Method for detecting defects of a photovoltaic module and system for detecting defects of a photovoltaic module

Abstract

The application provides a defect detection method of a photovoltaic module and a defect detection system of the photovoltaic module. Any two adjacent tunneling magnetoresistance sensors in a sensing unit located on the back plate side of the photovoltaic module have a height difference in the thickness direction of the back plate. The method comprises the following steps: acquiring spatial magnetic field distribution of the photovoltaic module sensed by the plurality of tunneling magnetoresistance sensors within a preset time length, and the current of the photovoltaic module carries an alternating component of a target modulation frequency; extracting a spatial vertical gradient of the magnetic field generated by the current according to the spatial magnetic field distribution; performing synchronous demodulation processing on the spatial vertical gradient according to a reference clock signal corresponding to the target modulation frequency; performing image reconstruction processing on the demodulation result to obtain the current density distribution of the photovoltaic module; and determining a defect area of the photovoltaic module according to the current density distribution. The application solves the problem that the existing defect detection technologies of the photovoltaic module cannot simultaneously meet the comprehensive detection requirements of outdoor online, non-contact and high sensitivity.

Description

Technical Field

[0001] This application relates to the field of photovoltaic module testing technology, and more specifically, to a method and system for detecting defects in photovoltaic modules. Background Technology

[0002] During the manufacturing and long-term operation of crystalline silicon photovoltaic modules, factors such as process fluctuations, mechanical stress, and environmental aging can easily lead to defects such as microcracks, main grid / fine grid fractures, potential-induced degradation (PID), and hot spot short circuits. These defects essentially manifest as interruptions or distortions in local current paths, directly resulting in a decrease in module output power and even safety hazards.

[0003] Currently, defect detection technologies for photovoltaic modules include electroluminescence / photoluminescence imaging technology. These methods are based on the principle of radiative recombination, using an applied bias current or photoexcitation to induce carrier recombination and release near-infrared light. A CCD (Charge-Coupled Device) camera is then used to capture images of the light intensity distribution, which can identify microcracks and grid breaks, making it the mainstream method for factory inspection. However, its limitations are significant: it requires strict shading conditions and cannot be used outdoors in situ; electroluminescence (EL) detection requires applying a forward bias current to the module, necessitating the interruption of power generation and connection to an external power source in grid-connected systems, making operation complex and posing safety risks; furthermore, it lacks sufficient response to early defects such as interface defects and weak leakage currents that generate weak light signals, limiting its quantitative capabilities.

[0004] Current defect detection technologies for photovoltaic modules also include infrared thermal imaging. This type of method is based on the Joule heating effect and identifies hot spots and areas of abnormal power by detecting differences in surface temperature distribution caused by abnormal resistance or leakage current in defective areas. It can be operated under outdoor natural light conditions. However, the thermal hysteresis effect of this type of method limits early defect response. The heat from defects needs to be conducted to the surface through the encapsulation layer, making it difficult to capture weak defects in the early stages in time. In addition, the spatial resolution is insufficient, and it is easy to miss microcracks, hidden cracks, or fine grid fractures below the millimeter level. Furthermore, the detection results of this type of method are significantly affected by external factors such as ambient temperature and wind speed, resulting in insufficient stability and repeatability.

[0005] In summary, existing defect detection technologies for photovoltaic modules cannot simultaneously meet the comprehensive requirements of outdoor online, non-contact, and high-sensitivity testing. There is an urgent need to develop a detection method that can perceive internal defect conditions non-contactly without interrupting operation while the module is generating power normally. Summary of the Invention

[0006] The main objective of this application is to provide a defect detection method and a defect detection system for photovoltaic modules, so as to at least solve the problem that the existing defect detection technologies for photovoltaic modules cannot simultaneously meet the comprehensive requirements of outdoor online, non-contact, and high-sensitivity detection.

[0007] To achieve the above objectives, according to one aspect of this application, a defect detection method for a photovoltaic module is provided. A sensing unit is disposed on one side of the photovoltaic module having a backsheet. The sensing unit includes multiple tunnel magnetoresistive sensors (TMS), which overlap in a first direction, and any two adjacent TMS sensors in the first direction have a height difference. The first direction is the thickness direction of the backsheet. The method includes: acquiring the spatial magnetic field distribution of the photovoltaic module sensed by the multiple TMS sensors within a preset time period, wherein the sensed current of the photovoltaic module carries an AC component of a target modulation frequency; extracting the spatial vertical gradient of the magnetic field generated by the current based on the spatial magnetic field distribution of the photovoltaic module, the spatial vertical gradient being used to characterize the rate of change of the magnetic field strength in the first direction; performing synchronous demodulation processing on the spatial vertical gradient according to a reference clock signal corresponding to the target modulation frequency to extract a demodulation result reflecting local current density vector distortion; performing image reconstruction processing on the demodulation result to obtain the current density distribution of the photovoltaic module; and determining the defect region of the photovoltaic module based on the current density distribution of the photovoltaic module.

[0008] Optionally, the method further includes: outputting a target excitation signal that varies periodically at the target modulation frequency to the photovoltaic module by modulating the excitation source, so that the current of the photovoltaic module carries the AC component of the target modulation frequency.

[0009] Optionally, by modulating the excitation source to output the target excitation signal that periodically varies with the target modulation frequency to the photovoltaic module, the method includes: modulating the target light source according to a preset pulse signal so that the light intensity applied to the photovoltaic module by the target light source varies periodically with the target modulation frequency; or modulating the load switching circuit electrically connected to the photovoltaic module according to a preset pulse signal so that the output current of the photovoltaic module varies periodically with the target modulation frequency.

[0010] Optionally, based on the spatial magnetic field distribution of the photovoltaic module, extracting the spatial vertical gradient of the magnetic field generated by the current includes: performing differential processing on the spatial magnetic field distribution of multiple tunnel magnetoresistive sensors to obtain differential results, and using the differential results as the spatial vertical gradient of the magnetic field.

[0011] Optionally, the spatial vertical gradient is synchronously demodulated according to a reference clock signal corresponding to the target modulation frequency to obtain the demodulation result, including: using the reference clock signal as a phase reference, extracting the target amplitude and target phase at the target modulation frequency from the differential result.

[0012] Optionally, image reconstruction processing is performed on the demodulation result to obtain the current density distribution of the photovoltaic module, including: establishing a linear mapping relationship between the spatial vertical gradient of the magnetic field and the current density distribution based on the target amplitude and the target phase; constructing a system response matrix based on the Biot-Savart law and sensor geometric parameters; and using the system response matrix, the linear mapping relationship, and matrix multiplication operations to invert the spatial vertical gradient of the magnetic field to obtain the current density vector field.

[0013] Optionally, determining the defect region of the photovoltaic module based on the current density distribution of the photovoltaic module includes: calculating the deviation between the amplitude of the current density vector at the target pixel in the current density vector field and the average amplitude of the current density vector at all pixels in a preset neighborhood of the target pixel; and determining the region where the deviation exceeds a preset threshold as the defect region.

[0014] According to another aspect of this application, a defect detection system for a photovoltaic module is provided, comprising: an excitation unit for modulating the photovoltaic module to make the current of the photovoltaic module carry an AC component of a target modulation frequency; a sensing unit located on one side of the photovoltaic module having a backsheet, for sensing the spatial magnetic field distribution of the photovoltaic module within a preset time period, the sensing unit including a plurality of tunnel magnetoresistive sensors, the plurality of tunnel magnetoresistive sensors overlapping in a first direction, and any two adjacent tunnel magnetoresistive sensors in the first direction having a height difference, the first direction being the thickness direction of the backsheet; a preprocessing unit for extracting the spatial vertical gradient of the magnetic field generated by the current according to the spatial magnetic field distribution of the photovoltaic module, and performing synchronous demodulation processing on the spatial vertical gradient according to a reference clock signal corresponding to the target modulation frequency to extract a demodulation result reflecting local current density vector distortion, wherein the spatial vertical gradient is used to characterize the rate of change of magnetic field strength in the first direction; and an image reconstruction unit for performing image reconstruction processing on the demodulation result to obtain the current density distribution of the photovoltaic module, and determining the defect region of the photovoltaic module according to the current density distribution of the photovoltaic module.

[0015] Optionally, the excitation unit includes: a light-emitting device module, a first pulse width modulation controller, and a synchronous clock source, wherein the light-emitting device module is used to output a target light source, the first pulse width modulation controller is used to modulate the target light source according to a preset pulse signal so that the light intensity applied to the photovoltaic module by the target light source changes periodically with the target modulation frequency, and the synchronous clock source is used to provide a reference clock signal corresponding to the target modulation frequency; or a load switching circuit, a second pulse width modulation controller, and a synchronous clock source, wherein the load switching circuit is electrically connected to the photovoltaic module, the second pulse width modulation controller is used to modulate the load switching circuit according to a preset pulse signal so that the output current of the photovoltaic module changes periodically with the target modulation frequency, and the synchronous clock source is used to provide a reference clock signal corresponding to the target modulation frequency.

[0016] Optionally, the sensing unit further includes: a circuit board on which the tunnel magnetoresistive sensor is disposed; a rigid limiting layer disposed on one side surface of the photovoltaic module having the backplate, for separating the rigid limiting layer from the tunnel magnetoresistive sensor; and a multiplexing circuit for sequentially acquiring the output signal of the tunnel magnetoresistive sensor according to a preset timing sequence.

[0017] Applying the technical solution of this application, the spatial magnetic field distribution of the photovoltaic module sensed by multiple tunnel magnetoresistive sensors within a preset time period is first obtained. This step utilizes the response characteristics of the tunnel magnetoresistive sensors to alternating magnetic fields to capture the AC component of the current carrying the target modulation frequency caused by defects in the photovoltaic module under test. The above response does not depend on external light sources or thermal effects, but is based on the interruption or detour of local current paths caused by defects, so that the detection signal directly corresponds to the vector distortion of local current density, thereby eliminating the sensitivity bottleneck caused by weak light intensity. Subsequently, a spatial gradient difference structure is constructed based on the spatial height difference between adjacent sensors to reduce the far-field background. By conserving near-field difference signals while minimizing the contribution of the signal, the effects of lift-off errors and environmental noise are eliminated, ensuring the stability and repeatability of detection results in complex outdoor environments and directly improving the signal-to-noise ratio and detection sensitivity. Next, the extracted spatial vertical gradient is synchronously demodulated using a reference clock as the phase reference, converting static interference into filterable background components. This yields a demodulated result for local current density vector distortion, enabling early defect response without interrupting operation and through non-contact sensing. Finally, image reconstruction is performed based on the demodulated result to generate a current density distribution map, which is then used to determine the defect region. Therefore, this solution, through the aforementioned collaborative steps, solves the problem that existing photovoltaic module defect detection technologies cannot simultaneously meet the comprehensive requirements of outdoor online, non-contact, and high-sensitivity detection. It realizes the transformation from laboratory research to engineering applications, improving the stability and accuracy of online detection of photovoltaic modules in complex environments. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 A hardware structure block diagram of a mobile terminal for performing a defect detection method for photovoltaic modules according to an embodiment of this application is shown. Figure 2 A schematic flowchart of a defect detection method for a photovoltaic module according to an embodiment of this application is shown. Figure 3 This diagram illustrates the spatial vertical gradient of the magnetic field between the first TMR sensor and the second TMR sensor in the defect detection method for photovoltaic modules provided in the embodiments of this application. Figure 4 A schematic diagram showing the positional relationship of the rigid limiting layer within the sensing unit in the defect detection method for photovoltaic modules provided in the embodiments of this application is illustrated. Figure 5A top view of the array layout of tunnel magnetoresistive sensors in the sensing unit of the defect detection method for photovoltaic modules provided in the embodiments of this application is shown. Figure 6 A structural block diagram of a defect detection system for a photovoltaic module provided according to an embodiment of this application is shown.

[0019] The above figures include the following reference numerals: 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device; 100. Rigid limiting layer; 200. Backplate; 300. Circuit board; 400. Multiplexing circuit. Detailed Implementation

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] As described in the background section, existing crystalline silicon photovoltaic modules are prone to defects such as microcracks, grid breaks, PID (Potential Instability), and hot spots during manufacturing and operation. These defects manifest as interruptions or distortions in local current paths, leading to power reduction and even safety hazards. Among existing mainstream defect detection methods, electroluminescence / photoluminescence imaging technology, while possessing high image resolution, requires strict shading and a forward bias current applied to the module. This limitation prevents its outdoor online use and necessitates interrupting power generation, creating an engineering obstacle in operational power plants. Infrared thermal imaging technology, although operable outdoors, suffers from insufficient sensitivity for detecting early micro-defects below the millimeter level due to the thermal conduction delay caused by multi-layer encapsulation structures and environmental temperature fluctuations. Therefore, there is an urgent need to develop a non-contact detection method for internal defects that can be detected without interrupting normal module power generation. To address the issue that existing technologies cannot simultaneously meet the comprehensive requirements of outdoor online, non-contact, and high-sensitivity detection, embodiments of this application provide a defect detection method and system for photovoltaic modules.

[0024] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0025] The methods and embodiments provided in this application can be executed on a mobile terminal, a computer terminal, or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a defect detection method for photovoltaic modules according to an embodiment of this application. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0026] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the photovoltaic module defect detection method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thus implementing the above-described method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0027] This embodiment provides a defect detection method for photovoltaic modules that runs on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0028] Figure 2 This is a flowchart of a defect detection method for a photovoltaic module according to an embodiment of this application. A sensing unit is disposed on one side of the photovoltaic module having a backsheet. The sensing unit includes multiple tunneling magnetoresistive sensors (TMS sensors). The multiple TMS sensors overlap in a first direction, and any two adjacent TMS sensors in the first direction have a height difference. The first direction is the thickness direction of the backsheet. Figure 2 As shown, the method includes the following steps: Step S201: Obtain the spatial magnetic field distribution of the photovoltaic module sensed by multiple tunnel magnetoresistive sensors within a preset time period, wherein the current of the sensed photovoltaic module carries an AC component of the target modulation frequency. Specifically, by acquiring the spatial magnetic field distribution sensed by multiple tunnel magnetoresistive sensors within a preset time period and filtering out the signal components carrying the target modulation frequency AC component, the signal containing weak local current distortion information is physically separated from background noise such as static geomagnetic field, low-frequency power grid interference and asynchronous broadband noise, thereby achieving accurate extraction of early defect signals.

[0029] Step S202: Based on the spatial magnetic field distribution of the photovoltaic module, extract the spatial vertical gradient of the magnetic field generated by the current. The spatial vertical gradient is used to characterize the rate of change of the magnetic field intensity in the first direction. Specifically, based on the spatial distribution differences of multiple tunnel magnetoresistive sensors, the spatial vertical gradient of the magnetic field generated by the target current is calculated and extracted. The dual height difference structure has a natural common-mode suppression capability against the uniform background magnetic field in the far field, which reduces the background contribution to near zero while retaining the significant difference signal generated by the local current anomaly in the near field, thereby obtaining the demodulation result reflecting the local current density vector distortion.

[0030] Step S203: Based on the reference clock signal corresponding to the target modulation frequency, perform synchronous demodulation processing on the spatial vertical gradient to extract the demodulation result reflecting the distortion of the local current density vector. Specifically, by using a reference clock corresponding to the target modulation frequency as a phase reference, the extracted spatial vertical gradient is synchronously demodulated, responding only to signal components that are strictly synchronized with the preset frequency, and converting static geomagnetic field, low-frequency power grid interference and asynchronous broadband noise into filterable background components, thereby achieving physical separation of the target signal and environmental interference in the frequency domain.

[0031] Step S204: Perform image reconstruction processing on the demodulation results to obtain the current density distribution of the photovoltaic module; Specifically, the demodulation results after synchronous demodulation are subjected to image reconstruction processing. This step makes the far-field background contribution approach zero, while retaining the significant difference signals generated by the near-field local current anomalies, and finally obtaining image data reflecting the current density distribution of the photovoltaic module.

[0032] Step S205: Determine the defect area of ​​the photovoltaic module based on the current density distribution of the photovoltaic module; Specifically, based on the current density distribution map obtained above, the deviation of each region from the reference value is compared and analyzed to determine the defective region, thus realizing the accurate location and identification of local minor current distortion defect regions such as early grid line damage.

[0033] Since defects in photovoltaic modules, such as microcracks and grid line breaks, are essentially vector distortions of local current density in electrical terms, the current path in the defect area is interrupted or detours, causing the local current direction or amplitude to deviate from the normal distribution. According to the Biot-Savart law, any change in current distribution will instantly map to a corresponding change in the distribution of the magnetic field in the surrounding space. Unlike infrared detection, which relies on heat diffusion and conduction through the encapsulation layer of the photovoltaic module, magnetic field signals do not require a medium for conduction and can directly penetrate the encapsulation layer and non-magnetic materials such as the backsheet, achieving an instantaneous response to the internal current state. This characteristic theoretically gives magnetic field signals a faster response speed and more direct defect sensing capability than thermal imaging, making them a superior physical carrier for early defect detection. However, applying magnetic field imaging to photovoltaic defect detection faces a severe sensitivity challenge. The operating current of a single solar cell is approximately 8–10 A, and the magnitude of the local current disturbance in the defect area is approximately 1%–10% of the normal current, i.e., ΔI ≈ 0.1–1 A. According to the Biot-Savart law, this magnitude of current disturbance generates a magnetic field signal of approximately tens to hundreds of nanoteslas (nT) at a distance of 1–2 mm. Hall sensors typically have a minimum detectable magnetic field in the microtesla (μT) range, while anisotropic magnetoresistance (AMR) and giant magnetoresistance (GMR) sensors can reach the nanotesla range. However, in actual detection bandwidths, their equivalent noise floor is on the same order of magnitude as the target signal, resulting in a severely insufficient signal-to-noise ratio. This causes the defective magnetic field characteristics to be submerged in the sensor's inherent noise floor.

[0034] Based on this, embodiments of this application introduce a tunnel magnetoresistance (TMR) sensor. The equivalent magnetic noise substrate of a TMR sensor can reach the order of 1–10 nT at a typical detection bandwidth (10 Hz–10 kHz). Combined with phase-locked loop detection technology, the equivalent noise can be further reduced by one to two orders of magnitude, enabling the stable extraction of defect magnetic field signals on the order of tens of nT. This provides an effective hardware foundation for magnetic field-based photovoltaic defect imaging.

[0035] In this embodiment, the spatial magnetic field distribution of the photovoltaic module sensed by multiple tunnel magnetoresistive (TMR) sensors within a preset time period is first obtained. This step utilizes the response characteristics of the TMR sensors to alternating magnetic fields to capture the AC component of the current carrying the target modulation frequency caused by defects in the photovoltaic module under test. The above response does not depend on external light sources or thermal effects, but is based on the interruption or detour of local current paths caused by defects, so that the detection signal directly corresponds to the vector distortion of local current density, thereby eliminating the sensitivity bottleneck caused by weak light intensity. Subsequently, a spatial gradient difference structure is constructed based on the spatial height difference between adjacent sensors, so that the far-field back The near-field difference signal is preserved while the background contribution approaches zero, thereby eliminating the effects of lift-off error and environmental noise. This ensures the stability and repeatability of the detection results in complex outdoor environments and directly improves the signal-to-noise ratio and detection sensitivity. Next, the extracted spatial vertical gradient is synchronously demodulated using a reference clock as the phase reference, converting static interference into a filterable background component. This yields a demodulated result for local current density vector distortion, enabling early defect response without interrupting operation and through non-contact sensing. Finally, image reconstruction is performed based on the demodulated result to generate a current density distribution map, which is then used to determine the defect region. Therefore, this solution, through the above collaborative steps, solves the problem that existing photovoltaic module defect detection technologies cannot simultaneously meet the comprehensive requirements of outdoor online, non-contact, and high-sensitivity detection. It realizes the transformation from laboratory research to engineering applications and improves the stability and accuracy of online detection of photovoltaic modules in complex environments.

[0036] In some optional implementations, the defect detection method for photovoltaic modules in this application embodiment further includes: outputting a target excitation signal that changes periodically at a target modulation frequency to the photovoltaic module by modulating the excitation source, so that the current of the photovoltaic module carries the AC component of the target modulation frequency.

[0037] Specifically, by modulating the excitation signal output by the excitation source, which periodically varies at the target modulation frequency, the static magnetic field generated by the photovoltaic module is converted into a dynamic magnetic field. This allows the light source to be modulated using a preset pulse signal, causing the applied light intensity to fluctuate at a specific frequency over time. This, in turn, excites an alternating current carrying an AC component of the target modulation frequency within the module. These steps not only achieve the physical conversion from a static to a dynamic magnetic field but also provide a clear phase reference for subsequent frequency domain coding and synchronous demodulation. By accurately extracting this specific frequency AC component, local current density vector distortion caused by defects can be more sensitively detected, effectively filtering out environmental noise interference and ensuring quantitative assessment of weak early defect signals under complex operating conditions.

[0038] In the above optional embodiments, outputting a target excitation signal that varies periodically at a target modulation frequency to the photovoltaic module by modulating the excitation source may include: modulating the target light source according to a preset pulse signal so that the light intensity applied to the photovoltaic module by the target light source varies periodically at the target modulation frequency.

[0039] Specifically, the LED array can be periodically switched on and off using a PWM drive controller. A preset frequency pulse signal precisely regulates the switching between high and low light intensity states, adding a periodic fluctuation component with an amplitude of ΔL to the average incident light intensity. This causes the photocurrent generated within the photovoltaic module to synchronously carry an AC component with the same frequency as the light source. This process not only transforms static illumination into a dynamic excitation source but also establishes a traceable frequency domain reference within the photovoltaic module. This provides a clear phase synchronization condition for subsequent phase-locked demodulation, ensuring accurate conversion and extraction of the photocurrent into a magnetic field signal.

[0040] In the above optional embodiments, photovoltaic modules can also be used as the main light source, with auxiliary light sources including near-infrared (NIR) lasers to excite the thermal effect inside the battery and enhance the magnetic field gradient. Specifically, utilizing the high thermal response characteristics of NIR lasers, even under strong light interference, the abnormal current density caused by local high temperature can be accurately captured. At the same time, the total light intensity change rate generated by multispectral excitation is greater, which allows subsequent phase-locked demodulation to more clearly extract the weak modulation frequency f (i.e., the preset frequency) component. Thus, this embodiment can also be applied to scenarios with complex ambient lighting conditions, such as insufficient visible light due to cloud cover.

[0041] For example, in an offline full inspection scenario, a PWM drive controller is used to periodically control the on / off state of a high-frequency modulated LED array. A preset frequency f is used as a reference to rapidly switch the LED output light intensity between high and low brightness states. The incident light intensity is adjusted periodically using a square wave function, as shown in the following formula: ; Where sq(•) represents the square wave function, L0 is the average light intensity of the two states, and ΔL is half of the light intensity switching amplitude.

[0042] The aforementioned non-uniform illumination excitation mechanism acts directly on the interior of the photovoltaic module, causing the response signal generated by the photocurrent to also carry an AC component of the same frequency f, forming a periodic current change with f as the fundamental frequency: ; Where I0 is the average current under both states, and ΔI is half of the current switching amplitude.

[0043] The above process transforms static illumination into a dynamic excitation source and establishes a traceable frequency domain reference within the component, providing clear phase synchronization conditions for subsequent phase-locked demodulation and ensuring accurate conversion and extraction from photogenerated current to magnetic field signals.

[0044] In the example above, the LED array is periodically switched on and off by the PWM drive controller to precisely regulate the switching between high and low light intensity. This causes the incident light intensity to be superimposed with a periodic fluctuation component with an amplitude of ΔL on the average value, so that the photocurrent generated inside the photovoltaic module synchronously carries an AC component with the same frequency f.

[0045] In the above optional embodiments, the method of outputting a target excitation signal that periodically changes at a target modulation frequency to the photovoltaic module by modulating the excitation source may further include: modulating a load switching circuit electrically connected to the photovoltaic module according to a preset pulse signal so that the output current of the photovoltaic module changes periodically at the target modulation frequency.

[0046] Specifically, when photovoltaic modules continuously generate electricity under natural sunlight, the light intensity fluctuates randomly due to environmental factors such as cloud cover and wind speed. This fluctuation frequency is unpredictable and cannot be directly used as a reference for phase-locked loop (PLL) detection. This embodiment can also use a PWM drive controller to periodically control the switching of the MOSFET controllable load circuit, ensuring that the electrical signal output by the module synchronously carries an AC component consistent with the excitation frequency. This modulation method based on a preset pulse signal transforms the static power generation current into a dynamic excitation source. This not only provides a traceable frequency domain reference for subsequent PLL detection, effectively solving the problem of environmental noise interference, but also accurately extracts the weak signal generated by local current density distortion, thereby achieving quantitative evaluation in online monitoring scenarios.

[0047] For example, in an outdoor power station online monitoring scenario, a controllable load switching circuit composed of MOSFETs is connected in series with the module output circuit. A PWM drive controller periodically switches the switching circuit on and off at a preset frequency f, causing the photovoltaic module's output current to periodically switch between two operating points. This eliminates the need for an additional light source, relying entirely on the module's own power generation current for excitation, without interfering with the normal power generation operation of the photovoltaic module. The modulated current expression is: ; Where I0 is the average current in both states, ΔI is half of the current switching amplitude, and sq(•) represents the square wave function.

[0048] Since photovoltaic modules continuously generate electricity under natural light, the photocurrent they produce also carries an AC component with the same frequency f. This signal serves as a reference for subsequent phase-locked demodulation, thus effectively overcoming measurement errors caused by environmental noise interference and lift-off effects.

[0049] In the above example, the PWM drive controller periodically controls the switching of the MOSFET controllable load circuit, and uses a square wave function to precisely regulate the switching of the output current between two operating points. This modulation method based on a preset pulse signal enables the electrical signal output by the photovoltaic module to synchronously carry an AC component consistent with frequency f, thereby converting the static power generation current into a dynamic excitation source. This provides a traceable frequency domain reference for subsequent phase-locked loop (PLL) detection, effectively solving the problem of environmental noise interference. It also accurately extracts weak signals generated by local current density distortion, enabling quantitative evaluation in online monitoring scenarios.

[0050] To achieve defect signal extraction in online monitoring scenarios while avoiding environmental interference and sensor noise, the modulation frequency f can be subject to the following three constraints: First, it should avoid the 50Hz power frequency and its harmonic frequencies to prevent false detection due to power grid background noise; second, it should be lower than the current response cutoff frequency determined by the time constant of the solar cell (approximately several hundred Hz to several kHz) to ensure that the excitation signal can fully reflect the internal dynamic characteristics of the module without distortion; third, it should be placed in the white noise range above the 1 / f noise corner frequency of the TMR sensor to maximize the signal-to-noise ratio. Considering these constraints, the recommended operating range for the modulation frequency f is 100Hz to 1kHz. Within this frequency band, it can stably extract weak magnetic field signals generated by local current density distortion, effectively filtering out background noise such as static geomagnetism, low-frequency power grid noise, and metal structure induction, providing a reference benchmark for subsequent phase-locked demodulation.

[0051] This application embodiment achieves flexible control of the excitation source of photovoltaic modules by controlling the light source intensity or the switching circuit. It can dynamically adjust the excitation mode according to the application scenario requirements. It can adapt to the control of light intensity waveform in offline full inspection scenarios, and can also switch to the modulation of current waveform characteristics in online operation and maintenance scenarios. Thus, it can continuously provide high-quality dynamic input for subsequent magnetic field signal extraction without interrupting power generation operation, and improve the ability to perceive defect signals in complex environments.

[0052] In this embodiment of the application, a sensing unit is provided on one side of the photovoltaic module with a backsheet. The sensing unit includes multiple tunnel magnetoresistive (TMR) sensors. The multiple TMR sensors overlap in a first direction, and any two adjacent TMR sensors in the first direction have a height difference. The first direction is the thickness direction of the backsheet.

[0053] For example, the sensing unit includes multiple detection pixel units, each of which includes a first TMR sensor S1 and a second TMR sensor S2. The projection positions of the two sensors in the XY plane coincide, and there is a fixed height difference Δh in the Z direction (the thickness direction of the backsheet in the photovoltaic module). S1 is closer to the backsheet, and S2 is farther from the backsheet. The two sensors output magnetic field measurement values ​​B1(x,y,t) and B2(x,y,t) at their respective heights.

[0054] In some optional implementations, the spatial vertical gradient of the magnetic field generated by the current is extracted based on the spatial magnetic field distribution of the photovoltaic module, including: performing differential processing on the spatial magnetic field distribution of multiple tunnel magnetoresistive (TMR) sensors to obtain differential results, and using the differential results as the spatial vertical gradient of the magnetic field.

[0055] Due to uneven backsheet thickness, installation deformation, and thermal expansion and contraction during photovoltaic module operation, the actual lift-off height of the sensor continuously fluctuates. Furthermore, the near-field magnetic field strength varies significantly with distance (attenuating approximately two to three times the distance). Even small changes in lift-off height introduce non-negligible systematic measurement errors. Therefore, this application proposes a dual-height differential sensing structure. Specifically, in each detection pixel unit, two TMR sensors with a fixed height difference are arranged vertically. Their output signals are differentially processed. The differential result is equivalent to extracting the spatial vertical gradient of the magnetic field. The gradient is more sensitive to local current anomalies in the near field and has a natural common-mode suppression capability against the uniform background magnetic field in the far field. Thus, the actual lift-off height can be inverted in real time through the measurement ratio of the two sensors, allowing for dynamic correction of the magnetic field measurement value and achieving adaptive compensation for the lift-off effect.

[0056] In the above optional implementation, the common-mode effect generated by the uniform background magnetic field in the far field is generated by the dual height difference structure, which completely cancels the background noise. Thus, only the significant difference signal caused by the local current distortion in the near field is retained in the extracted spatial vertical gradient, and the differential output of the TMR sensor is converted into a quantitative index reflecting the change of current density vector.

[0057] Specifically, by differentially processing the spatial magnetic field distribution of multiple tunnel magnetoresistive sensors, the dual height difference structure can be used to naturally suppress the common mode of the uniform background magnetic field in the far field. This makes the result after differential operation approach zero while retaining the significant difference signal generated by the local current anomaly in the near field. This provides an effective input for subsequent frequency domain coding and synchronous demodulation, thereby ensuring that subsequent steps can accurately extract weak defect signals and achieve early identification.

[0058] For example, the first TMR sensor S1 and the second TMR sensor S2, which have a height difference Δh, output the magnetic field measurement values ​​B1(x,y,t) and B2(x,y,t) at their respective heights, as follows: Figure 3 As shown, by performing a difference operation on the two, the estimated value of the spatial vertical gradient of the magnetic field is obtained: .

[0059] The aforementioned height difference Δh ranges from 0.2mm to 2.0mm. The lower limit of Δh can be determined by the precision of the sensor packaging process. When it is below 0.2mm, the difference in the spatial sampling positions of the two sensors is small, which may affect the generation of effective gradient resolution. The upper limit of Δh can be determined by the spatial correlation length of the near-field signal. When it is above 2.0mm, the difference in the magnetic field response of the two sensors to the same defect source is large, which may cause the differential result to deviate from the linear gradient approximation, introducing nonlinear errors. Within the working range of lift-off height h = 1.0 to 2.0mm, the value of Δh can range from 25% to 75% of h.

[0060] It should be noted that the embodiments of this application are not limited to the case of two TMR sensors in the Z direction, but can also be a multi-sensor array that is uniformly or non-uniformly distributed along the Z direction, and high-order gradient extraction can be achieved through differential extraction of adjacent layers or weighted differential extraction.

[0061] In the above optional implementation, a definite functional relationship between the differential ratio of the sensor array and the lift-off height h is established based on the gradient-distance mapping model. This allows the actual current distance to be deduced based on the sensor's geometric spacing and material properties. During actual detection, when backplane deformation or environmental temperature changes cause fluctuations in the distance between the sensor and the battery layer, the differential ratio ΔB / B1 will change accordingly. Using a pre-calibrated mapping model, the current differential ratio is inverted back to the actual lift-off height h, thereby achieving dynamic correction of the original magnetic field measurement value. This process effectively offsets the lift-off error caused by backplane deformation or thermal expansion and contraction, ensuring that the defect signal extracted based on the magnetic field gradient is not affected by environmental interference.

[0062] For example, in the calibration, a standard straight conductor with known current is used as a reference source. A precision displacement stage is used to gradually change the lifting height of the sensor array within the range of 0.5 to 3.0 mm. The ΔB / B1 ratio at each height is recorded, and the gradient-distance mapping function is fitted and stored as the system calibration parameters.

[0063] Multiple TMR sensors are arranged in an array. When the center-to-center distance between adjacent TMR sensors is 2.0 to 2.5 mm, the positioning accuracy varies with the lift-off height as follows: when the lift-off height is 1.0 mm, the positioning accuracy is approximately 3.0 to 5.0 mm; when the lift-off height is 1.5 mm, the positioning accuracy is approximately 4.0 to 7.0 mm; and when the lift-off height is 2.0 mm, the positioning accuracy is approximately 5.0 to 10 mm.

[0064] A TMR sensor array is deployed on the backsheet side of the photovoltaic module to collect the spatial magnetic field distribution caused by local current density distortion within a near-field sensing range of 1.0–2.0 mm from the backsheet. By modulating the excitation of the light source / controllable modulation of the MOSFET, the defect-related magnetic field signal carries specific frequency characteristics. Combined with digital phase-locked loop detection, frequency domain separation between the target signal and the environmental background is achieved. The spatial gradient of the magnetic field is extracted through a dual-height differential sensing structure. When the photovoltaic module is in normal power generation, the background magnetic field (such as the geomagnetic field or grid interference) exhibits a uniform distribution on the sensor array. At this time, differential operation is performed on the outputs of adjacent sensors. The far-field uniform component is canceled out, and only the local gradient change caused by the interruption or detour of the near-field current path is retained. The final differential result is the demodulated data that can accurately reflect the location and degree of the defect.

[0065] The aforementioned TMR sensor array can be integrated onto a circuit board (such as a PCB substrate) using a bare-chip direct mounting method. To ensure that the distance between the TMR sensor array and the photovoltaic module backplane remains stable within the near-field sensing range, in some optional implementations, such as... Figure 4 As shown, the sensing unit also includes a rigid limiting layer 100, which is disposed on one side surface of the photovoltaic module having a backplate 200, and is used to separate the rigid limiting layer 100 from the circuit board 300 on which TMR sensors (such as a first TMR sensor S1 and a second TMR sensor S2) are disposed.

[0066] In the above optional embodiments, the rigid limiting layer can be made of low-permeability alumina ceramic (relative permeability ≈ 1) to avoid introducing additional magnetic field distortion; the nominal thickness can be 0.5 mm, which, in conjunction with the backplate thickness tolerance (typical value 0–0.5 mm), maintains the actual distance between the closest end of the sensor (S1) and the battery layer within the range of 0.5–1.0 mm. The rigid limiting layer can simultaneously prevent physical contact between the sensor array and the backplate and protect the device.

[0067] In the above optional implementations, such as Figure 5As shown, the sensing unit may also include a multiplexing circuit 400, which sequentially acquires the output signals of each channel sensor (S1 / S2) according to a preset timing sequence. The acquisition period must meet the timing requirement of completing the full array scan within the preset modulation period.

[0068] Taking a modulation frequency of f=300Hz and an array size of 40×24=960 pixel units as an example, one modulation period T=1 / 300≈3.3ms. Using an 8-channel parallel multiplexing circuit, 8 channels are acquired simultaneously each time, requiring a total of 960 / 8=120 gating switches. Each switch includes approximately 5μs of multiplexing circuit stabilization time and approximately 83μs of ADC single sampling time (corresponding to a sampling rate of 12kHz), for a total switching time of approximately 88μs. The total time for 120 switches is approximately 10.6ms. This duration exceeds the 3.3ms of a single modulation period; therefore, the actual acquisition spans approximately 4 modulation periods to complete the full array scan. Phase-locked demodulation of each pixel is performed independently within its corresponding acquisition time window, without affecting demodulation accuracy.

[0069] In some optional implementations, the spatial vertical gradient is synchronously demodulated according to a reference clock signal corresponding to the target modulation frequency to obtain a demodulation result, including: using the reference clock signal as a phase reference, extracting the target amplitude and target phase at the target modulation frequency from the differential result.

[0070] In the above optional implementation, by performing synchronous demodulation processing using a reference clock signal as the phase reference, the target amplitude and corresponding target phase information that are strictly synchronized with the target modulation frequency can be accurately extracted from the differential results. The target amplitude reflects the magnetic field component that reflects the local current density vector distortion. This not only effectively filters out non-target background components such as static geomagnetic field, low-frequency power grid interference, and asynchronous broadband noise, but also retains the weak phase characteristics carried by the near-field defect signal. The extracted target amplitude and target phase together constitute the demodulation result that can characterize the local current density vector distortion, providing input data for subsequent inversion and reconstruction using a forward physical model. This ensures that the two-dimensional current density vector field inside the solar cell can be accurately reconstructed, achieving precise quantitative imaging characterization of the defect location, morphology, and severity.

[0071] Specifically, when the photovoltaic module is in normal power generation, the background magnetic field (such as the geomagnetic field or power grid interference) exhibits a uniform distribution on the sensor array. At this time, the output reference clock is the phase reference. The signal component that is strictly synchronized with the frequency is selected from the differential results, and all asynchronous broadband noise and low-frequency interference are ignored. The final demodulation result is the data that can accurately reflect the location and extent of the defect.

[0072] It should be noted that the specific implementation of synchronous demodulation in the embodiments of this application is not limited to specific technical means, including but not limited to digital quadrature phase-locked demodulation, Fast Fourier Transform (FFT) narrowband spectrum extraction and correlation integration method, signal processing methods that can selectively extract the target signal at frequency f and whose equivalent detection bandwidth is narrower than 10% of the modulation frequency f are all within the protection scope of the embodiments of this application.

[0073] In some optional implementations, the demodulation results are processed by image reconstruction to obtain the current density distribution of the photovoltaic module, including: establishing a linear mapping relationship between the spatial vertical gradient of the magnetic field and the current density distribution based on the target amplitude and target phase; constructing a system response matrix based on the Biot-Savart law and sensor geometric parameters; and using the system response matrix, the linear mapping relationship, and matrix multiplication operations to invert the spatial vertical gradient of the magnetic field to obtain the current density vector field.

[0074] In the above optional implementation, converting the weak magnetic field signal after frequency-domain synchronous demodulation into a current density distribution map can include two key processes: establishing a physical model and constructing a mathematical mapping. First, using the target amplitude and phase as anchor points, a linear mapping relationship is established between the spatial vertical gradient and the current density. Second, based on the Biot-Savart law and sensor geometric parameters, a system response matrix is ​​constructed, and the inversion from magnetic field to current density is completed through matrix multiplication. This linear mapping relationship ensures that the weak defect signal does not lose key feature information during reconstruction and transforms abstract physical quantities into specific two-dimensional vector fields. This not only eliminates common-mode interference caused by height differences in the sensor array but also compensates for lift-off errors through differential ratios. Ultimately, the originally imperceptible millivolt-level magnetic field signal is accurately reconstructed into a clear current density distribution map, which can then be used to accurately locate defect areas such as broken grids, microcracks, and hot spots.

[0075] For example, when the amplitude and phase of a pixel are detected to be strictly synchronized with a preset frequency f, the current density value corresponding to that point is calculated according to the Biot-Savart law and filled into the corresponding matrix unit. Pixels in the surrounding area that do not respond due to lack of defects or weak signals are set to zero. After matrix multiplication, the original magnetic field data is transformed into a visualized defect characterization image, where highlighted areas represent current path interruptions (broken grids), and color variations represent the degree of current density attenuation (hidden cracks), thus intuitively displaying the internal defects of the photovoltaic module.

[0076] In the example above, when the amplitude and phase of a pixel are detected to be strictly synchronized with a preset frequency f, the local current density value corresponding to that point is calculated according to the Biot-Savart law. If a defect exists around the point causing the current path to be interrupted or detoured, the resulting magnetic field signal will deviate significantly from the background noise, thus being extracted as a highlight area. For areas without obvious fractures but with microcracks, the defect causes the local current density to attenuate rather than be completely interrupted. By calculating the ratio or difference between the response amplitude of that point and the surrounding normal area, the weak abnormal signal is converted into a specific color level. The darker the color, the lower the current density and the more severe the defect; conversely, the lighter the color, the higher the health level. Therefore, the contrast in color depth reflects the location and severity of the defect area (broken grid / microcrack).

[0077] In some optional implementations, the defect region of the photovoltaic module is determined based on the current density distribution of the photovoltaic module, including: calculating the deviation between the amplitude of the current density vector at the target pixel in the current density vector field and the average amplitude of the current density vector at all pixels in the preset neighborhood of the target pixel; and determining the region where the deviation exceeds a preset threshold as the defect region.

[0078] In the above optional embodiments, the defect areas of photovoltaic modules are accurately located by using the local anomaly characteristics of the current density vector field through quantitative analysis. This can include calculating the deviation between the magnitude (i.e., absolute value) of the current density of a pixel and the average value of its surrounding neighborhood, and using this as a judgment criterion. When the current density vector of a certain point deviates from its neighborhood average value by more than a preset threshold, that point is determined to be a defect area. This effectively captures the weak leakage current or path distortion caused by early defects such as microcracks, grid line breaks, or hot spot short circuits, and accurately locates the abnormal position. Compared with the traditional image comparison method, the embodiments of this application do not rely on the global luminous intensity, but focus on the small perturbations of local electrical characteristics, effectively overcoming the problem that weak light signals caused by interface defects, weak leakage currents, and early degradation are difficult to identify.

[0079] For example, when a microcrack with a width of only 0.5 mm exists on the surface of a photovoltaic module, this crack causes a slight shift or detour in the local current path, resulting in the current density vector magnitude of a certain pixel in that region being significantly higher than the average value of the surrounding normal area. Based on the Biot-Savart law, a forward physical model is established, and the spatial magnetic field gradient data after dual-height difference is reconstructed into a two-dimensional current density vector field inside the cell using an inversion algorithm. Subsequently, defect identification is performed on the reconstructed current density vector field J(x,y): the deviation between the current density amplitude of each pixel and the mean of its neighborhood is calculated. If the deviation value of a specific pixel exceeds a preset threshold, it is marked as a potential defect area. During this process, combined with the characteristics of abrupt changes in current direction, typical defect patterns such as grid breakage, microcracks, and hot spots are further distinguished. Finally, the above identification results are directly mapped to a visualization interface, visually displaying the boundary position and morphology of the microcrack in color, with corresponding type labels superimposed, thus clearly presenting internal defects that are difficult to detect with the naked eye without the need for additional light source interference.

[0080] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the defect detection method for photovoltaic modules of this application will be described in detail below with reference to specific embodiments.

[0081] This embodiment relates to a specific method for defect detection of photovoltaic modules, including the following steps: Step S1: The controller sends a synchronization signal to the LED array or MOSFET controllable switch, modulates the external excitation source with a preset frequency, and transmits the magnetic field signal containing the target current density to the acquisition unit. Step S2: The TMR sensor array synchronously acquires the spatial magnetic field distribution data B1(x,y,t) and B2(x,y,t) of two adjacent sensors in real time; The TMR sensor array synchronously acquires the spatial magnetic field distribution data B1(x,y,t) and B2(x,y,t) of two adjacent sensors in real time to obtain the difference ratio ΔB / B1. Step S3: Perform difference operation: ; Based on the dual-height differential sensing structure, common-mode suppression is performed on the far-field uniform background magnetic field, and only the significant difference signal generated by the near-field local current anomaly is retained, thereby obtaining demodulated data reflecting the location and extent of the defect. Step S4: Extract the amplitude distribution A(x,y) and phase distribution from the demodulation results. (x,y) is further eliminated by digital phase-locked loop detection to improve the signal-to-noise ratio by eliminating environmental noise interference. Step S5: Based on the extracted amplitude and phase information, the local current density vector field J(x,y) is inverted and solved to realize the physical reconstruction from magnetic field measurement to current density; Step S6: Perform defect analysis, classification, and visualization output on the inverted current density distribution, and finally generate a detection report containing the defect location, level, and severity.

[0082] This application also provides a defect detection system for photovoltaic modules. It should be noted that the photovoltaic module defect detection system of this application can be used to execute the defect detection method for photovoltaic modules provided in this application. This system is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0083] The following describes the defect detection system for photovoltaic modules provided in the embodiments of this application.

[0084] Figure 6 This is a schematic diagram of a defect detection system for photovoltaic modules according to an embodiment of this application. Figure 6 As shown, the system includes: Excitation unit 10 is used to modulate the photovoltaic module so that the current of the photovoltaic module carries an AC component of the target modulation frequency. The sensing unit 20 is located on the side of the photovoltaic module with a backsheet and is used to sense the spatial magnetic field distribution of the photovoltaic module within a preset time period. The sensing unit includes multiple tunnel magnetoresistive sensors, which overlap in a first direction and any two adjacent tunnel magnetoresistive sensors in the first direction have a height difference. The first direction is the thickness direction of the backsheet. The preprocessing unit 30 is used to extract the spatial vertical gradient of the magnetic field generated by the current according to the spatial magnetic field distribution of the photovoltaic module, and to perform synchronous demodulation processing on the spatial vertical gradient according to the reference clock signal corresponding to the target modulation frequency, so as to extract the demodulation result reflecting the local current density vector distortion. The spatial vertical gradient is used to characterize the rate of change of the magnetic field strength in the first direction. The image reconstruction unit 40 is used to perform image reconstruction processing on the demodulation results to obtain the current density distribution of the photovoltaic module, and to determine the defect area of ​​the photovoltaic module based on the current density distribution of the photovoltaic module.

[0085] In this embodiment, the sensing unit first acquires the spatial magnetic field distribution of the photovoltaic module sensed by multiple tunnel magnetoresistive sensors within a preset time period. This step utilizes the response characteristics of the tunnel magnetoresistive sensors to alternating magnetic fields to capture the AC component of the current carrying the target modulation frequency caused by defects in the photovoltaic module under the action of the excitation unit. The above response does not depend on external light sources or thermal effects, but is based on the interruption or detour of local current paths caused by defects, so that the detection signal directly corresponds to the vector distortion of local current density, thereby eliminating the sensitivity bottleneck caused by weak light intensity. Subsequently, the preprocessing unit constructs a spatial gradient difference structure based on the spatial height difference between adjacent sensors, so that the far field... The background contribution approaches zero while preserving the near-field difference signal, thereby eliminating the effects of lift-off error and environmental noise. This ensures the stability and repeatability of the detection results in complex outdoor environments and directly improves the signal-to-noise ratio and detection sensitivity. Next, the preprocessing unit uses a reference clock as the phase reference to synchronously demodulate the extracted spatial vertical gradient, converting static interference into a filterable background component. This yields a demodulated result for local current density vector distortion, enabling early defect response without interrupting operation and through non-contact sensing. Finally, the image reconstruction unit performs image reconstruction based on the demodulation result, generating a current density distribution map and determining the defect region accordingly. Therefore, this solution, through the above collaborative steps, solves the problem that existing photovoltaic module defect detection technologies cannot simultaneously meet the comprehensive requirements of outdoor online, non-contact, and high-sensitivity detection. It realizes the transformation from laboratory research to engineering applications and improves the stability and accuracy of online detection of photovoltaic modules in complex environments.

[0086] In this embodiment, the excitation unit is used to make the internal current of the solar cell carry an AC component of a preset frequency f, thereby providing a traceable frequency domain reference for subsequent phase-locked detection. Depending on the application scenario, the physical composition of the excitation unit can be divided into offline full-inspection mode and online monitoring mode.

[0087] The above-mentioned offline full inspection mode is applicable to offline inspection scenarios. The excitation unit of the offline full inspection mode may include: a light-emitting device module, a first pulse width modulation controller, and a synchronous clock source. The light-emitting device module is used to output the target light source, the first pulse width modulation controller is used to modulate the target light source according to a preset pulse signal so that the light intensity applied to the photovoltaic module by the target light source changes periodically with the target modulation frequency, and the synchronous clock source is used to provide a reference clock signal corresponding to the target modulation frequency.

[0088] For example, the excitation unit in the offline full inspection mode includes: a high-frequency modulated LED array, a PWM drive controller, and a synchronous clock output interface; in offline testing scenarios such as factory inspection, the photovoltaic module under test is in a controlled environment, and the PWM drive controller periodically controls the LED array to switch on and off at a preset frequency f, so that the output light intensity periodically switches between two light intensities, forming a periodic light intensity change at frequency f: ; Where sq(•) represents the square wave function, L0 is the average light intensity of the two states, and ΔL is half of the light intensity switching amplitude.

[0089] Under the aforementioned modulated illumination excitation, the photocurrent inside the photovoltaic module exhibits periodic changes at the same frequency: ; Where I0 is the average current under both states, and ΔI is half of the current switching amplitude.

[0090] The above process transforms static illumination into a dynamic excitation source and establishes a traceable frequency domain reference within the component, providing clear phase synchronization conditions for subsequent phase-locked demodulation and ensuring accurate conversion and extraction from photogenerated current to magnetic field signals.

[0091] The above-mentioned online monitoring mode is applicable to online monitoring scenarios. The excitation unit of the online monitoring mode may include: a load switching circuit, a second pulse width modulation controller, and a synchronous clock source. The load switching circuit is electrically connected to the photovoltaic module. The second pulse width modulation controller is used to modulate the load switching circuit according to a preset pulse signal so that the output current of the photovoltaic module changes periodically at the target modulation frequency. The synchronous clock source is used to provide a reference clock signal corresponding to the target modulation frequency.

[0092] For example, the excitation unit includes: a controllable load switching circuit, a PWM drive controller, and a synchronous clock output interface. In the scenario of online monitoring of outdoor power plants, photovoltaic modules continuously generate electricity under natural sunlight. The intensity of sunlight is affected by environmental factors such as cloud cover and wind speed, resulting in random fluctuations. The frequency of these fluctuations is not fixed and is uncontrollable, making it unsuitable as a direct reference for phase-locked loop (PLL) detection. Therefore, in the online monitoring scenario, output current modulation is used to replace light source excitation. A controllable load switching circuit composed of MOSFETs is connected in series in the module output circuit. The PWM drive controller periodically switches the switching circuit on and off at a preset frequency f, causing the module output current to periodically switch between two operating points, forming a periodic current change with frequency f as the fundamental frequency. ; Where I0 is the average current in both states, ΔI is half of the current switching amplitude, and sq(•) represents the square wave function.

[0093] The above online monitoring mode requires no additional light source and relies entirely on the power generation current of the module itself for excitation, without interfering with the normal power generation operation of the module.

[0094] To achieve defect signal extraction in online monitoring scenarios while avoiding environmental interference and sensor noise, the modulation frequency f can be subject to the following three constraints: First, it should avoid the 50Hz power frequency and its harmonic frequencies to prevent false detection due to power grid background noise; second, it should be lower than the current response cutoff frequency determined by the time constant of the solar cell (approximately several hundred Hz to several kHz) to ensure that the excitation signal can fully reflect the internal dynamic characteristics of the module without distortion; third, it should be placed in the white noise range above the 1 / f noise corner frequency of the TMR sensor to maximize the signal-to-noise ratio. Considering these constraints, the recommended operating range for the modulation frequency f is 100Hz to 1kHz. Within this frequency band, it can stably extract weak magnetic field signals generated by local current density distortion, effectively filtering out background noise such as static geomagnetism, low-frequency power grid noise, and metal structure induction, providing a reference benchmark for subsequent phase-locked demodulation.

[0095] The aforementioned PWM drive controller synchronously outputs a reference clock signal that is strictly in phase with the load modulation, which is used by the phase-locked loop detection module of the preprocessing unit.

[0096] In this embodiment, the sensing unit is used to collect the spatial magnetic field distribution B(x,y,t) and its vertical gradient information in the near-field region on the backsheet side of the battery, providing raw measurement data for subsequent signal processing; the sensing unit may include a TMR sensor array, a circuit board, a rigid limiting layer, and a multiplexing circuit.

[0097] For example, the sensing unit includes multiple detection pixel units, each containing a first TMR sensor S1 and a second TMR sensor S2. Their projected positions in the XY plane coincide, and they have a fixed height difference Δh in the Z direction (the thickness direction of the backsheet in the photovoltaic module). S1 is closer to the backsheet, and S2 is farther away. The two sensors output magnetic field measurements B1(x,y,t) and B2(x,y,t) at their respective heights, as shown below. Figure 3 As shown, by performing a difference operation on the two, the estimated value of the spatial vertical gradient of the magnetic field is obtained: .

[0098] Specifically, for local current anomalies (such as defect areas) close to the sensor, the magnetic field attenuates drastically with distance, resulting in significant differences between the magnetic fields measured by the two sensors and a large differential result. For uniform background magnetic fields in the far field (such as the Earth's magnetic field or the magnetic field generated by the metal support), the magnetic fields measured by the two sensors are approximately equal, and the differential result approaches zero. This achieves selective enhancement of near-field defect signals and common-mode suppression of far-field background interference. For a standard current distribution with a known geometry, there is a definite functional relationship between the near-field magnetic field gradient and the lift-off height. By establishing a gradient-distance mapping model through pre-calibration, the actual lift-off height h can be estimated in real time using the differential ratio ΔB / B1, and the magnetic field measurement value can be dynamically corrected to compensate for lift-off errors caused by backplate deformation and thermal expansion and contraction.

[0099] For example, in the calibration, a standard straight conductor with known current is used as a reference source. A precision displacement stage is used to gradually change the lift-off height of the sensor array within the range of 0.5 to 3.0 mm. The ΔB / B1 ratio at each height is recorded, and the gradient-distance mapping function is fitted and stored as the system calibration parameters.

[0100] The height difference Δh between the first TMR sensor S1 and the second TMR sensor S ranges from 0.2 mm to 2.0 mm. The lower limit of Δh can be determined by the precision of the sensor packaging process. When it is below 0.2 mm, the spatial sampling position difference between the two sensors is small, which may affect the generation of effective gradient resolution. The upper limit of Δh can be determined by the spatial correlation length of the near-field signal. When it is above 2.0 mm, the magnetic field response of the two sensors to the same defect source is significantly different, which may cause the differential result to deviate from the linear gradient approximation and introduce nonlinear error. Within the working range of lift-off height h = 1.0 to 2.0 mm, the value of Δh can range from 25% to 75% of h.

[0101] It should be noted that the embodiments of this application are not limited to the case of two sensors, but can also be a multi-sensor array that is uniformly or non-uniformly distributed along the Z direction, and high-order gradient extraction can be achieved through differential extraction between adjacent layers or weighted differential extraction.

[0102] In some optional embodiments, the sensing unit further includes: a circuit board on which the tunnel magnetoresistive sensor is disposed; a rigid limiting layer disposed on one side surface of the photovoltaic module having a backsheet, for separating the rigid limiting layer from the tunnel magnetoresistive sensor; and a multiplexing circuit for sequentially acquiring the output signal of the tunnel magnetoresistive sensor according to a preset timing sequence.

[0103] Specifically, the aforementioned TMR sensor array can be integrated onto a circuit board (such as a PCB substrate) by mounting. In order to ensure that the distance between the TMR sensor array and the backsheet of the photovoltaic module is stable within the near-field sensing range, a rigid limiting layer 100 is disposed on one side surface of the photovoltaic module with the backsheet 200, which is used to separate the rigid limiting layer 100 from the circuit board 300 on which TMR sensors (such as the first TMR sensor S1 and the second TMR sensor S2) are disposed.

[0104] In the above optional embodiments, the rigid limiting layer can be made of low-permeability alumina ceramic (relative permeability ≈ 1) to avoid introducing additional magnetic field distortion; the nominal thickness can be 0.5 mm, which, in conjunction with the backplate thickness tolerance (typical value 0–0.5 mm), maintains the actual distance between the closest end of the sensor (S1) and the battery layer within the range of 0.5–1.0 mm. The rigid limiting layer can simultaneously prevent physical contact between the sensor array and the backplate and protect the device.

[0105] Multiple TMR sensors can be integrated into an array on a circuit board (such as a PCB substrate) using a bare-chip direct mounting method, with a center-to-center spacing of 2.0–2.5 mm between adjacent TMR sensors. Figure 5 As shown, the multiplexing circuit 400 sequentially acquires the output signals of each channel sensor (S1 / S2) according to a preset timing sequence. The acquisition period must meet the timing requirement of completing the full array scan within the preset modulation period.

[0106] Taking a modulation frequency of f=300Hz and an array size of 40×24=960 pixel units as an example, one modulation period T=1 / 300≈3.3ms. Using an 8-channel parallel multiplexing circuit, 8 channels are acquired simultaneously each time, requiring a total of 960 / 8=120 gating switches. Each switch includes approximately 5μs of multiplexing circuit stabilization time and approximately 83μs of ADC single sampling time (corresponding to a sampling rate of 12kHz), for a total switching time of approximately 88μs. The total time for 120 switches is approximately 10.6ms. This duration exceeds the 3.3ms of a single modulation period; therefore, the actual acquisition spans approximately 4 modulation periods to complete the full array scan. Phase-locked demodulation of each pixel is performed independently within its corresponding acquisition time window, without affecting demodulation accuracy.

[0107] In this embodiment of the application, the preprocessing unit may include: a high-speed analog-to-digital converter (ADC), a field-programmable gate array (FPGA) or a microcontroller unit (MCU) and a synchronous demodulation module.

[0108] The aforementioned preprocessing unit performs high-speed sampling and synchronous demodulation of the analog magnetic field signal output by the TMR sensor. Using the reference clock output by the excitation unit as the phase reference, it selectively extracts the amplitude A(x,y) and phase at the modulation frequency f from the differential magnetic field signal ΔB(x,y,t). (x,y) is used to suppress broadband noise and asynchronous interference outside the frequency f, which is then used by the image reconstruction unit. In some optional implementations, the ADC sampling rate is not less than 20 times the modulation frequency f.

[0109] The specific implementation of the above-mentioned synchronous demodulation module is not limited to specific technical means, including but not limited to digital quadrature phase-locked demodulation, Fast Fourier Transform (FFT) narrowband spectrum extraction and correlation integration method, signal processing methods that can selectively extract the target signal at frequency f and whose equivalent detection bandwidth is narrower than 10% of the modulation frequency f, are all within the protection scope of the embodiments of this application.

[0110] In this embodiment, the image reconstruction unit may include: a computational processing module, a forward physical model, an inversion algorithm module, a defect identification model, and a visualization interface. Based on the magnetic field gradient amplitude distribution A(x,y), the image reconstruction unit reconstructs the two-dimensional current density vector field J(x,y) inside the battery through a forward physical model and an inversion algorithm, and identifies and locates defects in abnormal regions.

[0111] Specifically, the aforementioned forward physical model establishes a linear mapping relationship between the magnetic field gradient distribution ΔB(x,y) and the current density distribution J(x,y) based on the Biot–Savart law. The system response matrix K is pre-calculated offline and stored. The inversion solution employs a linear inversion method based on regularization constraints. The regularization parameters are also determined offline and directly invoked during online detection, without consuming real-time computing resources. The specific algorithm implementation is not limited to a particular method; any inversion method capable of stably solving the underdetermined magnetic field-current inverse problem and achieving reconstruction errors that meet positioning accuracy requirements falls within the protection scope of this application's embodiments.

[0112] Specifically, the defect identification model is used to identify defects in the reconstructed current density vector field J(x,y). The deviation between the current density amplitude of each pixel and the mean of the neighborhood is calculated, and the area with the deviation exceeding the preset threshold is marked as a potential defect area. Combined with the current direction change feature, typical defect patterns such as broken grid (current path interruption), hidden crack (current detour) and hot spot (local current density abnormal concentration) are classified and identified. The current density amplitude distribution is displayed in the form of a color map through a visualization interface, and the location label and type label of the defect identification result are superimposed.

[0113] The photovoltaic module defect detection system includes a processor and a memory. The aforementioned modules are stored as program units in the memory, and the processor executes these program units to achieve the corresponding functions. All the aforementioned modules reside in the same processor; alternatively, the aforementioned modules may be located in different processors in any combination.

[0114] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and by adjusting kernel parameters, the limitations of current photovoltaic module defect detection technologies in simultaneously meeting the comprehensive requirements of outdoor online, non-contact, and high-sensitivity testing can be addressed.

[0115] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0116] This application provides a computer-readable storage medium including a stored program, wherein the program, when running, controls the device where the computer-readable storage medium is located to execute a defect detection method for photovoltaic modules.

[0117] This application provides a processor for running a program, wherein the program executes a defect detection method for photovoltaic modules.

[0118] This application provides an apparatus including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in a method for detecting defects in photovoltaic modules. The apparatus described herein can be a server, PC, tablet, mobile phone, etc.

[0119] This application also provides a computer program product that, when executed on a data processing device, is adapted to perform a program that initializes the steps of a defect detection method having at least photovoltaic modules.

[0120] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0121] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0122] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0123] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0124] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0125] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0126] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0127] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0128] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0129] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0130] As can be seen from the above description, the embodiments of this application achieve the following technical effects: 1) This application covers the full life cycle testing needs through two switchable excitation modes: the offline full inspection scenario uses a target light source such as a modulated LED array to apply periodic light intensity excitation, which is suitable for production line off-line inspection; the online monitoring scenario uses a controllable load switching circuit such as a MOSFET to periodically disturb the component operating point, which does not require an additional light source and does not interfere with normal power generation operation, and is suitable for long-term online deployment in power plants that have been put into operation. The two modes share the magnetic field acquisition, synchronous demodulation, inversion reconstruction and defect identification process, with only the excitation unit being different, and the deployment is flexible.

[0131] 2) In this application, the tunnel magnetoresistive sensor is deployed on the back panel of the module. It does not need to establish an electrical connection with the module. The magnetic field signal can directly penetrate the encapsulation layer and back panel of the photovoltaic module and other non-magnetic materials to achieve non-contact detection. In the online monitoring mode, the load switching circuit required for excitation is connected in series with the output circuit of the module. It only applies periodic disturbance to the output operating point and does not affect the normal power generation operation of the module. It does not need to be connected to an external power source. 3) In this application, magnetic field signals are used as the detection carrier. According to the Biot-Savart law, the change in current density distribution is instantly mapped to the change in the surrounding spatial magnetic field. No medium conduction or thermal diffusion delay is required. For early defects such as initial microcracks or early grid line damage that cause slight distortion of local current, they can be detected before the thermal effect is significant, and have a defect response capability earlier than infrared thermal imaging.

[0132] 4) This application employs a triple mechanism to synergistically suppress the aforementioned interference: First, both excitation modes encode the defect-related magnetic field signal to a preset modulation frequency, using a reference clock as the phase reference, responding only to signal components strictly synchronized with the frequency, thus converting static geomagnetic field, low-frequency power grid interference, and asynchronous broadband noise into filterable background components, achieving physical separation of the target signal from environmental interference in the frequency domain; Second, the dual-height differential structure has a natural common-mode suppression capability against the uniform background magnetic field in the far field, and differential operations make the far-field background contribution approach zero, while retaining the significant difference signal generated by the near-field local current anomaly, achieving selective enhancement of the near-field defect signal; Third, the actual lift-off height is inverted in real time through the differential ratio, and the magnetic field measurement value is dynamically corrected to compensate for the distance fluctuation error caused by backplate deformation and thermal expansion and contraction, ensuring the stability of the detection results under complex working conditions; The triple mechanism targets three independent sources of interference—frequency domain interference, spatial interference, and distance error—complementing each other to form a complete anti-interference system.

[0133] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for defect detection in photovoltaic modules, characterized in that, A sensing unit is disposed on one side of the photovoltaic module having a backsheet. The sensing unit includes multiple tunneling magnetoresistive sensors, which overlap in a first direction, and any two adjacent tunneling magnetoresistive sensors in the first direction have a height difference. The first direction is the thickness direction of the backsheet. The method includes: The spatial magnetic field distribution of the photovoltaic module sensed by multiple tunnel magnetoresistive sensors within a preset time period is obtained, wherein the current of the sensed photovoltaic module carries an AC component of the target modulation frequency. Based on the spatial magnetic field distribution of the photovoltaic module, the spatial vertical gradient of the magnetic field generated by the current is extracted. The spatial vertical gradient is used to characterize the rate of change of the magnetic field strength in the first direction. Based on the reference clock signal corresponding to the target modulation frequency, the spatial vertical gradient is synchronously demodulated to extract the demodulation result reflecting the distortion of the local current density vector. The demodulation results are subjected to image reconstruction processing to obtain the current density distribution of the photovoltaic module; Based on the current density distribution of the photovoltaic module, the defect area of ​​the photovoltaic module is determined.

2. The method according to claim 1, characterized in that, Also includes: By modulating the excitation source, a target excitation signal that varies periodically with the target modulation frequency is output to the photovoltaic module, so that the current of the photovoltaic module carries the AC component of the target modulation frequency.

3. The method according to claim 2, characterized in that, By modulating the excitation source, the photovoltaic module is output with a target excitation signal that varies periodically at the target modulation frequency, including: The target light source is modulated according to a preset pulse signal, so that the light intensity applied to the photovoltaic module by the target light source changes periodically at the target modulation frequency; or The load switching circuit electrically connected to the photovoltaic module is modulated according to a preset pulse signal so that the output current of the photovoltaic module changes periodically at the target modulation frequency.

4. The method according to any one of claims 1 to 3, characterized in that, Based on the spatial magnetic field distribution of the photovoltaic module, the spatial vertical gradient of the magnetic field generated by the current is extracted, including: The spatial magnetic field distribution of the multiple tunnel magnetoresistive sensors is differentially processed to obtain differential results, and the differential results are used as the spatial vertical gradient of the magnetic field.

5. The method according to claim 4, characterized in that, Based on a reference clock signal corresponding to the target modulation frequency, the spatial vertical gradient is synchronously demodulated to obtain the demodulation result, including: Using the reference clock signal as a phase reference, the target amplitude and target phase at the target modulation frequency are extracted from the differential result.

6. The method according to claim 5, characterized in that, The demodulation results are subjected to image reconstruction processing to obtain the current density distribution of the photovoltaic module, including: Based on the target amplitude and the target phase, a linear mapping relationship is established between the spatial vertical gradient of the magnetic field and the current density distribution; The system response matrix is ​​constructed based on the Biot-Savart law and the sensor's geometric parameters. The spatial vertical gradient of the magnetic field is then inverted using the system response matrix, the linear mapping relationship, and matrix multiplication to obtain the current density vector field.

7. The method according to claim 6, characterized in that, Based on the current density distribution of the photovoltaic module, the defect areas of the photovoltaic module are determined, including: Calculate the deviation between the magnitude of the current density vector at the target pixel in the current density vector field and the average magnitude of the current density vector at all pixels in a preset neighborhood of the target pixel; determine the region where the deviation exceeds a preset threshold as the defect region.

8. A defect detection system for photovoltaic modules, characterized in that, include: An excitation unit is used to modulate the photovoltaic module so that the current of the photovoltaic module carries an AC component of the target modulation frequency. A sensing unit, located on one side of the photovoltaic module with a backsheet, is used to sense the spatial magnetic field distribution of the photovoltaic module within a preset time period. The sensing unit includes multiple tunnel magnetoresistive sensors, which overlap in a first direction, and any two adjacent tunnel magnetoresistive sensors in the first direction have a height difference. The first direction is the thickness direction of the backsheet. The preprocessing unit is used to extract the spatial vertical gradient of the magnetic field generated by the current according to the spatial magnetic field distribution of the photovoltaic module, and to perform synchronous demodulation processing on the spatial vertical gradient according to the reference clock signal corresponding to the target modulation frequency, so as to extract the demodulation result reflecting the local current density vector distortion, wherein the spatial vertical gradient is used to characterize the rate of change of magnetic field strength in the first direction. The image reconstruction unit is used to perform image reconstruction processing on the demodulation result to obtain the current density distribution of the photovoltaic module, and to determine the defect area of ​​the photovoltaic module based on the current density distribution of the photovoltaic module.

9. The system according to claim 8, characterized in that, The excitation unit includes: The system includes a light-emitting device module, a first pulse width modulation controller, and a synchronous clock source. The light-emitting device module outputs a target light source. The first pulse width modulation controller modulates the target light source according to a preset pulse signal, causing the light intensity applied to the photovoltaic module by the target light source to change periodically at the target modulation frequency. The synchronous clock source provides a reference clock signal corresponding to the target modulation frequency. The system includes a load switching circuit, a second pulse width modulation controller, and a synchronous clock source. The load switching circuit is electrically connected to the photovoltaic module. The second pulse width modulation controller is used to modulate the load switching circuit according to a preset pulse signal so that the output current of the photovoltaic module changes periodically at the target modulation frequency. The synchronous clock source is used to provide a reference clock signal corresponding to the target modulation frequency.

10. The system according to claim 8, characterized in that, The sensing unit further includes: The circuit board, on which the tunnel magnetoresistive sensor is mounted; A rigid limiting layer is disposed on one side surface of the photovoltaic module having the back sheet, for separating the rigid limiting layer from the tunnel magnetoresistive sensor; A multiplexing circuit is used to sequentially acquire the output signals of the tunnel magnetoresistive sensor according to a preset timing sequence.

Images