Intelligent monitoring platform for photovoltaic system construction

Abstract

The present invention discloses an intelligent monitoring platform for photovoltaic system construction, and relates to the field of photovoltaic monitoring technology. The platform includes: obtaining photovoltaic construction plan information of the photovoltaic system; obtaining multiple simulated photovoltaic panel array monitoring images according to the photovoltaic construction plan information; collecting multiple actual photovoltaic panel array monitoring images after the photovoltaic panel array construction is completed; performing global construction angle verification to obtain multiple global verification results; when any global verification result does not meet the requirements, dividing multiple photovoltaic panel images, performing construction angle verification, obtaining multiple verification result sets, and monitoring and managing photovoltaic panels that do not meet the verification requirements. The present invention solves the technical problems of insufficient accuracy of traditional photovoltaic construction monitoring methods in the prior art, which affect system performance and stability, and achieves the technical effect of improving the accuracy and timeliness of photovoltaic system construction monitoring, thereby effectively optimizing the performance and stability of the photovoltaic system.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic monitoring, and in particular to an intelligent monitoring platform for photovoltaic system construction. Background Art

[0002] Under the general trend of global energy transformation, photovoltaic power generation, as an important way to utilize renewable energy, is being used more and more widely. However, in the construction process of photovoltaic systems, monitoring is crucial. Traditional photovoltaic construction monitoring methods have obvious deficiencies in accuracy. Usually, these methods rely on limited manual measurements and simple instruments and equipment, and it is difficult to fully and carefully obtain various key parameters and information in construction. For example, there are often large errors in the measurement of important indicators such as the installation angle and position accuracy of photovoltaic panels.

[0003] The existing technology has the technical problem that the traditional photovoltaic construction monitoring method is insufficient in accuracy, which affects the system performance and stability. Summary of the invention

[0004] The present application provides an intelligent monitoring platform for photovoltaic system construction, which is used to solve the technical problems of insufficient accuracy of traditional photovoltaic construction monitoring methods and affecting system performance and stability.

[0005] In view of the above problems, the present application provides an intelligent monitoring platform for photovoltaic system construction.

[0006] The present application provides an intelligent monitoring platform for photovoltaic system construction, the platform comprising:

[0007] A photovoltaic construction plan acquisition module, the photovoltaic construction plan acquisition module is used to acquire photovoltaic construction plan information of the photovoltaic system, wherein the photovoltaic construction plan information includes construction location information and an array of installation angles of the photovoltaic panel array; a simulated photovoltaic panel array monitoring image acquisition module, the simulated photovoltaic panel array monitoring image acquisition module is used to generate photovoltaic panel array monitoring images in multiple time periods after the construction of the photovoltaic system according to the photovoltaic construction plan information, and obtain multiple simulated photovoltaic panel array monitoring images, wherein each simulated photovoltaic panel array monitoring image includes a photovoltaic panel image shape information array and a reflective feature information array of the photovoltaic panel array; a photovoltaic panel array monitoring image acquisition module, the photovoltaic panel array monitoring image acquisition module is used to acquire multiple actual photovoltaic panel array monitoring images of the photovoltaic panel array in the multiple time periods after the construction of the photovoltaic panel array is completed; a global verification result acquisition module, the global verification result acquisition module is used to acquire A global construction angle verification is performed using the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images to obtain multiple global verification results; a photovoltaic panel image set acquisition module is used to divide the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images into multiple photovoltaic panel images when any global verification result does not meet the global verification requirements, so as to obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets; a photovoltaic panel verification result acquisition module is used to perform construction angle verification on the multiple actual photovoltaic panel image sets and the multiple simulated photovoltaic panel image sets, obtain multiple shape verification result sets and multiple reflection verification result sets, and calculate to obtain multiple photovoltaic panel verification results, and for photovoltaic panels corresponding to photovoltaic panel verification results that do not meet the photovoltaic panel verification requirements, a monitoring management plan is decided to be obtained to perform monitoring management.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] A photovoltaic construction plan acquisition module is used to acquire photovoltaic construction plan information of a photovoltaic system; a simulated photovoltaic panel array monitoring image acquisition module is used to generate photovoltaic panel array monitoring images in multiple time periods after the construction of the photovoltaic system according to the photovoltaic construction plan information, and obtain multiple simulated photovoltaic panel array monitoring images; a photovoltaic panel array monitoring image acquisition module is used to acquire multiple actual photovoltaic panel array monitoring images of the photovoltaic panel array in the multiple time periods after the construction of the photovoltaic panel array is completed; a global verification result acquisition module is used to use multiple actual photovoltaic panel array monitoring images and multiple simulated photovoltaic panel array monitoring images to perform global construction angle verification and obtain multiple global verification results; a photovoltaic panel image set acquisition module is used to divide multiple photovoltaic panel images when any global verification result does not meet the global verification requirements, and obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets; a photovoltaic panel verification result acquisition module is used to perform construction angle verification on multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets, obtain multiple shape verification result sets and multiple reflection verification result sets, and for photovoltaic panels corresponding to the photovoltaic panel verification results that do not meet the photovoltaic panel verification requirements, a monitoring management plan is decided and obtained to perform monitoring management. The technical effect of improving the accuracy and timeliness of photovoltaic system construction monitoring has been achieved, thereby effectively optimizing the performance and stability of the photovoltaic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A schematic diagram of the structure of an intelligent monitoring platform for photovoltaic system construction provided in an embodiment of the present application;

[0012] Figure 2 A schematic diagram of the flow of a photovoltaic construction plan acquisition module in an intelligent monitoring platform for photovoltaic system construction provided in an embodiment of the present application.

[0013] Explanation of the reference numerals: photovoltaic construction plan acquisition module 10, simulated photovoltaic panel array monitoring image acquisition module 20, photovoltaic panel array monitoring image acquisition module 30, global verification result acquisition module 40, photovoltaic panel image set acquisition module 50, photovoltaic panel verification result acquisition module 60. DETAILED DESCRIPTION

[0014] This application provides an intelligent monitoring platform for photovoltaic system construction, which is used to solve the technical problems of insufficient accuracy of traditional photovoltaic construction monitoring methods and affecting system performance and stability.

[0015] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0016] Embodiment 1, as Figure 1 As shown, the present application provides an intelligent monitoring platform for photovoltaic system construction, the platform comprising:

[0017] The photovoltaic construction plan acquisition module 10 is used to acquire photovoltaic construction plan information of the photovoltaic system, wherein the photovoltaic construction plan information includes construction location information and installation angle array of the photovoltaic panel array.

[0018] Specifically, the photovoltaic construction plan information covers two core parts: construction location information and the installation angle array of the photovoltaic panel array. The construction location information specifies in detail the specific location where the photovoltaic system will be installed, which includes not only geographical coordinates, but also various specific conditions of the installation site, such as terrain, landform, surrounding environment, etc. These factors have an important impact on the installation and operation of the photovoltaic system. For example, different terrains require different installation methods, and the shading of the surrounding environment will affect the lighting effect of the photovoltaic panel. The installation angle array of the photovoltaic panel array accurately specifies the inclination angle and orientation of each photovoltaic panel during installation. The installation angle of the photovoltaic panel directly determines its efficiency in receiving sunlight, because the incident angle of sunlight changes with time and seasons. The appropriate installation angle can ensure that the photovoltaic panel absorbs sunlight to the maximum extent, thereby improving the power generation performance of the photovoltaic system. For example, in the northern hemisphere, photovoltaic panels are usually tilted at a certain angle toward the south to better receive sunlight in different seasons. Ensure the accuracy of the construction of the photovoltaic system and improve the power generation efficiency of the photovoltaic system.

[0019] A simulated photovoltaic panel array monitoring image acquisition module 20 is used to generate photovoltaic panel array monitoring images in multiple time periods after the construction of the photovoltaic system according to the photovoltaic construction plan information, and obtain multiple simulated photovoltaic panel array monitoring images, wherein each simulated photovoltaic panel array monitoring image includes a photovoltaic panel image shape information array and a reflective feature information array of the photovoltaic panel array.

[0020] Specifically, the simulated photovoltaic panel array monitoring image acquisition module 20 will obtain photovoltaic construction plan information, including construction location information. By analyzing the construction location, detailed illumination feature information of the location at different times, such as illumination angle and illumination intensity, can be obtained. These illumination feature information is crucial for accurately simulating the working state of the photovoltaic panel. Next, the simulated photovoltaic panel array monitoring image acquisition module 20 uses the technical means of generating adversarial networks. The generated adversarial network can accurately generate multiple simulated photovoltaic panel array monitoring images at multiple time points according to the installation angle array of the photovoltaic panel array and the position of the image acquisition device. Each generated simulated photovoltaic panel array monitoring image contains rich information. Among them, the photovoltaic panel image shape information array can accurately present the shape characteristics of the photovoltaic panel, including the specific shapes of the photovoltaic panels at different distances and installation angles in the image. At the same time, the reflective feature information array records the reflection of the photovoltaic panel in detail, such as the reflection angle and reflection intensity. This information can truly reflect the optical properties of the photovoltaic panel in actual work. In general, the role of this module is to generate photovoltaic panel array monitoring images that are highly similar to the actual construction through precise calculations and simulations before construction, providing important reference for subsequent analysis and decision-making.

[0021] The photovoltaic panel array monitoring image acquisition module 30 is used to acquire a plurality of actual photovoltaic panel array monitoring images of the photovoltaic panel array within the plurality of time periods after the photovoltaic panel array construction is completed.

[0022] Specifically, the main function of the photovoltaic panel array monitoring image acquisition module 30 is to collect monitoring images of the photovoltaic panel array within multiple pre-set time periods after the construction of the photovoltaic panel array is completed. When the photovoltaic panel array is installed and put into use, the module will start the image acquisition work, and shoot or monitor the photovoltaic panel array at preset time intervals to obtain multiple actual photovoltaic panel array monitoring images. These time periods are determined according to actual needs and system design, including different times of the day, different seasons or other specific time periods. By collecting these actual monitoring images, the module can obtain the working status and performance of the photovoltaic panel array in real time. These image data will be used to compare and analyze with the simulated photovoltaic panel array monitoring images to evaluate whether the actual operation of the photovoltaic panel array meets expectations and whether there are any potential problems or abnormalities, providing important data on the actual operation of the photovoltaic panel array, which helps to ensure the stable operation and efficient performance of the photovoltaic system.

[0023] The global verification result acquisition module 40 is used to use the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images to perform global construction angle verification to obtain multiple global verification results.

[0024] Specifically, the main function of the global verification result acquisition module 40 is to verify the global construction angle using multiple actual photovoltaic panel array monitoring images and multiple simulated photovoltaic panel array monitoring images. This module uses a twin network to identify the similarity between simulated images and actual images in multiple time periods. A twin network is a deep learning model that can extract and compare features of two input images to determine their similarity. In this module, the twin network inputs the simulated image and the actual image of each time period, and then outputs their similarity scores. If the construction quality is high, the similarity between the simulated image and the actual image should be relatively high, for example, reaching 90%. Therefore, the module will judge whether the construction angle meets the requirements based on the similarity score, and obtain multiple global verification results. These results can be used to evaluate the construction quality of the photovoltaic panel array and to find existing problems. For example, if the similarity score of a certain time period is low, it means that the construction angle of the time period may be biased and needs further inspection and adjustment. Through global verification of multiple time periods, the construction of the photovoltaic panel array can be fully understood to ensure that it meets the design requirements and expected performance.

[0025] The photovoltaic panel image set acquisition module 50 is used to divide the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images into multiple photovoltaic panel images when any global verification result does not meet the global verification requirements, so as to obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets.

[0026] Specifically, the main function of the photovoltaic panel image set acquisition module 50 is to divide the actual and simulated photovoltaic panel array monitoring images into photovoltaic panel images when the global verification results do not meet the requirements, so as to obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets. When it is found that the similarity between the simulated image and the actual image in any time period does not meet the requirements, this indicates that there is a problem of insufficient angle installation quality of the photovoltaic panel. In order to more accurately find out these problematic photovoltaic panels, it is necessary to divide the simulated images and actual images of multiple photovoltaic panels. Since there are multiple time periods and multiple photovoltaic panels, multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets will be formed. In this way, the image of each photovoltaic panel can be analyzed and verified separately, so as to more accurately determine which photovoltaic panels have problems with the installation angle, thereby improving the accuracy and timeliness of verification.

[0027] The photovoltaic panel verification result acquisition module 60 is used to verify the construction angles of the multiple actual photovoltaic panel image sets and the multiple simulated photovoltaic panel image sets, obtain multiple shape verification result sets and multiple reflection verification result sets, and calculate to obtain multiple photovoltaic panel verification results. For the photovoltaic panels corresponding to the photovoltaic panel verification results that do not meet the photovoltaic panel verification requirements, a monitoring management plan is decided to be obtained to perform monitoring and management.

[0028] Specifically, the photovoltaic panel verification result acquisition module 60 uses the twin network as a technical tool to carry out the construction angle verification work on multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets. The twin network can accurately identify the shape and reflective feature information in the actual image and simulated image of the same photovoltaic panel in the same time period, calculate the similarity between the two, and use this similarity as the verification result. These verification results are strictly screened. If the similarity between the actual image and the simulated image of a photovoltaic panel fails to meet the predetermined requirements, it indicates that the installation angle of the photovoltaic panel is inaccurate. After the above processing, the module can obtain multiple shape verification result sets and multiple reflection verification result sets. Based on these sets, multiple comprehensive photovoltaic panel verification results are further calculated. For those photovoltaic panels corresponding to the photovoltaic panel verification results that do not meet the photovoltaic panel verification requirements, the module will make decisions quickly, obtain monitoring and management plans specially customized for such situations, and immediately carry out corresponding monitoring and management work to improve the accuracy and timeliness of photovoltaic system construction monitoring, thereby effectively optimizing the performance and stability of the photovoltaic system.

[0029] In one possible implementation, Figure 2 As shown, the photovoltaic construction plan acquisition module 10 also includes:

[0030] Collecting construction location information of planned construction of the photovoltaic system; collecting installation angle arrays of planned construction of photovoltaic panel arrays in the photovoltaic system; integrating the construction location information and installation angle arrays to obtain photovoltaic construction plan information.

[0031] Specifically, the process of collecting construction location information for the planned construction of a photovoltaic system usually requires the use of a variety of technologies and methods. For example, the global positioning system (GPS) is used to accurately obtain the geographic coordinates of the construction site, including information such as longitude, latitude and altitude. At the same time, the geographic information system (GIS) is used to obtain the topographic and geomorphic features of the location, such as flat plains, rolling mountains, or hills with slopes. In addition, it is also crucial to understand the surrounding environment of the construction location. This includes the distribution and height of surrounding buildings, the presence of objects such as trees and hills that may cause obstructions, and local climatic conditions, such as average sunshine hours, rainfall, wind speed, etc. In order to obtain this information, field surveys can be carried out by professionals using measuring instruments to conduct on-site measurements and records. Satellite remote sensing images and aerial images can also be used for analysis to obtain large-area, macroscopic location information.

[0032] Collect the installation angle array of the planned construction of the photovoltaic panel array in the photovoltaic system. The installation angle has a direct impact on the efficiency of the photovoltaic panel in receiving solar energy. Different latitudes, seasons and local sunshine conditions need to be considered to determine the appropriate installation angle. In the actual collection process, professional measurement tools and software can be used. For example, a sun angle meter is used to determine the position of the sun at different times, and then the appropriate installation angle array is calculated based on this data.

[0033] Finally, the collected construction location information and installation angle array are integrated. This integration process combines two key elements into a complete photovoltaic construction plan information. Through integration, the overall picture of the photovoltaic system construction can be clearly presented, providing accurate and comprehensive guidance for subsequent construction, simulation, monitoring and verification. For example, the integrated information can make it clear that at a specific mountaintop location, the photovoltaic panel array should be installed according to a specific angle array to maximize power generation efficiency and adapt to the local environment.

[0034] In a possible implementation, the simulated photovoltaic panel array monitoring image acquisition module 20 further includes:

[0035] According to the photovoltaic construction plan information, multiple illumination feature information of the construction location information in the multiple time periods are collected; a simulation image generator for generating photovoltaic panel array monitoring images is constructed; the multiple illumination feature information and the photovoltaic construction plan information are input into the simulation image generator, and photovoltaic panel array monitoring images are generated respectively to obtain multiple simulation photovoltaic panel array monitoring images.

[0036] Specifically, a series of professional equipment and technical means are used to collect multiple illumination feature information of construction location information in multiple time periods based on photovoltaic construction plan information. For example, high-precision solar radiation monitors are used to measure the solar radiation intensity in different time periods in real time. These instruments can accurately record the value of solar energy reaching the construction location in each time period. At the same time, the illumination angle in different time periods is determined by combining astronomical calculations with geographic information systems (GIS). GIS can provide detailed information such as the geographic coordinates and terrain undulations of the construction location. Combined with astronomical algorithms, it can calculate the altitude and azimuth of the sun relative to the location in different seasons and at different time points. In addition, in order to obtain more comprehensive and accurate illumination feature information, the impact of local weather conditions on illumination will also be considered. For example, data such as cloud coverage and atmospheric transparency collected by meteorological stations are used to correct the measurement results of illumination intensity and angle. These illumination feature information are crucial for accurately simulating the working state of photovoltaic panel arrays.

[0037] A simulated image generator for photovoltaic panel array monitoring image generation is constructed based on a generative adversarial network (GAN). The generative adversarial network consists of a generator and a discriminator. During the training process, the generator and the discriminator continuously engage in adversarial games, and the generator strives to generate more realistic images so that it can deceive the discriminator. The discriminator continuously improves its discriminative ability to better distinguish between real images and generated simulated images. Through multiple iterative training, the performance of the generator gradually improves, and it can eventually generate highly realistic photovoltaic panel array monitoring images.

[0038] The collected and sorted multiple illumination feature information and detailed photovoltaic construction plan information are accurately input into the simulation image generator built based on the generative adversarial network. These input information provide key parameters and constraints for the simulation image generator. The illumination feature information includes specific data such as illumination intensity, angle, and spectral distribution in different time periods, while the photovoltaic construction plan information covers important details such as the geographical environment of the construction location, the layout of the photovoltaic panel array, and the installation angle. After receiving these inputs, the simulation image generator will use its internal complex calculation and generation mechanism to comprehensively consider various factors and gradually generate the corresponding photovoltaic panel array monitoring image. Due to the diversity of input information and the change of time dimension, the generator will calculate and generate for each time period and different illumination conditions respectively, so as to obtain multiple simulated photovoltaic panel array monitoring images. In this way, the multiple simulated photovoltaic panel array monitoring images finally obtained can comprehensively and accurately simulate the actual monitoring effect of the photovoltaic panel array under different illumination conditions and construction environments, providing an important reference basis for subsequent analysis and verification, and improving the accuracy and timeliness of photovoltaic system construction monitoring.

[0039] In a possible implementation, the simulated photovoltaic panel array monitoring image acquisition module 20 further includes:

[0040] A sample illumination feature information set and a sample photovoltaic construction plan information set are collected, and photovoltaic panel array monitoring images under different illumination feature information and sample photovoltaic construction plan information are collected to obtain a sample simulated photovoltaic panel array monitoring image set; based on a generative adversarial network, a generation path and a discrimination path are constructed within a simulation image generator; the sample illumination feature information set, the sample photovoltaic construction plan information set and the sample simulated photovoltaic panel array monitoring image set are used to perform supervised training on the generation path and the discrimination path until the generation convergence requirements are met, thereby obtaining a trained simulation image generator.

[0041] Specifically, the collection of sample illumination feature information sets involves detailed measurement and recording of various illumination conditions, including information on changes in illumination intensity, illumination angle, and illumination spectrum distribution in different time periods. The collection of sample photovoltaic construction plan information sets covers many factors, including the layout planning of photovoltaic panel arrays, the geographical environment characteristics of the installation location, and the models and specifications of photovoltaic panels. Under different combinations of illumination feature information and sample photovoltaic construction plan information, the photovoltaic panel array is actually monitored and its monitoring images are collected. These images reflect the actual performance of the photovoltaic panel array under specific illumination and construction conditions. By integrating the images collected under these different conditions, a sample simulated photovoltaic panel array monitoring image set is finally obtained, which provides a rich and diverse data foundation for the subsequent construction and training of the simulation image generator.

[0042] It is a critical and complex task to construct the generative path and discriminative path in the simulated image generator based on the generative adversarial network. Construction of the generative path: Using a deep neural network structure, such as a multi-layer convolutional neural network (CNN), the network input is sample illumination feature information and sample photovoltaic construction plan information; the network gradually learns how to generate realistic photovoltaic panel array monitoring images from the input data through a series of convolution, pooling and activation function operations; the goal of the generative path is to simulate the features and details similar to the real monitoring images as much as possible. Construction of the discriminative path: The discriminative path is also a deep neural network with a similar structure to the generative path, but different functions; its input is the simulated image generated by the generative path or the real sample simulated photovoltaic panel array monitoring image; the discriminative path determines whether the input image is real or generated by learning the features of the real image; for example, the generative path will learn how to generate features such as the arrangement of photovoltaic panels, light reflection and shadow from the input light intensity and construction plan information, while the discriminative path can distinguish some parts of the image generated by the generative path that do not conform to the real physical laws or the details are not accurate enough. When constructing these two paths, it is necessary to carefully design the network's parameters such as the number of layers, number of neurons, activation functions, etc. to ensure that they can effectively learn and process data, thereby achieving the purpose of generating realistic simulated images and accurately judging their authenticity.

[0043] In this training process, the sample illumination feature information set, the sample photovoltaic construction plan information set and the sample simulated photovoltaic panel array monitoring image set are first used as training data. For the generation path, it receives the sample illumination feature information and the sample photovoltaic construction plan information, and tries to generate a simulated photovoltaic panel array monitoring image. Then, the discriminant path simultaneously receives the image generated by the generation path and the real sample simulated photovoltaic panel array monitoring image, and judges their authenticity. In the early stage of training, the image generated by the generation path is often of poor quality, and it is easy to be identified as a false image by the discriminant path. As the training progresses, according to the feedback given by the discriminant path, the generation path continuously adjusts its own parameters to generate more realistic and more consistent with the characteristics of the real image. Simultaneously, the discriminant path is also constantly optimizing its judgment ability to more accurately distinguish between the real image and the generated image. This process will continue. Through continuous iterative optimization, the performance of the generation path and the discriminant path will gradually improve. When the generation convergence requirements are met, it means that the image generated by the generation path is realistic enough, and the discriminant path is difficult to accurately distinguish between the generated image and the real sample image. At this point, a trained simulated image generator is obtained. For example, assuming that the generation convergence requirement is that the average difference between the generated image and the real sample image is less than a certain threshold, if during the training process, after multiple iterations, the calculated average difference reaches or is less than this threshold, it means that the generation convergence requirement is met and the training is completed. The trained simulation image generator can generate high-quality simulated photovoltaic panel array monitoring images based on the newly input lighting feature information and photovoltaic construction plan information, providing strong support for subsequent analysis and application, and improving the accuracy and timeliness of photovoltaic system construction monitoring.

[0044] In a possible implementation, the global verification result acquisition module 40 further includes:

[0045] Collect multiple sample photovoltaic panel array monitoring image combinations, and analyze the similarity of two sample photovoltaic panel array monitoring images in each sample photovoltaic panel array monitoring image combination to obtain a sample global similarity set; build a global verifier based on the twin network; use the multiple sample photovoltaic panel array monitoring image combinations and the sample global similarity set to train the global verifier until convergence; combine the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images respectively, and input them into the global verifier to obtain multiple global similarities as multiple global verification results.

[0046] Specifically, multiple combinations consisting of two sample photovoltaic panel array monitoring images are collected. For each such combination, the similarity characteristics between the two sample images are deeply analyzed. A method based on image feature extraction and comparison is used to analyze the similarity. The image is converted into a grayscale image to simplify processing, and the image features, such as the contour features, surface texture features, and light intensity distribution features of the photovoltaic panel, are extracted. For the contour features, an edge detection algorithm is used to obtain the edge information of the photovoltaic panel and calculate the similarity of the edges. For the texture features, the roughness and direction of the texture are described by calculating the grayscale co-occurrence matrix, and the texture similarity of the two images is compared. For the light intensity distribution feature, the image is divided into multiple regions, the average light intensity of each region is calculated, and then the light intensity difference of the corresponding regions of the two images is compared. By combining the comparison results of these features, a weighted average algorithm is used to give a similarity value between 0 and 1. 0 means completely dissimilar, and 1 means completely identical. Such analysis and processing are performed on all the collected sample combinations, and all the similarity values ​​obtained constitute the sample global similarity set.

[0047] A global verifier is built based on the architecture of the Siamese network. The Siamese network consists of two sub-networks with the same structure and shared parameters, which can process and compare two input images simultaneously.

[0048] Multiple sample photovoltaic panel array monitoring image combinations are input into the two branches of the twin network of the global verifier. At the same time, there is already a corresponding sample global similarity set, which contains the real similarity information of these sample combinations. During the training process, the global verifier will calculate the similarity value of the input sample image combination based on its current parameters and weights, and compare the calculated similarity value with the real similarity in the sample global similarity set to obtain the error. According to this error, the parameters of the twin network in the global verifier, such as the weight and bias of the convolution layer, are adjusted through the back propagation algorithm to make the calculated similarity closer to the real similarity. This process will be repeated continuously, and the network parameters will be optimized in each iteration. As the training progresses, the performance of the global verifier gradually improves, and the calculated similarity becomes more and more accurate. The global verifier is considered to have converged until the output of the global verifier no longer changes significantly after multiple iterations, or the error reaches a pre-set small threshold. For example, initially for a certain sample image combination, the similarity calculated by the global verifier is 0.7, while the real sample global similarity is 0.9. After adjusting the parameters through back propagation, the next calculated similarity becomes 0.8. After multiple iterations, the final calculated similarity is close to 0.9, and the changes in subsequent iterations are very small. At this time, the training can be considered to have converged.

[0049] Finally, multiple actual photovoltaic panel array monitoring images and multiple simulated photovoltaic panel array monitoring images are combined in pairs, and these combinations are input into the trained global verifier in turn. The global verifier will process and analyze each combination, and output a global similarity value for each combination through its internal calculation and comparison mechanism. These output global similarity values ​​constitute multiple global verification results, which can reflect the similarity between each pair of actual and simulated photovoltaic panel array monitoring images. By analyzing and comparing a large number of sample photovoltaic panel array monitoring images, the actual operating status of the photovoltaic panel array can be more accurately evaluated, potential problems can be discovered in a timely manner, and monitoring accuracy can be improved.

[0050] In a possible implementation, the photovoltaic panel image set acquisition module 50 further includes:

[0051] Determine respectively whether the global similarities within the multiple global verification results are greater than or equal to the global similarity threshold, if so, the global verification requirement is met, if not, the global verification requirement is not met; when any global verification result does not meet the global verification requirement, divide and segment the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images into multiple photovoltaic panel images to obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets of the multiple photovoltaic panels in the photovoltaic panel array.

[0052] Specifically, for each global similarity in the multiple global verification results obtained, it is compared with a pre-set global similarity threshold. For example, the threshold here is set to 0.9 or 90%. The setting of this threshold is based on actual conditions. It is not 1 because there are errors in image acquisition and processing. If the global similarity in a global verification result is greater than or equal to this threshold, it means that it meets the global verification requirements. This means that the corresponding actual photovoltaic panel array monitoring image and the simulated photovoltaic panel array monitoring image have a high degree of similarity as a whole and meet the expected standards. However, if the global similarity in a global verification result is less than the threshold, it means that it does not meet the global verification requirements.

[0053] When it is found that any global verification result fails to meet the global verification requirements, it is necessary to perform more detailed processing on the previous multiple actual photovoltaic panel array monitoring images and multiple simulated photovoltaic panel array monitoring images. First, determine the layout of the photovoltaic panel array and the boundary information of each photovoltaic panel. Based on prior knowledge, such as the specifications and installation methods of photovoltaic panels, or automatically detect the edges and positions of each photovoltaic panel through image recognition technology. Then, based on these boundary information, the overall actual photovoltaic panel array monitoring image and simulated photovoltaic panel array monitoring image are accurately divided and segmented. Each independent photovoltaic panel is extracted from the entire array image to form multiple separate photovoltaic panel images. Two sets of photovoltaic panel image sets corresponding to the actual situation and the simulated situation are obtained. One set is a set of multiple actual photovoltaic panel images of multiple photovoltaic panels in the photovoltaic panel array, which includes each photovoltaic panel image segmented from the actual monitoring image; the other set is a set of multiple simulated photovoltaic panel images, which is the corresponding photovoltaic panel image segmented from the simulated image. It can more accurately locate the specific photovoltaic panels that cause the global verification to fail to meet the requirements, accurately locate the problem, and achieve targeted optimization.

[0054] In a possible implementation, the photovoltaic panel image set acquisition module 60 further includes:

[0055] A set of sample photovoltaic panel images is obtained, and the photovoltaic panel shape information and reflective feature information in each sample photovoltaic panel image are extracted to obtain a set of sample photovoltaic panel shape information and a set of sample reflective feature information; the sample photovoltaic panel image set, the sample photovoltaic panel shape information set and the sample reflective feature information set are used to train a photovoltaic panel image recognizer; the photovoltaic panel image recognizer is used to respectively recognize the multiple actual photovoltaic panel image sets and the multiple simulated photovoltaic panel image sets to obtain multiple actual photovoltaic panel shape information sets, multiple actual reflective feature information sets, multiple simulated photovoltaic panel shape information sets and multiple simulated reflective feature information sets; similarity calculation is performed on the multiple actual photovoltaic panel shape information sets and the multiple simulated photovoltaic panel shape information sets to obtain multiple shape similarity sets as multiple shape verification result sets; similarity calculation is performed on the multiple actual reflective feature information sets and the multiple simulated reflective feature information sets to obtain multiple reflective similarity sets as multiple reflective verification result sets; weighted calculation is performed on the multiple shape verification result sets and the multiple reflective verification result sets to obtain multiple verification result sets; the mean of the multiple verification result sets is calculated to obtain multiple photovoltaic panel verification results.

[0056] Specifically, a large number of sample photovoltaic panel images in different states and environments are collected through various channels to form a sample photovoltaic panel image set. The sources of these images include monitoring images of actually operating photovoltaic power stations, test images in laboratories, etc. For each sample photovoltaic panel image in the set, image processing and analysis techniques are used to extract key information. When extracting the shape information of the photovoltaic panel, an edge detection algorithm is used to determine the outline of the photovoltaic panel, and then its geometric parameters such as length, width, area, and perimeter, as well as shape features such as bending degree and angle are calculated. For the extraction of reflective feature information, the light intensity distribution on the surface of the photovoltaic panel in the image, the color and hue of the reflection, the size and position of the reflective area, etc. are analyzed. The shape information extracted from all sample photovoltaic panel images is summarized to obtain a sample photovoltaic panel shape information set; similarly, the summary of reflective feature information forms a sample reflective feature information set.

[0057] Prepare a set of sample photovoltaic panel images, a set of sample photovoltaic panel shape information, and a set of sample reflective feature information, and build the architecture of the photovoltaic panel image recognizer, which is a model based on deep learning convolutional neural network (CNN). Input the sample photovoltaic panel image set into the recognizer. During the forward propagation of the network, the image data is processed by a series of convolution, pooling, and fully connected layers to extract the features of the image. At the same time, the corresponding sample photovoltaic panel shape information set and sample reflective feature information set are used as supervisory signals to compare with the features output by the network. By defining a loss function, such as mean square error or cross entropy loss, the difference between the network output and the true shape and reflective feature information is measured. The optimization algorithm stochastic gradient descent (SGD) is used to adjust the network parameters (such as the weight and bias of the convolution kernel) according to the gradient of the loss function. After multiple iterations of training, the network gradually learns how to accurately extract and associate shape and reflective feature information from the input photovoltaic panel image. For example, during the training process, if the shape parameters predicted by the network for a certain image are significantly different from the actual sample photovoltaic panel shape information, the value of the loss function will be larger, prompting the network to adjust the parameters to reduce this difference and improve the accuracy of the prediction. Continue training until the performance of the recognizer on the validation set reaches a satisfactory level, or the loss function converges to a smaller value, indicating that the recognizer has learned effective feature representations and patterns and can accurately recognize the shape and reflective feature information in the photovoltaic panel image.

[0058] After the photovoltaic panel image recognizer is trained, multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets are respectively input into the recognizer. When the recognizer processes the actual photovoltaic panel image set, each actual photovoltaic panel image is analyzed and interpreted. Through its internally learned patterns and feature extraction capabilities, the shape-related information of each actual photovoltaic panel, such as length, width, curvature, etc., is extracted. By summarizing this information, multiple actual photovoltaic panel shape information sets are formed. At the same time, the recognizer will also extract the reflective feature-related information of the actual photovoltaic panel, such as reflective intensity, distribution, color, etc., to form multiple actual reflective feature information sets. Similarly, when the recognizer processes multiple simulated photovoltaic panel image sets, similar operations are performed to obtain multiple simulated photovoltaic panel shape information sets and multiple simulated reflective feature information sets.

[0059] For the similarity calculation of the actual photovoltaic panel shape information set and the simulated photovoltaic panel shape information set: First, the shape information needs to be quantified and standardized for comparison. For example, the shape parameters such as the length, width, and angle of the photovoltaic panel are converted into numerical form. Then, for each pair of corresponding actual and simulated photovoltaic panel shape information, a similarity measurement method based on geometric shapes is used, such as the Shape Context algorithm. This algorithm calculates the similarity by comparing the edge distribution of the shape. For each set of comparisons, a shape similarity value is obtained. Repeat this process to calculate all combinations of actual and simulated shape information, and finally obtain multiple shape similarity sets as multiple shape verification result sets.

[0060] For the similarity calculation of the actual reflective feature information set and the simulated reflective feature information set: first extract and quantify the reflective feature information, such as the average and variance of the reflective intensity, the area of ​​the reflective region, etc., and then use a similarity algorithm suitable for feature vector comparison, such as cosine similarity. For each pair of actual and simulated reflective feature information, calculate a reflective similarity value. Traverse all combinations to obtain multiple reflective similarity sets, that is, multiple reflective verification result sets.

[0061] Weighted calculations are performed on multiple sets of shape verification results and multiple sets of reflective verification results: First, certain weights are assigned to the shape verification results and reflective verification results respectively. The size of the weight can be determined based on actual needs and the degree of emphasis on shape and reflective features. Assume that the weight of the shape verification result is w1, the weight of the reflective verification result is w2, and w1+w2=1. Then, for each set of corresponding shape verification results and reflective verification results, multiply them by their respective weights and add them together to obtain the weighted results. Through such weighted calculations, the effects of shape and reflective features on the similarity of photovoltaic panels are comprehensively considered, thereby obtaining multiple sets of verification results that can more comprehensively reflect the degree of similarity of photovoltaic panels.

[0062] The multiple verification results obtained are added together and then divided by the number of verification results to obtain the mean value, which is finally used as the verification result of multiple photovoltaic panels, comprehensively reflecting the overall similarity between the actual photovoltaic panels and the simulated photovoltaic panels.

[0063] In a possible implementation, the photovoltaic panel image set acquisition module 60 further includes:

[0064] Determine whether the multiple photovoltaic panel verification results meet the photovoltaic panel similarity threshold. If so, the photovoltaic panel verification requirements are met and no management is performed. If not, the photovoltaic panel verification results that do not meet the photovoltaic panel verification requirements are input into the photovoltaic panel angle construction management decision maker to make a decision to obtain a monitoring management plan, wherein the photovoltaic panel angle construction management decision maker is based on a decision tree and is constructed using sample photovoltaic panel verification results and sample monitoring management plans.

[0065] Specifically, first, the photovoltaic panel similarity threshold is clearly set. This threshold is a standard value to measure whether the photovoltaic panel verification result is qualified. For example, the threshold is set to 0.9 or 90%. It is not 1 because there are errors in image acquisition and processing. Multiple photovoltaic panel verification results are compared with the photovoltaic panel similarity threshold. If all photovoltaic panel verification results are greater than or equal to this threshold, it means that the photovoltaic panel verification requirements are met. In this case, no additional management measures are required because the actual situation of the photovoltaic panel is similar to the expected level to an acceptable degree.

[0066] When the verification result of a photovoltaic panel does not meet the photovoltaic panel similarity threshold, it means that the actual situation of the photovoltaic panel is not similar enough to the expected level, and further processing and management are required. At this time, these photovoltaic panel verification results that do not meet the requirements are input into the photovoltaic panel angle construction management decision maker, which is built based on the decision tree model. When building the decision maker, a large amount of sample data is used, including sample photovoltaic panel verification results under various different situations, and the corresponding sample monitoring and management plans. The decision tree model will make conditional judgments on the internal nodes according to the input features (i.e., the relevant parameters and features of the photovoltaic panel verification results that do not meet the requirements), traverse down along the branches, and finally reach the leaf nodes, thereby giving the corresponding monitoring and management plans. In this way, a targeted, data- and experience-based monitoring and management plan can be provided for photovoltaic panels that do not meet the requirements, so as to improve the performance and similarity of photovoltaic panels so that they can meet the expected requirements.

[0067] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0068] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0069] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. An intelligent monitoring platform for photovoltaic system construction, characterized in that: The platform includes: A photovoltaic construction plan acquisition module, the photovoltaic construction plan acquisition module is used to acquire photovoltaic construction plan information of the photovoltaic system, wherein the photovoltaic construction plan information includes construction location information and an installation angle array of a photovoltaic panel array; A simulated photovoltaic panel array monitoring image acquisition module, the simulated photovoltaic panel array monitoring image acquisition module is used to generate photovoltaic panel array monitoring images in multiple time periods after the photovoltaic system construction according to the photovoltaic construction plan information, and obtain multiple simulated photovoltaic panel array monitoring images, wherein each simulated photovoltaic panel array monitoring image includes a photovoltaic panel image shape information array and a reflective feature information array of the photovoltaic panel array; A photovoltaic panel array monitoring image acquisition module, wherein the photovoltaic panel array monitoring image acquisition module is used to acquire a plurality of actual photovoltaic panel array monitoring images of the photovoltaic panel array within the plurality of time periods after the photovoltaic panel array construction is completed; A global verification result acquisition module, wherein the global verification result acquisition module is used to perform global construction angle verification using the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images to obtain multiple global verification results; A photovoltaic panel image set acquisition module, wherein the photovoltaic panel image set acquisition module is used to divide the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images into multiple photovoltaic panel images when any global verification result does not meet the global verification requirements, so as to obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets; A photovoltaic panel verification result acquisition module, the photovoltaic panel verification result acquisition module is used to verify the construction angle of the multiple actual photovoltaic panel image sets and the multiple simulated photovoltaic panel image sets, obtain multiple shape verification result sets and multiple reflection verification result sets, and calculate to obtain multiple photovoltaic panel verification results. For photovoltaic panels corresponding to photovoltaic panel verification results that do not meet photovoltaic panel verification requirements, a monitoring management plan is decided to be obtained and monitoring management is performed; The photovoltaic panel verification result acquisition module is also used for: Acquire a set of sample photovoltaic panel images, extract photovoltaic panel shape information and reflective feature information in each sample photovoltaic panel image, and acquire a set of sample photovoltaic panel shape information and a set of sample reflective feature information; Using the sample photovoltaic panel image set, the sample photovoltaic panel shape information set and the sample reflection feature information set to train a photovoltaic panel image recognizer; The photovoltaic panel image recognizer is used to recognize the multiple actual photovoltaic panel image sets and the multiple simulated photovoltaic panel image sets respectively, to obtain multiple actual photovoltaic panel shape information sets, multiple actual reflection feature information sets, multiple simulated photovoltaic panel shape information sets and multiple simulated reflection feature information sets; Performing similarity calculation on the multiple actual photovoltaic panel shape information sets and the multiple simulated photovoltaic panel shape information sets to obtain multiple shape similarity sets as multiple shape verification result sets; Performing similarity calculation on the multiple actual reflective feature information sets and the multiple simulated reflective feature information sets to obtain multiple reflective similarity sets as multiple reflective verification result sets; Performing weighted calculation on the multiple shape verification result sets and the multiple reflection verification result sets to obtain multiple verification result sets; Calculate the mean of the multiple verification result sets to obtain multiple photovoltaic panel verification results.

2. The intelligent monitoring platform for photovoltaic system construction according to claim 1, characterized in that: The photovoltaic construction plan acquisition module is also used for: Collect construction location information for planned construction of photovoltaic systems; Collect the installation angle array of the planned construction of the photovoltaic panel array in the photovoltaic system; The construction location information and the installation angle array are integrated to obtain photovoltaic construction plan information.

3. The intelligent monitoring platform for photovoltaic system construction according to claim 1, characterized in that: The simulated photovoltaic panel array monitoring image acquisition module is also used for: According to the photovoltaic construction plan information, collecting a plurality of illumination characteristic information of the construction location information in the plurality of time periods; Construct a simulated image generator for photovoltaic panel array monitoring image generation; The plurality of illumination feature information and the photovoltaic construction plan information are input into the simulation image generator, and photovoltaic panel array monitoring images are generated respectively to obtain a plurality of simulation photovoltaic panel array monitoring images.

4. The intelligent monitoring platform for photovoltaic system construction according to claim 3 is characterized in that: The simulated photovoltaic panel array monitoring image acquisition module is also used for: Collecting a sample illumination feature information set and a sample photovoltaic construction plan information set, and collecting photovoltaic panel array monitoring images under different illumination feature information and sample photovoltaic construction plan information, to obtain a sample simulated photovoltaic panel array monitoring image set; Based on the generative adversarial network, the generation path and discrimination path in the simulated image generator are constructed; The sample illumination feature information set, the sample photovoltaic construction plan information set and the sample simulated photovoltaic panel array monitoring image set are used to perform supervised training on the generation path and the discrimination path until the generation convergence requirements are met, thereby obtaining a trained simulation image generator.

5. The intelligent monitoring platform for photovoltaic system construction according to claim 1, characterized in that: The global verification result acquisition module is also used for: Collect multiple sample photovoltaic panel array monitoring image combinations, and analyze the similarity between two sample photovoltaic panel array monitoring images in each sample photovoltaic panel array monitoring image combination to obtain a sample global similarity set; Build a global validator based on the twin network; Using the plurality of sample photovoltaic panel array monitoring image combinations and the sample global similarity set, training the global verifier until convergence; The multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images are respectively combined and input into the global verifier to obtain multiple global similarities as multiple global verification results.

6. The intelligent monitoring platform for photovoltaic system construction according to claim 5, characterized in that: The photovoltaic panel image set acquisition module is also used for: Determine whether the global similarities in the multiple global verification results are greater than or equal to a global similarity threshold, if so, the global verification requirement is met, and if not, the global verification requirement is not met; When any global verification result does not meet the global verification requirements, the multiple actual photovoltaic panel array monitoring images and the multiple simulated photovoltaic panel array monitoring images are divided and segmented into multiple photovoltaic panel images to obtain multiple actual photovoltaic panel image sets and multiple simulated photovoltaic panel image sets of the multiple photovoltaic panels in the photovoltaic panel array.

7. The intelligent monitoring platform for photovoltaic system construction according to claim 1, characterized in that: The photovoltaic panel verification result acquisition module is also used for: Determine whether the verification results of the plurality of photovoltaic panels meet a photovoltaic panel similarity threshold, and if so, the photovoltaic panel verification requirement is met and no management is performed; If not, the photovoltaic panel verification results that do not meet the photovoltaic panel verification requirements will be input into the photovoltaic panel angle construction management decision maker to make a decision to obtain a monitoring management plan, wherein the photovoltaic panel angle construction management decision maker is based on a decision tree and is constructed using sample photovoltaic panel verification results and sample monitoring management plans.

Images