Remote monitoring method, device and equipment of photovoltaic system and storage medium

Abstract

The invention relates to the field of remote monitoring, and particularly discloses a remote monitoring method and device for a photovoltaic system, equipment and a storage medium. The method comprises the following steps: firstly, acquiring a thermal infrared image of a photovoltaic base station acquired by a thermal infrared imager, a visible light image of the photovoltaic base station acquired by a camera and current values and voltage values of a plurality of preset position assemblies of the photovoltaic base station acquired by a sensor, and then performing feature extraction and correlation analysis on the three images by using a deep learning technology to obtain a correlation result; and finally, a classification result is obtained through a classifier to judge whether to send out a safety early warning of the photovoltaic base station, so that preventive maintenance is carried out, problem deterioration is avoided, the cost of manual intervention is reduced, and the safety and reliability of the photovoltaic base station are improved.

Description

Technical Field

[0001] The present application relates to the field of remote monitoring, and more specifically, to a remote monitoring method, device, equipment and storage medium for a photovoltaic system. Background Art

[0002] Photovoltaic power station is a kind of power station that uses solar energy to generate electricity, using photovoltaic technology to convert sunlight into electrical energy. It consists of a large number of photovoltaic modules (solar panels), and usually transmits the generated electricity to the power system or directly supplies it to users through grid-connected or off-grid methods.

[0003] Traditional photovoltaic system monitoring methods usually rely on manual inspections and regular checks. However, since photovoltaic modules are affected by factors such as shadows, dirt, and aging of panels, it is difficult to detect potential problems in the system in a timely manner by relying solely on manual inspections, especially in large-scale power stations, where it is difficult to efficiently monitor the operating status of each panel. In addition, photovoltaic power stations usually generate high voltages, which poses a potential safety threat of electric shock to system installers, maintenance personnel, and emergency rescue personnel.

[0004] Therefore, a remote monitoring solution for photovoltaic systems is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiments of this application provide a remote monitoring method, device, equipment and storage medium for a photovoltaic system, which first obtains a thermal infrared image of a photovoltaic base station collected by an infrared thermal imager, a visible light image of a photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by a sensor, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally obtains a classification result through a classifier to determine whether to issue a safety warning for the photovoltaic base station, so as to perform preventive maintenance, avoid the deterioration of the problem, and reduce the cost of manual intervention, thereby improving the safety and reliability of the photovoltaic base station.

[0006] According to one aspect of the present application, a remote monitoring method for a photovoltaic system is provided, which includes:

[0007] Acquire a thermal infrared image of the photovoltaic base station collected by an infrared thermal imager, a visible light image of the photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by a sensor;

[0008] Extracting a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor;

[0009] Based on the photovoltaic base station area monitoring multi-modal feature vector and the photovoltaic base station current-voltage association feature vector, it is determined whether to issue a photovoltaic base station safety warning.

[0010] According to another aspect of the present application, a remote monitoring device for a photovoltaic system is provided, comprising:

[0011] A photovoltaic base station data acquisition module is used to acquire a thermal infrared image of the photovoltaic base station collected by an infrared thermal imager, a visible light image of the photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by a sensor;

[0012] A photovoltaic base station data processing module is used to extract a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor;

[0013] The photovoltaic base station safety warning judgment module is used to judge whether to issue a photovoltaic base station safety warning based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector.

[0014] According to another aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the remote monitoring device for the photovoltaic system as described above.

[0015] According to another aspect of the present application, a computer-readable medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the remote monitoring device for a photovoltaic system as described above.

[0016] Compared with the prior art, the present application provides a remote monitoring method, device, equipment and storage medium for a photovoltaic system, which first obtains a thermal infrared image of a photovoltaic base station collected by an infrared thermal imager, a visible light image of a photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by a sensor, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a classifier to obtain a classification result to determine whether to issue a safety warning for the photovoltaic base station, thereby performing preventive maintenance, avoiding the deterioration of problems, and reducing the cost of manual intervention, thereby improving the safety and reliability of the photovoltaic base station. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 4 is a flow chart of a remote monitoring method for a photovoltaic system according to an embodiment of the present application.

[0019] Figure 2 A flowchart for extracting a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor in a remote monitoring method for a photovoltaic system according to an embodiment of the present application.

[0020] Figure 3 The present invention is a flowchart of extracting features from the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager in the remote monitoring method of the photovoltaic system according to the embodiment of the present application to obtain the thermal infrared area tracking feature vector of the photovoltaic base station.

[0021] Figure 4 The present invention is a flowchart of extracting features of current values ​​and voltage values ​​of components at multiple predetermined positions of a photovoltaic base station collected by sensors in a remote monitoring method of a photovoltaic system according to an embodiment of the present application to obtain a current-voltage correlation feature vector of the photovoltaic base station.

[0022] Figure 5 4 is a block diagram of a remote monitoring device for a photovoltaic system according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0024] Figure 1 FIG. 1 is a flow chart of a remote monitoring method for a photovoltaic system according to an embodiment of the present application. Figure 1As shown, according to the remote monitoring method 100 of the photovoltaic system of the embodiment of the present application, the method includes: S110, acquiring a thermal infrared image of a photovoltaic base station collected by an infrared thermal imager, a visible light image of a photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by a sensor; S120, extracting a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, the visible light image of the photovoltaic base station collected by the camera, and the current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by the sensor; S130, judging whether to issue a photovoltaic base station safety warning based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector.

[0025] In the above-mentioned remote monitoring method 100 of the photovoltaic system, the step S110 obtains the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, the visible light image of the photovoltaic base station collected by the camera, and the current value and voltage value of the components at multiple predetermined positions of the photovoltaic base station collected by the sensor. It should be understood that the infrared thermal imager generates thermal images by detecting infrared radiation emitted from the surface of the object, and these images can show the temperature distribution of the photovoltaic components. The camera is used to collect visible light images of the photovoltaic base station. The visible light image provides the appearance information of the photovoltaic component, which can be used to observe whether there are dust, stains, cracks and other problems on the surface of the component. These images help to comprehensively evaluate the physical state of the photovoltaic base station and detect external factors that may affect the efficiency of photovoltaic power generation. Finally, the sensor is used to collect the current and voltage values ​​of the components at each predetermined position of the photovoltaic base station. The current and voltage sensors are usually installed at the output end of the photovoltaic panel to monitor the electrical performance of the component in real time. The current and voltage values ​​can be transmitted to the monitoring platform in real time through the data acquisition system for status evaluation. In this way, infrared thermal imagers, cameras and electrical sensors together constitute a multi-dimensional monitoring system for photovoltaic base stations, which can comprehensively monitor the temperature, appearance and electrical performance of photovoltaic base stations, helping maintenance personnel to promptly detect potential problems and perform maintenance and optimization.

[0026] Specifically, a photovoltaic power station is a facility that uses solar energy to generate electricity, using photovoltaic technology to convert sunlight into electrical energy. This type of power station is usually composed of a large number of photovoltaic components (i.e., solar panels), which transmit the generated electricity to the power system through grid-connected or off-grid means, or directly supply it to users. Traditional photovoltaic system monitoring mainly relies on manual inspections and regular inspections. However, since photovoltaic components are susceptible to factors such as shadows, dirt, and aging of solar panels, it is difficult to detect potential problems in a timely manner by manual inspection alone, especially in large photovoltaic power stations, where it is even more difficult to efficiently monitor the working status of each solar panel. In addition, the high voltage generated by photovoltaic power stations also increases the safety risk of electric shock to installation, maintenance, and emergency rescue personnel. Therefore, in the technical solution of the present application, by obtaining the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, the visible light image of the photovoltaic base station collected by the camera, and the current value and voltage value of multiple predetermined position components of the photovoltaic base station collected by the sensor, and combining deep learning technology, it is determined whether it is necessary to issue a safety warning for the photovoltaic power station, so as to carry out preventive maintenance in a timely manner, prevent the problem from further deteriorating, and reduce the cost of manual intervention, thereby improving the safety and stability of the photovoltaic power station.

[0027] In the above-mentioned remote monitoring method 100 of the photovoltaic system, the step S120 extracts the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector from the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, the visible light image of the photovoltaic base station collected by the camera, and the current values ​​and voltage values ​​of the components of multiple predetermined positions of the photovoltaic base station collected by the sensor. It should be understood that by fusing data of different modes, the health status of the photovoltaic base station can be fully reflected. Not only can a comprehensive evaluation be made from multiple aspects such as temperature, appearance and electrical performance, but the details of the electrical performance can also be further revealed through the current-voltage correlation feature vector, helping to accurately diagnose possible faults or performance degradation problems. Ultimately, through multimodal data fusion, accurate monitoring and real-time early warning of photovoltaic base stations can be achieved, the stability and reliability of the system can be improved, maintenance costs can be reduced, and the service life of the equipment can be extended.

[0028] Figure 2 A flowchart of extracting a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor in the remote monitoring method of the photovoltaic system according to the embodiment of the present application. Figure 2As shown, in a specific embodiment of the present application, the step S120 extracts the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of the photovoltaic base station multiple predetermined position components collected by the sensor, including: S121, feature extraction of the photovoltaic base station thermal infrared image collected by the infrared thermal imager to obtain the photovoltaic base station thermal infrared area tracking feature vector; S122, feature extraction of the photovoltaic base station visible light image collected by the camera to obtain the photovoltaic base station visible light tracking feature vector; S123, associating the photovoltaic base station thermal infrared area tracking feature vector and the photovoltaic base station visible light tracking feature vector to obtain the photovoltaic base station area monitoring multimodal feature vector; S124, feature extraction of the current values ​​and voltage values ​​of the photovoltaic base station multiple predetermined position components collected by the sensor to obtain the photovoltaic base station current-voltage correlation feature vector.

[0029] It should be understood that the infrared thermal imager can capture the temperature changes on the surface of the photovoltaic base station. These temperature changes are often related to faults or damage (such as short circuits, poor contact, component aging, etc.). Therefore, feature extraction of the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager can help the system detect and track possible fault areas. Specifically, in a specific embodiment of the present application, feature extraction includes image preprocessing, hot spot detection, region segmentation, and feature calculation steps. First, the image quality is improved by image denoising and enhancement processing to ensure that the thermal information in the image is clear and discernible. Then, the image segmentation algorithm (such as threshold segmentation, edge detection or region growth) is used to identify the hot spot areas in the image. These hot spots usually correspond to the fault points of the photovoltaic components. Then, the shape, area, temperature distribution, local temperature difference and other features of the hot spot are extracted. These features can help determine the severity and type of the fault. Finally, the dynamic changes of the hot spot area are tracked by comparing multiple frames of images, and the thermal infrared area tracking feature vector is generated. The information of the time dimension and the space dimension is combined to reflect the health status of the photovoltaic base station in real time.

[0030] Further, in step S122, feature extraction is performed on the visible light image of the photovoltaic base station collected by the camera, with the purpose of generating a visible light tracking feature vector of the photovoltaic base station, so as to realize accurate monitoring and fault identification of the photovoltaic base station status. It should be understood that the health status of the photovoltaic base station directly affects its power generation efficiency, and the visible light image can provide appearance information on the surface of the photovoltaic module, such as stains, cracks, aging and other problems. By extracting features from these images, the visual information can be converted into a digital feature vector for subsequent state analysis and fault diagnosis. Specifically, in a specific embodiment of the present application, first, the image quality is improved by image preprocessing (such as denoising, contrast enhancement, etc.) to reduce the interference of environmental factors on the image. Then, an image segmentation algorithm (such as edge-based segmentation or threshold segmentation) is used to separate the photovoltaic module from the background and identify the area of ​​concern. Then, the key features reflecting the health status of the photovoltaic module are extracted from the image through a feature extraction algorithm, such as the shape of the crack, the area of ​​contamination, or the surface temperature difference of the photovoltaic panel. These extracted features reflect the appearance of the photovoltaic base station, which can help detect possible faults or damage, and then generate a visible light tracking feature vector for the photovoltaic base station.

[0031] Furthermore, infrared thermal imagers and cameras reflect the status of photovoltaic base stations from the aspects of temperature and appearance respectively. Thermal infrared images can reveal temperature changes at the fault point, while visible light images provide information on damage, stains or cracks on the appearance of the equipment. Combining these two types of features can help improve the accuracy and timeliness of fault detection. Common methods include weighted fusion, splicing, or automatic fusion through deep learning models (such as multimodal neural networks). Through multimodal feature vectors, the system can not only more accurately identify and locate the fault area of ​​the photovoltaic base station, but also track and evaluate its operating status in real time, which helps to improve the operation and maintenance efficiency and reliability of the photovoltaic system.

[0032] In particular, considering that current and voltage are important parameters for measuring the working status of photovoltaic modules, their changes directly reflect the health status of the modules, such as poor contact, module aging or shadowing. Therefore, by extracting the features of these current and voltage data, the operating efficiency of photovoltaic base stations can be effectively monitored and potential problems can be discovered in time.

[0033] Figure 3 The present invention is a flowchart of extracting features from the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager to obtain the thermal infrared area tracking feature vector of the photovoltaic base station in the remote monitoring method of the photovoltaic system according to the embodiment of the present application. Figure 3As shown, in a specific embodiment of the present application, the step S121, performing feature extraction on the photovoltaic base station thermal infrared image collected by the infrared thermal imager to obtain a photovoltaic base station thermal infrared area tracking feature vector, includes: S1211, passing the photovoltaic base station thermal infrared image collected by the infrared thermal imager through a photovoltaic base station thermal infrared target detection network to obtain multiple photovoltaic base station thermal infrared regions of interest; S1212, passing the multiple photovoltaic base station thermal infrared regions of interest through a photovoltaic base station thermal infrared region significant target detector to obtain the photovoltaic base station thermal infrared region tracking feature vector.

[0034] It should be understood that the thermal image provided by the infrared thermal imager can reflect the changes in the surface temperature distribution of the photovoltaic module, and the temperature anomaly is often related to the failure of the photovoltaic module (such as short circuit, poor contact or local overheating, etc.). Through the thermal infrared target detection network, the hot spot area in the image can be automatically analyzed, thereby improving the efficiency and accuracy of fault detection and monitoring. In a specific embodiment of the present application, first, the thermal image collected by the infrared thermal imager is used as input data. These images usually have a high spatial resolution and can clearly show the temperature distribution of the photovoltaic module. Next, these images are input into the thermal infrared target detection network of the photovoltaic base station, which is usually based on a deep learning model, such as a convolutional neural network (CNN) or a regional convolutional neural network (R-CNN). After training, these networks can automatically identify and extract thermal anomaly areas (i.e., heat source areas or hot spots) in the image. These areas are usually manifested as temperature anomaly points in the image, which may be caused by equipment failure or other problems. Among them, the network extracts multi-level feature representations from the original thermal image through multiple levels of convolution and pooling operations, thereby achieving efficient target detection. During the detection process, the network will output multiple candidate boxes (bounding boxes), each box represents a possible region of interest, which usually corresponds to abnormal hotspots or potential fault locations in the image. Through subsequent algorithms such as non-maximum suppression (NMS), redundant boxes are removed, and finally multiple high-confidence regions of interest are obtained. These regions of interest can accurately mark the abnormal parts of the photovoltaic components, helping operation and maintenance personnel to quickly locate and diagnose problem areas. By continuously monitoring these areas, the system can track the health status of the photovoltaic base station in real time, detect faults in time, and perform effective operation and maintenance management. Specifically, in the technical solution of the present application, the photovoltaic base station thermal infrared target detection network is used to process the photovoltaic base station thermal infrared image collected by the infrared thermal imager with the following target detection formula to obtain the multiple photovoltaic base station thermal infrared regions of interest;

[0035] Among them, the target detection formula is:

[0036] ROI=(cls(ψ det ,B),Regr(ψdet ,B))

[0037] Among them, ψ det is the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, B is the anchor frame, ROI is the thermal infrared region of interest of multiple photovoltaic base stations, cls(ψ det ,B) indicates classification, Regr(ψ det ,B) indicates regression.

[0038] Furthermore, considering that the regions of interest in infrared thermal images usually represent possible fault points or hot spots of the equipment, the recognition accuracy of these regions can be enhanced through further processing by the salient object detector, providing strong data support for subsequent fault location and diagnosis. Specifically, first, the multiple regions of interest extracted by the thermal infrared target detection network represent the areas of abnormal temperature distribution in the image, which may correspond to problems such as overheating and local failure of photovoltaic modules. Next, these regions of interest are input into the salient object detector of the thermal infrared region of the photovoltaic base station. The salient object detector is a deep learning model specially used to extract the most representative and significant areas in the image, usually based on convolutional neural networks (CNN) or other efficient target detection frameworks. The role of this detector is to find out the thermal abnormal areas that have a greater impact on the system status by further analyzing the spatial characteristics, temperature patterns and texture information of the regions of interest, and ignoring possible background noise or irrelevant information. In this process, the salient object detector uses the size, shape, temperature change and other characteristics of each heat source in the image to perform a fine analysis of multiple regions of interest and identify the most significant abnormal hot spots. These areas may be caused by electrical failures, poor contact, occlusion or aging, and reflect the potential problems of the components through temperature changes. After being processed by the detector, the precise features of each region of interest can be obtained, including its position, temperature, shape, size and other information, which constitute the photovoltaic base station thermal infrared area tracking feature vector. Specifically, each layer of the photovoltaic base station thermal infrared area salient target detector is used to perform the following operations on the input data in the forward pass of the layer: convolution processing is performed on the input data using a first convolution kernel to obtain a first convolution feature map; convolution processing is performed on the first convolution feature map using a second convolution kernel to obtain a second convolution feature map, wherein the size of the first convolution kernel is larger than the size of the second convolution kernel; and, pooling processing is performed on the second convolution feature map to obtain a pooling feature map; and activation processing is performed on the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the photovoltaic base station thermal infrared area salient target detector is the photovoltaic base station thermal infrared area tracking feature vector, and the input of the first layer of the photovoltaic base station thermal infrared area salient target detector is the multiple photovoltaic base station thermal infrared regions of interest.

[0039] In a specific embodiment of the present application, the step S122, performing feature extraction on the photovoltaic base station visible light image collected by the camera to obtain a photovoltaic base station visible light tracking feature vector, includes: passing the photovoltaic base station visible light image collected by the camera through a photovoltaic base station visible light image denoising module to obtain a photovoltaic base station visible light denoising enhanced image; passing the photovoltaic base station visible light denoising enhanced image through a photovoltaic base station visible light feature extractor based on a spatial attention mechanism to obtain the photovoltaic base station visible light tracking feature vector.

[0040] It should be understood that visible light images are easily disturbed by various factors such as ambient light, weather changes, sensor noise, etc. during the acquisition process, resulting in noise, blur or unclear parts in the image. These noises will affect subsequent analysis and judgment, especially in tasks such as appearance inspection and fault diagnosis of photovoltaic modules, reducing the accuracy and efficiency of recognition. In the technical solution of the present application, the photovoltaic base station visible light image denoising module is usually based on advanced image processing technologies, such as filtering algorithms, deep learning models, etc., to remove random noise and background interference in the image, while retaining important details in the image as much as possible. In traditional methods, commonly used denoising techniques include median filtering, Gaussian filtering, bilateral filtering, etc., which remove high-frequency noise in the image by smoothing the pixel values ​​of the image. However, traditional methods may cause the loss of image details, especially at the edges of photovoltaic modules or small fault areas. In order to solve this problem, the technical solution of the present application adopts a deep learning method, especially a convolutional neural network (CNN) or a generative adversarial network (GAN), to retain more image details. By training the deep network, the model can learn how to dynamically adjust the noise reduction strategy according to the different features of the image, reducing the noise while retaining important information in the image (such as cracks, stains, damage, etc.). This deep learning-based noise reduction method can effectively remove complex noise, improve image quality, and retain image details more accurately.

[0041] Furthermore, although the denoised and enhanced visible light image contains more clear details and less noise, different areas in the image contribute unequally to the health status or fault characteristics of the photovoltaic base station. For example, the surface cracks, stains or damaged areas of photovoltaic modules are often small and have strong local features, while other irrelevant areas or background noise contribute less to the overall image analysis. At this time, the spatial attention mechanism can effectively guide the feature extractor to focus on these areas with important information, thereby improving the extraction quality of key features. Among them, the core idea of ​​the spatial attention mechanism is to emphasize the key parts of the image by assigning different weights to different spatial areas of the image. In this process, the photovoltaic base station visible light feature extractor based on the spatial attention mechanism first extracts features from the image through a convolutional neural network (CNN) to obtain a preliminary feature representation of each pixel or region. Then, the spatial attention module assigns a weight value to each position by calculating the significance of different positions in the image. These weights reflect the contribution of the region to the overall image analysis, so that the feature extraction process can focus more on important areas in the image, such as cracks, fault points, or temperature anomalies, while ignoring background or irrelevant parts. Through this process, the feature extractor can eventually generate a photovoltaic base station visible light tracking feature vector, which contains the key areas in the image. The rich feature information accurately reflects the health status of photovoltaic components. Specifically, the convolutional coding part of the photovoltaic base station visible light feature extractor based on the spatial attention mechanism is used to perform deep convolutional coding on the photovoltaic base station visible light denoising and enhancement image to obtain an initial convolutional feature map; the initial convolutional feature map is input into the spatial attention part of the photovoltaic base station visible light feature extractor based on the spatial attention mechanism to obtain a spatial attention map; the spatial attention map is passed through the Softmax activation function to obtain a spatial attention feature map; the spatial attention feature map and the initial convolutional feature map are multiplied by the position points to obtain the photovoltaic base station visible light tracking feature vector.

[0042] Figure 4 This is a flow chart of extracting features of the current values ​​and voltage values ​​of multiple components at predetermined locations of the photovoltaic base station collected by sensors in a remote monitoring method of a photovoltaic system according to an embodiment of the present application to obtain a current-voltage correlation feature vector of the photovoltaic base station. Figure 4As shown, in a specific embodiment of the present application, the step S124, performing feature extraction on the current values ​​and voltage values ​​of the components at multiple predetermined positions of the photovoltaic base station collected by the sensor to obtain the current-voltage associated feature vector of the photovoltaic base station, includes: S1241, arranging the current values ​​and voltage values ​​of the components at multiple predetermined positions of the photovoltaic base station collected by the sensor into a photovoltaic base station current input vector and a photovoltaic base station voltage input vector; S1242, calculating the transfer matrix between the photovoltaic base station current input vector and the photovoltaic base station voltage input vector to obtain the photovoltaic base station current-voltage associated feature vector.

[0043] It should be understood that a photovoltaic base station is usually composed of multiple photovoltaic modules, which are distributed in different locations. The current and voltage values ​​of each module are affected by its local environment, light, temperature, and the state of the module itself. In order to fully understand the working conditions of the entire photovoltaic base station, it is necessary to unify the current and voltage data of all modules and input them into subsequent models or algorithms in the form of input vectors. Arranging the current and voltage values ​​as input vectors can make the data more standardized in format, which is convenient for machine learning or deep learning models to process and analyze.

[0044] Furthermore, considering that current and voltage are key electrical parameters, there is a certain correlation between them, especially when the load, light, temperature and other conditions of the components change, the changes in current and voltage often affect each other. Therefore, by constructing the correlation feature vector between current and voltage, the working state of the photovoltaic system can be more comprehensively reflected, assisting subsequent fault diagnosis and performance evaluation.

[0045] In the above-mentioned remote monitoring method 100 for the photovoltaic system, the step S130 determines whether to issue a photovoltaic base station safety warning based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector. It should be understood that the photovoltaic base station is composed of multiple components, and their operating status is affected by environmental factors such as light, temperature, and wind speed. At the same time, performance degradation or failure may also occur due to hardware failure or improper maintenance. Monitoring and analysis based on these multimodal data can effectively judge the safety status of the photovoltaic base station, issue early warnings, and avoid system losses. Such an early warning system not only improves the operating efficiency of the photovoltaic base station, but also enhances the reliability and safety of the system.

[0046] In a specific embodiment of the present application, the step S130, based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage association feature vector, determines whether to issue a photovoltaic base station safety warning, including: S131, performing hidden disturbance optimization based on the intrinsic decomposition space on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage association feature vector to obtain a base station safety warning judgment feature vector; S132, passing the base station safety warning judgment feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a photovoltaic base station safety warning.

[0047] It should be understood that the fusion of the multimodal feature vector of photovoltaic base station regional monitoring and the current-voltage correlation feature vector to obtain the base station safety warning judgment feature vector is mainly to more comprehensively and accurately evaluate the operating status of the photovoltaic base station and promptly discover potential safety hazards. Among them, the safe operation of photovoltaic base stations not only depends on the changes in current and voltage, but also is affected by environmental factors (such as temperature, humidity, light intensity, etc.). Therefore, relying solely on current-voltage data may not fully reflect the health status of the system, especially under complex external environmental changes. By integrating the multimodal features of regional monitoring with the current-voltage correlation features, data of different dimensions can be fully utilized to provide more accurate safety warnings.

[0048] In particular, considering that the images collected by infrared thermal imagers and cameras obtain information from thermal radiation and visible light bands respectively, this leads to significant differences in the distribution of their image data in terms of color, texture, and brightness; while the current and voltage values ​​obtained by the sensor are based on physical electrical characteristics and are usually presented in the form of numerical vectors. Therefore, although these data are all related to the operating status of the PV base station, their features have different distribution patterns in their respective fields. When trying to apply the source domain (such as data obtained from the environment of a different PV base station or device) model to the target domain (a new PV base station or a different environment), there is usually a large difference in the feature space of the source domain and the target domain. This distribution difference will lead to a decrease in the generalization ability of traditional models, which is manifested as a significant decrease in performance. For example, some patterns of infrared thermal images or voltage and current features in the source domain may not be accurately recognized in the target domain because the background, lighting conditions, or device status of the two are different. Moreover, traditional feature fusion methods often simply splice or directly weight and synthesize data of different modes, fail to deeply understand and model the interaction between various data sources, and ignore the more complex semantic associations between various modes, resulting in the fused base station safety warning judgment feature vector being unable to fully express the complex semantic connections between various modes, thereby affecting downstream tasks, namely, the accuracy of safety warning judgment. Therefore, in the technical solution of the present application, the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector are subjected to hidden perturbation optimization based on the intrinsic decomposition space to obtain the base station safety warning judgment feature vector.

[0049] Specifically, the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage associated feature vector are subjected to hidden disturbance optimization based on the intrinsic decomposition space to obtain the base station safety early warning judgment feature vector, including: mapping the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage associated feature vector to the intrinsic decomposition space to obtain the photovoltaic base station area monitoring intrinsic feature vector and the photovoltaic base station current-voltage associated intrinsic feature vector; performing inter-domain feature calibration on the photovoltaic base station area monitoring intrinsic feature vector and the photovoltaic base station current-voltage associated intrinsic feature vector to obtain the photovoltaic base station area monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector; constructing a photovoltaic base station multimodal fine-grained semantic information field between the photovoltaic base station area monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector; and fusing the photovoltaic base station area monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector based on the photovoltaic base station multimodal fine-grained semantic information field to obtain the base station safety early warning judgment feature vector.

[0050] More specifically, constructing a photovoltaic base station multimodal fine-grained semantic information field between the photovoltaic base station regional monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector, including: multiplying the photovoltaic base station regional monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector after transposition, and then dividing by the scale of the photovoltaic base station current-voltage associated modulated intrinsic feature vector to obtain a photovoltaic base station multimodal fine-grained interaction matrix; and convolutionally encoding the photovoltaic base station multimodal fine-grained matrix to obtain the photovoltaic base station multimodal fine-grained semantic information field.

[0051] More specifically, based on the multimodal fine-grained semantic information field of the photovoltaic base station, the photovoltaic base station regional monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector are fused to obtain a base station safety early warning judgment feature vector, including: multiplying the photovoltaic base station multimodal fine-grained semantic information field by the photovoltaic base station regional monitoring modulated intrinsic feature vector to obtain the photovoltaic base station regional monitoring multimodal feature vector after information field mapping; multiplying the photovoltaic base station multimodal fine-grained semantic information field by the photovoltaic base station current-voltage associated modulated intrinsic feature vector to obtain the photovoltaic base station current-voltage associated feature vector after information field mapping; weighted fusion of the photovoltaic base station regional monitoring multimodal feature vector after information field mapping and the photovoltaic base station current-voltage associated feature vector after information field mapping to obtain the base station safety early warning judgment feature vector.

[0052] In the embodiment of the present application, specifically, the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage association feature vector are subjected to hidden disturbance optimization based on the intrinsic decomposition space to obtain the base station safety early warning judgment feature vector, including: the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage association feature vector are processed by the following formula to obtain the base station safety early warning judgment feature vector; wherein, the formula is:

[0053]

[0054] v c =αv kt +βv st

[0055] Among them, V1 represents the multimodal feature vector of PV base station regional monitoring, V2 represents the current-voltage correlation feature vector of PV base station, PCA(·) represents principal component analysis, U1 represents the sequence of eigendecomposition vectors of PV base station regional monitoring, Λ1 represents the diagonal matrix of PV base station regional monitoring, λ 11 and λ 1mThey represent the eigenvalues ​​of the first and mth positions of the diagonal matrix of the PV base station area monitoring, T represents the transpose of the vector, concat{·;·} represents the vector concatenation, and v 11 、v 12 、v 1m They represent the first, second and mth eigenvectors of the sequence of PV base station regional monitoring intrinsic decomposition vectors, V k represents the PV base station regional monitoring intrinsic eigenvector, U2 represents the sequence of PV base station current-voltage correlation intrinsic decomposition vectors, Λ2 represents the PV base station regional monitoring diagonal matrix, λ 21 and λ 2m They represent the eigenvalues ​​of the first and mth positions of the current-voltage correlation diagonal matrix of the photovoltaic base station, respectively. 21 、v 22 、u 2m They represent the first, second and mth eigenvectors of the sequence of the current-voltage correlation eigendecomposition vectors of the photovoltaic base station, V s represents the PV base station current-voltage associated intrinsic eigenvector, v i represents the ith eigenvalue of the PV base station regional monitoring intrinsic eigenvector, v j represents the jth eigenvalue of the PV base station current-voltage correlation intrinsic eigenvector, represents the square of the vector norm, exp represents the natural exponential function, V' k represents the intrinsic eigenvector after modulation of the photovoltaic base station area monitoring, V' s represents the intrinsic eigenvector after the current-voltage correlation modulation of the photovoltaic base station, represents matrix multiplication, S represents the scale of the vector, M x Represents the multi-modal fine-grained interaction matrix of the photovoltaic base station, Conv 3×3 (M x ) indicates that M x A convolutional encoding operation based on a convolution kernel of 3×3 is performed, Ω represents the multimodal fine-grained semantic information field of the photovoltaic base station, and v kt represents the multi-modal feature vector of PV base station area monitoring after information field mapping, v st represents the current-voltage correlation feature vector of the photovoltaic base station after information field mapping, α and β represent weighted hyperparameters, and v c Represents the base station safety warning judgment feature vector.

[0056] That is, in the technical solution of the present application, the photovoltaic base station regional monitoring multimodal feature vector and the photovoltaic base station current-voltage associated feature vector are subjected to hidden disturbance optimization based on the intrinsic decomposition space. The process first maps the photovoltaic base station regional monitoring multimodal feature vector and the photovoltaic base station current-voltage associated feature vector to the intrinsic decomposition space to obtain the photovoltaic base station regional monitoring intrinsic feature vector and the photovoltaic base station current-voltage associated intrinsic feature vector. Those of ordinary skill in the art should know that the process of mapping to the intrinsic decomposition space is a combination of dimensionality reduction and feature extraction. In a specific example, the eigenvalue decomposition technology of principal component analysis (PCA) is used to find the intrinsic decomposition space, and the intrinsic decomposition space mapping is performed. The photovoltaic base station regional monitoring intrinsic feature vector and the photovoltaic base station current-voltage associated intrinsic feature vector after information field mapping not only represent the projection of the original feature vector in the new coordinate system, but also reflect the main direction of change of the data, which helps to avoid the "dimensionality disaster", that is, the problem of a sharp drop in sample density as the number of features increases.

[0057] Next, after mapping the feature vector to the intrinsic decomposition space, it is necessary to perform inter-domain feature calibration on the photovoltaic base station regional monitoring intrinsic feature vector and the photovoltaic base station current-voltage correlation intrinsic feature vector to adjust the source domain feature vector to better match the data distribution of the target domain. This operation aims to capture and compensate for the systematic deviation between the two domains to reduce the hidden disturbance of the class expression of the local feature distribution of the photovoltaic base station regional monitoring intrinsic feature vector and the photovoltaic base station current-voltage correlation intrinsic feature vector to the overall class expression. After modulation, the modulated intrinsic feature vector of the photovoltaic base station regional monitoring and the modulated intrinsic feature vector of the photovoltaic base station current-voltage correlation can retain the source domain characteristics while being more in line with the actual needs of the target domain, enhancing the model's understanding and application of the target domain data.

[0058] After completing the scale modulation of the feature vector, a multimodal fine-grained semantic information field of the photovoltaic base station is constructed between the intrinsic feature vector after the photovoltaic base station regional monitoring modulation and the intrinsic feature vector after the photovoltaic base station current-voltage association modulation. The multimodal fine-grained semantic information field of the photovoltaic base station is used to describe a more detailed relationship between feature vectors. The introduction of the multimodal fine-grained semantic information field of the photovoltaic base station enables the model to go beyond simple point-to-point comparison and consider more complex patterns and contextual information, thereby enhancing the model's sensitivity to subtle differences and improving the quality of feature representation. Finally, based on the multimodal fine-grained semantic information field of the photovoltaic base station, the intrinsic feature vector after the photovoltaic base station regional monitoring modulation and the intrinsic feature vector after the photovoltaic base station current-voltage association modulation are fused to obtain the base station safety warning judgment feature vector.

[0059] Furthermore, the base station safety warning judgment feature vector is passed through a classifier to obtain a classification result in order to realize automatic monitoring and real-time warning of the operating status of the photovoltaic base station. Among them. Common classification algorithms, such as support vector machine (SVM), random forest, and neural network, can extract effective information from complex features to determine whether the current state is normal. Specifically, when a new feature vector is input into the classifier, the classifier will give a classification result, such as "normal" or "fault" according to the pattern learned during training. If the classification result indicates that there is a potential risk or fault, the system will trigger a safety warning and notify the operation and maintenance personnel to conduct inspections or take corresponding preventive measures, thereby avoiding accidents and ensuring the safe and stable operation of the photovoltaic base station. In this way, intelligent monitoring and efficient response of photovoltaic base stations can be achieved, greatly improving the safety and operation and maintenance efficiency of the system.

[0060] In summary, the embodiment of the present application first obtains the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, the visible light image of the photovoltaic base station collected by the camera, and the current and voltage values ​​of multiple predetermined position components of the photovoltaic base station collected by the sensor, and then uses deep learning technology to perform feature extraction and correlation analysis on the three. Finally, a classifier is used to obtain a classification result to determine whether to issue a safety warning for the photovoltaic base station, so as to perform preventive maintenance, avoid the deterioration of the problem, and reduce the cost of manual intervention, thereby improving the safety and reliability of the photovoltaic base station.

[0061] As described above, the photovoltaic system remote monitoring method 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the photovoltaic system remote monitoring method 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the photovoltaic system remote monitoring method 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the photovoltaic system remote monitoring method 100 can also be one of the many hardware modules of the terminal device.

[0062] Alternatively, in another example, the photovoltaic system remote monitoring method 100 and the terminal device may also be separate devices, and the photovoltaic system remote monitoring method 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0063] Figure 5 FIG. 1 is a block diagram of a remote monitoring device for a photovoltaic system according to an embodiment of the present application. Figure 5As shown, according to the remote monitoring device of the photovoltaic system of the embodiment of the present application, it includes: S110, a photovoltaic base station data acquisition module 110, which is used to obtain the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager, the visible light image of the photovoltaic base station collected by the camera, and the current value and voltage value of multiple predetermined position components of the photovoltaic base station collected by the sensor; a photovoltaic base station data processing module 120, which is used to extract the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current value and voltage value of multiple predetermined position components of the photovoltaic base station collected by the sensor; a photovoltaic base station safety warning judgment module 130, which is used to judge whether to issue a photovoltaic base station safety warning based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector.

[0064] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned remote monitoring device for photovoltaic systems have been described in detail above. Figures 1 to 4 The description of the remote monitoring method of a photovoltaic system has been described in detail, and therefore, its repeated description will be omitted.

Claims

1. A remote monitoring method for a photovoltaic system, characterized in that: include: Acquire a thermal infrared image of the photovoltaic base station collected by an infrared thermal imager, a visible light image of the photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by a sensor; Extracting a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor; Based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage association feature vector, it is determined whether to issue a photovoltaic base station safety warning.

2. The remote monitoring method of a photovoltaic system according to claim 1, characterized in that: Extracting a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor, including: Performing feature extraction on the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager to obtain a thermal infrared area tracking feature vector of the photovoltaic base station; Performing feature extraction on the photovoltaic base station visible light image collected by the camera to obtain a photovoltaic base station visible light tracking feature vector; Associating the photovoltaic base station thermal infrared area tracking feature vector and the photovoltaic base station visible light tracking feature vector to obtain the photovoltaic base station area monitoring multimodal feature vector; Feature extraction is performed on the current values ​​and voltage values ​​of the components at the plurality of predetermined positions of the photovoltaic base station collected by the sensor to obtain a current-voltage correlation feature vector of the photovoltaic base station.

3. The remote monitoring method of a photovoltaic system according to claim 2, characterized in that: Performing feature extraction on the thermal infrared image of the photovoltaic base station collected by the infrared thermal imager to obtain a thermal infrared area tracking feature vector of the photovoltaic base station includes: The thermal infrared image of the photovoltaic base station collected by the infrared thermal imager is passed through a thermal infrared target detection network of the photovoltaic base station to obtain multiple thermal infrared regions of interest of the photovoltaic base station; The multiple photovoltaic base station thermal infrared regions of interest are passed through a photovoltaic base station thermal infrared region salient target detector to obtain the photovoltaic base station thermal infrared region tracking feature vector.

4. The remote monitoring method of a photovoltaic system according to claim 3, characterized in that: Performing feature extraction on the photovoltaic base station visible light image collected by the camera to obtain a photovoltaic base station visible light tracking feature vector includes: Passing the photovoltaic base station visible light image collected by the camera through a photovoltaic base station visible light image noise reduction module to obtain a photovoltaic base station visible light noise reduction enhanced image; The photovoltaic base station visible light denoising and enhanced image is passed through a photovoltaic base station visible light feature extractor based on a spatial attention mechanism to obtain the photovoltaic base station visible light tracking feature vector.

5. The remote monitoring method of a photovoltaic system according to claim 4, characterized in that: The current values ​​and voltage values ​​of the components at the plurality of predetermined positions of the photovoltaic base station collected by the sensor are subjected to feature extraction to obtain a current-voltage correlation feature vector of the photovoltaic base station, including: Arranging the current values ​​and voltage values ​​of the components at the plurality of predetermined positions of the photovoltaic base station collected by the sensor into a photovoltaic base station current input vector and a photovoltaic base station voltage input vector; A transfer matrix between the photovoltaic base station current input vector and the photovoltaic base station voltage input vector is calculated to obtain the photovoltaic base station current-voltage association feature vector.

6. The remote monitoring method of a photovoltaic system according to claim 5, characterized in that: Based on the photovoltaic base station area monitoring multi-modal feature vector and the photovoltaic base station current-voltage correlation feature vector, determining whether to issue a photovoltaic base station safety warning includes: Performing hidden disturbance optimization based on the intrinsic decomposition space on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector to obtain a base station safety early warning judgment feature vector; The base station safety warning judgment feature vector is passed through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a photovoltaic base station safety warning.

7. The remote monitoring method of a photovoltaic system according to claim 6, characterized in that: The photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector are subjected to hidden disturbance optimization based on the intrinsic decomposition space to obtain a base station safety early warning judgment feature vector, including: Mapping the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage association feature vector to an intrinsic decomposition space to obtain a photovoltaic base station area monitoring intrinsic feature vector and a photovoltaic base station current-voltage association intrinsic feature vector; Performing inter-domain feature calibration on the photovoltaic base station regional monitoring intrinsic feature vector and the photovoltaic base station current-voltage associated intrinsic feature vector to obtain the photovoltaic base station regional monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector; Constructing a photovoltaic base station multimodal fine-grained semantic information field between the photovoltaic base station regional monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage correlation modulated intrinsic feature vector; Based on the multimodal fine-grained semantic information field of the photovoltaic base station, the photovoltaic base station regional monitoring modulated intrinsic feature vector and the photovoltaic base station current-voltage associated modulated intrinsic feature vector are fused to obtain the base station safety warning judgment feature vector.

8. A remote monitoring device for a photovoltaic system, characterized in that: include: A photovoltaic base station data acquisition module is used to acquire a thermal infrared image of the photovoltaic base station collected by an infrared thermal imager, a visible light image of the photovoltaic base station collected by a camera, and current values ​​and voltage values ​​of components at multiple predetermined positions of the photovoltaic base station collected by a sensor; A photovoltaic base station data processing module is used to extract a photovoltaic base station area monitoring multimodal feature vector and a photovoltaic base station current-voltage correlation feature vector from the photovoltaic base station thermal infrared image collected by the infrared thermal imager, the photovoltaic base station visible light image collected by the camera, and the current values ​​and voltage values ​​of multiple components at predetermined positions of the photovoltaic base station collected by the sensor; The photovoltaic base station safety warning judgment module is used to judge whether to issue a photovoltaic base station safety warning based on the photovoltaic base station area monitoring multimodal feature vector and the photovoltaic base station current-voltage correlation feature vector.

9. An electronic device, characterized in that: include: at least one processor; And, a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the remote monitoring method for a photovoltaic system as described in any one of claims 1 to 7.

10. A readable storage medium, wherein a plurality of instructions are stored in the readable storage medium; the plurality of instructions are used for a processor to load and execute the remote monitoring method for a photovoltaic system according to any one of claims 1 to 7.

Images