Intelligent inspection system of distributed photovoltaic power station

Abstract

The embodiment of the invention provides an intelligent inspection system for a distributed photovoltaic power station. The intelligent inspection system comprises the following steps: acquiring surface temperature heat distribution images of a to-be-analyzed photovoltaic cell module in the distributed photovoltaic power station at a plurality of preset time points in a preset time period; performing heat distribution feature extraction on the surface temperature heat distribution images at the plurality of preset time points to obtain a sequence of heat distribution feature vectors of the battery assembly; extracting a time domain correlation feature between the sequences of the heat distribution feature vectors of the battery assembly to obtain a heat distribution time sequence feature vector of the battery assembly; and determining whether a fault early warning prompt is generated or not based on the battery assembly heat distribution time sequence feature vector. Therefore, the surface temperature heat distribution images can be analyzed and processed in combination with a deep learning algorithm, so that fault identification and early warning prompt of the photovoltaic cell module are carried out, and intelligent inspection is realized.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of intelligent inspection, and more specifically, to an intelligent inspection system for a distributed photovoltaic power station. Background Art

[0002] A distributed photovoltaic power station (distributed photovoltaic power generation system) refers to a photovoltaic power generation system installed on the user side or the distribution network side, which can directly supply electricity to users, or transmit electric energy to the distribution network, or achieve the balance of both. The basic equipment of a distributed photovoltaic power station includes photovoltaic cell components, photovoltaic array brackets, DC busbar boxes, DC distribution cabinets, grid-connected inverters, AC distribution cabinets, etc.

[0003] Photovoltaic cell components are the core components of a distributed photovoltaic power station, and their working status and performance directly affect the power generation and safety of the entire power station. Due to the dispersion and large scale of distributed photovoltaic power stations, traditional manual inspection methods are difficult to meet the requirements of real-time, accurate, and efficient monitoring. Therefore, an optimized inspection scheme is expected. Summary of the Invention

[0004] To solve the above technical problems, the present invention is proposed. Embodiments of the present invention provide an intelligent inspection system for a distributed photovoltaic power station, which acquires surface temperature thermal distribution images of photovoltaic cell components to be analyzed in a distributed photovoltaic power station at multiple predetermined time points within a predetermined time period; extracts thermal distribution characteristics from the surface temperature thermal distribution images at the multiple predetermined time points to obtain a sequence of battery component thermal distribution feature vectors; extracts temporal correlation characteristics between the sequences of the battery component thermal distribution feature vectors to obtain a battery component thermal distribution time series feature vector; and determines whether to generate a fault warning prompt based on the battery component thermal distribution time series feature vector. In this way, these surface temperature thermal distribution images can be analyzed and processed by combining deep learning algorithms to identify faults and give warning prompts for photovoltaic cell components, realizing intelligent inspection.

[0005] In a first aspect, embodiments of the present invention provide an intelligent inspection system for a distributed photovoltaic power station, which includes:

[0006] A thermal distribution image acquisition module, configured to acquire surface temperature thermal distribution images of photovoltaic cell components to be analyzed in a distributed photovoltaic power station at multiple predetermined time points within a predetermined time period;

[0007] A thermal distribution feature extraction module, configured to extract thermal distribution characteristics from the surface temperature thermal distribution images at the multiple predetermined time points to obtain a sequence of battery component thermal distribution feature vectors;

[0008] A time-domain correlation feature extraction module for extracting time-domain correlation features between sequences of the thermal distribution feature vectors of the battery assembly to obtain a time-series feature vector of the thermal distribution of the battery assembly; and

[0009] A fault warning prompt generation module for determining whether to generate a fault warning prompt based on the time-series feature vector of the thermal distribution of the battery assembly.

[0010] In some possible embodiments, the thermal distribution feature extraction module includes:

[0011] A feature extraction unit for extracting thermal distribution features from the surface temperature thermal distribution images at the multiple predetermined time points by using a deep learning network model to obtain a sequence of the thermal distribution feature vectors of the battery assembly.

[0012] In some possible embodiments, the deep learning network model is a battery assembly thermal distribution feature extractor based on a convolutional neural network model.

[0013] In some possible embodiments, the feature extraction unit is configured to:

[0014] Pass the surface temperature thermal distribution images at the multiple predetermined time points through the battery assembly thermal distribution feature extractor based on the convolutional neural network model to obtain a sequence of the thermal distribution feature vectors of the battery assembly.

[0015] In some possible embodiments, the battery assembly thermal distribution feature extractor based on the convolutional neural network model includes: an input layer, a convolutional layer, a pooling layer, an activation layer, and an output layer.

[0016] In some possible embodiments, the time-domain correlation feature extraction module includes:

[0017] A transformation coding unit for passing the sequence of the thermal distribution feature vectors of the battery assembly through a thermal distribution full-time-domain feature extractor based on a transformer to obtain the time-series feature vector of the thermal distribution of the battery assembly.

[0018] In some possible embodiments, the transformation coding unit is configured to:

[0019] Use the thermal distribution full-time-domain feature extractor based on the transformer to capture the full-time-domain correlation information between the sequences of the thermal distribution feature vectors of the battery assembly to obtain the time-series feature vector of the thermal distribution of the battery assembly.

[0020] In some possible embodiments, the fault warning prompt generation module includes:

[0021] A feature distribution optimization unit for optimizing the feature distribution of the time-series feature vector of the thermal distribution of the battery assembly to obtain an optimized time-series feature vector of the thermal distribution of the battery assembly; and

[0022] A classification unit is used to obtain a classification result by passing the optimized thermal distribution time series feature vector of the battery assembly through a multi-task classification head module, and the classification result is used to indicate whether a fault warning prompt is generated.

[0023] In some possible embodiments, the classification unit is configured to:

[0024] Pass the optimized thermal distribution time series feature vector of the battery assembly through a fine-grained classifier to obtain multiple fault category probability values;

[0025] Pass the optimized thermal distribution time series feature vector of the battery assembly through a coarse-grained classifier to obtain a first probability value and a second probability value;

[0026] Fuse the multiple fault category probability values, the first probability value, and the second probability value in a probability-weighted fusion manner to obtain a comprehensive expression probability value; and

[0027] Obtain the classification result based on the comprehensive expression probability value.

[0028] In a second aspect, an intelligent inspection method for a distributed photovoltaic power station according to an embodiment of the present invention includes:

[0029] Obtain surface temperature thermal distribution images of a photovoltaic battery assembly to be analyzed in a distributed photovoltaic power station at multiple predetermined time points within a predetermined time period;

[0030] Extract thermal distribution features from the surface temperature thermal distribution images at the multiple predetermined time points to obtain a sequence of battery assembly thermal distribution feature vectors;

[0031] Extract time-domain correlation features between the sequences of the battery assembly thermal distribution feature vectors to obtain a battery assembly thermal distribution time series feature vector; and

[0032] Determine whether to generate a fault warning prompt based on the battery assembly thermal distribution time series feature vector.

[0033] Compared with the prior art, the traditional manual inspection method requires a large amount of time and human resources. The remote monitoring system and method for a distributed photovoltaic power station provided by the embodiments of the present invention, while intelligent inspection uses an infrared thermal imager and deep learning algorithms to achieve automated and rapid detection of a large number of photovoltaic cell modules, greatly improving the efficiency of fault detection. By real-time monitoring and analysis of the surface temperature thermal distribution, potential fault signs can be detected in a timely manner, and early warning prompts can be given in advance, which helps to avoid the further expansion and damage of faults, and improve the reliability and lifespan of the photovoltaic cell modules. Intelligent inspection can accurately locate the fault position, help maintenance personnel quickly locate and repair the fault, reducing unnecessary repair and maintenance costs. By detecting and repairing faults in a timely manner, the loss of photovoltaic cell modules can be reduced, and the power generation efficiency and production capacity of the photovoltaic power station can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0035] Figure 1 FIG. is a block diagram of an intelligent inspection system for a distributed photovoltaic power station according to an embodiment of the present invention;

[0036] Figure 2 FIG. is a flowchart of an intelligent inspection method for a distributed photovoltaic power station according to an embodiment of the present invention;

[0037] Figure 3 FIG. is a schematic diagram of the architecture of an intelligent inspection method for a distributed photovoltaic power station according to an embodiment of the present invention;

[0038] Figure 4 FIG. is an application scenario diagram of an intelligent inspection system for a distributed photovoltaic power station according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] To enable those skilled in the art to better understand the technical solutions of the present invention, the following will further describe the present invention in detail in conjunction with the drawings and specific embodiments. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0040] Unless otherwise specifically stated, the technical terms or scientific terms used in the embodiments of the present invention shall have the ordinary meanings understood by those of ordinary skill in the art to which the present invention pertains. The use of "including" or "comprising" in the embodiments of the present invention neither limits the mentioned shapes, numbers, steps, actions, operations, components, elements, and / or their groups, nor excludes the occurrence or addition of one or more other different shapes, numbers, steps, actions, operations, components, elements, and / or their groups, or the addition of these. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity and order of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present invention, "a plurality" means two or more, unless otherwise specifically and clearly defined.

[0041] Unless otherwise specifically stated, the relative settings, numerical expressions, and numerical values of the components and steps described in these embodiments do not limit the scope of the present invention. At the same time, it should be understood that, for the sake of convenience of description, the dimensions of the various parts shown in the drawings are not drawn in actual proportional relationships. For technologies, methods, and devices known to those of ordinary skill in the relevant fields, they may not be discussed in detail, but in appropriate cases, the shown technologies, methods, and devices should be regarded as part of the authorization specification. In all the examples shown and discussed here, any specific other examples may have different values. It should be noted that: similar symbols and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0042] In the description of the embodiments of the present invention, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in the embodiments of the present invention and the features of different embodiments or examples.

[0043] Next, exemplary embodiments of the present invention will be described in detail with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited by the exemplary embodiments described here.

[0044] A distributed PV power station (distributed PV power generation system) is a power generation system that distributes the PV power generation system on the user side or the distribution network side. Compared with centralized PV power stations, distributed PV power stations are more decentralized and flexible, and can better meet the needs of different regions and users. A distributed PV power station consists of multiple PV power generation systems. Each PV power generation system usually consists of PV cell modules, brackets, inverters, and power distribution equipment, etc. The PV cell module is the core component, which converts solar energy into DC electrical energy; the bracket is used to support and fix the PV cell module to ensure that it can face solar radiation normally; the inverter converts DC electrical energy into AC electrical energy for users to use electricity or transmit electrical energy to the distribution network; the power distribution equipment is used to distribute electrical energy and protect the circuit.

[0045] Distributed PV power stations can be distributed in different regions and on the user side, reducing transmission losses and grid pressure. At the same time, the decentralized layout also reduces the demand for large areas of land and lowers land costs. Distributed PV power stations can be scaled up according to demand. Users can gradually increase the number of PV cell modules according to their own electricity demand and available space to achieve flexible capacity expansion. Distributed PV power stations use solar energy as a renewable energy source, do not produce greenhouse gases and pollutants, and are environmentally friendly. Distributed PV power stations are distributed on the user side or the distribution network side, which can reduce the dependence on the traditional power grid and improve the robustness and anti-interference ability of the power system. Distributed PV power stations can provide independent power supply for users, reduce the dependence on the traditional power system, and improve energy autonomy and power supply reliability.

[0046] PV cell modules are the core components of distributed PV power stations. Their working status and performance directly affect the power generation and safety of the entire power station. PV cell modules are mainly composed of PV cells, encapsulation materials, and support structures, which jointly complete the process of converting solar energy into electrical energy.

[0047] PV cells are the key components of PV cell modules and the core for converting solar energy into electrical energy. Common PV cells include monocrystalline silicon, polycrystalline silicon, and thin-film cells, etc. Monocrystalline silicon cells have higher conversion efficiency and better stability. Polycrystalline silicon cells have lower costs. Thin-film cells have better adaptability and flexibility. The working principle of PV cells is that when light shines on the cells, photons excite electrons in the cells to generate current.

[0048] PV cells need to be encapsulated and protected to improve their weather resistance and durability. Common encapsulation materials include glass, polymers, and backsheets, etc. Glass is commonly used for the front encapsulation of PV cell modules and has good light transmittance and protection performance. Polymer materials are commonly used for the back encapsulation and have good flexibility and weather resistance. The backsheet is used to protect the back of the cells and usually uses aluminum plates or plastic materials.

[0049] Photovoltaic cell modules need to have a certain mechanical strength and stability to withstand the influence of the external environment. The support structure is usually made of aluminum alloy or stainless steel and is used to fix and support photovoltaic cells and encapsulation materials. The support structure can also adjust the angle and orientation of the photovoltaic cell module to receive solar radiation to the greatest extent.

[0050] The conversion efficiency of a photovoltaic cell module refers to the ability to convert solar energy into electrical energy. High-efficiency photovoltaic cell modules can increase power generation and reduce the installation cost per unit area. Therefore, choosing photovoltaic cell modules with high conversion efficiency is crucial for improving the economy and power generation efficiency of distributed photovoltaic power stations. Photovoltaic cell modules need to work under various harsh environmental conditions, such as high temperature, humidity, ultraviolet radiation, wind, and rain. Therefore, the encapsulation materials and support structures of photovoltaic cell modules need to have good durability and reliability to ensure the long-term stable operation of the modules. The temperature of the photovoltaic cell module has an important impact on its conversion efficiency and lifespan. Excessive temperature will cause a decrease in the efficiency of the solar cells, and too low temperature may cause the solar cells to freeze or fail. Therefore, reasonably designing the heat dissipation system and controlling the temperature are crucial for optimizing the performance of photovoltaic cell modules. Pollution and shadow effects on the surface of photovoltaic cell modules will reduce the power generation efficiency of the modules. Therefore, regularly cleaning the surface of photovoltaic cell modules and avoiding shadow coverage can maximize the power generation efficiency of the modules.

[0051] Due to the dispersion and large scale of distributed photovoltaic power stations, traditional manual inspection methods are difficult to meet the real-time, accurate, and efficient monitoring requirements. Therefore, to improve the efficiency and reliability of monitoring, modern monitoring technologies and systems are widely used in distributed photovoltaic power stations. The remote monitoring system can, by using sensors and communication technologies, monitor and collect data of distributed photovoltaic power stations in real time. These data include key parameters such as power generation, voltage, current, and temperature. The remote monitoring system can transmit the data to the central monitoring center, and the operators can monitor the operation status of the power station through remote access and control and discover and solve problems in a timely manner.

[0052] Internet of Things technology can connect various devices and sensors in distributed photovoltaic power stations to the Internet to achieve data exchange and remote control between devices. Through Internet of Things technology, real-time monitoring and management of photovoltaic cell modules, inverters, power distribution equipment, etc. can be realized. The operators can analyze and process the data through the cloud platform for fault diagnosis and predictive maintenance to improve the reliability and operation efficiency of the power station.

[0053] By analyzing and processing a large amount of monitoring data, artificial intelligence and machine learning algorithms can be used to extract useful information and patterns, which can help predict the power generation of the power station, detect abnormal situations, optimize operation strategies, etc. Data analysis and artificial intelligence technologies can help operators better understand the performance and problems of the power station and take corresponding measures for optimization and improvement. The application of these modern monitoring technologies and systems can greatly improve the monitoring efficiency and reliability of distributed photovoltaic power stations.

[0054] In one embodiment of the present invention, Figure 1 is a block diagram of an intelligent inspection system for a distributed photovoltaic power station according to an embodiment of the present invention. As Figure 1 shown, the intelligent inspection system 100 for a distributed photovoltaic power station according to an embodiment of the present invention includes: a thermal distribution image acquisition module 110, configured to acquire surface temperature thermal distribution images of a photovoltaic cell module to be analyzed in a distributed photovoltaic power station at a plurality of predetermined time points within a predetermined time period; a thermal distribution feature extraction module 120, configured to perform thermal distribution feature extraction on the surface temperature thermal distribution images at the plurality of predetermined time points to obtain a sequence of battery module thermal distribution feature vectors; a time-domain correlation feature extraction module 130, configured to extract time-domain correlation features between the sequences of the battery module thermal distribution feature vectors to obtain a battery module thermal distribution time-series feature vector; and a fault warning prompt generation module 140, configured to determine whether to generate a fault warning prompt based on the battery module thermal distribution time-series feature vector.

[0055] In the thermal distribution image acquisition module 110, it is ensured to accurately acquire the surface temperature thermal distribution images of the photovoltaic cell module to be analyzed at a plurality of predetermined time points within a predetermined time period, which may involve selecting a suitable thermal imaging device, setting appropriate shooting time points and frequencies, and ensuring the clarity and accuracy of the images. In this way, the surface temperature thermal distribution images of the photovoltaic cell module are provided, providing the basic data for subsequent thermal distribution feature extraction.

[0056] In the thermal distribution feature extraction module 120, appropriate algorithms and methods are designed to perform feature extraction on the surface temperature thermal distribution images at a plurality of predetermined time points. These features may include the highest temperature, the lowest temperature, the average temperature, the temperature gradient, etc., for describing the thermal distribution of the battery module. In this way, the thermal distribution images are transformed into a sequence of battery module thermal distribution feature vectors, providing a quantitative description of the thermal state of the battery module.

[0057] In the time-domain correlation feature extraction module 130, the time-domain correlation features between the sequences of the thermal distribution feature vectors of the battery modules are analyzed, which may include calculating indexes such as the temperature change rate, temperature volatility, periodicity, etc., to reveal the time-series features of the thermal distribution of the battery modules. In this way, the time-series feature vectors of the thermal distribution of the battery modules are provided, which can more comprehensively understand the thermal behavior and change trend of the battery modules.

[0058] In the fault warning prompt generation module 140, based on the time-series feature vectors of the thermal distribution of the battery modules, it is determined whether to generate a fault warning prompt, which may involve setting appropriate fault warning thresholds and rules, as well as the corresponding relationship with the actual fault conditions. By analyzing the time-series feature vectors of the thermal distribution of the battery modules, potential fault conditions can be detected in a timely manner and warning prompts can be generated, which helps to improve the reliability of the battery modules and the accuracy of fault diagnosis.

[0059] The above modules provide a comprehensive monitoring and evaluation of the thermal behavior of the battery modules by obtaining thermal distribution images, extracting features, analyzing correlations, and generating warning prompts, which helps to detect and solve potential fault problems in a timely manner and improve the operation efficiency and reliability of the photovoltaic power station.

[0060] Aiming at the above technical problems, the technical concept of the present invention is to use an infrared thermal imager to collect the surface temperature thermal distribution images of the photovoltaic battery modules at different time points, and combine deep learning algorithms to analyze and process these surface temperature thermal distribution images, so as to identify and give warning prompts for the faults of the photovoltaic battery modules, and realize intelligent inspection.

[0061] The infrared thermal imager can measure the infrared radiation on the surface of an object and convert it into a thermal distribution image. During the operation of the photovoltaic battery module, if there are faults or abnormal conditions, such as hot spots, thermal spots, uneven temperature, etc., obvious features will be shown in the surface temperature distribution. Through the high-precision measurement of the infrared thermal imager, the thermal distribution image of the photovoltaic battery module can be obtained, providing a non-contact monitoring method for the working state of the battery module.

[0062] Combining the obtained thermal distribution images with deep learning algorithms can analyze and process the images to realize the fault identification and warning prompts of the photovoltaic battery modules. The deep learning algorithms can automatically extract the features in the images through learning and training on a large amount of image data, and perform fault classification and identification. For example, a deep learning model can be trained to identify abnormal thermal distribution patterns such as hot spots and thermal spots, as well as temperature change patterns related to the faults of the battery modules.

[0063] Traditional manual inspection methods require a large amount of time and human resources. In contrast, intelligent inspection using infrared thermal imagers and deep learning algorithms can achieve automated and rapid detection of large-scale photovoltaic cell modules, greatly improving the efficiency of fault detection. By real-time monitoring and analysis of the surface temperature thermal distribution, potential fault signs can be detected in a timely manner, and early warning prompts can be given, which helps to avoid the further expansion and damage of faults, and improve the reliability and lifespan of photovoltaic cell modules. Intelligent inspection can accurately locate the fault location, helping maintenance personnel quickly locate and repair faults, reducing unnecessary inspection and maintenance costs. By detecting and repairing faults in a timely manner, the losses of photovoltaic cell modules can be reduced, and the power generation efficiency and production capacity of photovoltaic power plants can be improved.

[0064] Using infrared thermal imagers and deep learning algorithms for intelligent inspection, analyzing and processing the surface temperature thermal distribution images of photovoltaic cell modules, can achieve fault identification and early warning prompts for photovoltaic cell modules, improving the operation efficiency and reliability of photovoltaic power plants. This intelligent inspection technology has important application value in the operation and maintenance management of photovoltaic power plants.

[0065] Based on this, in the technical solution of the present invention, first, surface temperature thermal distribution images of photovoltaic cell modules to be analyzed in a distributed photovoltaic power plant at multiple predetermined time points within a predetermined time period are obtained. In an embodiment of the present invention, an infrared thermal imager can be used as the data acquisition device to non-contactedly obtain the surface temperature information of photovoltaic cell modules, avoiding interference and damage to the equipment.

[0066] Then, the surface temperature thermal distribution images at the multiple predetermined time points are respectively passed through a thermal distribution feature extractor of a photovoltaic cell module based on a convolutional neural network model to obtain a sequence of thermal distribution feature vectors of the photovoltaic cell module. That is, the thermal distribution correlation features with neighborhood spatial relationships are mined from each of the surface temperature thermal distribution images to characterize the thermal distribution features of the photovoltaic cell module.

[0067] In a specific embodiment of the present invention, the thermal distribution feature extraction module includes: a feature extraction unit for using a deep learning network model to perform thermal distribution feature extraction on the surface temperature thermal distribution images at the multiple predetermined time points to obtain the sequence of thermal distribution feature vectors of the photovoltaic cell module.

[0068] Wherein, the deep learning network model is a thermal distribution feature extractor of a photovoltaic cell module based on a convolutional neural network model.

[0069] In a specific embodiment of the present invention, the feature extraction unit is used to: respectively pass the surface temperature thermal distribution images at the multiple predetermined time points through the thermal distribution feature extractor of the photovoltaic cell module based on the convolutional neural network model to obtain the sequence of thermal distribution feature vectors of the photovoltaic cell module.

[0070] Among them, the battery module thermal distribution feature extractor based on the convolutional neural network model includes: an input layer, a convolutional layer, a pooling layer, an activation layer, and an output layer.

[0071] It should be understood that using the feature extractor based on the convolutional neural network model can automatically learn and extract features related to the thermal distribution of the battery module from the thermal distribution image. Compared with manually designing the feature extraction algorithm, this automated method can capture important information in the image more comprehensively and accurately, avoiding subjectivity and human bias. The convolutional neural network model can learn complex feature representations, transforming the input thermal distribution image into a high-dimensional feature vector. The high-dimensional feature vector can better describe the thermal distribution of the battery module, including temperature distribution, hot spot location, temperature gradient, etc. By using the high-dimensional feature vector, more abundant and detailed information can be provided, which helps to more accurately evaluate the thermal state of the battery module.

[0072] By extracting the sequence of feature vectors at multiple time points, the temporal features of the battery module thermal distribution can be obtained, which can better understand the temperature change trend, periodicity, and abnormal conditions of the battery module. Temporal feature analysis can help discover potential fault modes and abnormal behaviors, providing more accurate fault warnings and maintenance suggestions. The feature extractor based on the convolutional neural network can efficiently process a large amount of thermal distribution image data. They can process multiple images in parallel, improving the processing speed and efficiency, which is very important for real-time monitoring and quick decision-making, especially in large-scale distributed photovoltaic power stations.

[0073] Through the battery module thermal distribution feature extractor based on the convolutional neural network model, automated feature extraction, high-dimensional feature representation, temporal feature analysis, and efficient data processing can be achieved. These beneficial effects help to improve the understanding and evaluation of the battery module thermal distribution, providing a more reliable and accurate basis for fault warning and maintenance decision-making.

[0074] Next, the sequence of the battery module thermal distribution feature vectors is passed through the thermal distribution full-time domain feature extractor based on the transformer to obtain the battery module thermal distribution temporal feature vector. Here, the battery module thermal distribution feature vector characterizes the thermal distribution feature information of each surface temperature thermal distribution image, and the thermal characteristics of the photovoltaic battery module should present dynamic changes in the time dimension. That is to say, there is a lack of communication and interaction of feature information between the sequences of the battery module thermal distribution feature vectors. Therefore, in the technical solution of the present invention, the thermal distribution full-time domain feature extractor based on the transformer is used to capture the full-time domain correlation information between the sequences of the battery module thermal distribution feature vectors, so as to characterize the thermal dynamic change process and thermal behavior pattern information of the photovoltaic battery module.

[0075] In a specific embodiment of the present invention, the time-domain correlation feature extraction module includes: a conversion encoding unit configured to obtain the time-series feature vector of the thermal distribution of the battery assembly by passing the sequence of the thermal distribution feature vectors of the battery assembly through a converter-based full-time-domain feature extractor for thermal distribution.

[0076] Wherein, the conversion encoding unit is configured to: utilize the converter-based full-time-domain feature extractor for thermal distribution to capture the full-time-domain correlation information among the sequences of the thermal distribution feature vectors of the battery assembly to obtain the time-series feature vector of the thermal distribution of the battery assembly.

[0077] Passing the sequence of the thermal distribution feature vectors of the battery assembly through the converter-based full-time-domain feature extractor for thermal distribution can obtain the time-series feature vector of the thermal distribution of the battery assembly. This method can further extract and represent the dynamic change information of the thermal distribution of the battery assembly, thereby obtaining a more comprehensive and detailed feature representation, which helps to more accurately perform fault identification and early warning.

[0078] The converter-based full-time-domain feature extractor for thermal distribution usually uses time series analysis methods, such as Fourier transform, wavelet transform, etc., to transform the sequence of the thermal distribution feature vectors of the battery assembly onto the full-time-domain frequency spectrum domain. By analyzing the time-domain frequency spectrum, the frequency characteristics and dynamic change patterns of the thermal distribution of the battery assembly can be captured.

[0079] The thermal distribution of the battery assembly changes over time. By extracting the time-series feature vector, the dynamic change information of the thermal distribution of the battery assembly can be captured, which helps to more accurately analyze and understand the working state of the battery assembly, including the evolution process and trend of faults. The time-series feature vector can comprehensively consider the information of multiple time points of the thermal distribution of the battery assembly, thereby providing a more comprehensive and accurate feature representation, which helps to distinguish the differences between the normal working state and the fault state and improve the accuracy of fault identification. By analyzing the trends and patterns of the time-series feature vector, the possible fault conditions of the battery assembly can be predicted and early warnings can be given in advance, which helps to take timely maintenance and repair measures to avoid the impact of faults on the photovoltaic power station. The time-series feature vector can provide richer information, which helps to provide more comprehensive and accurate decision-making support for the operation and maintenance management of the photovoltaic power station. For example, in terms of fault diagnosis and maintenance plan formulation, the time-series feature vector can provide a more detailed and reliable data basis.

[0080] By passing the sequence of the thermal distribution feature vectors of the battery assembly through the converter-based full-time-domain feature extractor for thermal distribution, the time-series feature vector of the thermal distribution of the battery assembly can be obtained. This method can capture the dynamic change information of the thermal distribution of the battery assembly, improve the accuracy of fault identification, support fault prediction and early warning, and enhance the decision-making support ability of operation and maintenance management.

[0081] Subsequently, the time-series feature vector of the thermal distribution of the battery assembly is passed through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated. In a specific example of the present invention, the process of encoding the time-series feature vector of the thermal distribution of the battery assembly through a multi-task classification head module to obtain a classification result for indicating whether a fault warning prompt is generated includes: first passing the time-series feature vector of the thermal distribution of the battery assembly through a fine-grained classifier to obtain multiple fault category probability values; at the same time, passing the time-series feature vector of the thermal distribution of the battery assembly through a coarse-grained classifier to obtain a first probability value and a second probability value; then, fusing the multiple fault category probability values, the first probability value, and the second probability value in a probability-weighted fusion manner to obtain a comprehensive expression probability value; and then obtaining the classification result based on the comprehensive expression probability value.

[0082] Here, fine-grained and coarse-grained fault classifications are performed simultaneously, and a comprehensive expression probability value is obtained through probability-weighted fusion, enabling the model to learn the differences in the feature information between the normally operating photovoltaic battery assembly and the photovoltaic battery assembly with anomalies or faults, clarifying the range boundaries of the presence and absence of anomalies or faults, optimizing the traditional judgment process using a single classifier, and alleviating the overfitting problem of the single classifier to a certain extent.

[0083] In a specific embodiment of the present invention, the fault warning prompt generation module includes: a feature distribution optimization unit for optimizing the time-series feature vector of the thermal distribution of the battery assembly to obtain an optimized time-series feature vector of the thermal distribution of the battery assembly; and a classification unit for passing the optimized time-series feature vector of the thermal distribution of the battery assembly through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated.

[0084] Specifically, the classification unit is configured to: pass the optimized time-series feature vector of the thermal distribution of the battery assembly through a fine-grained classifier to obtain multiple fault category probability values; pass the optimized time-series feature vector of the thermal distribution of the battery assembly through a coarse-grained classifier to obtain a first probability value and a second probability value; fuse the multiple fault category probability values, the first probability value, and the second probability value in a probability-weighted fusion manner to obtain a comprehensive expression probability value; and obtain the classification result based on the comprehensive expression probability value.

[0085] Through feature distribution optimization, the importance of useful features in the time-series feature vector can be enhanced, and the interference of irrelevant features can be reduced. This can improve the accuracy and robustness of the fault prediction model, enabling it to better capture features related to faults. Feature distribution optimization can reduce the dimensionality of the feature vector, remove redundant information, thereby reducing the complexity of the data and the computational cost, which is very important for processing large-scale data and improving the efficiency of the model in practical applications. The optimized feature distribution can better reflect the thermal state and fault features of the battery components, which can enhance the interpretation and understanding of the fault warning results and help operators better judge whether corresponding maintenance measures need to be taken.

[0086] The classification unit can determine whether to generate a fault warning prompt based on the optimized feature vector. Through the classification model, different feature patterns can be associated with fault situations, so as to timely detect potential faults and generate corresponding warning prompts for taking appropriate maintenance measures. The classification unit can simultaneously handle classification tasks for multiple fault types, which can improve the comprehensiveness and adaptability of the system, be able to identify multiple different types of faults, and generate corresponding warning prompts, which is very important for the situation where there are multiple fault modes in a distributed photovoltaic power station. The classification results generated by the classification unit can provide real-time fault warning prompts for operators, which can help operators take measures in a timely manner, reduce the impact of faults on the operation of the photovoltaic power station, and improve the reliability and safety of the system.

[0087] The feature distribution optimization unit and the classification unit play important roles in the thermal distribution monitoring and fault warning of battery components. The feature distribution optimization unit can enhance features, reduce dimensions, and improve interpretability, while the classification unit can generate fault warnings, handle multi-task classification, and support real-time decision-making, which helps to improve the accuracy and efficiency of fault prediction and maintenance decision-making.

[0088] In the technical solution of the present invention, each battery module thermal distribution feature vector in the sequence of battery module thermal distribution feature vectors represents the image semantic features of the surface temperature thermal distribution image of the photovoltaic battery module to be analyzed at each predetermined time point, and the feature distribution of each battery module thermal distribution feature vector follows the channel dimension distribution of the battery module thermal distribution feature extractor based on the convolutional neural network model. In this way, after passing the sequence of battery module thermal distribution feature vectors through the thermal distribution full-time domain feature extractor based on the transformer, the temporal correlation of the image semantic features of the surface temperature thermal distribution images of each photovoltaic battery module to be analyzed can be further extracted. However, the battery module thermal distribution temporal feature vector will also have an offset in the image semantic feature temporal distribution relative to the image semantic feature channel dimension distribution of each battery module thermal distribution feature vector. Therefore, it is desired to optimize the battery module thermal distribution temporal feature vector by fusing the battery module thermal distribution temporal feature vector and the sequence of battery module thermal distribution feature vectors.

[0089] Furthermore, when fusing the battery module thermal distribution temporal feature vector and the sequence of battery module thermal distribution feature vectors, since they have feature distributions with different dimensions, the distribution intensity of one of them will be greater than that of the other. Therefore, in order to avoid unbalanced fusion expression when the two are fused, the inventor of the present invention, for the battery module thermal distribution temporal feature vector, denoted as V 1 and the concatenated feature vector obtained by concatenating the sequence of battery module thermal distribution feature vectors, denoted as V 2 perform self-supervised balancing of the objective loss for feature interpolation fusion to obtain the optimized battery module thermal distribution temporal feature vector, denoted as V 1 '. It is expressed as: perform self-supervised balancing of the objective loss for feature interpolation fusion on the concatenated feature vector obtained by concatenating the battery module thermal distribution temporal feature vector and the sequence of battery module thermal distribution feature vectors with the following optimization formula to obtain the optimized battery module thermal distribution temporal feature vector; where the optimization formula is:

[0090]

[0091] where V 1 is the battery module thermal distribution temporal feature vector, V 2 is the concatenated feature vector obtained by concatenating the sequence of battery module thermal distribution feature vectors, and respectively represent the reciprocals of the global means of the battery module thermal distribution temporal feature vector V 1 and the concatenated feature vector V 2 , and I is the unit vector, V 1V' is the optimized thermal distribution time series feature vector of the battery module, ⊕ represents addition by position, represents subtraction by position, and ⊙ represents multiplication by position.

[0092] That is, considering feature fusion based on different feature distribution dimensions, if one of the thermal distribution time series feature vectors V 1 of the battery module to be fused and the cascaded feature vector V 2 is regarded as the input for enhancing the strong feature dimension of the other, the target distribution information of the target feature manifold of the other in the class space in the thermal distribution time series feature vector V 1 of the battery module and the cascaded feature vector V 2 may be lost, resulting in the loss of the purpose of class regression. Therefore, by cross-punishing the outlier distribution of the feature distributions relative to each other, it is possible to achieve a self-supervised balance of feature enhancement and regression robustness during feature interpolation fusion, so as to enhance the thermal distribution time series feature vector V 1 of the battery module and the cascaded feature vector V 2 of the battery module, thereby enhancing the expression effect of the optimized thermal distribution time series feature vector V 1 ', so as to improve the accuracy of the classification result obtained through the multi-task classification head module.

[0093] In summary, the intelligent inspection system 100 of the distributed photovoltaic power station based on the embodiments of the present invention is elucidated. It uses an infrared thermal imager to collect the surface temperature thermal distribution images of photovoltaic battery modules at different time points, and combines deep learning algorithms to analyze and process these surface temperature thermal distribution images, so as to perform fault identification and early warning prompts for photovoltaic battery modules and achieve intelligent inspection.

[0094] As described above, the intelligent inspection system 100 of the distributed photovoltaic power station according to the embodiments of the present invention can be implemented in various terminal devices, such as servers for intelligent inspection of distributed photovoltaic power stations. In one example, the intelligent inspection system 100 of the distributed photovoltaic power station according to the embodiments of the present invention can be integrated into the terminal device as a software module and / or a hardware module. For example, the intelligent inspection system 100 of the distributed photovoltaic power station can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the intelligent inspection system 100 of the distributed photovoltaic power station can also be one of the many hardware modules of the terminal device.

[0095] Alternatively, in another example, the intelligent inspection system 100 of the distributed photovoltaic power station and the terminal device may also be separate devices, and the intelligent inspection system 100 of the distributed photovoltaic power station can be connected to the terminal device through a wired and / or wireless network and transmit interaction information in accordance with a predefined data format.

[0096] In one embodiment of the present invention, Figure 2 is a flowchart of an intelligent inspection method for a distributed photovoltaic power station according to an embodiment of the present invention. Figure 3 is a schematic diagram of the architecture of an intelligent inspection method for a distributed photovoltaic power station according to an embodiment of the present invention. As Figure 2 and Figure 3 shown, the intelligent inspection method for the distributed photovoltaic power station includes: 210, obtaining surface temperature thermal distribution images of photovoltaic cell modules to be analyzed in a distributed photovoltaic power station at multiple predetermined time points within a predetermined time period; 220, performing thermal distribution feature extraction on the surface temperature thermal distribution images at the multiple predetermined time points to obtain a sequence of battery module thermal distribution feature vectors; 230, extracting temporal correlation features between the sequences of the battery module thermal distribution feature vectors to obtain a battery module thermal distribution time series feature vector; and 240, determining whether to generate a fault warning prompt based on the battery module thermal distribution time series feature vector.

[0097] Those skilled in the art can understand that the specific operations of each step in the above intelligent inspection method for the distributed photovoltaic power station have been described in detail in the description of the intelligent inspection system of the distributed photovoltaic power station above with reference to Figure 1 and thus, the repeated description thereof will be omitted.

[0098] Figure 4 is an application scenario diagram of the intelligent inspection system of the distributed photovoltaic power station according to an embodiment of the present invention. As Figure 4 shown, in this application scenario, first, surface temperature thermal distribution images of photovoltaic cell modules to be analyzed in a distributed photovoltaic power station at multiple predetermined time points within a predetermined time period are obtained (for example, as Figure 4 shown in C); then, the obtained surface temperature thermal distribution images are input into a server (for example, as Figure 4 shown in S) deployed with an intelligent inspection algorithm for the distributed photovoltaic power station, where the server can process the surface temperature thermal distribution images based on the intelligent inspection algorithm for the distributed photovoltaic power station to determine whether to generate a fault warning prompt.

[0099] The basic principles of the present invention have been described in connection with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present invention are merely examples and not limitations. It cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present invention. Additionally, the above-disclosed specific details are only for illustrative and easy-to-understand purposes, rather than limitations. The above details do not limit the present invention to necessarily implementing with the above specific details.

[0100] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present invention. Therefore, the present invention is not intended to be limited to the aspects shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

[0101] It can be understood that the above embodiments are merely exemplary embodiments adopted to illustrate the principles of the present invention. However, the present invention is not limited thereto. For those of ordinary skill in the art, various variations and improvements can be made without departing from the spirit and essence of the present invention, and these variations and improvements are also considered within the protection scope of the present invention.

Claims

1. An intelligent inspection system for distributed photovoltaic power stations, characterized in that: include: A thermal distribution image acquisition module is used to acquire thermal distribution images of surface temperatures of photovoltaic cell components to be analyzed in a distributed photovoltaic power station at multiple predetermined time points within a predetermined time period; A thermal distribution feature extraction module, used to extract thermal distribution features from the surface temperature thermal distribution images at the plurality of predetermined time points to obtain a sequence of thermal distribution feature vectors of the battery assembly; A time domain correlation feature extraction module, used to extract the time domain correlation features between the sequences of the battery component thermal distribution feature vectors to obtain the battery component thermal distribution time series feature vectors; as well as A fault warning prompt generation module is used to determine whether to generate a fault warning prompt based on the battery component thermal distribution time series feature vector.

2. The intelligent inspection system for distributed photovoltaic power stations according to claim 1 is characterized in that: The thermal distribution feature extraction module comprises: The feature extraction unit is used to extract thermal distribution features from the surface temperature thermal distribution images at the plurality of predetermined time points using a deep learning network model to obtain a sequence of thermal distribution feature vectors of the battery assembly.

3. The intelligent inspection system for distributed photovoltaic power stations according to claim 2 is characterized in that: The deep learning network model is a battery component thermal distribution feature extractor based on a convolutional neural network model.

4. The intelligent inspection system for distributed photovoltaic power stations according to claim 3 is characterized in that: The feature extraction unit is used to: The surface temperature thermal distribution images at the plurality of predetermined time points are respectively passed through the battery assembly thermal distribution feature extractor based on the convolutional neural network model to obtain a sequence of the battery assembly thermal distribution feature vectors.

5. The intelligent inspection system for distributed photovoltaic power stations according to claim 4 is characterized in that: The battery component thermal distribution feature extractor based on the convolutional neural network model includes: an input layer, a convolution layer, a pooling layer, an activation layer and an output layer.

6. The intelligent inspection system for distributed photovoltaic power stations according to claim 5 is characterized in that: The time domain correlation feature extraction module comprises: The conversion coding unit is used to pass the sequence of the thermal distribution feature vectors of the battery assembly through a converter-based thermal distribution full-time domain feature extractor to obtain the thermal distribution time series feature vectors of the battery assembly.

7. The intelligent inspection system for distributed photovoltaic power stations according to claim 6, characterized in that: The conversion coding unit is used for: The converter-based thermal distribution full-time domain feature extractor is used to capture the full-time domain correlation information between the sequences of the thermal distribution feature vectors of the battery assembly to obtain the thermal distribution time series feature vectors of the battery assembly.

8. The intelligent inspection system for distributed photovoltaic power stations according to claim 7, characterized in that: The fault warning prompt generating module includes: A feature distribution optimization unit, used for performing feature distribution optimization on the battery assembly thermal distribution timing feature vector to obtain an optimized battery assembly thermal distribution timing feature vector; and A classification unit is used to pass the optimized battery assembly thermal distribution time series feature vector through a multi-task classification head module to obtain a classification result, and the classification result is used to indicate whether a fault warning prompt is generated.

9. The intelligent inspection system for distributed photovoltaic power stations according to claim 8, characterized in that: The classification unit is used to: Passing the optimized battery assembly thermal distribution time series feature vector through a fine-grained classifier to obtain multiple fault category probability values; Passing the optimized battery assembly thermal distribution time series feature vector through a coarse-grained classifier to obtain a first probability value and a second probability value; fusing the plurality of fault category probability values, the first probability value, and the second probability value in a probability weighted fusion manner to obtain a comprehensive expression probability value; as well as The classification result is obtained based on the comprehensive expression probability value.

Images