Photovoltaic power station cloud side collaborative operation and maintenance regulation and control system and method based on digital twinning

Abstract

The invention relates to a photovoltaic power station cloud side collaborative operation and maintenance regulation and control system and method based on digital twinning, provides digital twinning, and relates to the field of program control systems in industrial internet. The system comprises a synchronous identification device which is arranged at a cloud end and uses a multi-dimensional digital twin of a target photovoltaic power station to synchronously identify current subfissure coverage data, current hot spot coverage data and current dust coverage data of a photovoltaic panel array according to on-site multi-class data; and the operation and maintenance regulation and control device is arranged at the cloud end, executes corresponding operation and maintenance alarm operation based on the synchronous identification result and provides a corresponding operation and maintenance regulation and control strategy. Aiming at the technical problem that a photovoltaic power station operation and maintenance regulation and control mechanism lacks a targeted multi-dimensional state change synchronous analysis mechanism and a cloud-side collaborative operation and maintenance regulation and control mechanism, a multi-dimensional digital twinborn body corresponding to a target photovoltaic power station is established, and a cloud-side collaborative operation and maintenance regulation and control system is established at a local end and a cloud end of the target photovoltaic power station; the technical problem is solved.

Description

Technical Field

[0001] The digital twin proposed in this invention relates to the field of program control systems in the Industrial Internet, and particularly to a cloud-edge collaborative operation and maintenance control system and method for photovoltaic power plants based on digital twins. Background Technology

[0002] Digital twins are simulation processes that integrate multiple disciplines, multiple physical quantities, multiple scales, and multiple probabilities. They make full use of data such as physical models, sensor updates, and operating history to reflect the state and changes of physical objects in real time and dynamically. Through data analysis, modeling, and simulation, they provide decision support for the optimization, prediction, and maintenance of physical objects.

[0003] Digital twins are frequently used in program control systems within the Industrial Internet, such as those in photovoltaic power plants. They are used to perform corresponding program control on various operating states of the photovoltaic power plant, including operation and maintenance alarm operations and operation and maintenance control strategies, in order to minimize manual operation and achieve automated management of the photovoltaic power plant.

[0004] For example, Chinese invention patent publication CN120710211A proposes a method and system for remote monitoring of photovoltaic power plants based on three-dimensional digital twin technology. The method includes: constructing a three-dimensional twin photovoltaic power plant; synchronizing data from the physical photovoltaic power plant using the three-dimensional twin photovoltaic power plant, and monitoring the real-time operating parameters of the twin devices, comparing the real-time monitored operating parameters with corresponding preset threshold parameters of the twin devices, and determining whether to trigger an initial alarm based on the comparison result; determining the twin device corresponding to the fault source and generating a fault alarm for the corresponding twin device or an independent twin device based on the fault propagation chain law of the twin devices; predicting independent twin devices with fault risk or twin devices belonging to the fault source based on the monitored twin device operating parameters, automatically generating inspection tasks, and obtaining the inspection path using an optimal path planning algorithm. This application can improve the visualization effect and fault device location accuracy of remote monitoring of photovoltaic power plants.

[0005] For example, Chinese invention patent publication CN118763801A proposes a comprehensive management and control method and system for the operation and maintenance safety of photovoltaic power plants, involving the field of operation and maintenance management technology. The method includes: acquiring health status parameters, electrical characteristic data, and environmental parameters based on a distributed optical fiber sensing system and combining them to obtain multimodal heterogeneous data; preprocessing and extracting features to generate multimodal heterogeneous features; performing weighted fusion to generate fused features; adding these features to an edge server for fault diagnosis and fault location, generating photovoltaic diagnostic data; reconstructing and learning inverter data to generate early warning information; performing correlation analysis, fault source tracing, and impact domain analysis to obtain a fault analysis report; performing digital modeling to generate a twin photovoltaic power plant; integrating the fault analysis report into the twin photovoltaic power plant; generating a maintenance resource allocation plan; generating an intervention plan based on operating rules; executing and generating intelligent processing guidance; and feeding back fault handling process data to the twin photovoltaic power plant to determine the fault handling effect.

[0006] However, existing technical solutions only involve fault analysis and processing of the electrical status of photovoltaic power plants based on digital twins, and the analysis and processing are all for single faults. They cannot perform synchronous analysis of the multi-dimensional state changes of photovoltaic panel array data such as microcrack coverage, hot spot coverage, and dust coverage using the same digital twin model. Consequently, they cannot provide data support for whether to implement operation and maintenance alarms and specific operation and maintenance control strategies for photovoltaic power plants. At the same time, the operation and maintenance control of photovoltaic power plants based on digital twins lacks a targeted cloud-edge collaboration mechanism, which makes it impossible to balance the saving of local computing resources, the decoupling of devices on both the cloud and edge sides, and the efficiency of multi-dimensional state data parsing. Summary of the Invention

[0007] To address the technical problems in existing technologies, this invention provides a cloud-edge collaborative operation and maintenance control system and method for photovoltaic power plants based on digital twins. On one hand, by establishing a multi-dimensional digital twin corresponding to the target photovoltaic power plant, the system reflects the multi-dimensional state changes of the target photovoltaic power plant in real time and dynamically, including data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic panel array. This provides data support for whether to implement operation and maintenance alarms and for specific operation and maintenance control strategies, thereby improving the intelligence level of operation and maintenance control of the target photovoltaic power plant. On the other hand, a cloud-edge collaborative operation and maintenance control system is established at both the local end and the cloud end of the target photovoltaic power plant. The construction and operation of the multi-dimensional digital twin corresponding to the target photovoltaic power plant are placed in the cloud end of the target photovoltaic power plant, while the collection of various types of data from the target photovoltaic power plant is placed at the edge end. This balances the saving of local computing resources, the decoupling of devices on both the cloud and edge sides, and the efficiency of multi-dimensional state data parsing.

[0008] According to one aspect of the present invention, a cloud-edge collaborative operation and maintenance control system for photovoltaic power plants based on digital twins is provided, the system comprising:

[0009] The classification and acquisition device is set at the edge and is used to simultaneously acquire visible light, infrared and near-infrared overhead images of the target photovoltaic power station wirelessly transmitted by the visible light camera, infrared camera and near-infrared camera on the drone.

[0010] The content parsing device, located at the edge, is connected to the classification acquisition device and is used to parse the multi-camera visual content corresponding to the target photovoltaic power station;

[0011] The power plant capture device, located at the edge, is used to capture the photovoltaic panel unfolding angle, number of photovoltaic panels, unfolded planar area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels of the target photovoltaic power plant as various configuration information of the target photovoltaic power plant.

[0012] The object building device, set up in the cloud, is used to output a multi-level convolutional neural network as a multi-dimensional digital twin of the target photovoltaic power station;

[0013] The synchronous identification device, set up in the cloud, is connected to the content parsing device, the power station capture device, and the object assembly device, respectively. It is used to use the multi-dimensional digital twin of the target photovoltaic power station to synchronously identify the current hidden crack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station based on the current ambient light, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station.

[0014] The operation and maintenance control device is located in the cloud and connected to the synchronization identification device. It is used to perform corresponding operation and maintenance alarm operations based on the synchronization identification results and provide corresponding operation and maintenance control strategies.

[0015] According to another aspect of the present invention, a cloud-edge collaborative operation and maintenance control method for photovoltaic power plants based on digital twins is provided, the method comprising:

[0016] At the edge, the visible light camera, infrared camera, and near-infrared camera on the drone are simultaneously captured wirelessly, providing overhead images of the target photovoltaic power station.

[0017] Analyze the multi-camera visual content corresponding to the target photovoltaic power station at the edge;

[0018] Capture the photovoltaic panel deployment angle, number of photovoltaic panels, unfolded planar area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels at the edge of the target photovoltaic power station as various configuration information of the target photovoltaic power station;

[0019] In the cloud, a multi-level convolutional neural network is used as the output of a multi-dimensional digital twin of the target photovoltaic power station;

[0020] Using a multi-dimensional digital twin of the target photovoltaic power station in the cloud, the current data on microcracks, hot spots, and dust coverage of the photovoltaic panel array of the target photovoltaic power station are simultaneously identified based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station.

[0021] Based on the synchronous identification results, the cloud performs corresponding operation and maintenance alarm operations and provides corresponding operation and maintenance control strategies.

[0022] Therefore, it can be seen that the present invention has at least the following prominent substantive features:

[0023] The first step is to establish a multi-dimensional digital twin corresponding to the target photovoltaic power station to reflect the multi-dimensional status changes of the target photovoltaic power station, including the data on microcracks, hot spots, and dust covering of the photovoltaic panel array, in real time and dynamically. This will provide data support for whether the operation and maintenance alarm of the target photovoltaic power station is implemented and for specific operation and maintenance control strategies, thereby improving the level of intelligence of the operation and maintenance control of the target photovoltaic power station.

[0024] The second point: Specifically, when any of the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station exceeds its corresponding threshold data, the corresponding operation and maintenance alarm operation is executed and the corresponding operation and maintenance control strategy is provided. Specifically, when the current microcrack coverage data / current hot spot coverage data exceeds its corresponding threshold data, the recommended number of panels to be replaced is determined based on the current microcrack coverage data / current hot spot coverage data and is positively correlated with the current microcrack coverage data / current hot spot coverage data. And when the current dust coverage data exceeds its corresponding threshold data, the decision is made based on the current dust coverage data to determine whether to start the unfolded plane cleaning of the photovoltaic panel array of the target photovoltaic power station.

[0025] The third point: Establish a cloud-edge collaborative operation and maintenance control system at both the local end and the cloud end of the target photovoltaic power station. Place the construction and operation of the multi-dimensional digital twin corresponding to the target photovoltaic power station in the cloud end of the target photovoltaic power station, and place the collection of various types of data at the target photovoltaic power station at the edge end. This balances the saving of local computing resources, the decoupling of devices on both the cloud and edge ends, and the efficiency of multi-dimensional status data parsing.

[0026] Fourthly, to simultaneously identify the multi-dimensional states of the photovoltaic array, including current microcrack coverage data, current hot spot coverage data, and current dust coverage data, a multi-dimensional digital twin with a customized structure for the target photovoltaic power station was introduced. The multi-dimensional digital twin of the target photovoltaic power station is a multi-level convolutional neural network, which includes convolutional neural networks at each level in a cascaded state, with the number of convolutional neural networks in each level being equal, and the number of levels of the convolutional neural network having the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station. The customized structural designs at each of these points ensure the effectiveness and stability of the simultaneous identification results of the multi-dimensional states of the target photovoltaic power station.

[0027] Fifthly: In each training session of the multidimensional digital twin, the known data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic array of the target photovoltaic power station at a certain past moment are used as the output data of the multidimensional digital twin. The current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content of the target photovoltaic power station at the same past moment, and the configuration information of the target photovoltaic power station are used as the input data of the multidimensional digital twin to complete the training, thereby ensuring the training effect of each training session of the multidimensional digital twin.

[0028] The sixth point: To synchronously identify the multi-dimensional status of the photovoltaic array's current microcrack coverage data, current hot spot coverage data, and current dust coverage data, multiple types of data from the target photovoltaic power station were introduced. These include the current ambient light intensity, the target photovoltaic power station's floor area, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station. The multi-camera visual content corresponding to the target photovoltaic power station consists of the CMYK component values ​​and coordinate values ​​of each pixel in the visible light overhead image where the target photovoltaic power station occupies a sub-frame, the red component values ​​and coordinate values ​​of each edge pixel in the infrared overhead image where the target photovoltaic power station occupies a sub-frame, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the near-infrared overhead image where the target photovoltaic power station occupies a sub-frame. The various configuration information of the target photovoltaic power station includes the photovoltaic panel unfolding angle, the number of photovoltaic panels, the unfolded planar area of ​​each photovoltaic panel, and the gap between two adjacent photovoltaic panels. Thus, a customized data structure design was carried out for the multiple types of data from the target photovoltaic power station. Attached Figure Description

[0029] The embodiments of the present invention will now be described with reference to the accompanying drawings, wherein:

[0030] Figure 1 This is a schematic diagram of the working scenario of the cloud-edge collaborative operation and maintenance control system and method for photovoltaic power plants based on digital twins according to the present invention.

[0031] Figure 2This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the first embodiment of the present invention.

[0032] Figure 3 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the second embodiment of the present invention.

[0033] Figure 4 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the third embodiment of the present invention.

[0034] Figure 5 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the fourth embodiment of the present invention.

[0035] Figure 6 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the fifth embodiment of the present invention.

[0036] Figure 7 The following is a flowchart illustrating the steps of a cloud-edge collaborative operation and maintenance control method for photovoltaic power plants based on digital twins, as shown in the sixth embodiment of the present invention. Detailed Implementation

[0037] Digital twin technology involves multiple fields, including the Internet of Things, big data, cloud computing, artificial intelligence, and machine learning. Through the integrated application of these technologies, digital twins can achieve the following main functions:

[0038] Real-time monitoring: Collect data on physical objects through sensors and update the status of the digital twin model in real time;

[0039] Predictive analytics: using historical data and algorithmic models to predict the future state and potential problems of physical objects;

[0040] Optimize decision-making: Find the best operating plan by simulating different operating conditions and strategies;

[0041] Maintenance support: Analyze equipment failures, develop maintenance plans, and reduce unexpected downtime.

[0042] Digital twin technology has wide applications in various fields such as industrial manufacturing, urban planning, healthcare, and aerospace. For example, in manufacturing, digital twins can be used to optimize production processes, improve product quality, and reduce maintenance costs; in urban planning, they can simulate the environmental impact of urban development, assisting in making more rational planning decisions. With continuous technological advancements, digital twins are expected to play an even more important role in the future.

[0043] like Figure 1The diagram illustrates a working scenario of the cloud-edge collaborative operation and maintenance control system and method for photovoltaic power plants based on digital twins, as presented in this invention. The digital twin proposed in this invention relates to the field of program control systems in the Industrial Internet.

[0044] The specific technical process of this invention is as follows:

[0045] Technical Process A: To simultaneously identify the multi-dimensional status of the photovoltaic array, including current microcrack coverage data, current hot spot coverage data, and current dust coverage data, a multi-dimensional digital twin, custom-designed for the target photovoltaic power station structure, is introduced, such as... Figure 1 As shown;

[0046] Specifically, the customized structural design of the multidimensional digital twin of the target photovoltaic power station is mainly reflected in the following aspects:

[0047] Aspect 1: The multidimensional digital twin of the target photovoltaic power station is a multi-level convolutional neural network, which includes convolutional neural networks at each level in a cascaded state, and the number of convolutional neural networks in each level is equal;

[0048] Aspect 2: In the multi-level convolutional neural network architecture of the multi-dimensional digital twin of the target photovoltaic power station, the number of levels of the convolutional neural network and the number of photovoltaic panels in the target photovoltaic power station have the same numerical trend.

[0049] In this way, through the customized structural design of aspects one and two, it is possible to design multi-dimensional digital twins with different structures for different photovoltaic power plants;

[0050] Thirdly, in each training session of the multidimensional digital twin, the known data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic array of the target photovoltaic power station at a certain past moment are used as the output data of the multidimensional digital twin. The current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content of the target photovoltaic power station at the same past moment, and the configuration information of the target photovoltaic power station are used as the input data of the multidimensional digital twin to complete the training. This ensures the training effect of each training session of the multidimensional digital twin.

[0051] Thus, through the customized structural design in the above aspects, the effectiveness and stability of the synchronous identification results of the multi-dimensional state of the target photovoltaic power station are guaranteed;

[0052] Technical Process B: To simultaneously identify the multi-dimensional status of the photovoltaic panel array, including current microcrack coverage data, current hot spot coverage data, and current dust coverage data, multiple types of data from the target photovoltaic power station site are introduced;

[0053] Specifically, such as Figure 1As shown, the various types of data introduced from the target photovoltaic power station site include the current ambient light intensity, the area occupied by the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station;

[0054] More specifically, the multi-camera visual content corresponding to the target photovoltaic power station includes the CMYK component values ​​and coordinate values ​​of each pixel in the sub-frame occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the near-infrared overhead image. The configuration information of the target photovoltaic power station includes the photovoltaic panel unfolding angle, the number of photovoltaic panels, the unfolded plane area of ​​each photovoltaic panel, and the gap between two adjacent photovoltaic panels. Thus, a customized data structure design was carried out for multiple types of data at the target photovoltaic power station site.

[0055] In this way, by fully and comprehensively selecting various types of data from the target photovoltaic power station site, the effectiveness and stability of the synchronous identification results of the multi-dimensional status of the target photovoltaic power station are further guaranteed;

[0056] Technical Process C: Using the multi-dimensional digital twin of the target photovoltaic power station's customized structural design based on Technical Process A, and based on the comprehensive selection of various types of on-site data from the target photovoltaic power station in Technical Process B, synchronous identification of the multi-dimensional status of the photovoltaic panel array, including current microcrack coverage data, current hot spot coverage data, and current dust coverage data, is completed. Figure 1 As shown;

[0057] Specifically, the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array are simultaneously identified and acquired as the current microcrack coverage area ratio, current hot spot coverage area ratio, and current dust coverage area ratio of the photovoltaic panel array.

[0058] For example, the current microcrack coverage area ratio is the ratio of the area occupied by all microcrack regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station; the current hot spot coverage area ratio is the ratio of the area occupied by all hot spot regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station; and the current dust coverage area ratio is the ratio of the area occupied by all dust regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station.

[0059] Technical Process D: Based on the synchronous identification results of Technical Process C, execute the corresponding operation and maintenance alarm operations and provide the corresponding operation and maintenance control strategies;

[0060] Specifically, based on the synchronously identified and acquired data on current microcrack coverage, current hot spot coverage, and current dust coverage of the photovoltaic array, corresponding operation and maintenance alarm operations are performed and corresponding operation and maintenance control strategies are provided.

[0061] More specifically, when the current microcrack coverage data exceeds its corresponding threshold data, a recommended number of panels to be replaced that is positively correlated with the current microcrack coverage data is determined based on the current microcrack coverage data; when the current hot spot coverage data exceeds its corresponding threshold data, a recommended number of panels to be replaced that is positively correlated with the current hot spot coverage data is determined based on the current hot spot coverage data; when the current dust coverage data exceeds its corresponding threshold data, it is determined whether to initiate the unfolded planar cleaning of the photovoltaic panel array for the target photovoltaic power station based on the current dust coverage data.

[0062] Crucially, this invention establishes a cloud-edge collaborative operation and maintenance control system for photovoltaic power plants based on digital twins at both the local end and the cloud end of the target photovoltaic power plant. The construction and operation of the multi-dimensional digital twin corresponding to the target photovoltaic power plant are placed in the cloud end of the target photovoltaic power plant, while the collection of various types of data from the target photovoltaic power plant is placed at the edge end. This balances the saving of local computing resources, the decoupling of devices on both the cloud and edge sides, and the efficiency of multi-dimensional status data parsing.

[0063] Therefore, through the coordinated operation of the above-mentioned technical processes, this invention establishes a multi-dimensional digital twin corresponding to the target photovoltaic power station. The multi-dimensional digital twin is used to reflect the multi-dimensional state changes of the target photovoltaic power station in real time and dynamically, including the data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic panel array. This provides data support for whether the operation and maintenance alarm of the target photovoltaic power station is implemented and for specific operation and maintenance control strategies, thereby improving the intelligent level of operation and maintenance control of the target photovoltaic power station. At the same time, the cloud-edge separation of the construction and operation of the multi-dimensional digital twin and the data acquisition further improves the operation and maintenance control performance of the target photovoltaic power station.

[0064] The key points of this invention are: synchronous identification of multi-dimensional status data of photovoltaic panel arrays in photovoltaic power plants, including current microcrack coverage data, current hot spot coverage data, and current dust coverage data; directional design of multi-dimensional digital twins with different customized structures in different photovoltaic power plants; full and comprehensive selection of multiple types of on-site data of the target photovoltaic power plant for synchronous identification; cloud-edge separation of the construction, operation and data acquisition of multi-dimensional digital twins; and automated analysis of operation and maintenance alarm operations and operation and maintenance control strategies based on multi-dimensional status data obtained through synchronous identification.

[0065] The following will describe in detail the cloud-edge collaborative operation and maintenance control system and method for photovoltaic power plants based on digital twins of the present invention through embodiments.

[0066] First Embodiment

[0067] Figure 2 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the first embodiment of the present invention.

[0068] like Figure 2 As shown, the photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins includes the following components:

[0069] The classification and acquisition device is set at the edge and is used to simultaneously acquire visible light, infrared and near-infrared overhead images of the target photovoltaic power station wirelessly transmitted by the visible light camera, infrared camera and near-infrared camera on the drone.

[0070] Specifically, a photovoltaic power station is a power generation system that utilizes solar energy and uses special materials such as crystalline silicon panels and electronic components such as inverters. A photovoltaic power station is a photovoltaic power generation system that can be connected to the power grid and transmit electricity to the grid.

[0071] The content parsing device, located at the edge, is connected to the classification acquisition device and is used to parse the multi-camera visual content corresponding to the target photovoltaic power station;

[0072] Specifically, the drone hovering directly above the photovoltaic panel array of the target photovoltaic power station uses a synchronous camera system for its onboard visible light camera, infrared camera, and near-infrared camera to obtain different camera images at the current moment, providing basic data for the analysis of the multi-camera visual content corresponding to the target photovoltaic power station.

[0073] The power plant capture device, located at the edge, is used to capture the photovoltaic panel unfolding angle, number of photovoltaic panels, unfolded planar area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels of the target photovoltaic power plant as various configuration information of the target photovoltaic power plant.

[0074] Specifically, the photovoltaic panel array includes photovoltaic panels of equal size, each photovoltaic panel has the same unfolding angle, which is the photovoltaic panel unfolding angle of the target photovoltaic power station, and the gap between two adjacent photovoltaic panels is equal.

[0075] The object building device, set up in the cloud, is used to output a multi-level convolutional neural network as a multi-dimensional digital twin of the target photovoltaic power station;

[0076] For example, using a multi-level convolutional neural network as the output of a multi-dimensional digital twin of a target photovoltaic power station includes: optionally using numerical simulation mode to test and simulate the modeling process of using a multi-level convolutional neural network as a multi-dimensional digital twin of a target photovoltaic power station;

[0077] The synchronous identification device, set up in the cloud, is connected to the content parsing device, the power station capture device, and the object assembly device, respectively. It is used to use the multi-dimensional digital twin of the target photovoltaic power station to synchronously identify the current hidden crack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station based on the current ambient light, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station.

[0078] Specifically, the current ambient light intensity, the area occupied by the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station reflect the different states of the photovoltaic panel array of the target photovoltaic power station. Therefore, it can be used to simultaneously identify the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station. The simultaneous identification of these multi-dimensional state data is also one of the key contents that distinguish this invention from the prior art.

[0079] The operation and maintenance control device is set up in the cloud and connected to the synchronization identification device. It is used to perform corresponding operation and maintenance alarm operations based on the synchronization identification results and provide corresponding operation and maintenance control strategies.

[0080] For example, the corresponding operation and maintenance alarm operation and the corresponding operation and maintenance control strategy can be performed based on the synchronous identification result. This includes: using an operation and maintenance alarm unit to perform the corresponding operation and maintenance alarm operation and the strategy parsing unit to provide the corresponding operation and maintenance control strategy.

[0081] Among them, using a multi-level convolutional neural network as the output of a multi-dimensional digital twin of the target photovoltaic power station includes: the multi-level convolutional neural network includes each level of convolutional neural network in a cascaded state and the number of convolutional neural networks in each level is equal, and the number of levels of the convolutional neural network has the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station.

[0082] For example, the number of levels of the convolutional neural network and the number of photovoltaic panels in the target photovoltaic power station also show the same numerical trend, including: the number of convolutional neural network levels is 3;

[0083] For example, the number of levels of the convolutional neural network has the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station, including: 1000 photovoltaic panels in the target photovoltaic power station, 2 levels of the convolutional neural network; 2000 photovoltaic panels in the target photovoltaic power station, 4 levels of the convolutional neural network; 3000 photovoltaic panels in the target photovoltaic power station, 6 levels of the convolutional neural network; 4000 photovoltaic panels in the target photovoltaic power station, 8 levels of the convolutional neural network, and so on.

[0084] In each training iteration of the multidimensional digital twin, the known data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic array of the target photovoltaic power station at a certain past moment are used as the output data of the multidimensional digital twin. The current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content of the target photovoltaic power station at the same past moment, and the configuration information of the target photovoltaic power station are used as the input data of the multidimensional digital twin to complete the training.

[0085] Among them, the cloud and the edge are the remote and local ends of any network in the Internet of Things or cloud computing network deployed for the target photovoltaic power station, respectively.

[0086] In this way, by placing the construction and operation of the multi-dimensional digital twin corresponding to the target photovoltaic power station in the cloud of the target photovoltaic power station, and placing the collection of various types of data from the target photovoltaic power station at the edge, the goal of saving local computing resources, decoupling the devices on both the cloud and edge sides, and improving the efficiency of multi-dimensional status data parsing is achieved.

[0087] Among them, the analysis of the multi-camera visual content corresponding to the target photovoltaic power station includes: analyzing the CMYK component values ​​and coordinate values ​​of each pixel point occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel point occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel point occupied by the target photovoltaic power station in the near-infrared overhead image, as the multi-camera visual content corresponding to the target photovoltaic power station.

[0088] Specifically, in the CMYK component values ​​of each pixel, the value of any component ranges from 0 to 255;

[0089] Among them, capturing the photovoltaic panel unfolding angle, number of photovoltaic panels, unfolded plane area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels of the target photovoltaic power station as various configuration information of the target photovoltaic power station includes: the target photovoltaic power station includes a photovoltaic panel array composed of multiple photovoltaic panels, and the multiple photovoltaic panels have the same photovoltaic panel unfolding angle, which is the angle between the photovoltaic panel screen and the ground plane, and the multiple photovoltaic panels have the same structure;

[0090] Among them, the corresponding operation and maintenance alarm operation and corresponding operation and maintenance control strategy based on the synchronous identification result includes: when any of the current hidden crack coverage data, current hot spot coverage data and current dust coverage data exceeds its corresponding threshold data, the corresponding operation and maintenance alarm operation and corresponding operation and maintenance control strategy is executed and the corresponding operation and maintenance control strategy is provided. The coverage data is the coverage area percentage.

[0091] Specifically, the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array are simultaneously identified and acquired as the current microcrack coverage area ratio, current hot spot coverage area ratio, and current dust coverage area ratio of the photovoltaic panel array.

[0092] Specifically, when any of the current microcrack coverage data, current hot spot coverage data, and current dust coverage data exceeds its corresponding threshold data, corresponding operation and maintenance alarm operations are executed and corresponding operation and maintenance control strategies are provided. The coverage data refers to the coverage area percentage, including: when the current microcrack coverage data exceeds its corresponding threshold data, determining the recommended number of panels to be replaced based on the current microcrack coverage data; when the current hot spot coverage data exceeds its corresponding threshold data, determining the recommended number of panels to be replaced based on the current hot spot coverage data; and when the current dust coverage data exceeds its corresponding threshold data, determining whether to initiate the unfolded planar cleaning of the photovoltaic panel array for the target photovoltaic power station based on the current dust coverage data.

[0093] For example, the current microcrack coverage area ratio is the ratio of the area occupied by all microcrack regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station; the current hot spot coverage area ratio is the ratio of the area occupied by all hot spot regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station; and the current dust coverage area ratio is the ratio of the area occupied by all dust regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station.

[0094] Second Embodiment

[0095] Figure 3 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the second embodiment of the present invention.

[0096] like Figure 3 As shown, compared to Figure 2 The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins also includes:

[0097] The real-time display device, located at the edge, is connected to the synchronization identification device to receive and synchronously display the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station.

[0098] For example, an LCD display array is used to realize a real-time display device, which is set at the edge and connected to a synchronization identification device to receive and synchronously display the current microcrack coverage data, current hot spot coverage data and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station.

[0099] The real-time display device is also connected to the operation and maintenance control device to receive and display the operation and maintenance control strategy corresponding to the synchronous identification result.

[0100] Third Embodiment

[0101] Figure 4 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the third embodiment of the present invention.

[0102] like Figure 4 As shown, compared to Figure 2 The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins also includes:

[0103] The model storage device, located in the cloud and connected to the object building device, is used to receive and store the multidimensional digital twin of the target photovoltaic power station.

[0104] For example, TF storage chips or MMC storage chips can be selected to replace the model storage device, set up in the cloud, and connected to the object building device to receive and store the multi-dimensional digital twin of the target photovoltaic power station;

[0105] The process of receiving and storing the multidimensional digital twin of the target photovoltaic power station includes: completing the model storage of the multidimensional digital twin of the target photovoltaic power station by storing various model parameters of the multidimensional digital twin of the target photovoltaic power station.

[0106] Fourth embodiment

[0107] Figure 5 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the fourth embodiment of the present invention.

[0108] like Figure 5 As shown, compared to Figure 2 The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins also includes:

[0109] A wireless remote control device, located at the edge, is connected to a classification and acquisition device. It is used to remotely control a drone to fly directly above the target photovoltaic power station and hover directly above the target photovoltaic power station using wireless communication mode.

[0110] Specifically, using wireless communication mode to remotely control a drone to fly directly above the target photovoltaic power station and hover directly above the target photovoltaic power station includes: the wireless communication mode is based on frequency division duplex communication mechanism or time division duplex communication mechanism;

[0111] Among them, the method of using wireless communication to remotely control a drone to fly directly above the target photovoltaic power station and hover directly above the target photovoltaic power station includes: providing navigation data directly above the target photovoltaic power station based on the navigation unit built into the wireless remote control device.

[0112] Fifth Embodiment

[0113] Figure 6 This is an internal structure diagram of a cloud-edge collaborative operation and maintenance control system for a photovoltaic power plant based on digital twins, as shown in the fifth embodiment of the present invention.

[0114] like Figure 6 As shown, compared to Figure 2 The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins also includes:

[0115] The shooting control device is located at the edge and connected to the classification and acquisition device. It is used to configure the shooting parameters of the visible light camera, infrared camera and near-infrared camera on the UAV respectively.

[0116] The shooting control device, located at the edge and connected to the classification and acquisition device, is used to configure the shooting parameters of the visible light camera, infrared camera and near-infrared camera on the UAV respectively. This includes configuring the shooting parameters of the visible light camera, infrared camera and near-infrared camera on the UAV respectively so that the visible light camera, infrared camera and near-infrared camera on the UAV have the same shooting frame rate and the same shooting resolution.

[0117] For example, the visible light camera, infrared camera, and near-infrared camera on the drone have the same shooting frame rate of 30 frames per second, and the visible light camera, infrared camera, and near-infrared camera on the drone have the same shooting resolution of 2K.

[0118] Next, various embodiments of the present invention will be further described.

[0119] Optionally, within the above embodiments, in the digital twin-based photovoltaic power plant cloud-edge collaborative operation and maintenance control system:

[0120] The CMYK component values ​​and coordinate values ​​of each pixel in the sub-frame occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the near-infrared overhead image are analyzed to serve as the multi-camera visual content corresponding to the target photovoltaic power station. This includes: analyzing the sub-frame occupied by the target photovoltaic power station in the visible light overhead image based on the reference outline pattern of the target photovoltaic power station, analyzing the sub-frame occupied by the target photovoltaic power station in the visible light overhead image, and analyzing the sub-frame occupied by the target photovoltaic power station in the near-infrared overhead image.

[0121] Specifically, the analysis of the target photovoltaic power station's occupancy in the visible light overhead image based on the reference outline pattern of the target photovoltaic power station, the analysis of the target photovoltaic power station's occupancy in the visible light overhead image, and the analysis of the target photovoltaic power station's occupancy in the near-infrared overhead image include: the three analysis operations are synchronous analysis operations;

[0122] Among them, the analysis of the target photovoltaic power station's occupancy in the visible light overhead image based on the reference outline pattern of the target photovoltaic power station, the analysis of the target photovoltaic power station's occupancy in the visible light overhead image, and the analysis of the target photovoltaic power station's occupancy in the near-infrared overhead image include: the reference outline pattern of the target photovoltaic power station is an imaging pattern that only includes the target photovoltaic power station from the overhead view.

[0123] For example, the baseline outline pattern of the target photovoltaic power station is an imaging pattern that includes only the target photovoltaic power station from an overhead view. The imaging pattern that includes only the target photovoltaic power station from an overhead view may be more than one frame.

[0124] The analysis of the CMYK component values ​​and coordinate values ​​of each pixel in the sub-image occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel in the sub-image occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the sub-image occupied by the target photovoltaic power station in the near-infrared overhead image, as the multi-camera visual content corresponding to the target photovoltaic power station, further includes: the red component value of each edge pixel in the sub-image occupied by the target photovoltaic power station in the infrared overhead image is the R component value of the edge pixel in the RGB color space, and the CMYK component values ​​of each pixel in the sub-image occupied by the target photovoltaic power station in the visible light overhead image are the C component value, M component value, Y component value and K component value of the pixel in the CMYK color space;

[0125] Specifically, for each pixel in the CMYK color space, the value of any one of the C, M, Y and K components is between 0 and 255.

[0126] The analysis of the CMYK component values ​​and coordinate values ​​of each pixel in the visible light overhead image occupying the sub-image, the red component values ​​and coordinate values ​​of each edge pixel in the infrared overhead image occupying the sub-image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the near-infrared overhead image occupying the sub-image, to serve as the multi-camera visual content corresponding to the target photovoltaic power station, further includes: the coordinate values ​​of each pixel being the vertical and horizontal coordinate values ​​of the pixel, and the coordinate values ​​of each edge pixel being the vertical and horizontal coordinate values ​​of the edge pixel.

[0127] And, optionally, within the above embodiments, in the digital twin-based photovoltaic power plant cloud-edge collaborative operation and maintenance control system:

[0128] Using a multi-dimensional digital twin of the target photovoltaic power station, the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station are simultaneously identified based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station. This includes inputting the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin in parallel.

[0129] For example, programmable logic devices can be used to implement the parallel input of the current ambient light intensity, the target photovoltaic power station's footprint, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin;

[0130] The process of using a multi-dimensional digital twin of the target photovoltaic power station to synchronously identify the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station also includes: running the multi-dimensional digital twin to obtain the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station output by the multi-dimensional digital twin;

[0131] The parallel input of the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin includes: performing numerical normalization processing on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station before inputting them into the multi-dimensional digital twin in parallel.

[0132] For example, the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and the various configuration information of the target photovoltaic power station are respectively processed by octal numerical conversion and then input into the multi-dimensional digital twin in parallel.

[0133] The parallel input of the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin also includes: using different programmable logic devices to realize numerical normalization processing and parallel input respectively;

[0134] For example, different programmable logic devices are used to implement octal value conversion and parallel input respectively.

[0135] Sixth Embodiment

[0136] Figure 7 The following is a flowchart illustrating the steps of a cloud-edge collaborative operation and maintenance control method for photovoltaic power plants based on digital twins, as shown in the sixth embodiment of the present invention.

[0137] like Figure 7 As shown, the cloud-edge collaborative operation and maintenance control method for photovoltaic power plants based on digital twins includes the following steps:

[0138] At the edge, the visible light camera, infrared camera, and near-infrared camera on the drone are simultaneously captured wirelessly, providing overhead images of the target photovoltaic power station.

[0139] Specifically, a photovoltaic power station is a power generation system that utilizes solar energy and uses special materials such as crystalline silicon panels and electronic components such as inverters. A photovoltaic power station is a photovoltaic power generation system that can be connected to the power grid and transmit electricity to the grid.

[0140] Analyze the multi-camera visual content corresponding to the target photovoltaic power station at the edge;

[0141] Specifically, the drone hovering directly above the photovoltaic panel array of the target photovoltaic power station uses a synchronous camera system for its onboard visible light camera, infrared camera, and near-infrared camera to obtain different camera images at the current moment, providing basic data for the analysis of the multi-camera visual content corresponding to the target photovoltaic power station.

[0142] Capture the photovoltaic panel deployment angle, number of photovoltaic panels, unfolded planar area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels at the edge of the target photovoltaic power station as various configuration information of the target photovoltaic power station;

[0143] Specifically, the photovoltaic panel array includes photovoltaic panels of equal size, each photovoltaic panel has the same unfolding angle, which is the photovoltaic panel unfolding angle of the target photovoltaic power station, and the gap between two adjacent photovoltaic panels is equal.

[0144] In the cloud, a multi-level convolutional neural network is used as the output of a multi-dimensional digital twin of the target photovoltaic power station;

[0145] For example, using a multi-level convolutional neural network as the output of a multi-dimensional digital twin of a target photovoltaic power station includes: optionally using numerical simulation mode to test and simulate the modeling process of using a multi-level convolutional neural network as a multi-dimensional digital twin of a target photovoltaic power station;

[0146] Using a multi-dimensional digital twin of the target photovoltaic power station in the cloud, the current data on microcracks, hot spots, and dust coverage of the photovoltaic panel array of the target photovoltaic power station are simultaneously identified based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station.

[0147] Specifically, the current ambient light intensity, the area occupied by the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station reflect the different states of the photovoltaic panel array of the target photovoltaic power station. Therefore, it can be used to simultaneously identify the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station. The simultaneous identification of these multi-dimensional state data is also one of the key contents that distinguish this invention from the prior art.

[0148] Based on the synchronous identification results, execute corresponding operation and maintenance alarm operations and provide corresponding operation and maintenance control strategies in the cloud.

[0149] For example, the corresponding operation and maintenance alarm operation and the corresponding operation and maintenance control strategy can be performed based on the synchronous identification result. This includes: using an operation and maintenance alarm unit to perform the corresponding operation and maintenance alarm operation and the strategy parsing unit to provide the corresponding operation and maintenance control strategy.

[0150] Among them, using a multi-level convolutional neural network as the output of a multi-dimensional digital twin of the target photovoltaic power station includes: the multi-level convolutional neural network includes each level of convolutional neural network in a cascaded state and the number of convolutional neural networks in each level is equal, and the number of levels of the convolutional neural network has the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station.

[0151] For example, the number of levels of the convolutional neural network and the number of photovoltaic panels in the target photovoltaic power station also show the same numerical trend, including: the number of convolutional neural network levels is 3;

[0152] For example, the number of levels of the convolutional neural network has the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station, including: 1000 photovoltaic panels in the target photovoltaic power station, 2 levels of the convolutional neural network; 2000 photovoltaic panels in the target photovoltaic power station, 4 levels of the convolutional neural network; 3000 photovoltaic panels in the target photovoltaic power station, 6 levels of the convolutional neural network; 4000 photovoltaic panels in the target photovoltaic power station, 8 levels of the convolutional neural network, and so on.

[0153] In each training iteration of the multidimensional digital twin, the known data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic array of the target photovoltaic power station at a certain past moment are used as the output data of the multidimensional digital twin. The current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content of the target photovoltaic power station at the same past moment, and the configuration information of the target photovoltaic power station are used as the input data of the multidimensional digital twin to complete the training.

[0154] Among them, the cloud and the edge are the remote and local ends of any network in the Internet of Things or cloud computing network deployed for the target photovoltaic power station, respectively.

[0155] In this way, by placing the construction and operation of the multi-dimensional digital twin corresponding to the target photovoltaic power station in the cloud of the target photovoltaic power station, and placing the collection of various types of data from the target photovoltaic power station at the edge, the goal of saving local computing resources, decoupling the devices on both the cloud and edge sides, and improving the efficiency of multi-dimensional status data parsing is achieved.

[0156] Among them, the analysis of the multi-camera visual content corresponding to the target photovoltaic power station includes: analyzing the CMYK component values ​​and coordinate values ​​of each pixel point occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel point occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel point occupied by the target photovoltaic power station in the near-infrared overhead image, as the multi-camera visual content corresponding to the target photovoltaic power station.

[0157] Specifically, in the CMYK component values ​​of each pixel, the value of any component ranges from 0 to 255;

[0158] Among them, capturing the photovoltaic panel unfolding angle, number of photovoltaic panels, unfolded plane area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels of the target photovoltaic power station as various configuration information of the target photovoltaic power station includes: the target photovoltaic power station includes a photovoltaic panel array composed of multiple photovoltaic panels, and the multiple photovoltaic panels have the same photovoltaic panel unfolding angle, which is the angle between the photovoltaic panel screen and the ground plane, and the multiple photovoltaic panels have the same structure;

[0159] Among them, the corresponding operation and maintenance alarm operation and corresponding operation and maintenance control strategy based on the synchronous identification result includes: when any of the current hidden crack coverage data, current hot spot coverage data and current dust coverage data exceeds its corresponding threshold data, the corresponding operation and maintenance alarm operation and corresponding operation and maintenance control strategy is executed and the corresponding operation and maintenance control strategy is provided. The coverage data is the coverage area percentage.

[0160] Specifically, the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array are simultaneously identified and acquired as the current microcrack coverage area ratio, current hot spot coverage area ratio, and current dust coverage area ratio of the photovoltaic panel array.

[0161] Specifically, when any of the current microcrack coverage data, current hot spot coverage data, and current dust coverage data exceeds its corresponding threshold data, corresponding operation and maintenance alarm operations are executed and corresponding operation and maintenance control strategies are provided. The coverage data refers to the coverage area percentage, including: when the current microcrack coverage data exceeds its corresponding threshold data, determining the recommended number of panels to be replaced based on the current microcrack coverage data; when the current hot spot coverage data exceeds its corresponding threshold data, determining the recommended number of panels to be replaced based on the current hot spot coverage data; and when the current dust coverage data exceeds its corresponding threshold data, determining whether to initiate the unfolded planar cleaning of the photovoltaic panel array for the target photovoltaic power station based on the current dust coverage data.

[0162] For example, the current microcrack coverage area ratio is the ratio of the area occupied by all microcrack regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station; the current hot spot coverage area ratio is the ratio of the area occupied by all hot spot regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station; and the current dust coverage area ratio is the ratio of the area occupied by all dust regions currently existing on the photovoltaic panel array of the target photovoltaic power station to the total planar area of ​​the photovoltaic panel array of the target photovoltaic power station.

[0163] Furthermore, in the cloud-edge collaborative operation and maintenance control system and method for photovoltaic power plants based on digital twins according to the present invention:

[0164] The multi-level convolutional neural network includes convolutional neural networks at each level in a cascaded state, with the number of convolutional neural networks in each level being equal, and the number of levels of the convolutional neural network having the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station. This includes: using a quantity change curve to represent the numerical trend of the number of photovoltaic panels in the target photovoltaic power station, and using a level change curve to represent the numerical trend of the number of levels of the convolutional neural network.

[0165] Among them, the multi-level convolutional neural network includes convolutional neural networks at each level in a cascaded state, and the number of convolutional neural networks in each level is equal. The number of levels of the convolutional neural network and the number of photovoltaic panels in the target photovoltaic power station have the same numerical change trend. It also includes that the curvature at each uniform interval on the quantity change curve is equal to the curvature at each uniform interval on the level change curve.

[0166] For example, the multi-level convolutional neural network includes convolutional neural networks at each level in a cascaded state, with the number of convolutional neural networks in each level being equal, and the number of levels of the convolutional neural network having the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station. It also includes: selecting the MATLAB toolbox to complete the synchronous simulation and testing of the quantity change curve and the level change curve.

[0167] Although embodiments of the invention have been shown and described with reference to exemplary embodiments thereof, those skilled in the art will understand that various modifications in form and detail may be made without departing from the spirit and scope of the embodiments of the invention as defined in the following claims.

Claims

1. A cloud-edge collaborative operation and maintenance control system for photovoltaic power plants based on digital twins, characterized in that: The system includes: The classification and acquisition device is set at the edge and is used to simultaneously acquire visible light, infrared and near-infrared overhead images of the target photovoltaic power station wirelessly transmitted by the visible light camera, infrared camera and near-infrared camera on the drone. The content parsing device, located at the edge, is connected to the classification acquisition device and is used to parse the multi-camera visual content corresponding to the target photovoltaic power station; The power plant capture device, located at the edge, is used to capture the photovoltaic panel unfolding angle, number of photovoltaic panels, unfolded planar area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels of the target photovoltaic power plant as various configuration information of the target photovoltaic power plant. The object building device, set up in the cloud, is used to output a multi-level convolutional neural network as a multi-dimensional digital twin of the target photovoltaic power station; The synchronous identification device, set up in the cloud, is connected to the content parsing device, the power station capture device, and the object assembly device, respectively. It is used to use the multi-dimensional digital twin of the target photovoltaic power station to synchronously identify the current hidden crack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station based on the current ambient light, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station. The operation and maintenance control device is located in the cloud and connected to the synchronization identification device. It is used to perform corresponding operation and maintenance alarm operations based on the synchronization identification results and provide corresponding operation and maintenance control strategies.

2. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twin as described in claim 1, characterized in that: The output of the multi-dimensional digital twin of the target photovoltaic power station using a multi-level convolutional neural network includes: the multi-level convolutional neural network includes each level of convolutional neural network in a cascaded state and the number of convolutional neural networks in each level is equal; and the number of levels of the convolutional neural network has the same numerical trend as the number of photovoltaic panels in the target photovoltaic power station. In each training iteration of the multidimensional digital twin, the known data on microcrack coverage, hot spot coverage, and dust coverage of the photovoltaic array of the target photovoltaic power station at a certain past moment are used as the output data of the multidimensional digital twin. The current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content of the target photovoltaic power station at the same past moment, and the configuration information of the target photovoltaic power station are used as the input data of the multidimensional digital twin to complete the training. Among them, the cloud and the edge are the remote and local ends of any network in the Internet of Things or cloud computing network deployed for the target photovoltaic power station, respectively.

3. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins as described in claim 2, characterized in that: The analysis of the multi-camera visual content corresponding to the target photovoltaic power station includes: analyzing the CMYK component values ​​and coordinate values ​​of each pixel in the sub-frame occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the near-infrared overhead image, as the multi-camera visual content corresponding to the target photovoltaic power station. Among them, capturing the photovoltaic panel unfolding angle, number of photovoltaic panels, unfolded plane area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels of the target photovoltaic power station as various configuration information of the target photovoltaic power station includes: the target photovoltaic power station includes a photovoltaic panel array composed of multiple photovoltaic panels, and the multiple photovoltaic panels have the same photovoltaic panel unfolding angle, which is the angle between the photovoltaic panel screen and the ground plane, and the multiple photovoltaic panels have the same structure; Among them, the corresponding operation and maintenance alarm operation and corresponding operation and maintenance control strategy based on the synchronous identification result includes: when any of the current hidden crack coverage data, current hot spot coverage data and current dust coverage data exceeds its corresponding threshold data, the corresponding operation and maintenance alarm operation and corresponding operation and maintenance control strategy is executed and the corresponding operation and maintenance control strategy is provided. The coverage data is the coverage area percentage. Specifically, when any of the current microcrack coverage data, current hot spot coverage data, and current dust coverage data exceeds its corresponding threshold data, corresponding operation and maintenance alarm operations are executed and corresponding operation and maintenance control strategies are provided. The coverage data refers to the coverage area ratio, including: when the current microcrack coverage data exceeds its corresponding threshold data, determining the recommended number of panels to be replaced based on the current microcrack coverage data; when the current hot spot coverage data exceeds its corresponding threshold data, determining the recommended number of panels to be replaced based on the current hot spot coverage data; and when the current dust coverage data exceeds its corresponding threshold data, determining whether to initiate the unfolded planar cleaning of the photovoltaic panel array for the target photovoltaic power station based on the current dust coverage data.

4. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twin as described in claim 3, characterized in that, The system also includes: The real-time display device, located at the edge, is connected to the synchronization identification device to receive and synchronously display the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station. The real-time display device is also connected to the operation and maintenance control device to receive and display the operation and maintenance control strategy corresponding to the synchronous identification result.

5. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twin as described in claim 3, characterized in that, The system also includes: The model storage device, located in the cloud and connected to the object building device, is used to receive and store the multidimensional digital twin of the target photovoltaic power station. The process of receiving and storing the multidimensional digital twin of the target photovoltaic power station includes: completing the model storage of the multidimensional digital twin of the target photovoltaic power station by storing various model parameters of the multidimensional digital twin of the target photovoltaic power station.

6. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twin as described in claim 3, characterized in that, The system also includes: A wireless remote control device, located at the edge, is connected to a classification and acquisition device. It is used to remotely control a drone to fly directly above the target photovoltaic power station and hover directly above the target photovoltaic power station using wireless communication mode. Among them, the method of using wireless communication to remotely control a drone to fly directly above the target photovoltaic power station and hover directly above the target photovoltaic power station includes: providing navigation data directly above the target photovoltaic power station based on the navigation unit built into the wireless remote control device.

7. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twin as described in claim 3, characterized in that, The system also includes: The shooting control device is located at the edge and connected to the classification and acquisition device. It is used to configure the shooting parameters of the visible light camera, infrared camera and near-infrared camera on the UAV respectively. The shooting control device, located at the edge and connected to the classification and acquisition device, is used to configure the shooting parameters of the UAV-borne visible light camera, infrared camera, and near-infrared camera respectively. This includes configuring the shooting parameters of the UAV-borne visible light camera, infrared camera, and near-infrared camera respectively so that the UAV-borne visible light camera, infrared camera, and near-infrared camera have the same shooting frame rate and the same shooting resolution.

8. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins as described in any one of claims 3-7, characterized in that: The CMYK component values ​​and coordinate values ​​of each pixel in the sub-frame occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the sub-frame occupied by the target photovoltaic power station in the near-infrared overhead image are analyzed to serve as the multi-camera visual content corresponding to the target photovoltaic power station. This includes: analyzing the sub-frame occupied by the target photovoltaic power station in the visible light overhead image based on the reference outline pattern of the target photovoltaic power station, analyzing the sub-frame occupied by the target photovoltaic power station in the visible light overhead image, and analyzing the sub-frame occupied by the target photovoltaic power station in the near-infrared overhead image. Among them, the analysis of the target photovoltaic power station's occupancy in the visible light overhead image based on the reference outline pattern of the target photovoltaic power station, the analysis of the target photovoltaic power station's occupancy in the visible light overhead image, and the analysis of the target photovoltaic power station's occupancy in the near-infrared overhead image include: the reference outline pattern of the target photovoltaic power station is an imaging pattern that only includes the target photovoltaic power station from the overhead view. The analysis of the CMYK component values ​​and coordinate values ​​of each pixel in the sub-image occupied by the target photovoltaic power station in the visible light overhead image, the red component values ​​and coordinate values ​​of each edge pixel in the sub-image occupied by the target photovoltaic power station in the infrared overhead image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the sub-image occupied by the target photovoltaic power station in the near-infrared overhead image, as the multi-camera visual content corresponding to the target photovoltaic power station, further includes: the red component value of each edge pixel in the sub-image occupied by the target photovoltaic power station in the infrared overhead image is the R component value of the edge pixel in the RGB color space, and the CMYK component values ​​of each pixel in the sub-image occupied by the target photovoltaic power station in the visible light overhead image are the C component value, M component value, Y component value and K component value of the pixel in the CMYK color space; The analysis of the CMYK component values ​​and coordinate values ​​of each pixel in the visible light overhead image occupying the sub-image, the red component values ​​and coordinate values ​​of each edge pixel in the infrared overhead image occupying the sub-image, and the brightness gradient values ​​and coordinate values ​​of each edge pixel in the near-infrared overhead image occupying the sub-image, to serve as the multi-camera visual content corresponding to the target photovoltaic power station, further includes: the coordinate values ​​of each pixel being the vertical and horizontal coordinate values ​​of the pixel, and the coordinate values ​​of each edge pixel being the vertical and horizontal coordinate values ​​of the edge pixel.

9. The photovoltaic power plant cloud-edge collaborative operation and maintenance control system based on digital twins as described in any one of claims 3-7, characterized in that: Using a multi-dimensional digital twin of the target photovoltaic power station, the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station are simultaneously identified based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station. This includes inputting the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin in parallel. The process of using a multi-dimensional digital twin of the target photovoltaic power station to synchronously identify the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station also includes: running the multi-dimensional digital twin to obtain the current microcrack coverage data, current hot spot coverage data, and current dust coverage data of the photovoltaic panel array of the target photovoltaic power station output by the multi-dimensional digital twin; The parallel input of the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin includes: performing numerical normalization processing on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station before inputting them into the multi-dimensional digital twin in parallel. The parallel input of the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station into the multi-dimensional digital twin also includes: using different programmable logic devices to implement numerical normalization processing and parallel input respectively.

10. A cloud-edge collaborative operation and maintenance control method for photovoltaic power plants based on digital twins, characterized in that: The method includes: At the edge, the visible light camera, infrared camera, and near-infrared camera on the drone are simultaneously captured wirelessly, providing overhead images of the target photovoltaic power station. Analyze the multi-camera visual content corresponding to the target photovoltaic power station at the edge; Capture the photovoltaic panel deployment angle, number of photovoltaic panels, unfolded planar area of ​​each photovoltaic panel, and gap between two adjacent photovoltaic panels at the edge of the target photovoltaic power station as various configuration information of the target photovoltaic power station; In the cloud, a multi-level convolutional neural network is used as the output of a multi-dimensional digital twin of the target photovoltaic power station; Using a multi-dimensional digital twin of the target photovoltaic power station in the cloud, the current data on microcracks, hot spots, and dust coverage of the photovoltaic panel array of the target photovoltaic power station are simultaneously identified based on the current ambient light intensity, the area of ​​the target photovoltaic power station, the multi-camera visual content corresponding to the target photovoltaic power station, and various configuration information of the target photovoltaic power station. Based on the synchronous identification results, the cloud performs corresponding operation and maintenance alarm operations and provides corresponding operation and maintenance control strategies.

Images