Laser scribing method and system for perovskite photovoltaic module based on real-time electrical feedback

Abstract

The present application relates to the technical field of photovoltaic device manufacturing, and particularly relates to a perovskite photovoltaic module laser scribing method and system based on real-time electrical feedback, which comprises scribing processing of a functional film layer on a perovskite photovoltaic module substrate to form a scribing area for sub-cell series interconnection or electrode isolation; in-situ and real-time micro-area electrical performance measurement of the scribing area to obtain electrical characteristic quantities; online construction or updating of a dynamic process model between laser process parameters and electrical performance based on the electrical characteristic quantities, the laser process parameters and substrate position information; prediction and generation of an optimized laser process parameter set according to the model before subsequent scribing, and the optimized laser process parameter set is used for subsequent processing, and then electrical measurement is performed again to verify the prediction effect and iteratively update the model, through taking electrical performance as a direct control target, closed-loop adaptive adjustment of the laser scribing process is realized, and the consistency of scribing electrical performance of a large-area perovskite module and production yield are improved.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic device manufacturing technology, and in particular to a laser scribing method and system for perovskite photovoltaic modules based on real-time electrical feedback. Background Technology

[0002] Perovskite photovoltaic modules are thin-film photovoltaic devices that integrate multiple sub-cells in series by forming a multilayer functional thin-film structure on a substrate. To achieve module-level series integration, it typically involves P1 scribing to isolate the bottom electrode, P2 scribing to connect the upper and lower electrodes, and P3 scribing to isolate the top electrode. Specific scribing regions need to be formed along a predetermined path in the functional film layer to achieve electrical connectivity between sub-cells or electrical isolation between electrodes. Laser scribing is a commonly used precision processing method that uses a controllable laser beam to remove material or expose the target layer in a localized area, thereby forming the required functional scribing lines in the thin-film structure. For perovskite photovoltaic modules, the scribing region not only serves as a geometric separator but also directly determines whether the upper and lower electrodes form a reliable electrical connection or effective isolation. Therefore, the electrical performance of the scribing process is a key factor affecting module efficiency, yield, and long-term reliability.

[0003] In existing technologies, laser scribing of perovskite photovoltaic modules typically involves scribing the entire panel using preset fixed process parameters. The processing quality is then evaluated through offline optical inspection or sampling electrical testing. To improve process controllability, existing solutions have attempted to introduce online monitoring methods, such as judging the material removal status based on laser-induced breakdown spectroscopy, or using machine vision and confocal measurement to provide feedback adjustment on the scribing morphology. However, these technologies mainly monitor the processing from the perspective of material composition or geometric morphology, making it difficult to conduct in-situ, real-time, and quantitative micro-area electrical performance evaluation of the scribed area after scribing. Since qualified morphology or composition does not necessarily correspond to qualified electrical performance, existing technologies are still prone to electrical performance fluctuations in large-area module processing, making the consistency of module performance largely dependent on experience-based control and post-screening.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a laser scribing method and system for perovskite photovoltaic modules based on real-time electrical feedback, thereby effectively solving the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback, comprising the following steps: On the substrate of a perovskite photovoltaic module, at least one functional film layer is scribed along a predetermined scribing path using a laser to form scribing regions for sub-cell series interconnection or electrode isolation. After the scribing region is formed, the in-situ, real-time micro-area electrical performance measurement is immediately performed on the scribing region to obtain electrical characteristic quantities used to characterize the interconnection interface formation state or isolation state of the scribing region. Based on the processing data, including the electrical characteristics, corresponding laser process parameters, and substrate position information, a dynamic process model reflecting the mapping relationship between laser process parameters and predicted electrical performance is constructed or updated online. Before proceeding with the next or subsequent scribing process, based on the dynamic process model and combined with the context information of the current processing position, an optimized set of laser process parameters suitable for the processing position is predicted and generated. The optimized laser process parameter set is used to perform subsequent scribing, and the micro-area electrical performance measurement is performed again after processing to verify the prediction effect and iteratively update the dynamic process model.

[0007] Furthermore, at least during the P2 scribing process to connect the upper and lower electrodes, the laser beam is controlled to move along the main scribing direction while simultaneously superimposed with a transverse periodic oscillating motion, thereby forming a microgroove structure on the exposed surface of the lower electrode.

[0008] Furthermore, the physical state of the marked area is simultaneously sensed in situ and in real time to obtain its morphological feature data and / or material composition data as physical state data.

[0009] Furthermore, micro-area electrical performance measurements are adaptively performed based on the type of the scribed region: For the P2 scribed area used for sub-cell series interconnection, perform micro-area contact resistance measurement; For the P1 or P3 marked areas used for electrode isolation, perform micro-area insulation resistance measurements.

[0010] Furthermore, the micro-area contact resistance measurement performed on the P2 scribing area is performed in situ by a micro-area four-probe module integrated with and moving synchronously with the laser scribing unit; The probes of the micro-area four-probe module contact the exposed electrode surface of the scribed area with a constant micro-contact force during measurement.

[0011] Furthermore, a dynamic process model reflecting the mapping relationship between laser process parameters and predicted electrical performance is constructed or updated online. The steps include: After each acquisition of the electrical feature quantity, the currently acquired electrical feature quantity, the corresponding laser process parameters, and the substrate position information are used as a training sample. Based on the sequentially input training samples, the parameters of the dynamic process model are incrementally updated using recursive least squares or online gradient descent.

[0012] Furthermore, the dynamic process model is constructed or updated as a multi-task learning model, and the steps include: The multi-task learning model uses the laser process parameters and substrate position information as shared inputs, and outputs a joint prediction of the electrical characteristics and the physical state data. The update of the dynamic process model is based on a weighted common loss of electrical prediction error and physical state prediction error.

[0013] Furthermore, before proceeding to the next segment or subsequent scribing process, based on the dynamic process model and combined with the context information of the current processing position, an optimized laser process parameter set suitable for the processing position is predicted and generated. The steps include: Based on the dynamic process model, an optimization problem is constructed for the next processing position, wherein the objective function is to minimize the deviation between the predicted value of the electrical characteristic quantity of the processing position and the preset target value, the decision variable is the laser process parameter, and the constraint condition includes at least the equipment feasible region of the laser process parameter; Solving the optimization problem yields a set of laser process parameters that minimizes the objective function, which serves as the optimized laser process parameter set.

[0014] Furthermore, the optimized laser process parameter set is used to perform subsequent scribing, and the micro-area electrical performance measurement is performed again after processing to verify the prediction effect and iteratively update the dynamic process model. The steps include: The next section of scribing is performed using the optimized laser process parameter set, and the actual laser process parameters and processing position used are recorded. Perform the micro-area electrical performance measurement on the processed scribing area to obtain the measured electrical characteristic quantities; Calculate the prediction error between the measured electrical characteristic quantity and the predicted electrical characteristic quantity for the processing position; The dynamic process model is iteratively updated based at least on the prediction error, the actual laser process parameters used, and the processing location.

[0015] The present invention also includes an intelligent laser processing system, the system comprising: A motion platform used to support and position the substrate of perovskite photovoltaic modules; An integrated intelligent processing head, wherein the integrated intelligent processing head has a common optical path or is tightly integrated with the following: A laser scribing unit is used to generate and focus a scribing laser beam to form scribing areas on the film layer of the substrate. The micro-area electrical measurement unit is used to immediately perform in-situ, real-time micro-area electrical performance measurement on the scribing area after the scribing area is formed. The sensing unit is used to acquire multimodal monitoring data during the processing. The central controller is communicatively connected to both the integrated intelligent processing head and the motion platform; it is used to receive electrical characteristic quantities from the micro-area electrical measurement unit and multimodal monitoring data from the sensing unit. Based on the electrical characteristics, corresponding laser process parameters, and substrate position information, a dynamic process model is constructed or updated online. Based on the dynamic process model, an optimized laser process parameter set is generated for the subsequent unmarked areas; The optimized laser process parameter set is sent to the laser scribing unit and the motion platform to perform adaptive processing, and the dynamic process model is iteratively updated based on the verification data after processing.

[0016] The beneficial effects of this invention are as follows: By using the electrical properties of the scribing area as a direct control target, in-situ, real-time micro-area electrical property measurement is introduced during the laser scribing process. Combined with processing data, a dynamically updatable process model is constructed, thereby enabling the prediction and adaptive adjustment of subsequent scribing process parameters. This allows the laser scribing process to no longer rely solely on preset parameters or indirect morphology and composition information, but to achieve closed-loop control based on the electrical results that determine the module's function. This effectively reduces process deviations caused by factors such as substrate warping, film thickness fluctuations, or uneven thermal fields. In the processing of large-area perovskite photovoltaic modules, it can improve the consistency and controllability of the electrical properties of the scribing area, reduce the probability of poor connectivity or isolation failure, and thus help improve the overall performance, production yield, and long-term operational reliability of the module.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 The flowchart shows the laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback. Figure 2 This is a schematic diagram of traditional linear scanning P2 line drawing; Figure 3 This is a schematic diagram of the P2 line drawn using oscillating scanning. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1: like Figure 1 As shown, this application provides a laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback. The method includes: S10: On the substrate of a perovskite photovoltaic module, at least one functional film layer is scribed along a predetermined scribing path using a laser to form a scribing area for sub-cell series interconnection or electrode isolation. S20: Immediately after the formation of the scribing region, perform in-situ, real-time micro-area electrical performance measurements on the scribing region to obtain electrical characteristic quantities used to characterize the formation state or isolation state of the interconnection interface of the scribing region. S30: Based on processing data including electrical characteristics, corresponding laser process parameters, and substrate position information, a dynamic process model reflecting the mapping relationship between laser process parameters and predicted electrical performance is constructed or updated online. S40: Before proceeding with the next segment or subsequent scribing process, based on the dynamic process model and combined with the context information of the current processing position, predict and generate an optimized set of laser process parameters suitable for the processing position; S50: The subsequent scribing process is performed using an optimized set of laser process parameters, and the micro-area electrical performance is measured again after processing to verify the prediction effect and iteratively update the dynamic process model.

[0023] Specifically, firstly, on the substrate of the perovskite photovoltaic module, at least one functional film layer is scribed along a predetermined scribing path using a laser to form a corresponding scribing region. The laser scribing process can be based on preset initial laser process parameters. Immediately after scribing, without transferring the substrate, in-situ, real-time micro-area electrical performance measurements are performed on the scribing region to obtain electrical characteristic quantities that can directly characterize the electrical connectivity or isolation state of the scribing region. Specifically, for scribing regions used for connecting upper and lower electrodes, the electrical characteristic quantity can be the micro-area contact resistance or related conduction parameters; for scribing regions used for electrode isolation, the electrical characteristic quantity can be the micro-area insulation resistance or leakage current-related parameters. Subsequently, the electrical characteristic quantities, the corresponding laser process parameters, and the position information of the scribing region on the substrate are used as processing data input processing units to construct or update a dynamic process model online describing the relationship between laser process parameters and electrical performance, enabling the model to reflect... The influence of laser scribing on electrical performance under different spatial locations and process conditions is investigated. Before proceeding to the next stage or subsequent scribing process, the electrical performance of the upcoming processing location is predicted based on a dynamic process model and contextual information of the current processing location. This generates an optimized set of laser process parameters suitable for that location, which is then used to execute subsequent scribing processes. After subsequent scribing is completed, in-situ, real-time micro-area electrical performance measurements are performed on the newly formed scribing area. The measured results are used to verify the prediction effect, and the dynamic process model is further iteratively updated. This forms a closed-loop control process of scribing-measurement-modeling-prediction-re-scibing throughout the entire scribing process. Through this method, even when there is warping, uneven film thickness, or changes in thermal field distribution on a large-area substrate, the laser scribing process parameters can be dynamically and adaptively adjusted with electrical performance as the direct control target, thereby achieving consistency and reliability improvement of the electrical performance of the scribing area.

[0024] By using the electrical properties of the scribing area as a direct control target, in-situ, real-time micro-area electrical property measurements are introduced during the laser scribing process. Combined with processing data, a dynamically updated process model is constructed, enabling the prediction and adaptive adjustment of subsequent scribing process parameters. This overall technical solution allows the laser scribing process to move beyond relying solely on preset parameters or indirect morphological and compositional information. Instead, it enables closed-loop control based on the electrical results that determine the module's function, effectively reducing process deviations caused by factors such as substrate warping, film thickness fluctuations, or uneven thermal fields. In the processing of large-area perovskite photovoltaic modules, this improves the consistency and controllability of the electrical properties of the scribing area, reducing the probability of poor connectivity or isolation failures. This, in turn, helps improve the overall performance, production yield, and long-term operational reliability of the module, while reducing reliance on manual experience and post-processing screening.

[0025] As a preferred embodiment of the above embodiment, in step S10, at least during the P2 scribing for connecting the upper and lower electrodes, the laser beam is controlled to move along the main scribing direction while simultaneously superimposed with a transverse periodic oscillation motion, thereby forming a microgroove structure on the exposed lower electrode surface. The frequency of the transverse periodic oscillation motion is 1 kHz to 100 kHz, and the amplitude is 0.5 μm to 5 μm. By introducing this transverse periodic oscillation into the laser scanning trajectory, the laser action point generates a periodic transverse displacement during the advancement of the main scribing direction, thereby causing the laser energy to exhibit a transversely distributed distribution characteristic within the scribing area. The transverse periodic oscillation can be achieved through a scanning galvanometer, an acousto-optic deflection device, or a piezoelectric driven deflection element, and its oscillation parameters include oscillation frequency, oscillation amplitude, and oscillation waveform. The oscillation waveform can be a sine wave, a triangular wave, or other periodic function. By setting the oscillation parameters, while removing the upper functional film layer in the P2 scribing area, a micro-groove structure periodically distributed along the main scribing direction is formed on the surface of the lower conductive electrode. This micro-groove structure is formed by multiple transverse sweeps and has certain depth, width, and periodic characteristics, thereby significantly increasing the actual contact area between the upper and lower electrodes and improving the three-dimensional interlocking state of the electrode interface. In some embodiments, the activation of transverse periodic oscillation and its parameters can be adjusted according to the real-time micro-area electrical performance measurement results. For example, when the contact resistance in the P2 scribing area is detected to be too high, the interconnection effect can be enhanced by increasing the oscillation amplitude or adjusting the oscillation frequency, so that the formed micro-groove structure is coordinated with the electrical performance target.

[0026] In this embodiment, in step S20, the physical state of the scribing area is simultaneously sensed in situ and in real time to obtain its morphological feature data and / or material composition data as physical state data. Specifically, the physical state sensing and electrical measurement are completed at the same processing station, and the detection position corresponds spatially to the scribing area that has just been completed, so that the obtained electrical feature quantities and physical state data are consistent in the time and spatial dimensions. The morphological feature data may include one or more of the width, depth, edge morphology, and bottom flatness of the scribing groove, which can be obtained through confocal displacement measurement, white light interferometry, or other non-contact methods. The material composition data is obtained through optical detection methods. It is used to characterize the exposure or residual state of different functional films in the scribing area. It can be obtained through spectral analysis. In-situ detection of the bottom of the scribing can be performed using laser-induced breakdown spectroscopy, Raman spectroscopy, or reflection spectroscopy. The obtained morphological feature data and / or material composition data can be used alone or correlated with the synchronously obtained electrical feature quantities to help determine the physical cause of changes in electrical performance. It can also be used as input data to participate in the construction or updating of dynamic process models, so that the laser scribing process based on real-time electrical feedback has a comprehensive perception capability of electrical results and their physical causes.

[0027] As a preferred embodiment of the above, in step S20, the micro-area electrical performance measurement is adaptively performed according to the type of the scribing region: For the P2 scribed area used for sub-cell series interconnection, perform micro-area contact resistance measurement; For the P1 or P3 marked areas used for electrode isolation, perform micro-area insulation resistance measurements.

[0028] Specifically, when the scribing area is the P2 scribing area used for series interconnection of sub-cells, micro-area contact resistance measurement is performed on the scribing area. By applying a test current to the newly formed interconnection area and measuring the voltage response, the electrical connection state between the upper electrode and the lower conductive electrode is quantitatively characterized. When the scribing area is the P1 or P3 scribing area used for electrode isolation, micro-area insulation resistance measurement is performed on the scribing area. By applying a test voltage between the isolated electrodes and detecting the leakage current or equivalent insulation resistance value, the insulation reliability of the isolation area is quantitatively characterized. Different types of micro-area electrical performance measurements can be achieved through the same micro-area electrical measurement unit. This measurement unit switches the corresponding measurement mode or test parameters according to the scribing area type, thereby completing the electrical performance testing of scribing areas with different functions without changing the hardware structure.

[0029] Among them, the micro-area contact resistance measurement performed on the P2 scribing area is performed in situ by a micro-area four-probe module that is integrated with the laser scribing unit and moves synchronously. During measurement, the probes of the micro-area four-probe module contact the exposed electrode surface of the scribing area with a constant micro-contact force. The micro-area four-probe module and the laser scribing unit are mounted on the same processing head or rigid connection structure, keeping their spatial relative position with the laser focus fixed. Thus, after completing the P2 scribing process, electrical measurements can be performed on the newly formed interconnected area without re-alignment. The micro-area four-probe module includes four micro probes arranged in one direction. The outer probe is used to apply the test current, and the inner probe is used to collect the voltage signal to obtain the micro-area contact resistance value reflecting the quality of the upper and lower electrode interconnection interface. During the measurement process, each probe contacts the exposed lower conductive electrode surface of the scribing area with a constant micro-contact force. The micro-contact force is adjusted through an elastic support structure, piezoelectric drive, or force feedback control, so that the probes form a stable electrical contact while avoiding mechanical damage to the electrodes and the functional film layer below.

[0030] The micro-area insulation resistance measurement performed on the P1 or P3 marked area is achieved by applying a test voltage to the intact electrodes on both sides of the marked groove and measuring the micro-area leakage current in situ.

[0031] As a preferred embodiment of the above, in step S30, a dynamic process model reflecting the mapping relationship between laser process parameters and predicted electrical performance is constructed or updated online. The steps include: After acquiring each electrical feature, the currently acquired electrical feature, the corresponding laser process parameters, and the substrate position information are used as a training sample. Based on sequentially input training samples, the parameters of the dynamic process model are incrementally updated using recursive least squares or online gradient descent.

[0032] Specifically, an online dynamic process model reflecting the mapping relationship between laser process parameters and predicted electrical performance is constructed or updated. The process is as follows: After each in-situ, real-time micro-area electrical performance measurement of the scribing region is completed, the currently acquired electrical feature quantity, the corresponding laser process parameter, and the position information of the scribing region on the substrate are input into the processing unit as a training sample. The electrical feature quantity can be micro-area contact resistance or micro-area insulation resistance. The laser process parameter can include one or more of laser power, scanning speed, pulse frequency, and oscillation parameters. The substrate position information is used to characterize the influence of different spatial regions on the processing results. The dynamic process model is a parametric prediction model used to predict the corresponding electrical performance results under given process parameters and position conditions. Based on the sequentially input training samples, the model parameters are incrementally updated using recursive least squares or online gradient descent. The recursive least squares method recursively corrects the model parameters by combining historical samples and the latest samples to achieve rapid convergence. The online gradient descent method continuously and adaptively updates the model parameters by making small adjustments along the gradient direction of the loss function based on the prediction error of the current sample.

[0033] In this embodiment, in step S30, the dynamic process model constructed or updated is a multi-task learning model, and the steps include: The multi-task learning model uses laser process parameters and substrate position information as shared inputs, and outputs a joint prediction of electrical characteristic quantities and physical state data. The dynamic process model is updated based on a weighted common loss of electrical prediction error and physical state prediction error.

[0034] Specifically, the constructed or updated dynamic process model is a multi-task learning model. This model uses laser process parameters and substrate position information as shared input features. The laser process parameters include one or more of laser power, scanning speed, pulse frequency, and oscillation parameters. The substrate position information is used to characterize the spatial distribution of the scribing area on the substrate. Based on the shared input, the multi-task learning model sets multiple parallel output tasks to simultaneously predict the electrical characteristics and physical state data of the scribing area. The electrical characteristics include micro-area contact resistance or micro-area insulation resistance, and the physical state data includes morphological feature data and / or material composition data. By sharing some model parameters, the common influence of laser process parameters and substrate position on electrical performance and physical state is learned. At the same time, the differences between electrical changes and physical state changes are characterized by task-related output structures. During the model update process, after obtaining new training samples, the prediction errors of the model for electrical features and physical state data are calculated separately. The electrical prediction errors and physical state prediction errors are weighted according to preset weights to form a common loss function for model parameter updates. Based on the weighted common loss, the shared parameters and task-related parameters of the model are incrementally updated using online gradient descent or recursive least squares method.

[0035] As a preferred embodiment of the above, in step S40, before performing the next segment or subsequent scribing process, based on the dynamic process model and combined with the context information of the current processing position, an optimized laser process parameter set suitable for the processing position is predicted and generated. The steps include: Based on the dynamic process model, an optimization problem is constructed for the next processing position, where the objective function is to minimize the deviation between the predicted value of the electrical characteristic quantity of the processing position and the preset target value, the decision variable is the laser process parameter, and the constraint condition includes at least the equipment feasible region of the laser process parameter; Solving the optimization problem yields a set of laser process parameters that minimize the objective function, serving as the optimized laser process parameter set. Specifically, based on a dynamic process model, a parameter optimization problem is constructed for the next processing position, using laser process parameters as decision variables. These parameters include one or more of the following: laser output power, scanning speed, pulse frequency, spot size, and oscillation parameters. The objective function is defined as minimizing the deviation between the predicted electrical characteristic value and the preset target value at that position. The predicted electrical characteristic value is output by the dynamic process model, and the preset target value is the target contact resistance or target insulation resistance range set for the corresponding scribing area. Constraints, including at least the feasible region of the laser process parameters, are set to ensure the executability of the parameter solution. By solving the parameter optimization problem, a set of laser process parameters that minimizes the objective function is obtained, serving as the optimized laser process parameter set applicable to the current processing position and applied to subsequent scribing processes. After processing, electrical measurements are used to verify the prediction effect, and the results are fed back to continuously update the dynamic process model, thus forming a prediction-execution-verification closed-loop control process guided by electrical performance objectives.

[0036] In this embodiment, in step S50, the subsequent scribing process is performed using an optimized laser process parameter set, and the micro-area electrical performance is measured again after processing to verify the prediction effect and iteratively update the dynamic process model. The steps include: S51: Use the optimized laser process parameter set to execute the next section of scribing, and record the actual laser process parameters used and the processing position; S52: Perform micro-area electrical performance measurement on the scribed area after processing to obtain the measured electrical characteristic quantities; S53: Calculate the prediction error between the measured electrical characteristic quantity and the electrical characteristic quantity predicted for the machining position; S54: The dynamic process model is iteratively updated based at least on the prediction error, the actual laser process parameters used, and the processing position.

[0037] Specifically, firstly, an optimized set of laser process parameters is used to perform laser scribing on the next or subsequent scribing area. During the process, the actual laser process parameters used and the corresponding processing position information are recorded simultaneously. This recorded information accurately reflects the execution of the model's predicted parameters under real processing conditions, thus providing basic data for subsequent error analysis. Next, in-situ, real-time micro-area electrical performance measurements are performed again on the completed scribing area to obtain the measured electrical characteristics of the scribing area. The type of measured electrical characteristics is consistent with the electrical measurement method used for this scribing area in step S20. For example, micro-area contact resistance is obtained in scribing area P2, and micro-area insulation resistance is obtained in scribing areas P1 or P3, thus ensuring the consistency between the predicted and measured results in a physical sense. Then, based on the measured electrical characteristics, the prediction error between the measured values ​​and the electrical characteristics predicted for this processing position in step S40 is calculated. The prediction error is used to quantitatively characterize the dynamic process model's prediction under the current processing position and current process conditions. The accuracy can be expressed as the difference between the two, relative deviation, or other error measures. Finally, based at least on the prediction error, the actual laser process parameters used, and the corresponding processing position information, the dynamic process model is iteratively updated. Specifically, the prediction error can be used as new training information and introduced into the parameter update process of the dynamic process model. This allows the model to retain historical processing experience while gradually correcting its response to the current process conditions and spatial position. The model update process can be combined with the aforementioned recursive least squares method, online gradient descent method, or multi-task learning model update mechanism, so that the prediction error directly drives the adaptive adjustment of the model parameters. Through the above implementation method, after performing subsequent scribing processing with optimized laser process parameters, the prediction results are verified using measured electrical performance. The dynamic process model is continuously iteratively updated based on the prediction error, so that the laser scribing process forms a self-learning closed-loop control mechanism with real electrical results as the feedback core, thereby continuously improving the accuracy of process parameter prediction and the stability of the overall processing process.

[0038] Example 2: The present invention also includes an intelligent laser processing system, the system comprising: A motion platform used to support and position the substrate of perovskite photovoltaic modules; An integrated intelligent processing head, wherein the integrated intelligent processing head has a common optical path or is tightly integrated with the following: A laser scribing unit is used to generate and focus a scribing laser beam to form scribing areas on a film layer of a substrate. The micro-area electrical measurement unit is used to perform in-situ, real-time micro-area electrical performance measurements on the scribing area immediately after the scribing area is formed. The sensing unit is used to acquire multimodal monitoring data during the processing. The central controller is communicatively connected to the integrated intelligent processing head and the motion platform; it is used to receive electrical characteristic quantities from the micro-area electrical measurement unit and multimodal monitoring data from the sensing unit. Based on electrical characteristics, corresponding laser process parameters, and substrate position information, a dynamic process model is constructed or updated online. Based on the dynamic process model, an optimized laser process parameter set is generated for the subsequent unmarked areas; The optimized laser process parameter set is sent to the laser scribing unit and motion platform to perform adaptive processing, and the dynamic process model is iteratively updated based on the verification data after processing.

[0039] The intelligent laser processing system in this embodiment mainly includes the following parts: 1. Motion platform: Composed of a vacuum adsorption platform and a precision linear motor, it is used to adsorb and fix glass substrates with a maximum size of 1.2m × 0.6m and achieve nanometer-level precision motion positioning.

[0040] 2. Integrated intelligent processing head: fixed to the gantry frame. Its core innovation lies in the tight integration of multiple functional modules: Laser scribing unit: It adopts a nanosecond pulse laser with a wavelength of 532nm. After beam expansion and shaping, the beam is focused on the substrate surface by the f-θ lens scanning galvanometer. The laser parameters (power, frequency, scanning speed, oscillation mode) can be adjusted in real time by the central control unit.

[0041] Micro-area electrical measurement unit: A micro-area four-probe module fabricated using microelectromechanical systems (MEMS) technology. Four diamond probes are arranged in a straight line with a tip spacing of 50 μm. This module achieves synchronized raising and lowering with the laser focus and constant contact force control through piezoelectric ceramic actuators. Its spatial position relative to the laser scribing unit is calibrated at the factory using a precision reference plate to obtain a fixed spatial offset vector. During processing, the central controller calculates the precise placement of the probes in real time based on the laser processing coordinates and achieves positioning through closed-loop motion control, ensuring that the measurement point is precisely the scribing area that has just been completed.

[0042] Multimodal sensing unit: a) Confocal displacement sensor: used to measure the height of the substrate surface in real time, and the data is fed back to the central controller to dynamically compensate for substrate warping and ensure that the laser focus position is constant.

[0043] b) Laser-induced breakdown spectroscopy module: A low-energy probe laser is used to irradiate the bottom of the scribing. By collecting and analyzing the generated plasma spectrum, the material composition is determined (e.g., to confirm whether the P2 scribing has completely removed the perovskite and hole transport layer, exposing the underlying ITO), providing compositional basis for electrical diagnostics.

[0044] 3. Central Control Unit: Contains a built-in industrial computer and customized software. Its logical architecture corresponds to steps S30-S50 in the claims. The core software includes: Data fusion module: Real-time synchronous processing of electrical characteristics, morphological data (from sensors), composition data and location information.

[0045] Model learning engine: Configured to execute the algorithm as described in claim 7 or 8, a pre-trained model is loaded during system initialization. During processing, for each "electrical feature quantity-process parameter-position" data sample obtained from S20, the model learning engine incrementally updates the parameters online using recursive least squares to optimize the parameters of the dynamic process model. When multimodal perception is enabled, the engine runs as the multi-task learning model described above, simultaneously predicting electrical performance and physical state.

[0046] Optimization Solver: Before each line segment begins, the optimization solver constructs an optimization problem for the current position based on the latest dynamic process model. The objective function of this problem is to make the electrical characteristic quantity predicted by the model as close as possible to the preset target value. The constraints include the feasible domain of laser parameters and the acceptable range of physical state. The solver quickly solves the problem using a sequential quadratic programming algorithm to generate an optimized set of laser process parameters.

[0047] Execution Control and Iteration Module: Controls the execution of the processing unit, calculates the prediction error after processing, and triggers iterative updates of the model.

[0048] Example 3: Application of closed-loop scribing method in the fabrication of large-area perovskite components; Using the system of Example 1, lines P1, P2, and P3 were scribed on a 1.2m × 0.6m perovskite semi-finished substrate with a fully deposited thin film structure. This example focuses on describing the closed-loop control process of scribing line P2, and its process strictly corresponds to that of claim 1 and dependent claims.

[0049] 1. Initialization: The system loads the line drawing pattern and initial laser parameters based on historical data.

[0050] 2. Perform line marking, measurement, and sensing: The laser head performs the first segment P2 tracing along the preset path. In this embodiment, the above-mentioned oscillation scanning mode (frequency 20kHz, amplitude 2μm) is enabled.

[0051] After the line is drawn, the laser head is raised and the micro-area four-probe unit is immediately lowered to measure the micro-area contact resistance of the line segment that was just drawn, and obtain the electrical characteristic quantity Rmeasure.

[0052] Simultaneously, the confocal sensor and the LIBS module acquire morphological and compositional data of the region, respectively.

[0053] 3. Data modeling and optimization decision-making: The central controller's model learning engine uses the measured Rmeasure, the process parameters used, the location information, and the physical state data as training samples to update the dynamic process model online.

[0054] The optimizer then starts for the next scribbled position, calling the updated model to solve the optimization problem. For example, if the model predicts that the resistance at the next position may be too high, the solver will calculate a set of higher laser power and appropriate scanning speed as the optimization parameter set.

[0055] 4. Perform verification and iterative learning: The system uses an optimized parameter set to execute the next line drawing.

[0056] Immediately after processing, the micro-area contact resistance was measured again to obtain the actual value.

[0057] The system calculates the prediction error ΔR = Rawtal - Rpredicted (where Rpredicted is the predicted value in S40).

[0058] Based on this prediction error and the actual process parameters used, the model learning engine is triggered again to iteratively update the dynamic process model, thereby completing a complete closed loop of "perception-modeling-decision-execution-verification-learning".

[0059] Performance comparison results: During the execution of the method of the present invention, the system automatically triggered a total of 15 power increases (average +6.2%) in the four edge regions of the substrate and triggered 2 power decreases (average -3%) in the central region to compensate for the uneven distribution of the thermal field.

[0060] The final assembled component was subjected to electrical performance mapping tests. The short-circuit current density (J / L) of the 20 series-connected sub-cells of the component (401) of this invention was measured. sc The distribution is extremely uniform, with a standard deviation of 3.8%, while the J of the comparative component (402) is much smaller. sc The distribution is discrete, with a standard deviation as high as 14.5%.

[0061] The average fill factor (FF) of the component of this invention reaches 82.3%, which is significantly higher than the 74.1% of the comparative example. The overall component efficiency (APCE) is improved by 1.8% in absolute value.

[0062] This demonstrates that the method of the present invention achieves uniformity and systematic improvement of component performance through real-time electrical feedback closed-loop control.

[0063] Example 4: Component interface structure and reliability verification; The P2 scribed region of the component of the present invention and the comparative component in Example 2 was cut by focused ion beam (FIB) and observed by electron microscopy.

[0064] like Figure 2 As shown, the comparative example uses straight lines to form a relatively smooth ITO surface, which allows the subsequently deposited metal electrode to form a planar contact with it, resulting in a limited contact area.

[0065] like Figure 3 As shown, this invention uses oscillating scribing to form a clear periodic wavy microgroove structure on the ITO surface. The subsequently deposited metal electrode material completely fills these grooves, forming a "root-like" three-dimensional interlocking interface. This periodic microgroove interlocking structure significantly increases the effective contact area of ​​the upper and lower electrodes and reduces the contact resistance. At the same time, its mechanical interlocking effect enhances the interface adhesion, which is one of the fundamental reasons for the improved component performance and stability.

[0066] Accelerated aging tests were conducted (85°C, 85% relative humidity). After 1000 hours of testing, the fill factor of the component of this invention decreased by only 6.5% of its initial value, while the comparative component decreased by 20.1%. This indicates that the P2 interface with a microgroove interlocking structure prepared by the method of this invention can effectively suppress interface degradation under humid and hot conditions, and significantly improve the long-term reliability of the component.

[0067] The adjustment system described above in this invention can effectively realize the laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0068] Similarly, the above-mentioned optimization schemes for the system can also achieve the optimization effects corresponding to the methods in Embodiment 1, which will not be repeated here.

[0069] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and accompanying drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback, characterized in that, The method includes: On the substrate of a perovskite photovoltaic module, at least one functional film layer is scribed along a predetermined scribing path using a laser to form scribing regions for sub-cell series interconnection or electrode isolation. After the scribing region is formed, the in-situ, real-time micro-area electrical performance measurement is immediately performed on the scribing region to obtain electrical characteristic quantities used to characterize the interconnection interface formation state or isolation state of the scribing region. Based on the processing data, including the electrical characteristics, corresponding laser process parameters, and substrate position information, a dynamic process model reflecting the mapping relationship between laser process parameters and predicted electrical performance is constructed or updated online. Before proceeding with the next or subsequent scribing process, based on the dynamic process model and combined with the context information of the current processing position, an optimized set of laser process parameters suitable for the processing position is predicted and generated. The optimized laser process parameter set is used to perform subsequent scribing, and the micro-area electrical performance measurement is performed again after processing to verify the prediction effect and iteratively update the dynamic process model.

2. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 1, characterized in that, At least during the P2 scribing process to connect the upper and lower electrodes, the laser beam is controlled to move along the main scribing direction while simultaneously superimposed with a transverse periodic oscillating motion, thereby forming a microgroove structure on the exposed surface of the lower electrode.

3. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 1, characterized in that, Simultaneously, the physical state of the marked area is sensed in situ and in real time to obtain its morphological feature data and / or material composition data as physical state data.

4. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 1, characterized in that, Micro-area electrical performance measurements are adaptively performed based on the type of the scribed region: For the P2 scribed area used for sub-cell series interconnection, perform micro-area contact resistance measurement; For the P1 or P3 marked areas used for electrode isolation, perform micro-area insulation resistance measurements.

5. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 4, characterized in that, The micro-area contact resistance measurement performed on the P2 scribing area is performed in situ by a micro-area four-probe module integrated with and moving synchronously with the laser scribing unit. The probes of the micro-area four-probe module contact the exposed electrode surface of the scribed area with a constant micro-contact force during measurement.

6. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 1, characterized in that, The steps for building or updating a dynamic process model online that reflects the mapping relationship between laser process parameters and predicted electrical properties include: After each acquisition of the electrical feature quantity, the currently acquired electrical feature quantity, the corresponding laser process parameters, and the substrate position information are used as a training sample. Based on the sequentially input training samples, the parameters of the dynamic process model are incrementally updated using recursive least squares or online gradient descent.

7. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 3, characterized in that, The dynamic process model being built or updated is a multi-task learning model, and the steps include: The multi-task learning model uses the laser process parameters and substrate position information as shared inputs, and outputs a joint prediction of the electrical characteristics and the physical state data. The update of the dynamic process model is based on a weighted common loss of electrical prediction error and physical state prediction error.

8. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 1, characterized in that, Before proceeding with the next segment or subsequent scribing process, based on the dynamic process model and combined with the context information of the current processing position, an optimized laser process parameter set suitable for the processing position is predicted and generated. The steps include: Based on the dynamic process model, an optimization problem is constructed for the next processing position, wherein the objective function is to minimize the deviation between the predicted value of the electrical characteristic quantity of the processing position and the preset target value, the decision variable is the laser process parameter, and the constraint condition includes at least the equipment feasible region of the laser process parameter; Solving the optimization problem yields a set of laser process parameters that minimizes the objective function, which serves as the optimized laser process parameter set.

9. The laser scribing method for perovskite photovoltaic modules based on real-time electrical feedback according to claim 1, characterized in that, The optimized laser process parameter set is used to perform subsequent scribing, and the micro-area electrical performance measurement is performed again after processing to verify the prediction effect and iteratively update the dynamic process model. The steps include: The next section of scribing is performed using the optimized laser process parameter set, and the actual laser process parameters and processing position used are recorded. Perform the micro-area electrical performance measurement on the processed scribing area to obtain the measured electrical characteristic quantities; Calculate the prediction error between the measured electrical characteristic quantity and the predicted electrical characteristic quantity for the processing position; The dynamic process model is iteratively updated based at least on the prediction error, the actual laser process parameters used, and the processing location.

10. An intelligent laser processing system, characterized in that, The system includes: A motion platform used to support and position the substrate of perovskite photovoltaic modules; An integrated intelligent processing head, wherein the integrated intelligent processing head has a common optical path or is tightly integrated with the following: A laser scribing unit is used to generate and focus a scribing laser beam to form scribing areas on the film layer of the substrate. The micro-area electrical measurement unit is used to immediately perform in-situ, real-time micro-area electrical performance measurement on the scribing area after the scribing area is formed. The sensing unit is used to acquire multimodal monitoring data during the processing. The central controller is communicatively connected to both the integrated intelligent processing head and the motion platform; it is used to receive electrical characteristic quantities from the micro-area electrical measurement unit and multimodal monitoring data from the sensing unit. Based on the electrical characteristics, corresponding laser process parameters, and substrate position information, a dynamic process model is constructed or updated online. Based on the dynamic process model, an optimized laser process parameter set is generated for the subsequent unmarked areas; The optimized laser process parameter set is sent to the laser scribing unit and the motion platform to perform adaptive processing, and the dynamic process model is iteratively updated based on the verification data after processing.

Images