Device and method for removing adhesive film through femtosecond laser welding of photovoltaic glass

Abstract

The invention relates to the technical field of photovoltaic module manufacturing, in particular to a device and method for removing an adhesive film through femtosecond laser welding of photovoltaic glass of a double-glass photovoltaic module. In view of the problems of aging, high cost, low efficiency and the like in the traditional adhesive film welding process, the device comprises a femtosecond laser energy regulation and control module, an optical focusing module, a path planning module and the like. In the implementation process, all the modules work cooperatively, and welding is achieved through the procedures of plasma treatment, nanometer texturing and the like. The final result is remarkable, the welding quality is improved by 35%, the reject ratio is reduced to below 0.7%, the cost of a single chip is reduced by 3 yuan, more than 5 million yuan is saved per year, the production efficiency is improved by 2.2 times, and the development of the photovoltaic industry is effectively promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic module manufacturing, and specifically relates to a device and method for removing adhesive film by femtosecond laser welding photovoltaic glass of a double-glass photovoltaic module. Background Art

[0002] Traditional double-glass photovoltaic modules use adhesive films to achieve the encapsulation connection between photovoltaic glass and solar cells. The adhesive films are very prone to aging under long-term ultraviolet radiation, high temperature and high humidity environments, and may turn yellow, crack, and debond, resulting in reduced module transmittance and power generation efficiency. At the same time, it affects the airtightness and mechanical stability of the modules, shortening the service life of the modules.

[0003] High production costs: As a key raw material for double-glass photovoltaic modules, film production accounts for a significant portion of procurement costs. Furthermore, the film application and curing processes require specialized equipment and energy consumption, increasing equipment investment and energy costs during production. Furthermore, the disposal of film waste after use also increases environmental costs.

[0004] Low production efficiency: The film welding process involves multiple steps, including film cutting, laying, lamination, and curing. These complex operations, and the lamination and curing process is time-consuming, resulting in long production cycles and making it difficult to meet the large-scale, high-efficiency production needs of the photovoltaic industry. Furthermore, defects such as bubbles and wrinkles are prone to appearing during the film laying process, requiring manual inspection and repair, further reducing production efficiency. Summary of the Invention

[0005] The purpose of this invention is to provide a device for removing adhesive film from photovoltaic glass in double-glass photovoltaic modules by femtosecond laser welding. This device aims to solve the problems of aging, high cost, and low efficiency of traditional adhesive film welding processes, and achieve efficient, high-quality adhesive film-free welding of photovoltaic glass. It includes:

[0006] Femtosecond laser multi-mode energy control welding module, used to adjust femtosecond laser energy parameters according to the characteristics of photovoltaic glass;

[0007] Adaptive optical focusing and path planning system, used to obtain glass information and plan welding paths, while achieving dynamic focusing;

[0008] Adhesive-free interface activation pretreatment device, used for cleaning, activation and nano-texturing of photovoltaic glass surfaces;

[0009] Real-time monitoring and closed-loop feedback control system, used to monitor welding process parameters in real time and perform feedback adjustment;

[0010] The integrated and efficient production line is used to realize the integrated production of photovoltaic glass loading, pretreatment, welding, testing and unloading.

[0011] Furthermore, the femtosecond laser multi-mode energy-controlled welding module includes a femtosecond laser generator, an energy control unit, a laser transmission optical path and a welding head; the femtosecond laser generator output pulse width is adjustable from 10 to 500 fs; the energy control unit integrates a pulse modulator, an attenuator and a frequency controller, which can achieve precise adjustment of pulse energy from 0 to 100 μJ and frequency from 1 to 100 kHz; the laser transmission optical path adopts a combination of polarization-maintaining fiber and a reflector; and the welding head is equipped with a replaceable focusing lens group.

[0012] Furthermore, the adaptive optical focusing,

[0013] Laser ranging sensor, image processor, motion control card and dynamic focusing lens; the industrial camera is a 5-megapixel, 100-fps high-speed camera with a telecentric lens; the laser ranging sensor has an accuracy of ±5μm; the image processor uses an algorithm based on convolutional neural network and reinforcement learning for image analysis and path planning, wherein the convolutional neural network is used to extract the surface features of photovoltaic glass, and the formula is W i is the convolution kernel weight, b i is the bias, n is the number of convolution kernels, * represents the convolution operation, and σ is the activation function. The Q-learning algorithm in reinforcement learning is used to make path decisions. The Q value obtained by executing action a in state s is calculated as follows:

[0014] α is the learning rate, r is the immediate reward, γ is the discount factor, s′ is the next state after executing action a, and a′ is the action in the next state; the motion control card drives the high-precision linear motor and the rotary motor; the dynamic focusing mirror is driven by a voice coil motor and can complete focal length adjustment within 2ms.

[0015] Furthermore, the adhesive-free film interface activation pretreatment device includes a plasma processing unit and a nano-texturing module; the plasma processing unit includes a radio frequency power supply, a gas supply system and a plasma generating chamber, adopts radio frequency glow discharge technology, the working gas is a mixture of argon and oxygen, and the processing power is adjustable from 50 to 500W; the nano-texturing module shares a laser source with the femtosecond laser welding module, introduces part of the laser into the pretreatment area through a spectrometer, and uses a galvanometer scanning system to achieve nano-scale processing of the glass surface.

[0016] Furthermore, the real-time monitoring and closed-loop feedback control system includes a sensor array, a data acquisition card, an industrial computer and a control actuator; the sensor array includes an infrared temperature sensor, a strain gauge stress sensor with a resolution of 0.1 MPa and a laser power meter with an accuracy of ±0.5%; the data acquisition card transmits the sensor data to the industrial computer at a sampling frequency of 10 kHz; the industrial computer has a built-in control algorithm based on PID parameter adaptive adjustment. The traditional PID control formula is Where u(t) is the control output, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(t) is the error; the proportional coefficient K in this system p The adaptive adjustment formula is K p0 is the initial proportional coefficient, △K p To adjust the step size, is related to the error e(t) and the error change rate Related functions, similarly for K i and K d Perform adaptive adjustment; realize parameter adjustment by controlling the actuator.

[0017] Furthermore, the integrated and efficient production line includes a loading robot arm, a pretreatment station, a welding station, an inspection station, a unloading robot arm and a conveyor belt; the loading robot arm is a six-axis collaborative robot equipped with a vacuum suction cup; the conveyor belt adopts a synchronous belt drive driven by a servo motor with a positioning accuracy of ±0.5mm; the inspection station integrates an optical microscope and an ultrasonic flaw detector.

[0018] Furthermore, the device comprises the following steps:

[0019] The photovoltaic glass is loaded using a loading robot and transported to the pre-processing station via a conveyor belt;

[0020] The photovoltaic glass is sequentially subjected to plasma cleaning, activation and nano-texturing treatment to complete the pre-treatment of the adhesive-free interface activation;

[0021] The pre-treated glass is transferred to the welding station. The adaptive optical focusing and path planning system uses an algorithm based on convolutional neural networks and reinforcement learning to plan the welding path. The femtosecond laser multi-mode energy control welding module performs femtosecond laser welding. At the same time, the real-time monitoring and closed-loop feedback control system uses a control algorithm based on PID parameter adaptive adjustment to adjust the welding parameters according to the monitored parameters.

[0022] The welded components are inspected for appearance and internal defects using an optical microscope and an ultrasonic flaw detector respectively;

[0023] The qualified products will be unloaded by the unloading robot arm.

[0024] Furthermore, the plasma cleaning activation treatment time is 30-60 seconds.

[0025] Furthermore, in the nano-texturing process, the processing speed of the femtosecond laser is 10-50 mm / s.

[0026] Furthermore, the integrated and efficient production line has an automatic mold changing function, which can quickly switch the production of photovoltaic glass of different specifications.

[0027] The present invention has the following beneficial effects:

[0028] 1: Femtosecond laser multi-mode energy control welding module

[0029] In view of the problem that traditional welding energy is difficult to accurately match different photovoltaic glass materials and welding requirements, the present invention designs a femtosecond laser multi-mode energy regulation welding module. The module integrates pulse width modulation, energy density adjustment and frequency switching functions. The optical properties of photovoltaic glass (such as refractive index and absorption rate) are detected in real time by the built-in spectrum analyzer, and the system automatically matches the preset energy regulation model. For example, for glass materials with high absorption rates, the module automatically increases the pulse energy and shortens the pulse width to reduce the heat-affected zone; for thinner glass, the energy density is reduced and the pulse frequency is increased to achieve fast and precise welding. Compared with traditional single-mode welding, this module can improve welding quality by more than 30%, avoiding defects such as cracking and deformation of glass due to improper energy input.

[0030] 2: Adaptive optics focusing and path planning system

[0031] In order to solve the problems of rough path planning and low welding accuracy of existing welding equipment, the present invention constructs an adaptive optical focusing and path planning system. The equipment is equipped with a high-resolution industrial camera and a laser ranging sensor to obtain the surface morphology, size and position information of photovoltaic glass in real time. Based on a deep learning algorithm, the system intelligently identifies surface defects and edge contours of the glass and generates an optimal welding path. During the welding process, the dynamic focusing mirror automatically adjusts the focal length according to the undulations of the glass surface to ensure that the femtosecond laser always acts on the welding area in the best focusing state. Through this system, the welding positioning accuracy can reach ±10μm, and the path planning efficiency is improved by 40%, effectively avoiding the problems of loose welding or excessive welding caused by welding deviations.

[0032] 3: Non-adhesive film interface activation pretreatment device

[0033] Traditional welding lacks effective treatment of the glass surface, resulting in insufficient bonding strength at the welding interface. The present invention provides a film-free interface activation pretreatment device, which includes a plasma treatment unit and a nano-texturing module. The plasma treatment unit generates low-temperature plasma through radio frequency discharge to clean and activate the surface of the photovoltaic glass, remove surface oxides and pollutants, and introduce active groups such as hydroxyl and carboxyl groups to enhance surface chemical activity. The nano-texturing module utilizes the high energy density characteristics of the femtosecond laser to etch nano-scale protrusions, grooves or hole structures on the glass surface, increasing the surface roughness and specific surface area. After pretreatment, the wettability of the glass surface is improved by 50%, and the bonding strength of the welding interface is improved by 45%, laying the foundation for high-quality welding.

[0034] 4: Real-time monitoring and closed-loop feedback control system

[0035] The existing welding process lacks dynamic monitoring and timely adjustment mechanisms, making it difficult to ensure the stability of welding quality. The present invention constructs a real-time monitoring and closed-loop feedback control system, which uses temperature sensors, stress sensors, laser power sensors, etc. distributed in the welding area to collect key parameters such as temperature field distribution, stress changes, and laser energy in real time during the welding process. The system compares and analyzes the collected data with the preset standard parameters, and when the parameters deviate, the feedback control mechanism is immediately triggered. For example, if the temperature is too high, the system automatically reduces the laser power and speeds up the welding speed; if the stress is abnormal, the welding path is adjusted or an annealing process is added. The system improves the stability of the welding process by 60% and reduces the product defect rate to below 1%.

[0036] 5: Integrated and efficient production line

[0037] The traditional adhesive film welding process has scattered procedures and poor connection, resulting in low production efficiency. The present invention designs an integrated and efficient production line, which integrates the processes of photovoltaic glass loading, pretreatment, welding, testing and unloading on the same equipment platform. Each process module is seamlessly connected through an intelligent robotic arm and a conveyor belt, and is uniformly scheduled using a PLC control system. For example, the loading robotic arm automatically feeds the glass into the pretreatment device, and after processing, it is directly transferred to the welding module. The welded components immediately enter the inspection unit, and qualified products are output by the unloading robotic arm. The assembly line also has an automatic mold changing function, which can quickly switch the production of photovoltaic glass of different specifications. The production cycle is shortened to 5 minutes per component, which is more than twice the efficiency of traditional processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 Overall process flow chart. DETAILED DESCRIPTION

[0039] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0040] Example 1

[0041] Traditional adhesive film welding processes, plagued by issues such as film aging, high costs, and low efficiency, have severely hampered the further development of the photovoltaic industry. This example, based on the apparatus and process for removing adhesive film from the photovoltaic glass of double-glass photovoltaic modules described in the claims, details the implementation of this technology and its significant efficiency gains through practical application cases.

[0042] 2. Implementation process of each module of the device

[0043] (1) Femtosecond laser multi-mode energy control welding module

[0044] The femtosecond laser welding device of the present invention has been put into use on the production line of a large photovoltaic module manufacturer. The femtosecond laser generator in the femtosecond laser multi-mode energy regulation welding module adopts a high-stability mode-locked laser, which can flexibly adjust the output pulse width according to different photovoltaic glass materials. When a batch of photovoltaic glass with high absorption rate is delivered to the production line, the built-in spectrometer quickly detects the optical properties of the glass and obtains its key parameters such as refractive index and absorption rate. According to the preset energy regulation model, the system automatically controls the energy regulation unit, increases the pulse energy to 80μJ, and shortens the pulse width to 50fs to reduce the heat-affected zone and avoid defects such as cracking and deformation of the glass due to improper energy input.

[0045] The laser transmission optical path utilizes a combination of polarization-maintaining fiber and reflectors to efficiently transmit the femtosecond laser light to the welding head. The welding head is equipped with a replaceable focusing lens assembly, allowing users to select lenses with the appropriate focal length based on welding requirements, ensuring precise laser focus on the welding area. Throughout the welding process, the module monitors laser energy parameters in real time and makes fine adjustments based on subtle changes in glass properties to ensure consistent welding quality. Compared to traditional single-mode welding, photovoltaic glass welded using this module has improved welding quality by 35%, effectively reducing scrap rates.

[0046] (2) Adaptive Optics Focusing and Path Planning System

[0047] A 5-megapixel, 100-fps high-speed industrial camera, paired with a telecentric lens, enables rapid and accurate image capture of the photovoltaic glass being transported to the welding station. A laser ranging sensor measures the glass surface height in real time with a high accuracy of ±5μm, capturing its surface topography, size, and position.

[0048] The image processor uses GPU-accelerated deep learning algorithms to process images and extracts photovoltaic glass surface features based on convolutional neural networks (CNN). Play a role through multiple convolution kernels W i Perform convolution operation on the input image I and add bias b i After that, it passes through the activation function σ (such as ReLU function) to obtain the feature map F, so as to accurately identify the defects, edge contours and other information on the glass surface.

[0049] Next, the Q-learning algorithm in reinforcement learning is used to make path decisions. The Q value obtained by executing action a in state s is calculated by the formula Computational updates, through continuous trial and error and learning, generate

[0050] Optimal welding path. Based on the generated path, the motion control card drives high-precision linear and rotary motors to control the three-dimensional motion of the welding head. Simultaneously, the dynamic focusing lens, driven by a voice coil motor, can adjust its focal length within 2ms, automatically adjusting the focus according to the undulations of the glass surface, ensuring that the femtosecond laser always maintains optimal focus on the weld area. Field testing has shown that this system achieves welding positioning accuracy of ±8μm, improving path planning efficiency by 45%, and effectively avoiding problems such as loose or excessive welds caused by welding deviations.

[0051] (3) Non-adhesive film interface activation pretreatment device

[0052] The photovoltaic glass first enters the plasma treatment unit, which includes an RF power supply, a gas supply system, and a plasma generation chamber. RF glow discharge technology is used, with the working gas being a mixture of argon and oxygen. The processing power is adjusted to 300W depending on the glass material and surface condition. The RF power generates a low-temperature plasma, which cleans and activates the glass surface for a set duration of 45 seconds. The plasma effectively removes oxides and contaminants from the glass surface while also introducing reactive groups such as hydroxyl and carboxyl groups, enhancing surface chemical activity.

[0053] The glass then enters the nanotexturing module, which shares the same laser source as the femtosecond laser welding module. A beam splitter directs a portion of the laser light into the pretreatment area. Using a galvanometer scanning system, nanometer-scale bumps, grooves, and holes are etched onto the glass surface at a processing speed of 30 mm / s, following a preset scanning pattern. This increases the surface roughness and specific surface area. After pretreatment, the wettability of the glass surface increases by 55%, and the bond strength of the weld interface increases by 50%, laying a solid foundation for subsequent high-quality welding.

[0054] (4) Real-time monitoring and closed-loop feedback control system

[0055] A sensor array consisting of infrared temperature sensors (accuracy ±1°C), strain gauge stress sensors (resolution 0.1 MPa), and laser power meters (accuracy ±0.5%) is distributed throughout the welding area. This array collects key parameters such as temperature field distribution, stress changes, and laser energy during the welding process in real time. A data acquisition card transmits the sensor data to an industrial computer at a sampling rate of 10 kHz.

[0056] Industrial computers have built-in control algorithms based on PID parameter adaptive adjustment. Traditional PID control formula

[0057] Used to calculate the control output, but in this system, the proportional coefficient K p By formula Perform adaptive adjustment, and similarly adjust the integral coefficient K i and differential coefficient K d The system also performs adaptive adjustments. If it detects excessively high temperatures in the welding area, it automatically reduces laser power and increases welding speed. If stress becomes abnormal, it promptly adjusts the welding path or adds an annealing step. This system has improved welding process stability by 65% ​​and reduced product defect rates to below 0.7%.

[0058] (5) Integrated and efficient production line

[0059] The loading arm utilizes a six-axis collaborative robot equipped with a vacuum suction cup. It automatically adjusts gripping force and angle based on the size and weight of the photovoltaic glass, accurately picking up the glass from the glass storage rack. After the glass is positioned using a vision positioning system, it is placed on the conveyor belt. The conveyor belt uses a servo motor-driven synchronous belt drive with a positioning accuracy of ±0.5mm, delivering the glass to the pre-processing station at a steady speed.

[0060] After the film-free interface activation pretreatment is completed at the pretreatment station, the glass is directly transferred to the welding station. The welded components immediately enter the inspection station, which integrates an optical microscope and an ultrasonic flaw detector to conduct a full-scale inspection of the welding quality. The test results are transmitted to the control system in real time. Unqualified products are sorted by the robotic arm to the rework area, and qualified products are grabbed by the unloading robotic arm and placed on the finished product storage rack or packaging line. The assembly line also has an automatic mold change function. When it is necessary to switch to the production of photovoltaic glass of different specifications, the equipment parameter adjustment and mold replacement can be completed within 5 minutes, and the production cycle is shortened to 4.5 minutes per component, which is more than 2.2 times more efficient than the traditional process.

[0061] 3. Process Implementation

[0062] (1) Photovoltaic glass loading

[0063] In the photovoltaic module production workshop, a loading robot arm picks up photovoltaic glass from glass storage racks according to a pre-set program. The robot arm's equipped visual positioning system identifies and locates the glass. Using image analysis technology, it determines the precise position and posture of the glass, ensuring accurate placement on the conveyor belt. Once the conveyor belt is activated, it transports the glass to the pre-processing station at a steady speed. The entire loading process is highly automated, requiring no human intervention, significantly improving loading efficiency and accuracy.

[0064] (2) Activation pretreatment of non-adhesive film interface

[0065] After the glass enters the plasma processing unit, the mixed gas of argon and oxygen is ionized under the action of the electric field generated by the radio frequency power supply, generating low-temperature plasma. The active particles in the plasma react chemically with the pollutants and oxides on the surface of the glass, decomposing and removing them, and at the same time introducing active groups such as hydroxyl and carboxyl groups on the glass surface. After the 45-second processing time is over, the glass enters the nano-texturing module. The femtosecond laser performs high-precision processing on the glass surface according to the preset scanning pattern, forming nano-scale protrusions, grooves or hole structures. During the processing process, the uniformity and consistency of the nanostructure are ensured by precisely controlling the energy, frequency and scanning speed of the laser, further enhancing the wettability of the glass surface and the bonding strength of the welding interface.

[0066] (3) Femtosecond laser welding

[0067] The pre-treated glass is transported to the welding station by a conveyor belt, and the adaptive optical focusing and path planning system starts working immediately. The industrial camera and laser ranging sensor quickly obtain the surface morphology, size and position information of the glass, and the image processor generates the optimal welding path through a deep learning algorithm. The femtosecond laser multi-mode energy control welding module automatically adjusts the energy parameters of the femtosecond laser, including pulse width, energy and frequency, according to the characteristics of the glass and the welding path. During the welding process, the real-time monitoring and closed-loop feedback control system continuously collects parameters such as temperature, stress, laser energy, etc. in the welding area, and compares and analyzes them with the preset standard parameters. Once a deviation in the parameters is found, the feedback control mechanism is immediately triggered to adjust the welding parameters to ensure the stability and consistency of the welding quality.

[0068] (4) Welding quality inspection

[0069] Welded components are transported along a conveyor belt to the inspection station. An optical microscope carefully examines the weld area for surface defects such as porosity, cracks, and lack of fusion. An ultrasonic flaw detector, utilizing the reflection and transmission principles of ultrasonic waves, verifies the weld strength and defects within the component, detecting voids, inclusions, and other problems. The test results are transmitted in real time to the control system, which classifies the components accordingly. Unqualified products are sorted to the repair area, while qualified products are conveyed to the unloading station.

[0070] (5) Finished product unloading

[0071] After receiving instructions from the control system, the unloading robot accurately grasps qualified products and places them on the finished product storage rack or packaging line. The robot's grasping action is precise and stable, avoiding damage to the finished product. The entire unloading process is closely coordinated with other links in the production line, ensuring efficient and continuous operation of photovoltaic module production.

[0072] IV. Qualitative description of efficiency improvement results

[0073] (1) Quality improvement

[0074] Through the femtosecond laser multi-mode energy control welding module, the laser energy parameters can be precisely adjusted according to the different photovoltaic glass materials, reducing the heat-affected zone, avoiding problems such as glass breakage and deformation, and improving the welding quality by 35% compared to traditional processes. The adaptive optical focusing and path planning system achieves high-precision welding positioning and optimal path planning, effectively avoiding welding deviations and further improving welding quality. The film-free interface activation pretreatment device enhances the wettability of the glass surface and the bonding strength of the welding interface, making the overall performance of the component more stable and reliable. The real-time monitoring and closed-loop feedback control system monitors and adjusts the welding process in real time to ensure the consistency of welding quality. The product defect rate is reduced to below 0.7%, which is far lower than the 5% of the traditional film welding process, greatly improving the product qualification rate and reliability.

[0075] (2) Cost reduction

[0076] Eliminating the use of adhesive film in the traditional adhesive film welding process has reduced the cost of a single module by 3 yuan. Furthermore, by eliminating the specialized equipment and energy consumption required for film laying and curing processes, equipment investment and energy costs in the production process have been reduced. Furthermore, the cost of film waste disposal has also been eliminated. For example, after one year of operation, this large-scale photovoltaic module manufacturer has achieved annual cost savings exceeding 5 million yuan, significantly reducing the company's production costs and improving its economic efficiency and market competitiveness.

[0077] (3) Improved efficiency

[0078] The integrated, highly efficient production line integrates photovoltaic glass loading, pretreatment, welding, testing, and unloading processes onto a single platform. Each process module is seamlessly connected via intelligent robotic arms and conveyor belts, and a PLC control system is used for unified scheduling. The production cycle has been shortened to 4.5 minutes per module, an efficiency improvement of more than 2.2 times compared to traditional processes. The application of an adaptive optical focusing and path planning system has increased path planning efficiency by 45%, reducing welding preparation time. Real-time monitoring and a closed-loop feedback control system ensure the stability of the welding process, avoiding rework and downtime caused by quality issues, further improving production efficiency and meeting the photovoltaic industry's demand for large-scale, high-efficiency production.

[0079] The embodiments of the present invention are presented for purposes of illustration and description and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those skilled in the art. The embodiments are chosen and described in order to better illustrate the principles of the invention and its practical application and to enable those skilled in the art to understand the invention and design various embodiments with various modifications as suited for specific applications.

Claims

1. A device for removing adhesive film by femtosecond laser welding photovoltaic glass of double-glass photovoltaic modules, characterized in that: include: Femtosecond laser multi-mode energy control welding module, used to adjust femtosecond laser energy parameters according to the characteristics of photovoltaic glass; Adaptive optical focusing and path planning system, used to obtain glass information and plan welding paths, while achieving dynamic focusing; Adhesive-free interface activation pretreatment device, used for cleaning, activation and nano-texturing of photovoltaic glass surfaces; Real-time monitoring and closed-loop feedback control system, used to monitor welding process parameters in real time and perform feedback adjustment; The integrated and efficient production line is used to realize the integrated production of photovoltaic glass loading, pretreatment, welding, testing and unloading.

2. The device according to claim 1, characterized in that The femtosecond laser multi-mode energy-controlled welding module includes a femtosecond laser generator, an energy control unit, a laser transmission optical path, and a welding head. The femtosecond laser generator outputs an adjustable pulse width of 10-500fs. The energy control unit integrates a pulse modulator, an attenuator, and a frequency controller, enabling precise adjustment of pulse energy from 0-100μJ and frequency from 1-100kHz. The laser transmission optical path utilizes a combination of polarization-maintaining fiber and a reflector. The welding head is equipped with a replaceable focusing lens assembly.

3. The device according to claim 1, characterized in that The adaptive optical focus, laser ranging sensor, image processor, motion control card and dynamic focusing mirror; the industrial camera is a high-speed camera with 5 million pixels and a frame rate of 100fps and is equipped with a telecentric lens; the laser ranging sensor has an accuracy of ±5μm; the image processor uses an algorithm based on convolutional neural network and reinforcement learning for image analysis and path planning, wherein the convolutional neural network is used to extract the surface features of photovoltaic glass, and the formula is W i is the convolution kernel weight, b i is the bias, n is the number of convolution kernels, * represents the convolution operation, and σ is the activation function. The Q-leaming algorithm in reinforcement learning is used to make path decisions. The Q value obtained by executing action a in state s is calculated as follows: α is the learning rate, r is the immediate reward, γ is the discount factor, s′ is the next state after executing action a, and a′ is the action in the next state; the motion control card drives the high-precision linear motor and the rotary motor; the dynamic focusing mirror is driven by a voice coil motor and can complete focal length adjustment within 2ms.

4. The device according to claim 1, characterized in that The adhesive-free film interface activation pretreatment device includes a plasma processing unit and a nano-texturing module; the plasma processing unit contains a radio frequency power supply, a gas supply system and a plasma generating chamber, adopts radio frequency glow discharge technology, the working gas is a mixture of argon and oxygen, and the processing power is adjustable from 50 to 500W; the nano-texturing module shares a laser source with the femtosecond laser welding module, introduces part of the laser into the pretreatment area through a spectrometer, and uses a galvanometer scanning system to achieve nano-scale processing of the glass surface.

5. The device according to claim 1, characterized in that The real-time monitoring and closed-loop feedback control system includes a sensor array, a data acquisition card, an industrial computer, and a control actuator; the sensor array includes an infrared temperature sensor, a strain gauge stress sensor with a resolution of 0.1 MPa, and a laser power meter with an accuracy of ±0.5%; the data acquisition card transmits sensor data to the industrial computer at a sampling frequency of 10 kHz; the industrial computer has a built-in control algorithm based on PID parameter adaptive adjustment. The traditional PID control formula is Where u(t) is the control output, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(t) is the error; the proportional coefficient K in this system p The adaptive adjustment formula is K p0 is the initial proportional coefficient, ΔK p To adjust the step size, is related to the error e(t) and the error change rate Related functions, similarly for K i and K d Perform adaptive adjustment; realize parameter adjustment by controlling the actuator.

6. The device according to claim 1, characterized in that The integrated and efficient production line includes a loading robot arm, a pretreatment station, a welding station, an inspection station, a unloading robot arm and a conveyor belt; the loading robot arm is a six-axis collaborative robot equipped with a vacuum suction cup; the conveyor belt adopts a synchronous belt drive driven by a servo motor with a positioning accuracy of ±0.5mm; the inspection station integrates an optical microscope and an ultrasonic flaw detector.

7. A process for removing adhesive film by femtosecond laser welding photovoltaic glass of double-glass photovoltaic modules, characterized in that: The device according to any one of claims 1 to 6 comprises the following steps: The photovoltaic glass is loaded using a loading robot and transported to the pre-processing station via a conveyor belt; The photovoltaic glass is sequentially subjected to plasma cleaning, activation and nano-texturing treatment to complete the pre-treatment of the adhesive-free interface activation; The pre-treated glass is transferred to the welding station. The adaptive optical focusing and path planning system uses an algorithm based on convolutional neural networks and reinforcement learning to plan the welding path. The femtosecond laser multi-mode energy control welding module performs femtosecond laser welding. At the same time, the real-time monitoring and closed-loop feedback control system uses a control algorithm based on PID parameter adaptive adjustment to adjust the welding parameters according to the monitored parameters. The welded components are inspected for appearance and internal defects using an optical microscope and an ultrasonic flaw detector respectively; The qualified products will be unloaded by the unloading robot arm.

8. The process according to claim 7, characterized in that: The plasma cleaning activation treatment time is 30-60 seconds.

9. The process according to claim 7, characterized in that: In the nano-texturing process, the processing speed of the femtosecond laser is 10-50 mm / s.

10. The process according to claim 7, characterized in that: The integrated and efficient production line has an automatic mold changing function, which can quickly switch the production of photovoltaic glass of different specifications.

Images