A device and method for real-time monitoring of el under dynamic load of a photovoltaic module

Abstract

This invention belongs to the field of photovoltaic module defect detection technology, and relates to an EL real-time monitoring device and method for photovoltaic modules under dynamic load. It includes: a dynamic load loading system connected to the photovoltaic module; a pulse power supply system connected to the electrodes of the photovoltaic module; an EL imaging system positioned facing the test surface of the photovoltaic module; a control system, with its input connected to the dynamic load loading system and its output connected to the pulse power supply system, the EL imaging system, and the dynamic load loading system respectively; and an image processing and analysis system, with its input connected to both the EL imaging system and the dynamic load loading system. This invention can monitor the initiation and propagation of internal defects in photovoltaic modules in real time and in situ throughout the entire dynamic load process, and quantitatively correlate damage evolution with load parameters, thereby accurately obtaining the critical load and propagation rate for microcrack initiation, providing direct basis for module reliability assessment and fatigue resistance design.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic module defect detection technology, and relates to a real-time EL monitoring device and method for photovoltaic modules under dynamic load. Background Technology

[0002] Electroluminescence (EL) testing, as a highly sensitive non-destructive testing method, has been widely used in the identification of defects such as microcracks, broken grids, and fragments in photovoltaic modules. A conventional EL testing system typically consists of a darkroom, a DC power supply, a near-infrared camera, and image processing software. During testing, the module is under static conditions with no mechanical load or a constant load. A forward bias voltage is applied to cause the cells to emit light, and EL images are acquired, thereby enabling the location and identification of defects. Based on this, some testing schemes employ a pre- and post-load comparison approach, performing EL imaging before and after the mechanical load test, and comparing the differences between the images to determine whether the load caused the formation of new microcracks or the expansion of existing defects. In addition, a few studies have proposed performing EL testing during static load maintenance to explore defect behavior under constant pressure.

[0003] However, the aforementioned existing technologies all have significant shortcomings. The dynamic loads (such as wind vibration, avalanche impact, and transportation vibration) that photovoltaic modules experience in actual service environments often exhibit transient, alternating, and fatigue-accumulating characteristics. The initiation and propagation of microcracks typically occur at the moment of load peak or the critical point of material fatigue, representing a rapidly evolving and irreversible damage process. Conventional pre- and post-load comparison tests can only reveal the defect state at two isolated moments: the start and end of the test. Key information such as when damage occurred, how cracks propagated, and whether multiple induced events occurred is completely missing, making the damage process a "black box." Due to the inability to observe in real time, researchers struggle to establish a direct correspondence between load parameters (amplitude, frequency, number of cycles, etc.) and the damage evolution process. They cannot accurately determine core mechanical parameters such as the critical load for microcrack initiation, crack propagation rate, and fatigue life, severely hindering in-depth module structural design and reliability assessment. Meanwhile, pre- and post-load comparison testing requires two clamping and two imaging operations, and typically necessitates transferring the component from the mechanical testing equipment to the EL darkroom. This process is cumbersome, and vibrations or stress release during transfer may introduce or mask additional damage, leading to inaccurate results. More importantly, some microcracks may partially close elastically after load unloading, causing the EL images after the load to fail to accurately reflect the maximum damage reached by the component during the loading period. This can result in missed or underestimated risks, affecting product quality assessment and failure analysis conclusions. Summary of the Invention

[0004] To address the problems in the prior art, this invention provides a real-time EL monitoring device and method for photovoltaic modules under dynamic load. This invention can monitor the initiation and propagation of internal defects in photovoltaic modules in real time and in situ throughout the entire dynamic load process, and quantitatively correlate damage evolution with load parameters, thereby accurately obtaining the critical load and propagation rate of microcrack initiation, providing a direct basis for module reliability assessment and fatigue resistance design.

[0005] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, the present invention provides a real-time EL monitoring device for photovoltaic modules under dynamic loads, comprising: A dynamic load loading system, which is connected to the photovoltaic module, is used to bear and apply dynamic loads to the photovoltaic module; A pulse power supply system, wherein the output terminal of the pulse power supply system is connected to the electrodes of the photovoltaic module; An EL imaging system, wherein the EL imaging system is positioned toward the surface of the photovoltaic module to be tested; The control system has its input terminal connected to the dynamic load loading system, and its output terminal connected to the pulse power supply system, the EL imaging system, and the dynamic load loading system, respectively. An image processing and analysis system, wherein the input terminals of the image processing and analysis system are connected to the EL imaging system and the dynamic load loading system, respectively.

[0006] Preferably, the dynamic load loading system includes a load actuator, a load sensor, and a displacement sensor; the force-applying end of the load actuator is connected to the photovoltaic module for applying mechanical force to the photovoltaic module; the load sensor is installed at the force-applying end of the load actuator for real-time load value detection; the displacement sensor is installed at the force-applying end of the load actuator for real-time displacement value detection; the output ends of the load sensor and the displacement sensor are both connected to the input end of the control system; the output end of the control system is connected to the load actuator.

[0007] Preferably, the pulse power supply system includes a programmable DC power supply, a high-speed switch array, and a voltage and current monitoring module; the output terminal of the programmable DC power supply is connected to the input terminal of the high-speed switch array; the output terminal of the high-speed switch array is connected to the electrodes of the photovoltaic module; the voltage and current monitoring module includes a voltage sampling unit and a current sampling unit, the two input terminals of the voltage sampling unit are respectively connected to the positive and negative electrodes of the photovoltaic module; the input terminal of the voltage sampling unit is connected in parallel to the two ends of the electrodes of the photovoltaic module, and the current sampling unit is connected in series on the connection line between the output terminal of the high-speed switch array and the electrodes of the photovoltaic module.

[0008] Preferably, the EL imaging system includes a near-infrared enhanced camera, an image acquisition card, and a near-infrared optical lens; the near-infrared optical lens is mounted on the incident light path of the near-infrared enhanced camera; the output of the near-infrared enhanced camera is connected to the input of the image processing and analysis system through the image acquisition card, and the input of the near-infrared enhanced camera is connected to the output of the control system.

[0009] Preferably, the control system includes a programmable timing controller, a signal conditioning module, and a clock synchronization module; the input terminal of the signal conditioning module is connected to the dynamic load loading system; the output terminal of the signal conditioning module is connected to the programmable timing controller; the clock synchronization module is connected to the programmable timing controller; and the output terminal of the programmable timing controller is connected to the pulse power supply system and the EL imaging system, respectively.

[0010] Preferably, the image processing and analysis system includes an image storage server, an image processing unit, and a data fusion unit; the input terminal of the image storage server is connected to the EL imaging system; the data reading terminal of the image processing unit is connected to the image storage server; and the input terminal of the data fusion unit is connected to both the image processing unit and the dynamic load loading system.

[0011] Secondly, the present invention provides a method for real-time monitoring of the photovoltaic module's electroluminescence (EL) under dynamic load, comprising the following steps: The photovoltaic module under test is mounted on the dynamic load loading system, and the output terminal of the pulse power system is connected to the electrode of the photovoltaic module. The EL imaging system is adjusted to align with the surface of the photovoltaic module under test. The dynamic load loading system is activated to apply dynamic loads to the photovoltaic module according to a preset load spectrum; at the same time, the control system acquires the load signals of the dynamic load loading system in real time and generates trigger signals according to the set triggering strategy. In response to the trigger signal, the control system first controls the pulse power supply system to output a pulse voltage to the photovoltaic module, causing the photovoltaic module to generate electroluminescence, and then controls the EL imaging system to perform exposure during the duration of the pulse voltage to acquire an electroluminescence image; The image processing and analysis system synchronously receives the electroluminescent images acquired by the EL imaging system and the load-time data recorded by the dynamic load loading system, processes the electroluminescent images in real time, extracts defect quantification indicators, and associates the defect quantification indicators with the corresponding load-time data to generate load-damage evolution information.

[0012] Preferably, the triggering strategy includes at least one of the following modes: Peak trigger mode: When the real-time acquired load signal value exceeds the preset load threshold, the trigger signal is generated; Equal-interval triggering mode: The trigger signal is generated once every preset time interval; Phase-locked mode: When the dynamic load is a periodic load, the trigger signal is generated when a preset specific phase of each cycle arrives.

[0013] Preferably, the real-time processing of the electroluminescent image includes: Dark field correction, flat field correction and background subtraction are performed on the acquired raw electroluminescent image to obtain the preprocessed current frame image; The preprocessed current frame image is registered with a static electroluminescent image pre-acquired before the application of dynamic load by using a feature point matching algorithm or cross-correlation algorithm to eliminate the influence of photovoltaic module displacement and deformation caused by dynamic load, and the registered current frame image is obtained. Calculate the difference image between the registered current frame image and the static electroluminescent image, extract the changed regions, and obtain the difference image; The differential image is segmented by thresholding, and at least one quantitative indicator among the area, length, and number of defects is extracted and calculated.

[0014] Preferably, the step of associating the defect quantification index with the corresponding load-time data to generate load-damage evolution information includes: The defect quantification index is correlated with the load amplitude at the corresponding moment obtained through the load-time data; Based on the correlated data, generate and output at least one of the following load-damage evolution curves: Curve showing the change in defect area with the number of dynamic load cycles; Crack length as a function of load amplitude; Curve showing the change in the number of defects or the rate of damage accumulation with the loading frequency.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The photovoltaic module dynamic load real-time EL monitoring device provided by this invention, by setting up a dynamic load loading system connected to the photovoltaic module, can continuously apply a preset dynamic load during the test, realistically simulating the stress state of the module under service environment; by setting up a pulse power supply system connected to the photovoltaic module electrodes and setting the EL imaging system to face the surface under test, and by connecting the control system to the dynamic load loading system, pulse power supply system and EL imaging system respectively, the control system can acquire load signals in real time and synchronously trigger pulse power supply and EL imaging according to a set triggering strategy, realizing accurate time-series acquisition of EL images under dynamic load for the first time, solving the problem that static or before-and-after load comparison detection cannot capture transient damage processes; at the same time, by setting up an image processing and analysis system connected to the EL imaging system and dynamic load loading system respectively, it can perform correlation analysis between the real-time acquired EL images and the corresponding load-time data, thereby transforming the originally isolated image information into a quantitative curve of load-damage evolution.

[0016] The present invention provides a real-time EL monitoring method for photovoltaic modules under dynamic load. First, by installing the photovoltaic module under test on a dynamic load loading system and adjusting the EL imaging system to align with the surface under test, the stability and consistency of the subsequent imaging field of view are ensured. Second, while applying the dynamic load through the dynamic load loading system, the control system acquires the load signal in real time and generates a trigger signal according to a triggering strategy, precisely linking the imaging action with the load state, avoiding data redundancy or missed acquisition at critical load moments due to blind image acquisition. Next, in response to the trigger signal, the control system first controls the pulse power supply system to output a pulse voltage to induce electroluminescence in the module, and then controls the EL imaging system to expose and acquire images during the duration of the pulse voltage. The image collection ensures high synchronization between EL luminescence intensity and exposure window, effectively avoiding component overheating during prolonged power-on and improving image signal-to-noise ratio. Finally, the image processing and analysis system simultaneously receives EL images and load-time data, processes the images in real time, extracts defect quantification indicators, and then correlates them with the corresponding load-time data to generate load-damage evolution information. This method maps dynamic mechanical parameters to the defect evolution process frame by frame, accurately obtaining the critical load for microcrack initiation, crack propagation rate, and damage accumulation law. It fundamentally overcomes the shortcomings of traditional load-before-and-after comparison detection, such as the inability to know the intermediate damage process, the ease of missing detection due to crack closure, and the inability to establish a quantitative relationship between load and damage. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of an EL real-time monitoring device for dynamic loads on photovoltaic modules according to the present invention.

[0019] The system includes: 1. Dynamic load loading system; 11. Load sensor; 12. Displacement sensor; 2. EL imaging system; 21. Near-infrared optical lens; 22. Near-infrared enhanced camera; 23. Image acquisition card; 3. Control system; 31. Programmable timing controller; 32. Signal conditioning module; 33. Clock synchronization module; 4. Image processing and analysis system; 41. Image storage server; 42. Image processing unit; 43. Data fusion unit; 5. Photovoltaic module. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0022] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0023] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply refers to its direction relative to "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0025] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0026] The present invention will now be described in further detail with reference to the accompanying drawings: The first objective of this invention is to provide a real-time EL monitoring device for photovoltaic modules under dynamic loads, such as... Figure 1 As shown, it includes: A dynamic load loading system 1 is connected to the photovoltaic module 5 and is used to bear and apply dynamic loads to the photovoltaic module 5. A pulse power supply system, the output terminal of which is connected to the electrodes of the photovoltaic module 5; An EL imaging system 2 is positioned facing the surface to be tested of the photovoltaic module 5. The control system 3 has its input terminal connected to the dynamic load loading system 1, and its output terminal connected to the pulse power supply system, the EL imaging system 2, and the dynamic load loading system 1, respectively. Image processing and analysis system 4, the input terminals of which are connected to EL imaging system 2 and dynamic load loading system 1 respectively.

[0027] This invention organically integrates a dynamic load loading system 1, a pulse power supply system, an EL imaging system 2, a control system 3, and an image processing and analysis system 4. During the application of dynamic loads to a photovoltaic module 5, the control system 3 precisely and synchronously triggers the pulse power supply to power the module based on the real-time load status, and simultaneously triggers the EL imaging system 2 to acquire high-frequency images. This achieves, for the first time, real-time, in-situ monitoring of the entire process of the initiation and expansion of internal defects (such as microcracks and grid breaks) under dynamic loads. Furthermore, the image processing and analysis system 4 fuses the EL images with corresponding load and displacement data, accurately establishing a quantitative correlation between damage evolution and mechanical parameters. This overcomes the shortcomings of traditional static or pre- and post-load comparison EL detection, which cannot capture transient damage processes, is prone to missed detections, and has low testing efficiency. This significantly improves the accuracy and efficiency of dynamic reliability assessment of the photovoltaic module 5.

[0028] For example, the dynamic load loading system 1 includes a loading actuator, a load sensor 11, and a displacement sensor 12. The loading actuator can be a conventional linear actuator in the art, such as an electro-hydraulic servo actuator, an electric cylinder, or an electromagnetic vibrator. Its force-applying end is rigidly connected to the surface or clamp of the photovoltaic module 5 and is used to apply mechanical forces such as tension, compression, or bending to the photovoltaic module 5 according to a preset load spectrum. The load sensor 11 is usually a strain gauge or piezoelectric force sensor and is installed between the force-applying end of the loading actuator and the photovoltaic module 5 to detect the instantaneous load value applied to the module in real time. The displacement sensor 12 can be a linear variable differential transformer (LVDT), a grating ruler, or a laser displacement meter and is also installed on the force-applying end of the loading actuator or its moving parts to detect the displacement value of the loading point in real time. The output terminals of the load sensor 11 and the displacement sensor 12 are both connected to the input terminal of the control system 3 to feed back the real-time acquired load and displacement signals to the control system 3. The output terminal of the control system 3 is connected to the drive unit (such as a servo valve or motor driver) of the load actuator to form a closed-loop control circuit, thereby enabling precise adjustment of the output of the load actuator according to the preset load spectrum or displacement spectrum, and achieving high-fidelity tracking of dynamic load waveforms (such as sine waves, random waves, and shock waves).

[0029] For example, the pulse power supply system includes a programmable DC power supply, a high-speed switching array, and a voltage and current monitoring module. The programmable DC power supply can be a high-precision programmable power supply commonly used in the art, whose output voltage (e.g., 0-100V) and current (e.g., 0-10A) can be programmed and set according to the specifications of the photovoltaic module 5 and the required EL brightness, and supports external trigger control to quickly establish a stable forward bias voltage. The output terminal of the programmable DC power supply is connected to the input terminal of the high-speed switching array, which is typically composed of semiconductor switching devices such as MOSFETs or IGBTs, possessing microsecond-level (<1μs) turn-on and turn-off response capabilities. Its output terminal is connected to the electrodes of the photovoltaic module 5, thereby realizing pulsed power supply. The voltage and current monitoring module includes a voltage sampling unit and a current sampling unit. The two input terminals of the voltage sampling unit are respectively connected to the positive and negative electrodes of the photovoltaic module 5 for real-time acquisition of the instantaneous voltage across the module. The current sampling unit is connected in series in the main circuit between the output terminal of the high-speed switching array and the electrodes of the photovoltaic module 5 for real-time monitoring of the transient current flowing through the module.

[0030] In addition to using a DC pulse power supply, an AC pulse power supply can also be used to study the electroluminescence characteristics of photovoltaic module 5 under reverse bias conditions. Although conventional EL detection relies on forward bias to excite minority carrier radiative recombination, certain special defects (such as local breakdown points or impurity deposits) can also produce weak luminescence characteristics under reverse bias. This design helps to identify specific types of latent defects.

[0031] During operation, the control system 3 can first turn on the high-speed switching array within the millisecond-level time window triggered by the dynamic load to apply a positive pulse voltage to the component using the programmable DC power supply. After the voltage and current monitoring module confirms that the voltage has stabilized to the set value, it immediately triggers the EL imaging system 2 to expose. After the exposure is completed, the switching array is quickly turned off to stop the power supply. This ensures the high-intensity, short-time stable electroluminescence excitation conditions required for EL image acquisition, and effectively avoids the component from overheating and being damaged due to prolonged power supply. At the same time, the real-time monitoring data of voltage and current can also be used to correct the power supply deviation caused by line impedance or component characteristic drift, ensuring that the electrical excitation conditions corresponding to each frame of EL image have good consistency and repeatability.

[0032] For example, the EL imaging system 2 includes a near-infrared enhanced camera 22, an image acquisition card 23, and a near-infrared optical lens 21; wherein, the near-infrared optical lens 21 is mounted on the incident light path of the near-infrared enhanced camera 22. This lens is typically designed with a large aperture (such as F1.4-F2.0) to maximize the collection of the weak near-infrared light emitted by the photovoltaic module 5, and the lens is coated with a near-infrared anti-reflection film to improve the transmittance in the 900-1100nm wavelength band. The focal length is selected according to the size of the module and the shooting distance, so that the surface of the module to be measured is completely imaged on the camera sensor; the near-infrared enhanced camera 22 generally adopts a scientific-grade back-illuminated C-type sensor. The MOS or deep-cooled CCD sensor has a typical quantum efficiency of ≥60% in the 900-1100nm band and millisecond-level or even microsecond-level electronic shutter control capability, which can effectively suppress noise and capture transient EL emission signals. The input end of the near-infrared enhanced camera 22 is connected to the output end of the control system 3, and receives external trigger signals from the control system 3, thereby achieving precise timing synchronization of exposure with pulse power supply and dynamic load status. The output end of the near-infrared enhanced camera 22 is connected to the input end of the image processing and analysis system 4 through the high-speed image acquisition card 23. The image acquisition card 23 supports high-speed interface protocols such as Camera Link, CoaXPress or 10 Gigabit Ethernet, and the data transmission rate can reach more than 10Gbps, ensuring that high-resolution, high-frame-rate EL images can be transmitted to the subsequent processing unit in real time without loss and with low latency.

[0033] Furthermore, for large-size photovoltaic modules (e.g., those over 2.5 m in length), multiple near-infrared cameras can be used to stitch together images. Software can then fuse the images of each sub-region into a complete high-resolution EL image, avoiding the possibility of missing tiny cracks due to insufficient resolution of a single camera. If the module undergoes significant translation during dynamic loading (e.g., on a linear reciprocating loading platform), a line-scan camera can be used in conjunction with a motion platform for continuous scanning imaging. The EL image in the stationary coordinate system can be acquired line by line during the module's movement, thereby eliminating motion blur introduced by a fixed field of view.

[0034] For example, the control system 3 includes a programmable timing controller 31, a signal conditioning module 32, and a clock synchronization module 33. The input of the signal conditioning module 32 is connected to the load sensor 11 and displacement sensor 12 in the dynamic load loading system 1. It receives analog voltage or current signals output by the sensors, filters, amplifies, levels, and shapes them to eliminate ambient noise interference and convert them into digital level trigger signals recognizable by the programmable timing controller 31. The output of the signal conditioning module 32 is connected to the programmable timing controller 31, sending the processed trigger signals to the programmable timing controller 31. The programmable timing controller 31 can be built based on a field-programmable gate array (FPGA) or a high-performance microcontroller (MCU). It has multiple preset logic strategies such as peak triggering, equal-interval triggering, phase-locked triggering, or event triggering. Based on the real-time load signal or external event signal input by the signal conditioning module 32, it accurately calculates and generates multiple timing control pulses. The clock synchronization module 33 is connected to the programmable timing controller 31 and can use a high-stability temperature-compensated crystal oscillator, a GPS timing module, or an IEEE 123 time synchronization module. The 1588 Precision Time Protocol (PTP) network clock source provides a unified, low-drift time reference for the entire control system 3, ensuring that the synchronization accuracy of the multi-channel trigger signals is better than 1 microsecond. The output of the programmable timing controller 31 is connected to the high-speed switching array in the pulse power supply system and the near-infrared enhanced camera 22 in the EL imaging system 2, respectively, to issue switching on commands and camera exposure commands at precise moments.

[0035] For example, the image processing and analysis system 4 includes an image storage server 41, an image processing unit 42, and a data fusion unit 43; wherein, the input end of the image storage server 41 is connected to the image acquisition card 23 in the EL imaging system 2, and is used to receive and persistently store the raw EL image sequence generated by high-speed acquisition, and it typically adopts a large-capacity NVMe SSD disk array or RAID. The system employs a 0 architecture to meet the high throughput and low latency requirements of millisecond-level continuous image writing. The data reading end of the image processing unit 42 is connected to the image storage server 41, which retrieves the current frame and reference frame images. Preprocessing operations such as dark field correction and flat field correction are performed sequentially. Then, the image registration module based on feature point matching or cross-correlation algorithms eliminates component displacement and elastic deformation caused by dynamic loads. Subsequently, the difference between the real-time image and the reference image is calculated using the differential analysis method. Defect regions such as hidden cracks and broken grids are extracted through threshold segmentation and morphological processing. Finally, indicators such as defect area, length, and quantity are quantified. The input end of the data fusion unit 43 is connected to the image processing unit 42 and the load sensor 11 and displacement sensor 12 in the dynamic load loading system 1, respectively. It is used to associate the time-stamped defect quantization data with the corresponding load value and displacement value to generate load-damage maps such as the defect area change curve with the number of cycles and the crack length evolution curve with the load amplitude.

[0036] The second objective of this invention is to provide a method for real-time monitoring of the photovoltaic module's electroluminescence (EL) under dynamic load, comprising the following steps: The photovoltaic module 5 to be tested is installed on the dynamic load loading system 1, and the output terminal of the pulse power system is connected to the electrode of the photovoltaic module 5. The EL imaging system 2 is adjusted to align with the surface of the photovoltaic module 5 to be tested. The dynamic load loading system 1 is activated to apply a dynamic load to the photovoltaic module 5 according to a preset load spectrum; at the same time, the control system 3 acquires the load signal of the dynamic load loading system 1 in real time and generates a trigger signal according to the set trigger strategy. In response to the trigger signal, the control system 3 first controls the pulse power system to output a pulse voltage to the photovoltaic module 5, causing the photovoltaic module 5 to generate electroluminescence, and then controls the EL imaging system 2 to perform exposure during the duration of the pulse voltage to acquire an electroluminescence image; The image processing and analysis system 4 synchronously receives the electroluminescent images acquired by the EL imaging system 2 and the load-time data recorded by the dynamic load loading system 1, performs real-time processing on the electroluminescent images, extracts defect quantification indicators, and associates the defect quantification indicators with the corresponding load-time data to generate load-damage evolution information.

[0037] This method, by introducing a control system 3 to precisely coordinate pulse power supply and high-speed imaging at the microsecond level during dynamic load loading, achieves for the first time real-time, in-situ capture of the defect initiation and propagation process of photovoltaic module 5 under real dynamic stress. This not only avoids the problem of missed detection caused by crack closure in traditional pre- and post-load comparison detection, but also improves the time resolution from static comparison to millisecond-level dynamic tracking. This allows researchers to accurately obtain the critical load for hidden crack initiation, crack propagation rate, and damage accumulation law related to loading frequency and waveform. Thus, it provides direct and high-fidelity experimental evidence for the fatigue design, reliability acceleration testing, and life prediction model of photovoltaic module 5, while significantly shortening the test cycle and reducing the additional errors introduced by multiple clamping.

[0038] For example, the triggering strategy can flexibly select at least one of the following modes according to different test objectives and dynamic load characteristics: Peak Trigger Mode: When the real-time acquired load signal value exceeds the preset load threshold, the trigger signal is generated; that is, the control system 3 monitors the instantaneous load value fed back by the load sensor 11 in real time, and when the value exceeds the preset load threshold (for example, set to 90% of the peak load or a certain absolute force value), the trigger signal is generated immediately. This mode is particularly suitable for studying the mechanism of hidden crack initiation of components under near-limit load conditions, and can accurately capture the EL response when the crack opens to its fullest extent under peak stress.

[0039] Equal-interval triggering mode: The triggering signal is generated once every preset time interval; that is, the control system 3 generates the triggering signal periodically at fixed time intervals (e.g., every 100 milliseconds or every 1 second), regardless of the current load value. This mode is suitable for studying the cumulative evolution of damage with loading time or number of cycles, and facilitates the generation of continuous damage development sequences.

[0040] Phase-locked mode: When the dynamic load is a periodic load, the trigger signal is generated when a preset specific phase is reached in each cycle; that is, when the dynamic load is a periodic waveform (such as a sine wave or a triangular wave), the control system 3 first identifies the period and phase zero point of the load signal, and then sets the required specific phase angle (for example, 90 degrees corresponds to the peak value, 0 degrees corresponds to the zero crossing point, or 180 degrees corresponds to the valley value). The trigger signal is precisely generated at the arrival time of the preset phase in each cycle. This mode is suitable for comparing and analyzing the periodic expansion law of defects in different cycles under the same phase, thereby eliminating the interference of stress state differences caused by different phases on defect performance.

[0041] In addition, it also includes an event-triggered mode, in which the control system 3, in addition to the load signal, also connects to an acoustic emission sensor or a load change detection circuit. When it detects an acoustic emission signal generated by the initiation or propagation of microcracks, or an abnormal drop / peak or other sudden change signal on the load curve, it immediately generates a trigger signal and continuously triggers high-speed continuous shooting at the moment the event occurs. This mode is particularly suitable for capturing sudden catastrophic failure processes, such as irreversible damage events such as instantaneous component breakage or large-area fragmentation.

[0042] For example, the real-time processing of the electroluminescent image includes: S1. Perform dark field correction, flat field correction, and background subtraction on the acquired original electroluminescent image to obtain the preprocessed current frame image. Dark field correction is performed by acquiring a dark field image under completely dark conditions and subtracting the image to eliminate sensor dark current noise. Flat field correction is performed by acquiring a flat field image under uniform light source and normalizing it to correct lens vignetting and sensor pixel response non-uniformity. Background subtraction is performed by removing stray light or fixed pattern noise in the test environment to obtain the preprocessed current frame image.

[0043] S2. Using a feature point matching algorithm or a cross-correlation algorithm, the preprocessed current frame image is registered with a static electroluminescent image pre-acquired before the application of dynamic load to eliminate the influence of displacement and deformation of photovoltaic module 5 caused by dynamic load, and a registered current frame image is obtained. Specifically, a feature point matching algorithm based on scale-invariant feature transform (SIFT), accelerated robust feature transform (SURF), or a cross-correlation algorithm based on the frequency domain is used to register the preprocessed current frame image with a static reference electroluminescent image pre-acquired before the application of dynamic load. By calculating the translation, rotation, and scaling transformation parameters between the two images and remapping the current frame, the influence of overall displacement, elastic bending deformation, or local warping of photovoltaic module 5 caused by dynamic load on defect location can be effectively eliminated, and a registered current frame image is obtained.

[0044] S3. Calculate the difference image between the registered current frame image and the static electroluminescent image, extract the changed areas, and obtain the difference image; specifically, calculate the pixel grayscale difference between the registered current frame image and the static reference image at the same coordinates, and generate the difference image. The areas in the difference image with significantly non-zero grayscale values ​​correspond to the areas of change such as newly appearing hidden cracks, the expansion of existing cracks, or broken grids during the dynamic loading process.

[0045] S4. Perform threshold segmentation on the differential image, and extract and calculate at least one quantitative indicator among the area, length, and number of defects; specifically, perform binarization processing on the differential image using Otsu's method or an adaptive threshold segmentation algorithm to separate the change region from the background, and calculate the pixel area, maximum Feret diameter (corresponding to crack length), and total number of defects for each defect based on connected component analysis.

[0046] This step effectively suppresses the noise and non-uniformity of the imaging system itself through dark-field and flat-field correction, significantly improving the image signal-to-noise ratio and the reliability of subsequent detection. Feature point matching or cross-correlation registration algorithms can accurately compensate for the rigid body displacement and non-uniform elastic deformation of components during dynamic loading, avoiding spurious differential signals caused by image misalignment and ensuring that only real defect changes are identified. The combination of differential analysis and threshold segmentation achieves clear separation between newly occurring damage and existing defects, and transforms subjective visual judgment into objective and quantifiable data indicators. This step enables the invention to output defect evolution data frame by frame during dynamic testing, providing immediate feedback for the control system 3 to adjust the loading strategy or for test personnel to determine the timing of shutdown.

[0047] For example, the step of associating the defect quantification index with the corresponding load-time data to generate load-damage evolution information includes: The defect quantification index is associated with the load amplitude at the corresponding moment obtained through the load-time data. Specifically, the data fusion unit 43 matches the defect quantification index (such as defect area, crack length, defect number, etc.) corresponding to each frame of electroluminescent image with the load amplitude, instantaneous phase, cycle number and loading frequency extracted from the load-time data at the same timestamp to ensure strict alignment of mechanical and damage data on the time axis. Based on the correlated data, generate and output at least one of the following load-damage evolution curves: Using the number of dynamic load cycles as the x-axis and the cumulative defect area or single new area as the y-axis, a curve is generated to show the change in defect area with the number of cycles, which is used to evaluate the fatigue life and damage accumulation rate of the component. Using the instantaneous load amplitude as the abscissa and the crack length at the corresponding moment as the ordinate, a curve of crack length versus load amplitude is generated to determine the critical load for the initiation of microcracks and the relationship between the speed of crack propagation and the load magnitude. By plotting the loading frequency on the x-axis and the number of new defects per unit cycle or the increase in damage area per unit time on the y-axis, curves are generated to show the change in the number of defects or the rate of damage accumulation with the loading frequency, in order to study the sensitivity of the component to dynamic loads at different frequencies.

[0048] This step deeply integrates the originally isolated massive EL image data with mechanical parameters, so that the damage evolution process under dynamic load is no longer limited to qualitative image observation, but is transformed into an engineering curve that can be quantitatively analyzed and mathematically modeled. Researchers can directly read the specific load value and number of cycles for the first appearance of the microcrack, obtain the quantitative relationship between crack propagation rate and load intensity, and clarify the accelerating or inhibiting effect of loading frequency on damage accumulation.

[0049] In summary, this invention achieves in-situ, real-time, and high-precision monitoring of the initiation and propagation of internal defects in photovoltaic module 5 throughout the entire dynamic loading process by establishing a real-time synchronous triggering mechanism for dynamic load and EL imaging, millisecond-level coordinated control of high-frequency pulse power supply and high-speed imaging, image registration and differential analysis methods under dynamic deformation, a multi-mode triggering strategy that flexibly adapts to various test scenarios, and a quantitative generation method for load-damage evolution curves. This method overcomes the fundamental shortcomings of traditional static or pre- and post-load comparison detection methods, such as the inability to capture transient damage processes, missed detections due to crack closure, and the lack of quantitative load-damage correlation. It can accurately obtain the critical load for microcrack initiation, crack propagation rate, and fatigue accumulation law, providing direct and quantifiable experimental evidence for the fatigue resistance design, accelerated reliability testing, and life prediction of photovoltaic module 5. Simultaneously, it significantly shortens the testing cycle and avoids additional errors introduced by multiple clamping operations.

[0050] In practical applications, for bare solar cells without electrodes or research scenarios that require eliminating electrode contact interference, photoluminescence technology can be combined with an external laser as the excitation source to irradiate the sample. With the help of appropriate filters, photoluminescence images can be collected, which can also reflect the defect distribution in crystalline silicon without the need for electrical contact.

[0051] Example This embodiment uses the EL real-time monitoring device under dynamic load of the photovoltaic module as the test object, and simulates the fatigue damage process of the module under wind vibration.

[0052] A 400W monocrystalline silicon photovoltaic module was selected and horizontally mounted on the test platform of the dynamic load loading system 1. The dynamic load loading system 1 uses an electro-hydraulic servo actuator, with the force application end contacting the surface of the photovoltaic module 5 via a pressure equalization beam. The positive output terminal of the pulse power supply system is connected to the positive lead electrode of the photovoltaic module 5, and the negative output terminal is connected to the negative lead electrode of the photovoltaic module 5. A near-infrared enhanced camera 22 (using a deep-cooled InGaAs sensor with a quantum efficiency greater than 70% in the 900-1100nm band) is mounted 1.5m directly above the module, using a 25mm focal length, F1.4 large aperture near-infrared lens to clearly fill the entire test surface of the module. The control system 3 uses an FPGA-based programmable timing controller 31, and a clock synchronization module 33 provides a global clock with 1μs accuracy.

[0053] Under no mechanical load, the programmable DC power supply outputs 85% (approximately 45V) of the component's open-circuit voltage, continuously supplying power for 100ms, followed by an 80ms camera exposure, acquiring an initial EL image as a reference image. The image processing and analysis system 4 annotates this reference image with defects, recording the original two minor microcracks and one broken grid location of the component. Simultaneously, the pulse power supply output voltage is set to 45V, the pulse width to 100ms (covering the 90ms camera exposure time with a 10ms stabilization margin), and the pulse rise time to less than 1ms. The dynamic load spectrum is set to a sine wave with an amplitude of ±1500N (simulating level 8 wind pressure), a frequency of 2Hz, and a total number of cycles of 10,000.

[0054] To address the need to study the microcrack propagation pattern under peak load in each cycle, given the periodic variation of the dynamic load, a phase-locked triggering strategy was selected. The control system 3 first locks the phase zero of the load waveform by performing zero-crossing detection and frequency tracking on the feedback signal from the load sensor 11. Then, a trigger signal is generated at the 90° phase of each cycle (i.e., the arrival time of the positive peak load plus 1500N). The timing control logic is as follows: at the instant the peak load arrives, the FPGA controller immediately outputs a T0 trigger signal; after a 1ms delay (power-up time), the high-speed switching array is turned on, and the pulse power supply applies a 45V positive bias to the component; after another 2ms delay (EL stabilization time), the camera's electronic shutter is triggered to begin exposure, with an exposure time set to 90ms; after the exposure ends, the high-speed switching array is immediately turned off, stopping the power supply. Only one frame of the EL image is acquired per cycle, continuing throughout the entire test process, for a total of 10,000 frames.

[0055] After the dynamic loading system is started, a dynamic load is continuously applied according to a 2Hz sine wave. Simultaneously, the control system 3 automatically triggers EL imaging once per cycle according to the phase-locked mode, and the image acquisition card 23 transmits the original EL image to the image storage server 41 in real time at a rate of 10Gbps. The image processing and analysis system 4 processes the image frame by frame: first, it performs dark field correction and flat field correction to eliminate sensor dark current and lens vignetting; then, it uses the SIFT feature point matching algorithm to register the current frame image with the initial static reference image, compensating for pixel offsets caused by component elastic bending and minute displacements; next, it calculates the difference image between the registered image and the reference image, performs median filtering for noise reduction, and uses Otsu's method for threshold segmentation to extract areas with significant grayscale changes. During the experiment, the system outputs defect quantification indicators in real time: for connected components in the difference image with an area greater than 5 pixels, they are marked as newly developed hidden cracks or extensions of existing cracks, and their pixel area is calculated and converted into the actual area (mm²). 2 ).

[0056] During the 328th iteration of the test, the image processing system detected for the first time an area of ​​1.2 mm in the differential image. 2 The system automatically recorded the change in the area at the tip of the initial microcrack. The load amplitude was +1498 N, and the number of cycles was 328. As the number of cycles increased, the area added per cycle at the crack tip increased from the initial 0.05 mm. 2 Gradually increase to 0.2 mm at the 5000th cycle. 2 .

[0057] The system generated a real-time evolution curve of the crack area with the number of cycles, showing that crack propagation exhibits three stages: initiation stage (1-328 cycles, no propagation), steady-state propagation stage (328-8000 cycles, linear area growth), and accelerated propagation stage (after 8000 cycles, exponential area growth). Simultaneously, the data fusion unit 43 correlated the defect area of ​​each frame with the corresponding load amplitude, plotting a scatter plot of crack propagation rate versus load amplitude. It was found that crack propagation begins when the absolute value of the instantaneous load exceeds 1200N, and the propagation rate is positively correlated with the fourth power of the load amplitude.

[0058] During the 9820th cycle of the test, the differential image detected three new radial cracks suddenly appearing in an area that was previously a broken grid, with the total area increasing by 15.6 mm in a single frame. 2 Meanwhile, load sensor 11 detected a small load drop (approximately 30N). Control system 3 determined this to be a sudden failure event in event-triggered mode and automatically increased the acquisition frequency to 5 frames per cycle (i.e., adding additional trigger points before and after the peak), continuously recording the rapid evolution images for the next 20 cycles; finally, after reaching the 10,000th cycle, the test automatically stopped.

[0059] After the test, the component was removed for post-static EL inspection, which verified the locations and expansion dimensions of all new microcracks detected by the real-time monitoring system, with a consistency of 97.3%. The generated complete test report includes: load spectrum parameters (waveform, amplitude, frequency, number of cycles), evolution video generated by compressing all 10,000 frames of EL images, crack area-cycle count curve, detailed log of the first expansion event, and high-resolution image sequence of the 10 cycles before sudden failure.

[0060] This embodiment fully demonstrates that the present invention can capture the entire process of the initiation and expansion of internal defects of photovoltaic module 5 in real time, in situ, and quantitatively during dynamic load, providing direct and reliable experimental basis for fatigue life assessment and wind vibration resistance design of the module.

[0061] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time EL monitoring device for photovoltaic modules under dynamic load, characterized in that, include: A dynamic load loading system (1) is connected to a photovoltaic module (5) and is used to bear and apply dynamic loads to the photovoltaic module (5). A pulse power supply system, the output terminal of which is connected to the electrodes of the photovoltaic module (5); An EL imaging system (2) is positioned facing the surface to be measured of the photovoltaic module (5); The control system (3) has its input terminal connected to the dynamic load loading system (1) and its output terminal connected to the pulse power supply system, the EL imaging system (2) and the dynamic load loading system (1) respectively. The image processing and analysis system (4) is connected to the EL imaging system (2) and the dynamic load loading system (1) respectively.

2. The photovoltaic module dynamic load real-time monitoring device according to claim 1, characterized in that, The dynamic load loading system (1) includes a load actuator, a load sensor (11), and a displacement sensor (12); the force-applying end of the load actuator is connected to the photovoltaic module (5) and is used to apply mechanical force to the photovoltaic module (5); the load sensor (11) is installed on the force-applying end of the load actuator and is used to detect the load value in real time; the displacement sensor (12) is installed on the force-applying end of the load actuator and is used to detect the displacement value in real time; the output ends of the load sensor (11) and the displacement sensor (12) are both connected to the input end of the control system (3); the output end of the control system (3) is connected to the load actuator.

3. The photovoltaic module dynamic load real-time monitoring device according to claim 1, characterized in that, The pulse power supply system includes a programmable DC power supply, a high-speed switch array, and a voltage and current monitoring module; the output terminal of the programmable DC power supply is connected to the input terminal of the high-speed switch array; the output terminal of the high-speed switch array is connected to the electrode of the photovoltaic module (5); the voltage and current monitoring module includes a voltage sampling unit and a current sampling unit, the two input terminals of the voltage sampling unit are respectively connected to the positive electrode and the negative electrode of the photovoltaic module (5); the input terminal of the voltage sampling unit is connected in parallel to the two ends of the electrode of the photovoltaic module (5), and the current sampling unit is connected in series on the connection line between the output terminal of the high-speed switch array and the electrode of the photovoltaic module (5).

4. The photovoltaic module dynamic load real-time monitoring device according to claim 1, characterized in that, The EL imaging system (2) includes a near-infrared enhanced camera (22), an image acquisition card (23), and a near-infrared optical lens (21); the near-infrared optical lens (21) is mounted on the incident light path of the near-infrared enhanced camera (22); the output end of the near-infrared enhanced camera (22) is connected to the input end of the image processing and analysis system (4) through the image acquisition card (23), and the input end of the near-infrared enhanced camera (22) is connected to the output end of the control system (3).

5. The photovoltaic module dynamic load real-time monitoring device according to claim 1, characterized in that, The control system (3) includes a programmable timing controller (31), a signal conditioning module (32), and a clock synchronization module (33); the input terminal of the signal conditioning module (32) is connected to the dynamic load loading system (1); the output terminal of the signal conditioning module (32) is connected to the programmable timing controller (31); the clock synchronization module (33) is connected to the programmable timing controller (31); the output terminal of the programmable timing controller (31) is connected to the pulse power supply system and the EL imaging system (2) respectively.

6. The photovoltaic module dynamic load real-time monitoring device according to claim 1, characterized in that, The image processing and analysis system (4) includes an image storage server (41), an image processing unit (42), and a data fusion unit (43); the input end of the image storage server (41) is connected to the EL imaging system (2); the data reading end of the image processing unit (42) is connected to the image storage server (41); the input end of the data fusion unit (43) is connected to the image processing unit (42) and the dynamic load loading system (1), respectively.

7. A method for real-time monitoring of the photovoltaic module's electroluminescence (EL) under dynamic load, characterized in that, The monitoring device according to any one of claims 1 to 6 includes the following steps: The photovoltaic module (5) to be tested is installed on the dynamic load loading system (1), and the output terminal of the pulse power system is connected to the electrode of the photovoltaic module (5). The EL imaging system (2) is adjusted to align with the surface of the photovoltaic module (5) to be tested. The dynamic load loading system (1) is activated to apply a dynamic load to the photovoltaic module (5) according to a preset load spectrum; at the same time, the control system (3) acquires the load signal of the dynamic load loading system (1) in real time and generates a trigger signal according to the set trigger strategy. In response to the trigger signal, the control system (3) first controls the pulse power system to output a pulse voltage to the photovoltaic module (5) so that the photovoltaic module (5) generates electroluminescence, and then controls the EL imaging system (2) to perform exposure during the duration of the pulse voltage and acquire an electroluminescence image; The image processing and analysis system (4) synchronously receives the electroluminescent image acquired by the EL imaging system (2) and the load-time data recorded by the dynamic load loading system (1), performs real-time processing on the electroluminescent image, extracts the defect quantification index, and associates the defect quantification index with the corresponding load-time data to generate load-damage evolution information.

8. The method for real-time monitoring of EL under dynamic load of a photovoltaic module according to claim 7, characterized in that, The triggering strategy includes at least one of the following modes: Peak trigger mode: When the real-time acquired load signal value exceeds the preset load threshold, the trigger signal is generated; Equal-interval triggering mode: The trigger signal is generated once every preset time interval; Phase-locked mode: When the dynamic load is a periodic load, the trigger signal is generated when a preset specific phase of each cycle arrives.

9. The method for real-time monitoring of EL under dynamic load of a photovoltaic module according to claim 7, characterized in that, The real-time processing of the electroluminescent image includes: Dark field correction, flat field correction and background subtraction are performed on the acquired raw electroluminescent image to obtain the preprocessed current frame image; The preprocessed current frame image is registered with the static electroluminescent image pre-acquired before the dynamic load is applied by using a feature point matching algorithm or cross-correlation algorithm to eliminate the influence of displacement and deformation of photovoltaic module (5) caused by dynamic load, and the registered current frame image is obtained. Calculate the difference image between the registered current frame image and the static electroluminescent image, extract the changed regions, and obtain the difference image; The differential image is segmented by thresholding, and at least one quantitative indicator among the area, length, and number of defects is extracted and calculated.

10. The method for real-time monitoring of EL under dynamic load of a photovoltaic module according to claim 7, characterized in that, The step of associating the defect quantification index with the corresponding load-time data to generate load-damage evolution information includes: The defect quantification index is correlated with the load amplitude at the corresponding moment obtained through the load-time data; Based on the correlated data, generate and output at least one of the following load-damage evolution curves: Curve showing the change in defect area with the number of dynamic load cycles; Crack length as a function of load amplitude; Curve showing the change in the number of defects or the rate of damage accumulation with the loading frequency.

Images