia. Optionally, without this limiting the scope of the present invention and as schematically illustrated in Figure 3, the method may comprise an additional stage of displaying (8) said image on a screen (81) operatively connected to said processor. (eleven ). Said display also comprises indicating, for each photovoltaic solar module recognized in said image, its corresponding percentage of power loss. Said indication can be made in any manner known to a person normally skilled in the art. Advantageously, without this limiting the scope of the present invention, said indication can be made using a graduated color scale. For this, a graphic scale can be displayed indicating the value corresponding to each color and each region of interest (21) can be colored with the color corresponding to its percentage of power loss.
La presente invención proporciona, además, un sistema para la medición de la pérdida de potencia de módulos solares fotovoltaicos que comprende un procesador (11 ) configurado para: adquirir (1 ) una imagen digital; reconocer (2), en dicha imagen, al menos una región de interés (21 ) correspondiente a un módulo solar fotovoltaico; adquirir (3) información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar (4) dicha región de interés y dicha información de irradiancia mediante la utilización de una red neuronal convolucional entrenada, obtener (5), luego de la utilización de dicha red neuronal convolucional, una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia; y determinar (6) dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad. The present invention also provides a system for measuring the power loss of photovoltaic solar modules comprising a processor (11) configured to: acquire (1) a digital image; recognize (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module; acquiring (3) irradiance information of the region in which said photovoltaic solar module is located; analyzing (4) said region of interest and said irradiance information by using a trained convolutional neural network, obtaining (5), after using said convolutional neural network, a plurality of probability values corresponding to a plurality of power loss percentage values; Y determining (6) said power loss percentage as the one that corresponds to a maximum probability value.
Todas las opciones para el procesador (11), para la cámara (12), para el piranómetro (31) y para la pantalla (81) descritas en el contexto del método que es objeto de la presente invención pueden ser aplicables a dichos elementos como parte del sistema que es objeto de la presente invención sin que esto limite el alcance de la misma. All the options for the processor (11), for the camera (12), for the pyranometer (31) and for the screen (81) described in the context of the method that is the object of the present invention may be applicable to said elements as part of the system that is the object of the present invention without this limiting its scope.
De acuerdo con la descripción previamente detallada, es posible medir el porcentaje de pérdida de potencia en módulos solares fotovoltaicos a partir de imágenes en el espectro visible. According to the previously detailed description, it is possible to measure the percentage of power loss in photovoltaic solar modules from images in the visible spectrum.
Todas las opciones para las diferentes características técnicas descritas pueden ser combinadas entre sí, o con otras opciones conocidas para una persona normalmente versada en la materia, sin que esto limite el alcance de la presente invención. All the options for the different technical features described can be combined with each other, or with other options known to a person normally skilled in the art, without this limiting the scope of the present invention.

Claims

REIVINDICACIONES
1. Un método para la medición de la pérdida de potencia de módulos solares fotovoltaicos, CARACTERIZADO porque comprende los pasos de: adquirir (1 ), mediante un procesador (11 ), una imagen digital; reconocer (2), en dicha imagen, al menos una región de interés (21) correspondiente a un módulo solar fotovoltaico, mediante dicho procesador (11); adquirir (3), mediante dicho procesador, información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar (4) dicha región de interés y dicha información de irradiancia mediante la utilización, por parte de dicho procesador (11), de una red neuronal convolucional entrenada (41), obtener (5), luego de la utilización de dicha red neuronal convolucional (41), una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia, mediante dicho procesador (11); y determinar (6) dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad, mediante dicho procesador (11). 1. A method for measuring the power loss of photovoltaic solar modules, CHARACTERIZED in that it comprises the steps of: acquiring (1), through a processor (11), a digital image; recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module, by means of said processor (11); acquiring (3), through said processor, irradiance information of the region in which said photovoltaic solar module is located; analyzing (4) said region of interest and said irradiance information by using, by said processor (11), a trained convolutional neural network (41), obtaining (5), after using said convolutional neural network (41), a plurality of probability values corresponding to a plurality of power loss percentage values, by said processor (11); and determining (6) said power loss percentage as the one that corresponds to a maximum probability value, by means of said processor (11).
2. El método de la reivindicación 1 , CARACTERIZADO porque comprende el paso de capturar (7) dicha imagen digital mediante una cámara (12), previo al paso de adquirir (1) dicha imagen digital mediante dicho procesador (11). 2. The method of claim 1, CHARACTERIZED in that it comprises the step of capturing (7) said digital image by means of a camera (12), prior to the step of acquiring (1) said digital image by means of said processor (11).
3. El método de la reivindicación 2, CARACTERIZADO porque dicho paso de adquirir (1) dicha imagen digital mediante dicho procesador (11) comprende transmitir dicha imagen digital desde dicha cámara (12) a dicho procesador (11 ). 3. The method of claim 2, CHARACTERIZED in that said step of acquiring (1) said digital image by said processor (11) comprises transmitting said digital image from said camera (12) to said processor (11).
4. El método de la reivindicación 3, CARACTERIZADO porque dicho paso de transmitir dicha imagen digital desde dicha cámara (12) a dicho procesador (11 ) comprende una transmisión cableada o una transmisión inalámbrica, así como una combinación entre ellas. 4. The method of claim 3, CHARACTERIZED in that said step of transmitting said digital image from said camera (12) to said processor (11) comprises a wired transmission or a wireless transmission, as well as a combination between them.
5. El método de la reivindicación 2, CARACTERIZADO porque comprende, adicionalmente, el paso de almacenar dicha imagen digital capturada en una memoria de almacenamiento (13) y porque dicho paso de adquirir (1) dicha imagen digital comprende leer dicha imagen digital desde dicha memoria de almacenamiento (13), mediante dicho procesador (11). 5. The method of claim 2, CHARACTERIZED in that it additionally comprises the step of storing said captured digital image in a storage memory (13) and in that said step of acquiring (1) said digital image comprises reading said digital image from said storage memory (13), by said processor (11).
6. El método de la reivindicación 1, CARACTERIZADO porque dicho paso de reconocer (2) dicha región de interés (21) comprende evaluar dicha imagen mediante una segunda red neuronal convolucional entrenada que es ejecutada mediante dicho procesador (11). The method of claim 1, CHARACTERIZED in that said step of recognizing (2) said region of interest (21) comprises evaluating said image by means of a second trained convolutional neural network that is executed by said processor (11).
7. El método de la reivindicación 6, CARACTERIZADO porque dicha segunda red neuronal convolucional entrega como resultado una matriz binaria, dicha matriz binaria que clasifica los pixeles de acuerdo a si corresponden o no a un módulo solar fotovoltaico. 7. The method of claim 6, CHARACTERIZED in that said second convolutional neural network results in a binary matrix, said binary matrix that classifies the pixels according to whether or not they correspond to a photovoltaic solar module.
8. El método de la reivindicación 6, CARACTERIZADO porque dicha segunda red neuronal convolucional, adicionalmente, entrega como resultado el número de módulos solares fotovoltaicos detectados. 8. The method of claim 6, CHARACTERIZED in that said second convolutional neural network, additionally, delivers the number of photovoltaic solar modules detected as a result.
9. El método de la reivindicación 1, CARACTERIZADO porque dicha información de irradiancia se obtiene mediante un piranómetro (31) que se encuentra en comunicación con dicho procesador (11 ). 9. The method of claim 1, CHARACTERIZED in that said irradiance information is obtained by means of a pyranometer (31) that is in communication with said processor (11).
10. El método de la reivindicación 1 , CARACTERIZADO porque comprende, adicionalmente, obtener una marca de tiempo correspondiente con dicha imagen digital y porque dicha información de irradiancia corresponde al momento indicado por dicha marca de tiempo. 10. The method of claim 1, CHARACTERIZED in that it further comprises obtaining a time stamp corresponding to said digital image and in that said irradiance information corresponds to the moment indicated by said time stamp.
11. El método de la reivindicación 1 , CARACTERIZADO porque se realiza sustancialmente en tiempo real. 11. The method of claim 1, CHARACTERIZED in that it is carried out substantially in real time.
12. El método de la reivindicación 1 , CARACTERIZADO porque, previo al paso de analizar (4) dicha región de interés (21 ) y dicha información de irradiancia, comprende corregir la perspectiva y dimensión de cada módulo solar reconocido, mediante dicho procesador (11). 12. The method of claim 1, CHARACTERIZED in that, prior to the step of analyzing (4) said region of interest (21) and said irradiance information, it comprises correcting the perspective and dimension of each recognized solar module, by means of said processor (11 ).
13. El método de la reivindicación 1 , CARACTERIZADO porque comprende, adicionalmente, desplegar (8) dicha imagen en una pantalla (81) conectada operativamente con dicho procesador (11), en donde dicho despliegue comprende indicar, para cada módulo solar fotovoltaico reconocido en dicha imagen, dicho porcentaje de pérdida de potencia. 13. The method of claim 1, CHARACTERIZED in that it further comprises displaying (8) said image on a screen (81) operatively connected to said processor (11), wherein said display comprises indicating, for each photovoltaic solar module recognized in said image, said power loss percentage.
14. El método de la reivindicación 13, CARACTERIZADO porque dicha indicación se realiza mediante una escala de colores graduada. 14. The method of claim 13, CHARACTERIZED in that said indication is made by means of a graduated color scale.
15. Un sistema para la medición de la pérdida de potencia de módulos solares fotovoltaicos, CARACTERIZADO porque comprende un procesador (11) configurado para: adquirir (1 ) una imagen digital; reconocer (2), en dicha imagen, al menos una región de interés (21) correspondiente a un módulo solar fotovoltaico; adquirir (3) información de irradiancia de la región en la cual se encuentra dicho módulo solar fotovoltaico; analizar (4) dicha región de interés y dicha información de irradiancia mediante la utilización de una red neuronal convolucional entrenada, - obtener (5), luego de la utilización de dicha red neuronal convolucional, una pluralidad de valores de probabilidad que se corresponden con una pluralidad de valores de porcentaje de pérdida de potencia; y determinar (6) dicho porcentaje de pérdida de potencia como aquél que se corresponde con un valor máximo de probabilidad. 15. A system for measuring the power loss of photovoltaic solar modules, CHARACTERIZED in that it comprises a processor (11) configured to: acquire (1) a digital image; recognizing (2), in said image, at least one region of interest (21) corresponding to a photovoltaic solar module; acquiring (3) irradiance information of the region in which said photovoltaic solar module is located; analyzing (4) said region of interest and said irradiance information by using a trained convolutional neural network, - obtaining (5), after using said convolutional neural network, a plurality of probability values corresponding to a plurality of power loss percentage values; and determining (6) said power loss percentage as the one that corresponds to a maximum probability value.
16. El sistema de la reivindicación 15, CARACTERIZADO porque comprende, adicionalmente, una cámara (12) en comunicación con dicho procesador (11 ). 16. The system of claim 15, CHARACTERIZED in that it additionally comprises a camera (12) in communication with said processor (11).
17. El sistema de la reivindicación 15, CARACTERIZADO porque comprende, adicionalmente, un piranómetro (31) en comunicación con dicho procesador (11 ). 17. The system of claim 15, CHARACTERIZED in that it additionally comprises a pyranometer (31) in communication with said processor (11).
18. El sistema de la reivindicación 15, CARACTERIZADO porque comprende, adicionalmente, una pantalla (81) conectada operativamente con dicho procesador (11). The system of claim 15, CHARACTERIZED in that it further comprises a screen (81) operatively connected to said processor (11).
PCT/CL2021/050046 2021-05-24 2021-05-24 Method and system for measuring power loss in photovoltaic solar modules WO2022246580A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CL2021/050046 WO2022246580A1 (en) 2021-05-24 2021-05-24 Method and system for measuring power loss in photovoltaic solar modules

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CL2021/050046 WO2022246580A1 (en) 2021-05-24 2021-05-24 Method and system for measuring power loss in photovoltaic solar modules

Publications (1)

Publication Number Publication Date
WO2022246580A1 true WO2022246580A1 (en) 2022-12-01

Family

ID=84229359

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CL2021/050046 WO2022246580A1 (en) 2021-05-24 2021-05-24 Method and system for measuring power loss in photovoltaic solar modules

Country Status (1)

Country Link
WO (1) WO2022246580A1 (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2020018046A (en) * 2018-07-24 2020-01-30 株式会社東芝 Solar power generation monitoring system
US20200358396A1 (en) * 2019-05-06 2020-11-12 Arizona Board Of Regents On Behalf Of Arizona State University Solar array fault detection, classification, and localization using deep neural nets
CN112465812A (en) * 2020-12-16 2021-03-09 合肥阳光智维科技有限公司 Dust detection device and method for photovoltaic module
EP3726729B1 (en) * 2017-12-14 2021-05-12 Acciona Energía, S.A. Automated photovoltaic plant inspection system and method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3726729B1 (en) * 2017-12-14 2021-05-12 Acciona Energía, S.A. Automated photovoltaic plant inspection system and method
JP2020018046A (en) * 2018-07-24 2020-01-30 株式会社東芝 Solar power generation monitoring system
US20200358396A1 (en) * 2019-05-06 2020-11-12 Arizona Board Of Regents On Behalf Of Arizona State University Solar array fault detection, classification, and localization using deep neural nets
CN112465812A (en) * 2020-12-16 2021-03-09 合肥阳光智维科技有限公司 Dust detection device and method for photovoltaic module

Similar Documents

Publication Publication Date Title
Kazantzidis et al. Cloud detection and classification with the use of whole-sky ground-based images
CN101614675B (en) Visibility measuring system and method thereof
US10585210B2 (en) Apparatus for radiometric correction and orthorectification of aerial imagery
Dev et al. Estimating solar irradiance using sky imagers
US10685231B2 (en) Computer system, and method and program for diagnosing plants
JPWO2007000999A1 (en) Image analysis apparatus and image analysis method
Lagüela et al. Aerial thermography from low-cost UAV for the generation of thermographic digital terrain models
US11317570B2 (en) Peanut maturity grading systems and methods
CN112649900A (en) Visibility monitoring method, device, equipment, system and medium
CN113486975A (en) Ground object classification method, device, equipment and storage medium for remote sensing image
CN111932519A (en) Weather prediction method and device, computer equipment and storage medium
CN115170792A (en) Infrared image processing method, device and equipment and storage medium
Yang et al. Comparison of hyperspectral imagery with aerial photography and multispectral imagery for mapping broom snakeweed
CN106556579A (en) Group's mist image all-weather self-adapting detection method based on laser
CN112949411B (en) Spectral image correction method and device
CN108181307B (en) Visibility measuring system and method
WO2022246580A1 (en) Method and system for measuring power loss in photovoltaic solar modules
US20200398316A1 (en) Peanut maturity grading systems and methods
CN116756560A (en) Dust aerosol recognition method, system, model training method, medium and equipment
CN114002704A (en) Laser radar monitoring method and device for bridge tunnel and medium
CN109658405A (en) Image data quality control method and system in a kind of observation of crops outdoor scene
KR102373278B1 (en) Distortion Method of Total Cloude Cover in Night Time using Ground Based Whole Sky Image Data
US20220236190A1 (en) Detection method and detection pad
CN114119408A (en) Express delivery detection method and device based on cat eye camera and cat eye camera
CN117095316B (en) Road surface inspection method, device, equipment and readable storage medium

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21942171

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE