A photovoltaic module detection method and apparatus

Abstract

This invention relates to the field of photovoltaic module testing technology, and provides a photovoltaic module testing method and apparatus. One photovoltaic module testing method includes the following steps: S1. Periodically performing hotspot detection on the surface of the photovoltaic panel; S2. Before hotspot detection, constructing multiple models based on real-time environmental parameters and thermal imager status parameters, and outputting the environmental impact index, thermal imager status index, and the detection accuracy confidence level of the photovoltaic panel hotspots from each model; S3. When the detection accuracy confidence level is higher than a preset threshold, performing photovoltaic panel hotspot detection. The photovoltaic module testing method provided by this invention performs a comprehensive pre-assessment of conditions before detection (step S2), constructs models based on real-time environmental parameters and thermal imager status parameters, and outputs the detection accuracy confidence level. This allows the system to dynamically quantify the impact of current detection conditions on the reliability of the results.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic module testing technology, and particularly relates to a photovoltaic module testing method and apparatus. Background Technology

[0002] Currently, hotspot detection in photovoltaic modules primarily relies on infrared thermal imaging technology. Thermal imagers scan the surface of photovoltaic panels to acquire temperature distribution images, thereby identifying areas of abnormal heating. However, in practical applications, the accuracy and reliability of thermal imager detection are constrained by multiple complex factors. Dynamic changes in solar irradiance can cause uneven heat distribution on the photovoltaic panel surface; fluctuations in ambient temperature and wind speed affect the heat conduction process; and differences in air humidity and dust concentration interfere with the accurate transmission of infrared radiation. All of these factors can lead to decreased thermal image quality, temperature signal distortion, or increased false alarm rates. These uncertainties in the environment and equipment conditions significantly reduce the reliability of detection results under certain conditions.

[0003] Furthermore, existing detection systems generally employ fixed time intervals or manually triggered detection modes, lacking the ability to dynamically optimize detection timing. When environmental conditions are harsh, such as strong winds, high humidity, or equipment malfunctions, the system still forces the detection task to proceed, generating a large amount of low-reliability data and causing unnecessary energy consumption and equipment wear. More importantly, for situations where detection conditions are not met, existing technologies can only simply skip the detection or record failure information, failing to establish an effective closed-loop feedback mechanism.

[0004] Therefore, there is an urgent need for a photovoltaic module testing solution that can integrate environmental parameters and equipment status in real time, intelligently assess the suitability of testing conditions, and proactively implement external adjustments to improve testing confidence when conditions are insufficient. Existing technologies urgently need improvement to address these issues. Summary of the Invention

[0005] The purpose of this invention is to provide a photovoltaic module testing method and apparatus to solve the above-mentioned problems.

[0006] This invention is implemented as follows: a photovoltaic module testing method, comprising the following steps:

[0007] S1. Regularly detect hot spots on the surface of photovoltaic panels;

[0008] S2. Before detecting hot spots on photovoltaic panels, multiple models are constructed based on real-time environmental parameters and thermal imager status parameters, and the environmental impact index, thermal imager status index, and the detection accuracy confidence level of photovoltaic panel hot spots are output from each model respectively.

[0009] S3. When the detection accuracy confidence level is higher than the preset threshold, the photovoltaic panel hotspot detection is performed at this time;

[0010] S4. When the detection accuracy confidence is lower than the preset threshold and the environmental impact index exceeds the preset threshold, the target unfolding angle is output and the shielding plate is controlled to change angle based on the detection accuracy confidence, environmental impact index and the current shielding plate unfolding angle (angle between the shielding plate and the photovoltaic panel). The detection accuracy confidence is then re-evaluated. If the detection accuracy confidence is still lower than the preset threshold, step S5 is executed.

[0011] S5. Based on the detection accuracy confidence level, the real-time surface cleanliness of the photovoltaic panel module, and the real-time air dust concentration, output the photovoltaic panel cleaning demand index. When the photovoltaic panel cleaning demand index exceeds the preset threshold, the cleaning device is started to clean, and the detection accuracy confidence level is re-evaluated. If the detection accuracy confidence level is still lower than the preset threshold, record "Insufficient conditions, skip this detection".

[0012] A further technical solution, in step S2, involves constructing an environmental data model based on solar irradiance, ambient temperature, ambient wind speed, and air humidity, and outputting an environmental impact index. Specifically:

[0013] The environmental impact index is obtained by weighted summation of normalized irradiance index, ambient temperature index, ambient wind speed index and air humidity index to obtain an intermediate value of environmental impact. The intermediate value of environmental impact is then multiplied by an adjustment scale coefficient and substituted into a negative exponential function with the natural constant e as the base for calculation. Finally, the result of the exponential function is obtained by subtracting 1 from the result of the calculation.

[0014] The weighted summation includes: the product of the irradiance index and its weight coefficient, the product of the ambient temperature index and its weight coefficient, the product of the square of the ambient wind speed index and its weight coefficient, the product of the air humidity index and its weight coefficient, and the product of the irradiance index and the ambient temperature index and its coupling weight coefficient.

[0015] The sum of all weight coefficients is 1, and all of them are positive numbers;

[0016] The irradiance index, ambient temperature index, ambient wind speed index, and air humidity index are obtained by normalizing the corresponding original parameters by substituting them into the maximum-minimum value formula.

[0017] A further technical solution involves calculating the irradiance index, ambient temperature index, ambient wind speed index, and air humidity index by substituting solar irradiance, ambient temperature, ambient wind speed, and air humidity sequentially into a maximum-minimum formula for normalization, thereby generating the irradiance index, ambient temperature index, ambient wind speed index, and air humidity index in sequence.

[0018] In a further technical solution, in step S2, a thermal imager state model is constructed based on the ground sampling distance, detection angle, and equivalent temperature difference of the thermal imager noise, and a thermal imager state index is output. Specifically, the thermal imager state index is equal to the result of the following expression:

[0019] (The product of the sampling distance index and the detection angle index, plus the product of the noise equivalent temperature difference index and the average of the sampling distance index and the detection angle index, plus the sum of a preset coupling term weight coefficient and the product of the sampling distance index, the detection angle index and the noise equivalent temperature difference index), divide the above sum by (1 plus the coupling term weight coefficient).

[0020] The sampling distance index, detection angle index, and noise equivalent temperature difference index are obtained by normalizing the corresponding original parameters by substituting them into the maximum-minimum value formula.

[0021] Further technical solutions, among which , as well as The calculation method is as follows: the ground sampling distance of the thermal imager, the detection angle of the thermal imager, and the noise equivalent temperature difference of the thermal imager are substituted into the maximum-minimum value formula for normalization, and the sampling distance index, detection angle index, and noise equivalent temperature difference index are generated in sequence.

[0022] A further technical solution, in step S2, involves constructing a detection accuracy confidence assessment model based on the hotspot temperature signal stability index, environmental impact index, thermal imager status index, and initial hotspot type factor, and outputting the detection accuracy confidence level; specifically:

[0023] The confidence level of detection accuracy is calculated by taking the square root of the weighted sum of the squares of the temperature signal stability index, environmental influence index, thermal imager status index, and hotspot initial judgment type factor, and then taking the negative exponential function. The sum of each adjustment coefficient is 1 and all are positive numbers.

[0024] The temperature signal stability index is calculated by dividing the standard deviation of the temperature of the hot spot area in the past several frames by its average value; the hot spot initial judgment type factor is a weight value assigned after the hot spot type is initially judged by a lightweight AI model, and the value range is 0 to 1.

[0025] In a further technical solution, the calculation method for the target deployment angle in step S4 is as follows:

[0026] First, based on the proportion of the difference between the detection accuracy confidence level and the confidence threshold to the confidence threshold, and then multiplied by the proportion of the environmental impact index to the sum of the environmental impact index and a small normal number, a confidence deficiency is calculated.

[0027] Then, the current shielding plate deployment angle index is added to the result of multiplying the confidence deficiency by an angle adjustment gain, the sum is limited to a value not exceeding 1, and finally multiplied by the maximum angle value to obtain the target deployment angle.

[0028] The current shielding plate unfolding angle index is calculated by dividing the current actual unfolding angle (range 0-90 degrees) by the maximum angle value of 90.

[0029] In a further technical solution, step S5, the calculation method for the photovoltaic panel cleaning demand index is as follows:

[0030] The surface cleanliness is calculated by multiplying the complement of the confidence level of the detection accuracy (1 minus the parameter), the complement of the surface cleanliness, and the air dust concentration index; the air dust concentration index is obtained by normalizing the actual air dust concentration by substituting it into the maximum-minimum value formula; the surface cleanliness is a dimensionless quantity with a value between 0 and 1.

[0031] A photovoltaic module testing device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the above-described photovoltaic module testing method.

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] This invention provides a photovoltaic module testing method that performs a comprehensive pre-assessment of conditions before testing (step S2). A model is constructed using real-time environmental parameters and thermal imager status parameters, and the detection accuracy confidence level is output. This allows the system to dynamically quantify the impact of current testing conditions on the reliability of the results. Compared to existing technologies that lack quantitative assessment of testing conditions, this method intelligently determines the timing of testing, performing actual testing only when the confidence level is higher than a preset threshold (step S3). This significantly improves the accuracy and reliability of the testing results and avoids invalid testing under unfavorable conditions.

[0034] This invention provides a photovoltaic module testing method that introduces an active adjustment mechanism when testing conditions are not met. When the environmental impact index is high, the system can intelligently output the target deployment angle and control the shielding plate to change angle based on the testing accuracy confidence level, the environmental impact index, and the shielding plate deployment angle (step S4). This ability to actively adjust external environmental factors is generally lacking in existing technologies. For example, in the above example, adjusting the shielding plate angle to optimize light or wind conditions aims to improve the testing environment and increase confidence, which is more beneficial than simply skipping the testing.

[0035] This invention provides a photovoltaic module testing method that further considers the impact of photovoltaic panel surface cleanliness on testing. When environmental adjustments are insufficient to improve confidence level, the system can output a photovoltaic panel cleaning demand index based on the detection accuracy confidence level, real-time surface cleanliness of the photovoltaic panel, and real-time air dust concentration, and activate the cleaning device in a timely manner (step S5). This mechanism effectively solves the problem of decreased detection accuracy caused by photovoltaic panel surface contamination, enabling the system to proactively improve the module's condition and create more favorable conditions for subsequent testing. Compared to the passive approach of simply recording "insufficient conditions" or waiting for manual intervention in existing technologies, this method provides a more complete closed-loop control, greatly improving the system's adaptability and operational efficiency.

[0036] This invention provides a photovoltaic module testing method that overcomes the shortcomings of existing technologies in photovoltaic module hotspot detection, such as low accuracy and lack of intelligent timing judgment and proactive adjustment, by constructing an intelligent pre-assessment, active adjustment, and closed-loop feedback mechanism. Therefore, this method can significantly improve the accuracy of photovoltaic module hotspot detection, system reliability, and operation and maintenance efficiency, providing strong technical support for the long-term stable operation of photovoltaic power plants. Attached Figure Description

[0037] Figure 1 This is a schematic diagram illustrating the steps of a photovoltaic module testing method. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0039] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0040] like Figure 1 As shown, a photovoltaic module testing method according to an embodiment of the present invention includes the following steps:

[0041] S1. Regularly detect hot spots on the surface of photovoltaic panels;

[0042] S2. Before detecting hot spots on photovoltaic panels, multiple models are constructed based on real-time environmental parameters and thermal imager status parameters, and the environmental impact index, thermal imager status index, and the detection accuracy confidence level of photovoltaic panel hot spots are output from each model respectively.

[0043] S3. When the detection accuracy confidence level is higher than the preset threshold, the photovoltaic panel hotspot detection is performed at this time;

[0044] S4. When the detection accuracy confidence is lower than the preset threshold and the environmental impact index exceeds the preset threshold, the target unfolding angle is output and the shielding plate is controlled to change angle based on the detection accuracy confidence, environmental impact index and the current shielding plate unfolding angle (angle between the shielding plate and the photovoltaic panel). The detection accuracy confidence is then re-evaluated. If the detection accuracy confidence is still lower than the preset threshold, step S5 is executed.

[0045] S5. Based on the detection accuracy confidence level, the real-time surface cleanliness of the photovoltaic panel module, and the real-time air dust concentration, output the photovoltaic panel cleaning demand index. When the photovoltaic panel cleaning demand index exceeds the preset threshold, the cleaning device is started to clean, and the detection accuracy confidence level is re-evaluated. If the detection accuracy confidence level is still lower than the preset threshold, record "Insufficient conditions, skip this detection".

[0046] In this embodiment, a photovoltaic module testing method is proposed. Its main feature is that it improves the accuracy and reliability of hotspot detection through an intelligent evaluation and adjustment mechanism.

[0047] First, the method includes periodically detecting hot spots on the surface of the photovoltaic panels. This step aims to establish a continuous monitoring cycle, ensuring that the operational status of the photovoltaic panels can be checked periodically. For example, it can be set to be checked daily, weekly, or monthly. In one implementation, a simple timer can be configured to automatically activate the thermal imager to collect data at fixed time intervals.

[0048] Secondly, before detecting hot spots on photovoltaic panels, multiple models are constructed based on real-time environmental parameters and thermal imager status parameters. The environmental impact index, thermal imager status index, and the detection accuracy confidence level of photovoltaic panel hot spots are output from each model.

[0049] Furthermore, when the detection accuracy confidence level is higher than a preset threshold, hotspot detection on the photovoltaic panel is performed. This step ensures that the actual hotspot detection operation is only performed when the detection conditions are assessed as reliable, thus avoiding invalid detection under unfavorable conditions. For example, a confidence threshold can be set; when the calculated detection accuracy confidence level exceeds this threshold, the system will activate the thermal imager to perform infrared scanning on the photovoltaic panel surface and analyze the acquired thermal images to identify potential hotspot areas.

[0050] However, when the detection accuracy confidence level is lower than a preset threshold, and the environmental impact index exceeds a preset threshold, the system outputs a target deployment angle based on the detection accuracy confidence level, the environmental impact index, and the current shielding plate deployment angle. The system then controls the shielding plate to change its angle and reassesses the detection accuracy confidence level. If the detection accuracy confidence level is still lower than the preset threshold, subsequent steps are executed. This step introduces an active adjustment mechanism designed to improve detection accuracy by changing external environmental conditions. For example, if the system determines that the current environment (such as strong direct sunlight or strong winds) significantly interferes with hotspot detection and the detection confidence level is insufficient, a new shielding plate deployment angle can be calculated using a simple proportional controller or lookup table method, based on the current confidence level, the degree of environmental impact, and the current position of the shielding plate. Subsequently, an actuator (such as a motor) drives the shielding plate to adjust to this target angle to optimize lighting conditions or reduce wind impact. After adjustment, the system executes the above evaluation steps again to recalculate the detection accuracy confidence level.

[0051] Finally, if the confidence level of the detection accuracy is still lower than the preset threshold after reassessment, a photovoltaic panel cleaning demand index is output based on the confidence level of the detection accuracy, the real-time surface cleanliness of the photovoltaic panel, and the real-time air dust concentration. When the photovoltaic panel cleaning demand index exceeds the preset threshold, the cleaning device is activated to clean the panel, and the confidence level of the detection accuracy is reassessed. If the confidence level of the detection accuracy is still lower than the preset threshold, "Insufficient conditions, skip this detection" is recorded. This step provides a solution to the problem of photovoltaic panel surface contamination and sets the final decision-making mechanism. For example, when environmental adjustments fail to effectively improve the confidence level, the system will further consider the cleanliness of the photovoltaic panel surface and the dust content in the air.

[0052] Based on this information, the system can calculate the photovoltaic panel cleaning demand index. If this index exceeds a preset cleaning trigger threshold, a command will be sent to the cleaning device to initiate the cleaning operation; for example, an automated cleaning robot will begin cleaning the surface of the photovoltaic panel. After cleaning is completed, the system will reassess the detection accuracy confidence level. If, after all active adjustments and cleaning, the detection accuracy confidence level still fails to meet the requirements, the system will record that this detection was skipped due to insufficient conditions and wait for the next detection cycle.

[0053] In a preferred embodiment of the present invention, in step S2, an environmental data model is constructed based on solar irradiance, ambient temperature, ambient wind speed, and air humidity, and an environmental impact index is output. Specifically:

[0054] ;

[0055] ;

[0056] in To adjust the scaling factor, This is the irradiance influence coefficient. This is the temperature influence coefficient. The wind speed influence coefficient, Humidity influence coefficient This is the coupling effect coefficient. ,and , , , , as well as All are greater than 0; The irradiance index, The ambient temperature index. This refers to the environmental wind speed index. The air humidity index. This is the median value for environmental impact. This is the environmental impact index.

[0057] In this embodiment, solar irradiance refers to the solar radiation power received per unit area, which is one of the key environmental factors affecting the operating status of photovoltaic panels and the detection effect of thermal imagers. It can be obtained in real time through solar irradiance sensors installed within the photovoltaic power station, such as thermopile-type or photoelectric irradiance meters; or by obtaining local real-time irradiance information through a meteorological station data interface.

[0058] Ambient temperature refers to the temperature of the air surrounding the photovoltaic panel, which directly affects the heat dissipation characteristics of the photovoltaic panel and the measurement accuracy of the thermal imager. Ambient temperature can be obtained in real time using temperature sensors such as thermistors and thermocouples; or it can be obtained by connecting to a local meteorological data service platform.

[0059] Ambient wind speed refers to the airflow speed around a photovoltaic panel. The wind speed affects the convective heat dissipation on the surface of the photovoltaic panel, and thus affects the displayed temperature of the hot spots. Ambient wind speed can be obtained in real time using an anemometer (such as a cup anemometer or ultrasonic anemometer) or through a meteorological data interface.

[0060] Air humidity refers to the water vapor content in the air. High humidity may affect the transmission of infrared signals, thus interfering with the detection results of thermal imagers. Air humidity can be obtained in real time using humidity sensors (such as capacitive humidity sensors or resistive humidity sensors) or through meteorological data interfaces.

[0061] An environmental data model is a mathematical or logical framework used to process and quantify the aforementioned real-time environmental parameters in order to assess their comprehensive impact on the accuracy of photovoltaic panel hotspot detection.

[0062] The Environmental Impact Index (EI) is a quantitative indicator used to represent the potential negative impact of current environmental conditions on the accuracy of photovoltaic (PV) panel hotspot detection. This index is typically designed with a value between 0 and 1, where 0 indicates no or minimal impact on detection accuracy, and 1 indicates a significant impact. The output can be a directly calculated value used as input to a subsequent detection accuracy confidence assessment model; or it can be presented as a visual chart to maintenance personnel for decision-making reference. (Formula follows) Used to calculate median environmental impact values It is a weighted sum that comprehensively considers multiple environmental factors and their interactions. Among them, The irradiance index, The ambient temperature index. This refers to the environmental wind speed index. These are air humidity indices, obtained by normalizing the original environmental parameters to ensure the comparability of different physical quantities in the model.

[0063] This is the irradiance influence coefficient, used to measure the weight of the influence of irradiance on detection accuracy; This is the temperature influence coefficient, used to measure the weight of the impact of ambient temperature on detection accuracy; The wind speed influence coefficient is used to measure the weight of the impact of ambient wind speed on detection accuracy. The wind speed term adopts... The form can better reflect the nonlinear effect of wind speed on heat dissipation and hotspot manifestation; This is the humidity influence coefficient, used to measure the weight of the impact of air humidity on detection accuracy; These are coupling effect coefficients, used to measure the weighting of the coupling effect between irradiance and ambient temperature on detection accuracy. This coupling term can capture the complex situation where the hotspot effect may be amplified or suppressed when both irradiance and ambient temperature are high. These weighting coefficients... , , , The sum of all factors is 1 and all are greater than 0, which ensures that the contribution of each factor is reasonably allocated. Furthermore, it can be adjusted and optimized through historical data analysis, expert experience, or optimization algorithms to adapt to the characteristics of different regions and seasons.

[0064] formula Used to adjust the median value of environmental impact Converted into the final environmental impact index This is an exponential function that can... Mapping it to a range of 0 to 1 makes it more consistent with the physical phenomenon of exponential decay or growth. The scaling factor, a constant greater than 0, is used to adjust the sensitivity and curve shape of the exponential function. This is achieved by adjusting... This can make the model exhibit different response speeds and intensities to environmental changes; for example, larger... It will make right The environmental impact index is more sensitive to changes, thus it will rise even when environmental conditions deteriorate slightly. It can rise rapidly.

[0065] This method addresses the problem of insufficient quantification of the impact of environmental parameters on the accuracy of photovoltaic module hotspot detection by constructing a refined environmental data model to output an environmental impact index. The model first collects key environmental parameters in real time, such as solar irradiance, ambient temperature, ambient wind speed, and air humidity. These raw parameters are then converted into comparable index forms, for example, through normalization, thereby eliminating the influence of different physical dimensions. After obtaining these environmental indices, this method uses a weighted summation mathematical expression to calculate the median environmental impact value. This expression not only considers the independent effects of various environmental factors but also specifically introduces a coupling term between irradiance and ambient temperature. This design can capture the complex interactions between factors in the real environment. For example, the hotspot effect may be more pronounced when strong irradiance and high temperature coexist, or the heat dissipation effect may weaken the hotspot manifestation under strong wind conditions. By assigning specific weight coefficients to each influencing factor and its coupling term, and ensuring that the sum of these coefficients is 1 and all are greater than 0, this method achieves flexible configuration and quantification of the degree of influence of different environmental factors. Subsequently, the intermediate values ​​of environmental impact are calculated. It is further fed into an exponential function to generate the final environmental impact index. The design of this exponential function makes the environmental impact index... It can reflect the impact of environmental conditions on detection accuracy in a non-linear manner, and its value is limited to between 0 and 1, facilitating subsequent unified evaluation and application. Adjustment scaling coefficient. The introduction of this feature allows the system to adjust the model's sensitivity to environmental changes based on actual needs. Through the construction and operation of the aforementioned environmental data model, this method can accurately and quantitatively assess the impact of current environmental conditions on the accuracy of photovoltaic panel hotspot detection. This environmental impact index... As one of the key parameters output from step S2, it will be directly input into the subsequent detection accuracy confidence assessment model, thereby making the assessment of detection accuracy confidence more comprehensive and accurate. When the environmental impact index... When the value is too high, it indicates that the environmental conditions are not conducive to detection. The system can decide whether to perform the detection based on this information, or trigger the shielding plate angle adjustment mechanism in step S4 to actively improve the detection conditions, thereby avoiding invalid detection under low confidence and improving the intelligence and reliability of the entire photovoltaic module detection method.

[0066] As a preferred embodiment of the present invention, wherein , , as well as The calculation method is as follows: the solar irradiance, ambient temperature, ambient wind speed and air humidity are substituted into the maximum-minimum value formula for normalization, and the irradiance index, ambient temperature index, ambient wind speed index and air humidity index are generated in sequence.

[0067] In this embodiment, , , as well as These represent the irradiance index, ambient temperature index, ambient wind speed index, and air humidity index, respectively. They are standardized values ​​used to quantify the impact of different environmental factors on the hotspot detection of photovoltaic modules. Their calculation method aims to convert the original physical measurements into unified, dimensionless values ​​to facilitate comprehensive evaluation in subsequent environmental impact index models. Normalization is achieved by substituting solar irradiance, ambient temperature, ambient wind speed, and air humidity sequentially into the maximum-minimum formula. This is a data preprocessing technique that aims to eliminate differences in the dimensions and numerical ranges of the original data, mapping data at different scales to a unified interval, such as [0,1] or [-1,1].

[0068] The maximum-minimum normalization formula is usually expressed as: X normalized = (XX) min ) / (X max -X min ), where X is the original data, X min and X max These are the minimum and maximum values ​​of the data in a specific dataset. This processing method ensures that all environmental parameters have the same weight or influence when participating in subsequent calculations, avoiding the dominance of certain parameters in the model due to differences in numerical values, thereby improving the fairness and accuracy of the model. Generating the irradiance index, ambient temperature index, ambient wind speed index, and air humidity index in sequence is a direct result of the normalization process; they are standardized representations of the original environmental parameters. These indices serve as inputs to environmental impact index models (such as the environmental data model mentioned above), ensuring the standardization and consistency of the model inputs.

[0069] In a preferred embodiment of the present invention, in step S2, a thermal imager state model is constructed based on the ground sampling distance of the thermal imager, the detection angle of the thermal imager, and the equivalent temperature difference of the thermal imager noise, and the thermal imager state index is output. Specifically:

[0070] ;

[0071] in These are the weighting coefficients of the coupling terms. The sampling distance index, To detect the angle index, The noise equivalent temperature difference index. This is the thermal imager status index.

[0072] In this embodiment, the ground sampling distance of the thermal imager refers to the actual physical size that a single pixel of the thermal imager can cover on the surface of the object being measured at a specific height and viewing angle. This parameter directly affects the spatial resolution of the thermal image and its ability to capture subtle hotspot features. It can be calculated using parameters such as the thermal imager's focal length, detector size, working distance, and pixel pitch, or obtained through actual measurement and calibration.

[0073] The detection angle of a thermal imager refers to the angle between the optical axis of the thermal imager and the normal to the surface of the photovoltaic panel being measured. This angle affects the accuracy of the thermal imager's measurement of the photovoltaic panel's surface temperature. An excessively large detection angle may lead to increased emissivity errors, geometric distortion, and a reduction in the effective pixel coverage area. The detection angle can be obtained in real time through the installation position, the pan-tilt control system, or the attitude sensor.

[0074] The noise equivalent temperature difference (HETD) of a thermal imager is an indicator of its ability to detect minute temperature differences. It represents the smallest detectable temperature difference corresponding to the output noise of the thermal imager when there is no signal input. The smaller the HETD value, the higher the sensitivity of the thermal imager and the stronger its ability to identify weak hotspot signals. This parameter is usually provided by the thermal imager manufacturer, but can also be evaluated through laboratory testing or field calibration.

[0075] A thermal imager state model is a mathematical or algorithmic model used to comprehensively evaluate the performance state of a thermal imager under current operating conditions. This model quantifies the applicability of the thermal imager to the detection task by integrating multiple key parameters. The thermal imager state index is a quantitative metric output by the thermal imager state model, used to characterize the degree to which the current performance of the thermal imager affects the accuracy of hotspot detection. This index is typically dimensionless, and its magnitude reflects the quality of the thermal imager's state; for example, a higher index value may indicate that the thermal imager's state is less conducive to high-precision detection.

[0076] Coupling term weight coefficients This is a weighting factor used to adjust the degree of interaction between parameters in the thermal imager's state model. This coefficient reflects the relative importance of the combined effects of three parameters—ground sampling distance, detection angle, and noise-equivalent temperature difference—on the overall performance of the thermal imager. This coefficient can be obtained through expert experience setting, historical data fitting, or optimization algorithm training.

[0077] Sampling distance index It is the result of normalizing the ground sampling distance of the thermal imager, used to convert the original physical quantity into a uniform and comparable value.

[0078] Detection Angle Index This is the result after normalizing the detection angle of the thermal imager, used to convert the original physical quantity into a uniform, comparable value. Noise Equivalent Temperature Difference Index It is the result of normalizing the equivalent temperature difference of thermal imager noise, used to convert the original physical quantity into a uniform and comparable value.

[0079] The solution in this application addresses the problem of incomplete parameter integration when evaluating thermal imager performance by constructing a thermal imager state model in step S2 based on the ground sampling distance, detection angle, and noise-equivalent temperature difference of the thermal imager, and outputting a thermal imager state index. Specifically, this solution quantifies three key parameters directly related to thermal image acquisition quality and temperature measurement reliability—the ground sampling distance, detection angle, and noise-equivalent temperature difference—through a comprehensive model. The sampling distance index... and detection angle index product term This demonstrates the amplifying effect between distance and angle; for example, long distances or angles exacerbate image distortion, making hotspot identification difficult; the noise equivalent temperature difference index... Average distance angle The combination reflects the nonlinear effect of noise under specific distance and angle conditions; that is, under unfavorable distances and angles, even if the equivalent temperature difference of thermal imager noise itself is not high, its impact on detection accuracy will be amplified; the coupling term weighting coefficient The introduced three-parameter coupling term This captures the complex interactions between parameters, such as the combined deterioration effect of distance and angle under high noise, ensuring that the model can fully consider the combined effects of these factors. (Denominator) The normalization process ensures a stable output of the index as parameters change, thus improving the thermal imager status index. It can comprehensively and accurately characterize the overall condition of the thermal imager. This thermal imager condition index... As an important input for constructing the detection accuracy confidence assessment model in step S2, it works together with the environmental impact index, the temperature signal stability index of hotspots, and the hotspot initial judgment type factor to enable the final output detection accuracy confidence to more accurately reflect the reliability of the current detection conditions. This avoids the problem of unreliable detection results caused by incomplete assessment of the thermal imager's condition, and significantly improves the accuracy and reliability of photovoltaic module hotspot detection.

[0080] in , as well as The calculation method is as follows: the ground sampling distance, detection angle, and noise equivalent temperature difference of the thermal imager are successively substituted into the maximum-minimum formula for normalization, and the sampling distance index, detection angle index, and noise equivalent temperature difference index are generated sequentially. Generating the sampling distance index, detection angle index, and noise equivalent temperature difference index means using the normalized values ​​as dimensionless indicators for subsequent calculations of the thermal imager state model. These indices are typically between 0 and 1, fairly reflecting the impact of each parameter on the thermal imager state.

[0081] In a preferred embodiment of the present invention, in step S2, a detection accuracy confidence assessment model is constructed based on the hotspot temperature signal stability index, environmental impact index, thermal imager status index, and hotspot initial judgment type factor, and the detection accuracy confidence is output. Specifically:

[0082] ;

[0083] in The signal conditioning coefficient, This is the environmental regulation coefficient. This is the thermal imager status adjustment coefficient. This is the hotspot adjustment coefficient. ,and , , as well as All are greater than 0. The temperature signal stability index. For environmental impact index, This refers to the thermal imager's condition index. Preliminary identification of hotspot type factors, , Dimensionless ( The type weights initially determined by the lightweight AI model are as follows: shading = 0.8, cell defect = 0.6, junction box association = 0.4. The confidence level for detection accuracy; where The calculation method is as follows: divide the standard deviation of the temperature of the hot spot area in the past N frames by its average value.

[0084] In this embodiment, the temperature signal stability index of the hotspot refers to the degree of fluctuation in the temperature data of the photovoltaic panel's hotspot area over time. A stable temperature signal generally indicates more reliable detection results, while a drastically fluctuating signal may indicate environmental interference or measurement error. Temperature signal stability index Its function is to quantify the reliability of hotspot temperature data, providing an intrinsic data quality indicator for confidence assessment. One approach is to continuously collect temperature data from hotspot areas, calculate their statistical dispersion (e.g., standard deviation), and then normalize the data by combining this with the mean.

[0085] Environmental Impact Index This is a quantitative indicator of the impact of external environmental factors (such as solar irradiance, ambient temperature, wind speed, and humidity) on the accuracy of hotspot detection. Its function is to reflect the degree of interference from external environmental conditions on the detection results of thermal imagers. One implementation method is to collect environmental parameters in real time through sensors and input them into a pre-trained environmental model. This model can assess the impact of the environment on detection based on historical data and expert experience. Another implementation method is to classify environmental impact into different levels based on threshold settings for different environmental parameters and assign a corresponding index value to each level.

[0086] Thermal Imager Status Index This is a quantitative indicator of the impact of the thermal imager's own operating conditions (such as ground sampling distance, detection angle, and noise-equivalent temperature difference) on the accuracy of hotspot detection. Its purpose is to reflect the influence of the thermal imager's own performance limitations and operating conditions on the detection results. One implementation method is to read the thermal imager's internal parameters or sensor data, such as focal length, distance to the target, and sensor noise level, and input them into the thermal imager's condition assessment model to calculate their impact on detection accuracy. Another implementation method is to periodically calibrate and test the thermal imager's performance, and dynamically adjust the thermal imager's condition index based on the test results.

[0087] Preliminary hotspot identification type factor This is a weighting factor assigned by a lightweight AI model to initially identify hotspot types, used to differentiate the difficulty or importance of different types of hotspots in detection and identification. Its purpose is to introduce consideration of the essential attributes of hotspots, making confidence assessment more targeted. One implementation method is to use pre-trained lightweight AI models such as convolutional neural networks to perform image recognition and classification of hotspot areas in thermal images, such as identifying them as shading, battery cell defects, junction box associations, etc., and assigning different weight values ​​according to the detection difficulty or potential hazard level of each type. Another implementation method is to preset an initial confidence influence factor for different hotspot types based on an expert knowledge base or historical fault data, and adjust it after the AI ​​model's judgment.

[0088] The detection accuracy confidence assessment model is a comprehensive mathematical model used to map multiple input parameters, such as the hotspot temperature signal stability index, environmental impact index, thermal imager status index, and hotspot initial judgment type factor, into a single detection accuracy confidence value through specific mathematical relationships. Its purpose is to provide a quantitative, multi-dimensional fusion index to comprehensively evaluate the reliability of current hotspot detection results. One implementation method is to use nonlinear functions, such as exponential or sigmoid functions, to weight and combine multiple influencing factors to ensure that the output confidence value is within a reasonable range and has good discriminative power.

[0089] Confidence level of detection accuracy This is a value between 0 and 1 output by the detection accuracy confidence assessment model, used to quantify the reliability of the current hotspot detection results. A higher confidence level indicates a more reliable detection result, and vice versa. Its role is to provide a crucial basis for subsequent detection decisions (such as whether to perform hotspot detection, whether to adjust detection conditions, etc.). One implementation method is to normalize the raw values ​​output by the model so that they fall within the range of 0 to 1, where 1 represents the highest confidence level and 0 represents the lowest confidence level.

[0090] Signal modulation coefficient Environmental regulation coefficient Thermal imager status adjustment coefficient Hotspot adjustment coefficient These coefficients are used to adjust the weights or influence of their respective factors in the confidence assessment model for detection accuracy. They collectively determine the relative contribution of each factor to the final confidence score. Their function is to allow the system to flexibly adjust the importance of different influencing factors based on the actual application scenario, equipment characteristics, or experiential knowledge. One implementation method is to initially set these coefficients through expert experience or historical data analysis; for example, in scenarios with significant environmental interference, the environmental adjustment coefficient can be appropriately increased. The weights. Another approach is to use optimization algorithms, such as genetic algorithms or particle swarm optimization, to automatically learn and optimize these coefficients through iterative training and validation, so that the confidence level of the model output best matches the actual detection accuracy.

[0091] The calculation method involves dividing the standard deviation of the temperature in the hotspot area over the past N frames by its average value. Its purpose is to provide a quantitative indicator reflecting the fluctuation of the hotspot temperature signal, thereby assessing the stability of the temperature data. One implementation involves the system continuously acquiring temperature data of the hotspot area captured by the thermal imager and maintaining a sliding window containing the most recent N frames of data. When new frame data arrives, the standard deviation and average value of all temperature values ​​within this window are calculated, and then a division operation is performed. Another implementation method is to use a weighted average or exponentially weighted moving average to calculate the average and standard deviation, giving more weight to recent data and thus more sensitively reflecting real-time changes in the temperature signal.

[0092] This application's solution achieves intelligent quantification of the reliability of photovoltaic module hotspot detection by constructing a multi-dimensional fusion detection accuracy confidence evaluation model. Before detecting hotspots on the photovoltaic panel, the system first acquires the hotspot's temperature signal stability index, environmental impact index, thermal imager status index, and hotspot initial type factor. Among these, the hotspot temperature signal stability index... This was calculated by examining the standard deviation and average value of temperature data from hotspot areas over the past N frames, quantifying the volatility of the hotspot temperature data and ensuring an intrinsic assessment of data quality. Environmental Impact Index Thermal imager status index The potential interferences to detection accuracy were quantified from two dimensions: the external environment and the equipment's own condition. Preliminary hotspot identification type factors. A lightweight AI model is used to initially determine the type of trending topic and assign weights, introducing consideration of the essential attributes of trending topics and making the assessment more targeted. Subsequently, these key parameters ( , , , The input is fed into the detection accuracy confidence assessment model. This model employs a non-linear function, namely... These parameters are then weighted and combined. Among them, the signal conditioning coefficient... Environmental regulation coefficient Thermal imager status adjustment coefficient and hotspot adjustment coefficient As weights, these factors allow the system to adjust their contribution to the final confidence level based on actual conditions. For example, when environmental interference is significant, the weights can be appropriately increased. The weighting of environmental factors makes their impact on confidence levels more significant. In this way, the model comprehensively considers data quality, external interference, device performance, and hotspot characteristics, outputting a detection accuracy confidence level between 0 and 1. This confidence level This serves as a key basis for determining whether to perform photovoltaic panel hotspot detection in step S3. When the temperature exceeds a preset threshold, it indicates that the current detection conditions are good and the detection results are highly reliable. Hotspot detection is only performed at this point, thus avoiding invalid or low-quality detection under unfavorable conditions. This multi-parameter fusion evaluation mechanism ensures that detection decisions are no longer simply triggered periodically, but are based on a real-time, comprehensive assessment of the detection environment and equipment status, greatly improving the accuracy and reliability of the detection. By incorporating key factors such as the temperature signal stability index of hotspots, environmental impact, thermal imager status, and hotspot type into a unified evaluation framework, this solution can more accurately reflect actual detection conditions, effectively avoiding misjudgments or omissions caused by insufficient consideration of single factors in traditional methods, thereby providing more reliable decision support for the operation and maintenance of photovoltaic modules.

[0093] In a preferred embodiment of the present invention, the target unfolding angle is calculated in step S4 as follows:

[0094] ;

[0095] ;

[0096] in To determine the confidence level of the detection accuracy, The confidence threshold. For environmental impact index, This is a measure of insufficient confidence. It is a positive constant. To adjust the gain for the angle, The shielding plate unfolding angle index. Expand the angle to the target; The calculation method is to divide the current shielding plate unfolding angle (range 0-90 degrees) by 90.

[0097] In this embodiment, the calculation method is intended to be based on the current detection conditions, including the detection accuracy confidence level. Environmental Impact Index Based on the current state of the shielding plate, it intelligently determines the final angle to which the shielding plate should be adjusted. Its function is to provide a quantifiable, executable instruction to guide the physical adjustment of the shielding plate, thereby improving the environmental conditions for photovoltaic panel hotspot detection. This calculation can be performed by the processor in the photovoltaic module detection device or by an external control unit connected to the device.

[0098] Among them, the confidence level is lacking. The calculation formula reflects the confidence level of the current detection accuracy. Compared with the preset confidence threshold The gap between them, combined with the environmental impact index Weighting is applied to quantify the difference between current and ideal detection conditions. This calculation can be performed using a floating-point unit or a dedicated digital signal processor, ensuring both accuracy and efficiency.

[0099] Target Deployment Angle The calculation formula is based on the current shielding plate deployment angle index. and the calculated confidence deficit Gain adjusted by angle Adjustments are made to ensure the final angle is within a reasonable range. This calculation can be performed by an arithmetic logic unit in a microcontroller or embedded system, or by a fast lookup using a pre-programmed lookup table. Detection accuracy and confidence level are then set. It is an indicator for measuring the reliability of current photovoltaic panel hotspot detection results and serves as the core basis for determining whether the detection conditions need to be adjusted. It can be evaluated in real time by an independent evaluation module or main control unit.

[0100] Confidence threshold It is a preset standard value used to determine whether the confidence level of the detection accuracy meets the requirements. This value serves as the basis for the trigger condition adjustment mechanism and can be set according to the actual application scenario. Environmental Impact Index It is an indicator that quantifies the impact of environmental factors on the accuracy of photovoltaic panel hotspot detection. Its function is to reflect the weight of the impact of environmental factors on the detection accuracy when calculating the confidence deficit. It can be calculated from the data collected by the environmental sensor array.

[0101] Positive constant It is a preset, positive-zero value used to stabilize the denominator. To avoid environmental impact index When the value is close to or equal to zero, division by zero errors or computational instability occur, thereby improving the robustness of the calculation.

[0102] Angle adjustment gain It is an adjustable parameter used to control the amount of confidence deficit. The impact of adjusting the shielding plate's unfolding angle can be configured according to the system's requirements for response speed and stability.

[0103] Shielding plate unfolding angle index This is a dimensionless value normalized to 0-1 for the current shielding plate deployment angle. It is calculated by dividing the current shielding plate deployment angle (range 0-90 degrees) by 90, thus converting the physical angle into a standardized parameter easy to calculate using formulas. This can be obtained by reading the output of the shielding plate angle sensor and performing a linear mapping. Target deployment angle. It is the final physical angle value that the shielding plate should be adjusted to after calculation, which serves as the instruction to control the shielding plate actuator.

[0104] This application's solution addresses the problem of accurately and adaptively adjusting the shielding plate angle to optimize detection conditions when the detection accuracy confidence level is insufficient by introducing an intelligent target deployment angle calculation mechanism. This is relevant when the detection accuracy confidence level for photovoltaic panel hotspot detection is low. Below the preset threshold And the environmental impact index When the threshold is exceeded, the system first calculates the confidence deficit. This calculation not only considers the relative difference between the current confidence level and the threshold, but also uses the environmental impact index. We weighted them so that the confidence deficit decreased as environmental conditions had a greater impact on the detection. The calculation results better reflect the actual degree of inadequacy. At the same time, a positive constant is introduced. This ensured the stability of the calculation. Subsequently, the system utilized this confidence deficiency. Current shielding plate deployment angle index and angle adjustment gain To calculate the target deployment angle Among them, the current shielding plate deployment angle index By normalizing the actual angle, different parameters can be calculated on a uniform scale. Angle adjustment gain. This allows the system to adjust the sensitivity of angle changes according to actual needs. Ultimately, through... The calculations ensured the target deployment angle. Within the physically permissible range of 0-90 degrees, and avoiding over-adjustment, the entire process forms a closed-loop feedback control. This means that based on the degree of inadequacy in the detection accuracy confidence level and environmental influences, the optimal adjustment angle of the shielding plate is intelligently calculated, guiding precise adjustments to the shielding plate to improve subsequent detection accuracy confidence. This dynamic and adaptive adjustment mechanism enables the photovoltaic module testing method to proactively optimize testing conditions based on real-time environment and testing status, significantly improving the reliability and efficiency of the testing.

[0105] In a preferred embodiment of the present invention, the photovoltaic panel cleaning demand index is calculated in step S5 as follows:

[0106] .

[0107] in To determine the confidence level of the detection accuracy, For surface cleanliness, , Dimensionless This refers to the air dust concentration index. Demand index for cleaning photovoltaic panels; The calculation method is as follows: substitute the actual air dust concentration into the maximum-minimum formula for normalization, and generate the air dust concentration index.

[0108] In this embodiment, the confidence level of detection accuracy It is a key indicator for evaluating the reliability of current testing conditions; a higher value indicates more reliable testing results. In this scheme, when the confidence level of testing accuracy is... A low level indicates poor current testing conditions, which may need to be improved through cleaning. Surface cleanliness Surface cleanliness is used to quantify the cleanliness of photovoltaic panel surfaces; values ​​are typically between 0 and 1, where 1 represents completely clean and 0 represents completely dirty. Air dust concentration index can be obtained in various ways. For example, sensors installed near the photovoltaic panel (such as optical sensors or image recognition systems) can be used to monitor the surface dirt of the photovoltaic panel in real time and convert it into a normalized value; alternatively, it can be estimated through regular manual inspections or preset empirical models. This reflects the dust content in the ambient air; a higher value indicates more severe air pollution and a greater likelihood of dust accumulation on the photovoltaic panel surface. Air Dust Concentration Index This can be generated by normalizing the actual airborne dust concentration using a maximum-minimum formula. For example, a laser scattering dust sensor or a piezoelectric balance dust sensor can be used to measure the concentration of particulate matter in the air in real time, and then normalized to an exponent between 0 and 1. Maximum-minimum normalization is a commonly used data processing method. Its principle is to linearly transform the original data to a specified range (usually 0 to 1), allowing data of different dimensions to be compared and calculated on the same scale.

[0109] The solution proposed in this application defines a photovoltaic panel cleaning demand index. The specific calculation method addresses the issue of accurately quantifying cleaning needs, ensuring that cleaning decisions are based on multi-factor collaborative evaluation, thus improving the objectivity and efficiency of decision-making. This calculation method integrates the confidence level of detection accuracy. Surface cleanliness and air dust concentration index Three key parameters are dynamically quantified through mathematical formulas to avoid bias from a single factor. Specifically, in In the design, based on the confidence level of detection accuracy generate This reflects the degree of insufficient detection accuracy; a higher value indicates lower confidence and a higher cleaning requirement, highlighting the necessity of prioritizing cleaning when confidence is insufficient. Based on surface cleanliness... generate This quantifies the degree of poor cleanliness; a higher value indicates more severe soiling and a higher cleaning requirement, highlighting the impact of cleanliness on testing conditions. Based on the air dust concentration index... This indicates the level of dust pollution; a higher value indicates a higher concentration of dust in the environment and a greater need for cleaning, emphasizing the real-time effect of external environmental factors. By multiplying these three contributing terms... It can integrate the synergistic effects of insufficient confidence, low cleanliness, and high dust concentration to ensure that the cleaning demand index fully reflects the actual scenario and avoids misjudgments caused by independent parameters. Among these, the air dust concentration index... The calculation method normalizes the actual air dust concentration, which unifies the dimensions and range of dust data, making it more consistent. It becomes a standardized index, making it easier to... and The combined calculations enhance the applicability and consistency of the overall formula. When the confidence level of the detection accuracy... If the reading is still below the preset threshold, the system will further assess the cleaning needs of the photovoltaic panels. At this point, the photovoltaic panel cleaning demand index will be calculated using the formula described above. .if If the threshold is exceeded, a cleaning device will be activated to clean the photovoltaic panel surface, thereby potentially improving the accuracy and confidence level of subsequent testing. If the accuracy and confidence level still do not meet the requirements after cleaning, "Insufficient conditions, skip this test" will be recorded to avoid invalid testing when reliable testing conditions are not available.

[0110] A photovoltaic module testing device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the photovoltaic module testing method described in any of the above embodiments.

[0111] This photovoltaic module testing device is a physical entity used to execute a photovoltaic module hotspot detection method. Its design integrates necessary hardware and software resources to achieve automated and intelligent detection of photovoltaic panel hotspots. The device can be a standalone embedded system, such as a system built on an industrial control computer or a dedicated data acquisition unit. Its core function is to provide a stable operating platform to support complex computational tasks and real-time data processing. The memory is a hardware unit used to store data and instructions, persistently storing the computer programs, model parameters, historical detection data, environmental parameter thresholds, and other configuration information required for the photovoltaic module testing method. The processor is the core computing unit of the device, responsible for parsing and executing the computer program instructions stored in the memory, performing data operations, logical judgments, and control operations. Its role is to drive the entire detection method, including but not limited to the acquisition and processing of real-time environmental parameters, analysis of thermal imager status parameters, construction and calculation of multiple models, evaluation of detection accuracy confidence, adjustment and control of the shielding plate angle, and start-up control of the cleaning device. The processor can be a microcontroller, suitable for embedded applications with high power consumption and cost requirements; or it can be a microprocessor, suitable for scenarios requiring more powerful computing capabilities and more complex operating system support. A computer program is a collection of instructions designed to guide a processor in performing a specific task, namely, implementing a photovoltaic module testing method. This program encodes all steps from S1 to S5, including logic for data acquisition, model calculation, conditional judgment, and control instruction generation. It can be written in a high-level programming language and compiled into processor-executable machine code. The program not only contains the core testing logic but may also include data management modules, communication modules, and fault diagnosis modules to ensure the stable and efficient operation of the entire testing system.

[0112] "Implementing a photovoltaic module testing method" refers to transforming an abstract testing method into a practically executable operational process through the collaborative work of the aforementioned memory and processor. Specifically, the processor reads the computer program from the memory and processes the real-time collected environmental data and thermal imager status data according to the logical sequence and calculation rules defined in the program. This process calculates key indicators such as the environmental impact index, thermal imager status index, and detection accuracy confidence level. When the detection accuracy confidence level falls below a preset threshold, the processor, based on the program logic, calculates the target deployment angle and controls the shielding plate to adjust its angle, or calculates the photovoltaic panel cleaning demand index and activates the cleaning device. This process ensures the automated and intelligent execution of the testing method, thereby optimizing the accuracy and efficiency of hotspot detection.

[0113] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for testing photovoltaic modules, characterized in that, Includes the following steps: S1. Regularly detect hot spots on the surface of photovoltaic panels; S2. Before detecting hot spots on photovoltaic panels, multiple models are constructed based on real-time environmental parameters and thermal imager status parameters, and the environmental impact index, thermal imager status index, and the detection accuracy confidence level of photovoltaic panel hot spots are output from each model respectively. S3. When the detection accuracy confidence level is higher than the preset threshold, the photovoltaic panel hotspot detection is performed at this time; S4. When the detection accuracy confidence is lower than the preset threshold and the environmental impact index exceeds the preset threshold, the target unfolding angle is output and the shielding plate is controlled to change angle based on the detection accuracy confidence, environmental impact index and the current shielding plate unfolding angle (angle between the shielding plate and the photovoltaic panel). The detection accuracy confidence is then re-evaluated. If the detection accuracy confidence is still lower than the preset threshold, step S5 is executed. S5. Based on the detection accuracy confidence level, the real-time surface cleanliness of the photovoltaic panel module, and the real-time air dust concentration, output the photovoltaic panel cleaning demand index. When the photovoltaic panel cleaning demand index exceeds the preset threshold, the cleaning device is started to clean, and the detection accuracy confidence level is re-evaluated. If the detection accuracy confidence level is still lower than the preset threshold, record "Insufficient conditions, skip this detection".

2. The photovoltaic module testing method according to claim 1, characterized in that, In step S2, an environmental data model is constructed based on solar irradiance, ambient temperature, ambient wind speed, and air humidity, and an environmental impact index is output. Specifically: The environmental impact index is obtained by weighted summation of normalized irradiance index, ambient temperature index, ambient wind speed index and air humidity index to obtain an intermediate value of environmental impact. The intermediate value of environmental impact is then multiplied by an adjustment scale coefficient and substituted into a negative exponential function with the natural constant e as the base for calculation. Finally, the result of the exponential function is obtained by subtracting 1 from the result of the calculation. The weighted summation includes: the product of the irradiance index and its weight coefficient, the product of the ambient temperature index and its weight coefficient, the product of the square of the ambient wind speed index and its weight coefficient, the product of the air humidity index and its weight coefficient, and the product of the irradiance index and the ambient temperature index and its coupling weight coefficient. The sum of all weight coefficients is 1, and all of them are positive numbers; The irradiance index, ambient temperature index, ambient wind speed index, and air humidity index are obtained by normalizing the corresponding original parameters by substituting them into the maximum-minimum value formula.

3. The photovoltaic module testing method according to claim 2, characterized in that, The calculation methods for the irradiance index, ambient temperature index, ambient wind speed index, and air humidity index are as follows: the solar irradiance, ambient temperature, ambient wind speed, and air humidity are successively substituted into the maximum-minimum value formula for normalization, and the irradiance index, ambient temperature index, ambient wind speed index, and air humidity index are generated in sequence.

4. The photovoltaic module testing method according to claim 1, characterized in that, In step S2, a thermal imager state model is constructed based on the ground sampling distance, detection angle, and noise equivalent temperature difference of the thermal imager, and a thermal imager state index is output. Specifically, the thermal imager state index is equal to the result of the following expression: (The product of the sampling distance index and the detection angle index, plus the product of the noise equivalent temperature difference index and the average of the sampling distance index and the detection angle index, plus the sum of a preset coupling term weight coefficient and the product of the sampling distance index, the detection angle index and the noise equivalent temperature difference index), divide the above sum by (1 plus the coupling term weight coefficient). The sampling distance index, detection angle index, and noise equivalent temperature difference index are obtained by normalizing the corresponding original parameters by substituting them into the maximum-minimum value formula.

5. The photovoltaic module testing method according to claim 4, characterized in that, in , as well as The calculation method is as follows: the ground sampling distance of the thermal imager, the detection angle of the thermal imager, and the noise equivalent temperature difference of the thermal imager are substituted into the maximum-minimum value formula for normalization, and the sampling distance index, detection angle index, and noise equivalent temperature difference index are generated in sequence.

6. The photovoltaic module testing method according to claim 1, characterized in that, In step S2, a detection accuracy confidence assessment model is constructed based on the hotspot temperature signal stability index, environmental impact index, thermal imager status index, and hotspot initial judgment type factor, and the detection accuracy confidence level is output; specifically: The confidence level of detection accuracy is calculated by taking the square root of the weighted sum of the squares of the temperature signal stability index, environmental influence index, thermal imager status index, and hotspot initial judgment type factor, and then taking the negative exponential function. The sum of each adjustment coefficient is 1 and all are positive numbers. The temperature signal stability index is calculated by dividing the standard deviation of the temperature of the hot spot area in the past several frames by its average value; the hot spot initial judgment type factor is a weight value assigned after the hot spot type is initially judged by a lightweight AI model, and the value range is 0 to 1.

7. The photovoltaic module testing method according to claim 1, characterized in that, In step S4, the target deployment angle is calculated as follows: First, based on the proportion of the difference between the detection accuracy confidence level and the confidence threshold to the confidence threshold, and then multiplied by the proportion of the environmental impact index to the sum of the environmental impact index and a small normal number, a confidence deficiency is calculated. Then, the current shielding plate deployment angle index is added to the result of multiplying the confidence deficiency by an angle adjustment gain, the sum is limited to a value not exceeding 1, and finally multiplied by the maximum angle value to obtain the target deployment angle. The current shielding plate unfolding angle index is calculated by dividing the current actual unfolding angle (range 0-90 degrees) by the maximum angle value of 90.

8. The photovoltaic module testing method according to claim 1, characterized in that, In step S5, the photovoltaic panel cleaning demand index is calculated as follows: The surface cleanliness is calculated by multiplying the complement of the confidence level of the detection accuracy (1 minus the parameter), the complement of the surface cleanliness, and the air dust concentration index; the air dust concentration index is obtained by normalizing the actual air dust concentration by substituting it into the maximum-minimum value formula; the surface cleanliness is a dimensionless quantity with a value between 0 and 1.

9. A photovoltaic module testing device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the photovoltaic module testing method according to any one of claims 1 to 8.

Images