Method for cleaning organic adhesive residues on surface of support plate

Abstract

The invention provides a method for cleaning organic glue residues on the surface of a carrier plate, which comprises the following steps of: acquiring surface cleanliness data in a cleaning process in real time by adopting a carrier plate surface cleanliness online monitoring device, analyzing a cleanliness change trend in real time through an algorithm, and dynamically adjusting soaking, oscillating and flushing treatment parameters and a circulation path, so as to realize cleaning of the organic glue residues on the surface of the carrier plate. Adaptively optimizing a cleaning scheme, and predicting a cleaning end point; deploying a colloidal particle agglomeration degree analyzer in the water tank, monitoring the agglomeration state data of colloidal particles in real time, analyzing the relevance between the agglomeration state and water quality parameters through an algorithm model, dynamically adjusting the pH value and ionic strength of ultrapure water, optimizing the water quality environment, and inhibiting redeposition of the colloidal particles.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a method for cleaning organic glue residues on a carrier surface. Background Art

[0002] The existing MicroLED display panel transfer solution is: attach a layer of anisotropic conductive adhesive ACF material (containing Au conductive particles) on the entire surface of the substrate, and then transfer the LED mass to the surface of the anisotropic conductive adhesive. Finally, a piece of quartz glass is pressed to break the Au particles, so that the LED electrode and the substrate CuPad are connected. In the above process, the existing carrier used to transfer the LED mass to the surface of the anisotropic conductive adhesive is also called a temporary carrier. The problem of organic glue residue in the repeated use of the temporary carrier after cleaning is a problem that needs to be improved.

[0003] During the cleaning process of the substrate in the dust-free zone, multiple technical problems are interrelated to form a complex system. First, the optimization of the stripping liquid immersion and ultrasonic oscillation parameters needs to consider the characteristics of the substrate material and the residue situation, and the best solution is obtained through algorithm model training. Secondly, the control of the water tank overflow and spray system is closely related to the layout of the substrate, and the algorithm needs to be iteratively optimized to achieve the best cleaning effect. Furthermore, the application of surface tension gradient analysis and Marangoni effect directly affects the efficiency of residue removal and needs to work in synergy with the water flow layout. At the same time, the real-time cleanliness monitoring system dynamically adjusts the parameters of each step through algorithm analysis and is linked to the dust-free zone control system. In addition, the analysis of the degree of agglomeration of colloid particles and the optimization of water quality parameters affect each other, and algorithm fusion analysis is required to achieve mutual promotion and optimization. Finally, the evaluation of particle removal ability and the analysis of residual stress on the surface of the substrate are mutually constrained, and the best balance point needs to be found through the algorithm. These problems constitute a network of interrelated technical problems that need to be comprehensively considered and solved through algorithm optimization. Summary of the invention

[0004] The purpose of the present invention is to solve the above defects and provide a method for cleaning organic glue residues on the surface of a carrier board, which realizes efficient cleaning of organic glue residues on the surface of the carrier board, effectively controls the environmental cleanliness during the cleaning process, prevents the redeposition of glue particles, and improves the cleaning quality and efficiency.

[0005] The object of the present invention is achieved in the following ways:

[0006] A method for cleaning organic glue residue on a carrier surface comprises the following steps:

[0007] The residual amount of organic glue on the surface of the carrier and the material size characteristic data are obtained, and the stripping liquid immersion sequence and duration combination are determined through algorithm model selection and parameter setting to obtain a multi-step immersion treatment plan. The immersion process is carried out in a dust-free area that meets the cleanliness level requirements, and an air filter device and anti-static measures are used to ensure the cleanliness of the environment;

[0008] According to the size and material property data of the carrier, the ultrasonic oscillation frequency and duration parameters are obtained through algorithm model training and evaluation, and the carrier is staggered in the horizontal direction and rotated in the vertical direction to form an oscillation arrangement. The oscillation process strictly controls the temperature and humidity conditions in the dust-free area;

[0009] According to the surface area and residual glue amount data of the carrier, the algorithm model is iteratively optimized to obtain the ultrapure water overflow and spray parameter combination, including flow rate, pressure and angle. The carrier is placed in a water tank at an angle and vertically spaced in multiple layers to form a water flow arrangement. The flushing process is evaluated in real time through dust particle monitoring and surface cleaning verification.

[0010] An online monitoring device for the surface cleanliness of the carrier plate is used to collect the surface cleanliness data in the cleaning process in real time, and an algorithm is used to analyze the cleanliness change trend in real time, dynamically adjust the soaking, shaking and rinsing treatment parameters and circulation path, adaptively optimize the cleaning scheme, and predict the cleaning endpoint;

[0011] A colloid agglomeration degree analyzer is deployed in the water tank to monitor the agglomeration state data of the colloid particles in real time, analyze the correlation between the agglomeration state and water quality parameters through an algorithm model, dynamically adjust the pH value and ionic strength of ultrapure water, optimize the water quality environment, and inhibit the re-deposition of colloid particles;

[0012] A particle removal capacity evaluation device is used to evaluate the removal efficiency of each particle size segment for particles of different sizes. The relationship between the removal efficiency, filtration accuracy and water flow rate is analyzed through an algorithm, the technical parameters of the filtration device are dynamically adjusted, and the horizontal arrangement spacing and staggered arrangement are optimized to form a particle removal process that is compatible with the degree of agglomeration of the colloid particles.

[0013] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0014] The present invention discloses a method for cleaning organic glue residues on the surface of a carrier. In view of the problems of cleanliness control and re-deposition of glue particles during the cleaning process of organic glue residues on the surface of the carrier, the amount of organic glue residues on the surface of the carrier and the material size characteristic data are obtained, and an algorithm model is used to determine a multi-step treatment scheme such as immersion, shaking and rinsing. In a dust-free area, the surface cleanliness data and the state of agglomeration of glue particles during the cleaning process are monitored in real time, and the treatment parameters and water quality environment are dynamically adjusted. The removal efficiency of particles of different sizes is evaluated by a particle removal capacity evaluation device, and the parameters of the filtering device and the layout of the carrier are optimized. The present invention realizes efficient cleaning of organic glue residues on the surface of the carrier, effectively controls the environmental cleanliness during the cleaning process, prevents the re-deposition of glue particles, and improves the cleaning quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flow chart of the method for cleaning organic glue residue on the surface of a carrier.

[0016] Figure 2 It is a schematic diagram of the method for cleaning organic glue residue on the surface of a carrier board of the present invention.

[0017] Figure 3 It is another schematic diagram of the method for cleaning organic glue residue on the surface of a carrier board of the present invention. DETAILED DESCRIPTION

[0018] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0019] like Figure 1-3 The method for cleaning organic glue residue on the surface of the carrier board in this embodiment may specifically include:

[0020] S101. Obtain data on the residual amount of organic glue on the surface of the carrier and the material size characteristics, determine the stripping solution immersion sequence and duration combination through algorithm model selection and parameter setting, and obtain a multi-step immersion treatment plan. The immersion process is carried out in a dust-free area that meets the cleanliness level requirements, and an air filter device and anti-static measures are used to ensure the cleanliness of the environment.

[0021] S101 specifically includes the following steps: Step 1, obtaining the data of the residual amount of organic glue on the surface of the carrier and the data of the material size characteristics as the input of the algorithm model. Step 2, according to the material size characteristics of the surface of the carrier, using the decision tree algorithm model, through parameter setting, determine the optimal stripping liquid immersion sequence and duration combination. Step 3, the carrier is immersed in different stripping liquids in a dust-free area that meets the cleanliness requirements according to the determined immersion sequence and duration. Step 3, during the immersion process, an air filter device is used to continuously filter the air in the dust-free area to remove particles and dust in the air. Step 5, during the immersion process, anti-static measures are taken, including the use of anti-static floors, anti-static clothing, anti-static gloves, etc., to prevent static electricity from damaging the surface of the carrier. Step 6, after the immersion is completed, the data of the residual amount of organic glue on the surface of the carrier is obtained to determine whether the residual amount is lower than the preset threshold. If so, the immersion process is completed. If not, return to step 2, adjust the immersion sequence and duration, and continue the immersion process. In step 7, the data of multiple batches of substrate immersion treatment are input into the algorithm model for training. Through parameter optimization, the accuracy of the model's prediction of the immersion sequence and duration combination is continuously improved to realize the intelligentization of the substrate immersion treatment process.

[0022] Specifically, the data of the amount of organic glue residue on the surface of the carrier and the data of the material size characteristics are obtained; according to the material size characteristic data, a decision tree algorithm model is used to determine the optimal stripping solution immersion sequence and duration combination through parameter setting; the carrier is immersed in different stripping solutions in a dust-free area that meets the cleanliness requirements according to the immersion sequence and duration; wherein, during the immersion process, an air filter device is used to continuously filter the air in the dust-free area to remove particles and dust in the air; at the same time, anti-static measures are adopted, including the use of anti-static floors, anti-static clothing, and anti-static gloves to prevent static electricity from damaging the surface of the carrier; after the immersion is completed, the data of the amount of organic glue residue on the surface of the carrier is obtained to determine whether the residue is lower than a preset threshold; if so, the immersion process is completed; if not, the step of determining the immersion sequence and duration combination is returned to adjust the immersion sequence and duration, and the immersion process is continued; the immersion process data of multiple batches of carriers are input into the decision tree algorithm model for training, and the accuracy of the model in predicting the immersion sequence and duration combination is continuously improved through parameter optimization.

[0023] First, the data on the amount of organic glue residue on the surface of the carrier is obtained through high-precision surface residue detection instruments such as X-ray fluorescence spectrometers, and the material and size characteristic data of the carrier is obtained by a three-dimensional scanner, and these data are input into the intelligent optimization model based on the decision tree algorithm. The model analyzes the characteristic parameters of the carrier such as material, size, residue, etc., and uses algorithms such as information gain ratio and Gini index to automatically calculate the optimal stripping solution immersion sequence and duration combination, such as soaking in isopropyl alcohol for 10 minutes, then soaking in acetone for 5 minutes, and finally soaking in deionized water for 2 minutes. According to the best solution determined by the model, the carrier is immersed in different stripping solutions in a Class 1000 dust-free workshop in turn. At the same time, the workshop is equipped with a high-efficiency air filtration system to filter particles with a diameter of ≥5μm in the air to ensure that the number of suspended particles is ≤1000 / m 3 In addition, anti-static floors, gloves, work clothes, etc. are used to make the electrostatic potential of the workshop ≤100V to avoid electrostatic damage. After the immersion is completed, the residual amount is tested again. If the residual amount is ≤1μg / cm 2 If the immersion solution is not satisfactory, the optimization model is returned to adjust the immersion solution until it is satisfactory. Multiple batches of immersion data, such as the type, concentration, duration, temperature, and residual amount of the immersion solution, are fed back to the algorithm model for deep learning. The model parameters are continuously corrected through methods such as gradient descent to improve the prediction accuracy, and finally the whole process of substrate immersion treatment is intelligentized.

[0024] S102. According to the size and material property data of the carrier, the ultrasonic oscillation frequency and duration parameters are obtained through algorithm model training and evaluation, and the carrier is staggered in the horizontal direction and rotated in the vertical direction to form an oscillation arrangement. The oscillation process strictly controls the temperature and humidity conditions in the dust-free area.

[0025] The size and material property data of the carrier are obtained as input parameters of the algorithm model. The machine learning algorithm is used to train and evaluate the input parameters through support vector machines or neural networks to obtain the optimal ultrasonic oscillation frequency and duration parameters. According to the size of the carrier, the staggered arrangement in the horizontal direction is determined, and the computer vision algorithm is used to determine whether the arrangement is reasonable. According to the material characteristics, the rotation angle in the vertical direction is determined, and the rotation effect is evaluated through the simulation algorithm. Combine horizontal staggering and vertical rotation to form the best oscillation arrangement method, and continuously iterate and optimize the arrangement through the optimization algorithm. In the dust-free area, temperature and humidity sensors are used to collect environmental data in real time, and the temperature and humidity parameters are adjusted through the PID control algorithm to ensure that the oscillation requirements are met. Comprehensive parameters such as oscillation frequency, duration, arrangement method and environmental conditions are used, and the final ultrasonic oscillation process parameters are obtained through simulation and experimental verification to guide actual production.

[0026] Specifically, the size and material property data of the carrier are obtained as input parameters of the algorithm model; the input parameters are trained and evaluated by a machine learning algorithm to obtain the optimal ultrasonic oscillation frequency and duration parameters; according to the size of the carrier, whether the staggered arrangement in the horizontal direction is reasonable is judged by a computer vision algorithm; according to the material properties, the effect of the rotation angle in the vertical direction is evaluated by a simulation algorithm; the horizontal staggered arrangement and the vertical rotation angle are combined to form an oscillation arrangement; the oscillation arrangement is iteratively optimized by an optimization algorithm to obtain the optimal oscillation arrangement; the environmental data in the dust-free area is obtained; according to the environmental data, the temperature and humidity parameters are adjusted by a PID control algorithm, and the adjusted temperature and humidity parameters are output; the ultrasonic oscillation frequency, duration, optimal oscillation arrangement and adjusted temperature and humidity parameters are comprehensively considered, and the final ultrasonic oscillation process parameters are obtained through simulation and experimental verification.

[0027] Obtain the dimensional data of the carrier such as length, width, thickness, and the characteristic parameters of the material such as density, hardness, and elastic modulus, and input these data into the support vector machine algorithm model. Through training of 100 historical data samples, the optimal hyperparameters of the model are: oscillation frequency range 20-50kHz, duration range 10-30 seconds. According to the length and width of the carrier, the edge detection algorithm of computer vision is used to determine that the staggered arrangement of 2 rows and 3 columns in the horizontal direction is the most reasonable, and the spacing between the carriers is 10mm. According to the hardness and elastic modulus of the material, through finite element simulation, it is determined that rotating 30° in the vertical direction is the best. Using the genetic optimization algorithm, the horizontal staggered arrangement and vertical rotation angle are iterated and optimized 100 times to obtain the best oscillation arrangement scheme. In the dust-free workshop, temperature and humidity sensors are used to collect data in real time, the temperature is controlled at 25±2℃, the humidity is controlled at 50%±5%, and the working status of the air conditioner and humidifier is adjusted in real time through the PID control algorithm. Based on the above parameters, simulation was performed in the simulation software and verified in the laboratory. Finally, the optimal process parameters of ultrasonic oscillation were determined as follows: frequency 40kHz, duration 20 seconds, horizontal staggered arrangement of carriers, vertical rotation 30°, temperature 25°C, and humidity 50%. Applying this process for mass production, the organic glue removal rate of the carrier can reach more than 95%.

[0028] S103. According to the surface area of ​​the carrier and the residual glue amount data, an algorithm model is iteratively optimized to obtain a combination of overflow and spray parameters of ultrapure water, including flow rate, pressure and angle. The carrier is tilted and vertically spaced in multiple layers in the water tank to form a water flow arrangement. The flushing process is evaluated in real time through dust particle monitoring and surface cleaning verification.

[0029] The surface area and residual glue amount data of the carrier are obtained, and the ultrapure water overflow and spray parameter combination, including flow rate, pressure and angle, is obtained through iterative optimization according to the preset algorithm model. The carrier is placed in the water tank in an inclined manner and vertically spaced in multiple layers, and the optimal water flow arrangement is formed according to the determined ultrapure water overflow and spray parameter combination. During the rinsing process, the dust particle concentration data in the water tank is obtained in real time through the dust particle monitoring device set in the water tank. At the same time, the surface cleanliness detection device is used to detect the cleanliness of the carrier surface in real time to obtain the surface cleanliness data. The dust particle concentration data and the surface cleanliness data are input into the preset cleaning effect evaluation model, and the current cleaning effect is evaluated in real time through model calculation. If the evaluation result does not reach the preset cleaning effect threshold, the ultrapure water overflow and spray parameter combination is adjusted, the water flow arrangement is re-formed, and the rinsing is continued until the evaluation result reaches the threshold requirement. If the evaluation result reaches the preset cleaning effect threshold, the rinsing is stopped, the carrier is taken out of the water tank, and the cleaning process of the carrier is completed.

[0030] Specifically, the surface area and residual glue amount data of the carrier are obtained; according to the surface area and residual glue amount data of the carrier, a combination of ultrapure water overflow and spray parameters is determined through a preset iterative optimization algorithm model, and the ultrapure water overflow and spray parameter combination includes flow rate, pressure and angle; the carrier is placed in a water tank in an inclined manner and in a multi-layer vertical spacing manner, and a water flow arrangement is formed according to the ultrapure water overflow and spray parameter combination; during the flushing process, the dust particle concentration data in the water tank is obtained in real time through a dust particle monitoring device arranged in the water tank; at the same time, a surface cleanliness detection device is used to monitor the surface cleanliness of the carrier. The cleanliness of the surface is detected in real time to obtain surface cleanliness data; the dust particle concentration data and the surface cleanliness data are input into a preset cleaning effect evaluation model, and the current cleaning effect is evaluated in real time through the cleaning effect evaluation model operation; if the cleaning effect evaluation result does not reach the preset cleaning effect threshold, the ultrapure water overflow and spray parameter combination are adjusted, the water flow arrangement is reformed, and flushing is continued; if the cleaning effect evaluation result reaches the preset cleaning effect threshold, the flushing is stopped, the carrier is taken out of the water tank, and the cleaning process of the carrier is completed.

[0031] First, the surface area data of the carrier is obtained by an optical measuring instrument, accurate to 0.1 square centimeter. Then, the amount of glue remaining on the surface of the carrier is measured by a chemical analysis instrument, accurate to 1 microgram. The surface area and residual glue data obtained are input into the preset mathematical model, which is based on the theory of fluid mechanics and cleaning dynamics. Through finite element analysis and numerical simulation, the optimal ultrapure water overflow rate range is 5-0 liters / minute, the spray pressure range is 80-100 Pa, and the spray angle range is 30-45 degrees. In the ultrasonic water tank, the carrier is tilted at an angle of 15 degrees and divided into three layers in the vertical direction with a layer spacing of 10 cm. According to the optimized parameter combination, the ultrapure water overflow rate is set to 8 liters / minute, the spray pressure is 90 Pa, and the spray angle is 37 degrees to form the best water flow arrangement. A laser dust particle counter is installed in the water tank, which can monitor the dust particle concentration in the water in real time, and can detect particles as small as 1 micron. At the same time, an optical surface detector is used to scan the cleanliness of the carrier surface in real time with a resolution of 5 microns. The dust particle concentration and surface cleanliness data are substituted into the preset fuzzy logic evaluation model, which comprehensively calculates the cleaning effect score through the membership function and the evaluation rule base. If the cleaning effect score is less than 85 points, the parameter adjustment program is triggered. Through genetic algorithms and neural networks, combined with previous cleaning data, the ultrapure water overflow and spray parameter combination is re-optimized to generate a new water flow arrangement and continue the flushing process. If the cleaning effect score reaches 85 points or more, the flushing is stopped, and the carrier is removed from the water tank by a robotic arm and sent to the next process. The entire process is automatically controlled by a computer without manual intervention, ensuring a stable and reliable cleaning effect.

[0032] S104, using an online monitoring device for the surface cleanliness of the carrier plate to collect surface cleanliness data during the cleaning process in real time, using an algorithm to analyze the cleanliness change trend in real time, dynamically adjusting the soaking, shaking and rinsing processing parameters and circulation path, adaptively optimizing the cleaning plan, and predicting the cleaning endpoint.

[0033] The cleanliness data of the carrier surface is obtained, and the cleanliness data is input into a pre-built cleanliness change trend analysis model to obtain the current cleanliness change trend; based on the cleanliness change trend obtained by analysis, it is determined whether the current cleaning process parameters need to be adjusted. If adjustment is required, the cleaning process parameters such as immersion time, oscillation frequency, and flushing pressure are dynamically adjusted; based on the adjusted cleaning process parameters, the next cleaning cycle path is determined, and the cleaning equipment is controlled to perform cleaning operations according to the determined cycle path; after each cleaning cycle, the cleanliness data of the carrier surface is re-obtained and input into the cleanliness change trend analysis model to adaptively optimize the cleaning plan; by comparing the cleanliness change trends after multiple cleaning cycles, the long short-term memory neural network algorithm is used to predict the cleaning time required to achieve the target cleanliness and obtain the cleaning endpoint; if the difference between the predicted cleaning endpoint and the current cleaning time is less than a preset threshold, it is determined that the cleaning is about to end, and the cleaning end preparation program is started; based on the cleaning endpoint prediction result, the remaining cleaning cycle paths and cleaning process parameters are dynamically adjusted to ensure that the cleaning operation is completed at the predicted endpoint.

[0034] Specifically, cleanliness data of the surface of the carrier are obtained, and the cleanliness data are input into a pre-constructed cleanliness change trend analysis model to obtain the current cleanliness change trend; according to the cleanliness change trend, it is determined whether the current cleaning process parameters need to be adjusted, and if adjustment is required, at least one of the cleaning process parameters of immersion time, oscillation frequency and flushing pressure is dynamically adjusted; according to the adjusted cleaning process parameters, the next cleaning cycle path is determined, and the cleaning equipment is controlled to perform cleaning operations according to the cleaning cycle path; after each cleaning cycle, the cleanliness data of the surface of the carrier is re-obtained, and input into the cleanliness change trend analysis model to adaptively optimize the cleaning scheme; by comparing the cleanliness change trends after multiple cleaning cycles, the long short-term memory neural network algorithm is used to predict the cleaning time required to achieve the target cleanliness, and obtain the cleaning endpoint; if the difference between the cleaning endpoint and the current cleaning time is less than a preset threshold, it is determined that the cleaning is about to end, and the cleaning end preparation program is started; according to the cleaning endpoint prediction result, the remaining cleaning cycle paths and cleaning process parameters are dynamically adjusted.

[0035] During the cleaning process, the cleanliness data of the carrier surface is obtained by a high-precision optical camera, and the cleanliness data is input into the cleanliness change trend analysis model built based on the support vector machine algorithm. The current cleanliness change rate is calculated to be 8% / min. According to the preset cleanliness change rate threshold of 5% / min, it is judged that the current cleaning process parameters need to be adjusted, and the immersion time is extended from 10min to 15min, the oscillation frequency is increased from 50Hz to 60Hz, and the flushing pressure is increased from 2MPa to 25MPa. According to the adjusted cleaning process parameters, the cleaning cycle path is optimized, and the cleaning equipment is controlled to perform cleaning operations according to the "immersion-oscillation-rinsing-immersion" cycle path. After each cleaning cycle, the cleanliness data of the carrier surface is re-obtained and input into the cleanliness change trend analysis model, and the cleaning scheme is adaptively optimized by the gradient descent algorithm. By comparing the cleanliness change trend after 5 cleaning cycles, the long short-term memory neural network algorithm is used to predict that the cleaning time required to reach the target cleanliness of 999% is 30min. If the difference between the predicted cleaning endpoint and the current cleaning time is less than 2 minutes, it is judged that the cleaning is about to end, and the cleaning end preparation procedures such as cleaning liquid recovery and carrier drying are started. According to the cleaning endpoint prediction results, the remaining cleaning cycles and the cleaning processing parameters of each cycle are dynamically adjusted, such as shortening the immersion time of the last two cycles to 5 minutes, reducing the oscillation frequency to 45Hz, and reducing the flushing pressure to 15MPa, to ensure that the cleaning operation is completed at the predicted endpoint, improving the cleaning efficiency and cleanliness stability.

[0036] S105. Deploy a colloid agglomeration degree analyzer in the water tank to monitor the agglomeration state data of the colloid particles in real time, analyze the correlation between the agglomeration state and water quality parameters through an algorithm model, dynamically adjust the pH value and ionic strength of ultrapure water, optimize the water quality environment, and inhibit the re-deposition of colloid particles.

[0037] A dedicated colloid agglomeration analyzer is deployed in the water tank. The device can monitor the agglomeration state of colloid particles in water in real time through high-precision sensors. After obtaining the data from the analyzer, it is processed by a pre-established algorithm model, which can analyze the relationship between the colloid agglomeration state and the pH value and ionic strength of water quality parameters. According to the analysis results of the algorithm model, the pH value and ionic strength in the water tank are automatically adjusted by the control system to optimize the water quality. The optimized water quality will be monitored again by the same analyzer to confirm whether the effect of the water quality adjustment meets expectations. If the adjusted water quality does not meet the expected effect, the algorithm model will re-analyze the data and adjust the control parameters until the optimal water quality state is achieved. Throughout the process, the system will continue to monitor the redeposition of colloid particles to ensure that the water quality remains in the optimal state for a long time. Through this dynamic adjustment mechanism, not only the quality of ultrapure water is improved, but also the redeposition of colloid particles is effectively prevented, ensuring the stable operation of the system and the continuous optimization of water quality.

[0038] Specifically, a special colloid agglomeration degree analyzer deployed in the water tank obtains the agglomeration state data of the colloids in the water through a high-precision sensor. According to the data, the data is processed by a pre-established algorithm model, and the algorithm model is used to analyze the relationship between the colloid agglomeration state and the water quality parameters. The pH value and ionic strength in the water tank are automatically adjusted according to the analysis results of the algorithm model by the control system to optimize the water quality. The optimized water quality is monitored again by the same analyzer to determine the effect of the water quality adjustment. If the adjusted water quality does not achieve the expected effect, the algorithm model will re-analyze the data and adjust the control parameters until the optimal water quality state is achieved. During the whole process, the system continuously monitors the redeposition of colloids. Through this dynamic adjustment mechanism, the quality of ultrapure water is improved, the redeposition of colloids is effectively prevented, and the stable operation of the system and the continuous optimization of water quality are guaranteed.

[0039] In the water treatment system, a dedicated colloid agglomeration analyzer is deployed. The device is equipped with a high-precision sensor that can monitor the agglomeration state of colloid particles in water in real time. For example, the sensor collects data once a minute with an accuracy of 0.1 micron. When the analyzer collects the data, the data will be sent to a pre-established algorithm model for processing. The model is based on the support vector machine (SVM) algorithm and can accurately analyze the complex relationship between the state of colloid agglomeration and water quality parameters, such as the change in pH value within the range of 5 to 5, and how the adjustment of ionic strength from 0.1 to 1 mol / L affects the stability of colloid particles. According to the results of the model analysis, the control system automatically adjusts the pH value and ionic strength in the water tank. For example, if the model predicts that the stability of colloid particles is reduced, the system may adjust the pH value to 8 and increase the ionic strength to 0.5 mol / L to optimize the water quality. The optimized water quality is monitored again by the analyzer to confirm whether the effect of the water quality adjustment is as expected. If the adjusted water quality does not achieve the expected effect, the algorithm model will re-analyze and adjust the control parameters based on the new data, such as further increasing the ion strength to 0.8 mol / L, until the optimal water quality is achieved. Throughout the process, the system will also continue to monitor the redeposition of colloid particles, and ensure that the water quality remains in the optimal state for a long time through periodic data analysis, thereby not only improving the quality of ultrapure water, but also effectively preventing the redeposition of colloid particles, ensuring the stable operation of the system and the continuous optimization of water quality.

[0040] By obtaining real-time monitoring data from the high-precision sensor of the colloid agglomeration degree analyzer, the pre-established algorithm model is used to analyze the relationship between the colloid agglomeration state and the pH value and ionic strength of the water quality parameters. The water quality is adjusted according to the analysis results. If the adjusted water quality does not achieve the expected effect, the algorithm model re-analyzes the data and adjusts the control parameters until the optimal water quality is obtained.

[0041] Real-time monitoring data of the state of colloid agglomeration is obtained through high-precision sensors, and the monitoring data is transmitted to the data processing module. The data processing module preprocesses the received monitoring data, extracts key characteristic parameters, and inputs the processed data into the pre-built model. The pre-built model analyzes the relationship between the state of colloid agglomeration and water quality parameters based on the input characteristic parameters, and obtains the water quality adjustment strategy. The water quality adjustment strategy is transmitted to the water quality control module, and the water quality control module adjusts the water quality parameters through the control equipment according to the adjustment strategy. After the adjustment is completed, the high-precision sensor continues to monitor the state of colloid agglomeration in real time and transmits the monitoring data to the data processing module. The data processing module compares and analyzes the data of the state of colloid agglomeration before and after the adjustment to determine whether the water quality adjustment effect meets expectations. If it does not meet expectations, the analysis results are fed back to the pre-built model. The pre-built model retrains and optimizes the model parameters based on the feedback analysis results, and generates a new water quality adjustment strategy.

[0042] Specifically, in this water treatment system, a high-precision sensor collects data on the state of colloid agglomeration once a second with an accuracy of up to 1 micron. The sensor transmits the real-time monitoring data to the data processing module via industrial Ethernet. The data processing module uses the Kalman filter algorithm to denoise and smooth the raw data, and extracts 10 key characteristic parameters such as the average particle size of the colloid particles and the number of agglomerates. The processed characteristic parameter data is input into a pre-built model based on the support vector machine (SVM) algorithm. The model analyzes the input characteristic parameters, establishes a nonlinear relationship between the state of colloid agglomeration and five water quality parameters such as pH value and ionic strength, and obtains the probability of colloid re-deposition under the current water quality. If the probability of re-deposition exceeds 5%, the model will generate a water quality adjustment strategy, such as adjusting the pH value from 0 to 5 and the ionic strength from 1 mol / L to 3 mol / L. The adjustment strategy is transmitted to the water quality control module via a digital signal. The control module adjusts the water quality parameters within 30 seconds by adjusting the flow rate of acid and alkali liquids and the amount of flocculant added according to the strategy instructions. After the adjustment is completed, the sensor continues to collect data on the state of colloid agglomeration at a frequency of once per second. The data processing module compares and analyzes the data within 1 minute before and after the adjustment, and calculates the rate of change of the probability of colloid re-deposition. If the rate of change does not reach the preset 80%, the analysis results are fed back to the pre-built model. The model uses optimization algorithms such as the gradient descent method to retrain the model parameters based on the feedback data and generate a new water quality adjustment strategy, such as further adjusting the ion strength to 5 mol / L. The water quality adjustment and effect evaluation are performed in this cycle until the probability of re-deposition is reduced to below 1% and the optimal water quality state is achieved. Through this intelligent control system, precise regulation of water quality and effective suppression of colloid re-deposition can be achieved, ensuring the stable and efficient operation of the ultrapure water treatment process.

[0043] S106. Use a particle removal capacity evaluation device to evaluate the removal efficiency of each particle size segment for particles of different sizes, analyze the relationship between the removal efficiency, filtration accuracy and water flow rate through an algorithm, dynamically adjust the technical parameters of the filtration device, and optimize the horizontal arrangement spacing and staggered arrangement to form a particle removal process that is compatible with the degree of agglomeration of the colloid particles.

[0044] The evaluation device is used to collect data on the removal efficiency of particles of different sizes, and the correlation between the removal efficiency, filtration accuracy and water flow rate is obtained through algorithm analysis. Based on these data, the technical parameters of the filtration device are dynamically adjusted to improve the filtration efficiency. The adjusted parameters are obtained to optimize the horizontal layout spacing and staggered arrangement to adapt to the degree of agglomeration of the colloid particles. Through this optimization, a new particle removal process is formed, which can effectively improve the removal efficiency. According to the implementation effect of the new process, the evaluation device is used again to test the effect, and the test results are analyzed by algorithm to determine whether it is necessary to further adjust the technical parameters or optimize the layout. If the test results are satisfactory, the process is standardized and used for the particle removal process in practical applications.

[0045] Specifically, the removal efficiency data of different particle sizes are obtained, and the data are collected by an evaluation device. Based on the data, an algorithm is used to analyze the correlation between the removal efficiency and the filtration accuracy and the water flow rate. The technical parameters of the filtration device are dynamically adjusted according to the results of the algorithm analysis. The adjusted technical parameters are obtained to optimize the horizontal layout spacing and the staggered arrangement. A new particle removal process is formed, which effectively improves the removal efficiency through optimized technical parameters and layout. An evaluation device is used to test the effect of the implementation of the new process, and the test results are analyzed by an algorithm to determine whether it is necessary to further adjust the technical parameters or optimize the layout. If the test results are satisfactory, the process is standardized and used for the particle removal process in actual applications.

[0046] In the implementation of the new particle removal process, the removal efficiency data of different particle sizes are first collected through a high-precision evaluation device. The correlation between the removal efficiency and the filtration accuracy (in the range of 1 micron to 10 microns) and the water flow rate (5 to 3 meters per second) is analyzed using machine learning algorithms, such as support vector machines (SVM) or decision trees. According to the results of the algorithm analysis, the technical parameters of the filtration device are dynamically adjusted, such as increasing the density of the filter layer or changing the pore size of the filter material to improve the filtration efficiency. After the adjustment, the horizontal arrangement spacing (adjusted to 5 to 20 cm) and the staggered arrangement (staggered at a 45-degree angle) are optimized to adapt to the degree of agglomeration of the colloid particles, thereby reducing the clogging of the colloid particles during the filtration process. Through this optimization, a new particle removal process is formed, which is tested by using the evaluation device again, collecting data and analyzing it through a deep learning algorithm to ensure that the removal efficiency reaches the expected target (at least 30% increase). If the test results show that the removal efficiency and water quality meet the set standards, the process is standardized and applied to the actual water treatment process to achieve efficient and stable particle removal effects.

[0047] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present application. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for cleaning organic glue residue on a carrier surface, characterized in that: The method comprises the following steps: S101, obtaining data on the residual amount of organic glue on the surface of the carrier and the material size characteristics, determining the stripping liquid immersion sequence and duration combination through algorithm model selection and parameter setting, and obtaining a multi-step immersion treatment plan, wherein the immersion process is carried out in a dust-free area that meets the cleanliness level requirements, and an air filter device and anti-static measures are used to ensure the cleanliness of the environment; S102, according to the size and material property data of the carrier, through algorithm model training and evaluation, the ultrasonic oscillation frequency and duration parameters are obtained, the carrier is staggered in the horizontal direction, and rotated in the vertical direction to form an oscillation arrangement, and the oscillation process strictly controls the temperature and humidity conditions of the dust-free zone; S103, according to the surface area of ​​the carrier and the residual glue amount data, an algorithm model is iteratively optimized to obtain a combination of overflow and spray parameters of ultrapure water, including flow rate, pressure and angle, and the carrier is placed in a water tank at an angle and vertically spaced in multiple layers to form a water flow arrangement. The rinsing process is evaluated in real time through dust particle monitoring and surface cleaning verification; S104, using an online monitoring device for the surface cleanliness of the carrier to collect surface cleanliness data in real time during the cleaning process, using an algorithm to analyze the cleanliness change trend in real time, dynamically adjusting the soaking, shaking and rinsing treatment parameters and circulation paths, adaptively optimizing the cleaning scheme, and predicting the cleaning endpoint; S105, deploying a colloid agglomeration degree analyzer in the water tank to monitor the agglomeration state data of the colloid particles in real time, analyzing the correlation between the agglomeration state and the water quality parameters through an algorithm model, dynamically adjusting the pH value and ionic strength of the ultrapure water, optimizing the water quality environment, and inhibiting the re-deposition of colloid particles; S106, using a particle removal capacity evaluation device to evaluate the removal efficiency of each particle size segment for particles of different sizes, analyzing the relationship between the removal efficiency and the filtration accuracy and the water flow rate through an algorithm, dynamically adjusting the technical parameters of the filtration device, and optimizing the horizontal arrangement spacing and staggered arrangement to form a particle removal process that is compatible with the degree of agglomeration of the colloid particles.

2. The method for cleaning organic glue residue on the surface of a carrier according to claim 1, characterized in that: The S101 also includes: Step 1: Obtain the data of organic glue residue on the substrate surface and material size characteristic data as input of the algorithm model; Step 2: Based on the material and size characteristics of the substrate surface, a decision tree algorithm model is used to determine the optimal stripping solution immersion sequence and duration combination through parameter setting; Step 3: Soak the substrate in different stripping solutions in a dust-free area that meets the cleanliness requirements according to the determined soaking order and duration; Step 4: During the soaking process, an air filter is used to continuously filter the air in the dust-free area to remove particles and dust in the air; Step 5: During the soaking process, anti-static measures are taken to prevent static electricity from damaging the surface of the carrier board. Step 6: After the immersion is completed, the data of the residual amount of organic glue on the surface of the carrier is obtained to determine whether the residual amount is lower than the preset threshold. If so, the immersion process is completed. If not, return to step 2, adjust the immersion order and duration, and continue the immersion process. In step 7, the data of multiple batches of substrate immersion treatment are input into the algorithm model for training. Through parameter optimization, the accuracy of the model's prediction of the immersion sequence and duration combination is continuously improved to realize the intelligentization of the substrate immersion treatment process.

3. The method for cleaning organic glue residue on the surface of a carrier according to claim 1, characterized in that: The S102 also includes: Obtain the size and material property data of the carrier board as input parameters of the algorithm model; Using machine learning algorithms, support vector machines or neural networks, the input parameters are trained and evaluated to obtain the optimal ultrasonic oscillation frequency and duration parameters; Determine the staggered arrangement in the horizontal direction according to the size of the carrier board, and use the computer vision algorithm to determine whether the arrangement is reasonable; According to the material characteristics, the rotation angle in the vertical direction is determined, and the rotation effect is evaluated through the simulation algorithm; Combine horizontal staggering and vertical rotation to form the best oscillation layout, and continuously iterate and optimize the layout through optimization algorithms; In the dust-free area, temperature and humidity sensors are used to collect environmental data in real time, and the temperature and humidity parameters are adjusted through the PID control algorithm to ensure that the vibration requirements are met; By comprehensively considering the oscillation frequency, duration, layout and environmental condition parameters, through simulation and experimental verification, the final ultrasonic oscillation process parameters are obtained to guide actual production.

4. The method for cleaning organic glue residue on the surface of a carrier according to claim 1, characterized in that: The S103 also includes: Obtain the data of the surface area and residual glue amount of the carrier plate, and obtain the ultrapure water overflow and spray parameter combination, including flow rate, pressure and angle, through iterative optimization according to the preset algorithm model; The carrier plate is placed in the water tank in an inclined manner and vertically spaced in multiple layers, and the optimal water flow arrangement is formed according to the determined ultrapure water overflow and spray parameter combination; During the flushing process, the dust particle concentration data in the water tank is obtained in real time through the dust particle monitoring device installed in the water tank; At the same time, a surface cleanliness detection device is used to detect the cleanliness of the carrier surface in real time to obtain surface cleanliness data; Input dust particle concentration data and surface cleanliness data into the preset cleaning effect evaluation model, and evaluate the current cleaning effect in real time through model calculation; If the evaluation result does not reach the preset cleaning effect threshold, the ultrapure water overflow and spray parameter combination is adjusted to re-form the water flow arrangement and continue flushing until the evaluation result reaches the threshold requirement; If the evaluation result reaches a preset cleaning effect threshold, the rinsing is stopped, the carrier is taken out of the water tank, and the cleaning process of the carrier is completed.

5. The method for cleaning organic glue residue on the surface of a carrier according to claim 1, characterized in that: The S104 also includes: Acquire cleanliness data of the carrier surface, input the cleanliness data into a pre-built cleanliness change trend analysis model, and obtain the current cleanliness change trend; According to the cleanliness change trend obtained by analysis, it is determined whether the current cleaning process parameters need to be adjusted. If adjustment is required, at least one cleaning process parameter of the soaking time, the oscillation frequency and the flushing pressure is dynamically adjusted; According to the adjusted cleaning processing parameters, the next cleaning cycle path is determined, and the cleaning equipment is controlled to perform the cleaning operation according to the determined cycle path; After each cleaning cycle, the cleanliness data of the substrate surface is re-obtained and input into the cleanliness trend analysis model to adaptively optimize the cleaning solution. By comparing the cleanliness change trends after multiple cleaning cycles, the long short-term memory neural network algorithm is used to predict the cleaning time required to reach the target cleanliness and obtain the cleaning endpoint. If the difference between the predicted cleaning end point and the current cleaning time is less than a preset threshold, it is determined that the cleaning is about to end, and the cleaning end preparation procedure is started; According to the cleaning endpoint prediction results, the remaining cleaning cycle paths and cleaning processing parameters are dynamically adjusted to ensure that the cleaning operation is completed at the predicted endpoint moment.

6. The method for cleaning organic glue residue on the surface of a carrier according to claim 1, characterized in that: The S105 also includes: Deploy a dedicated particle aggregation analyzer in the water tank. This device can monitor the aggregation state of particles in the water in real time through high-precision sensors; After acquiring the data from the analyzer, it is processed by a pre-established algorithm model, which can analyze the relationship between the state of colloidal aggregation and the pH value and ionic strength of water quality parameters; Based on the analysis results of the algorithm model, the pH value and ionic strength in the water tank are automatically adjusted through the control system to optimize the water quality; The optimized water quality will be monitored again by the same analyzer to confirm whether the effect of the water quality adjustment is as expected; If the adjusted water quality does not achieve the expected results, the algorithm model will reanalyze the data and adjust the control parameters until the optimal water quality is achieved; During the whole process, the system will continuously monitor the redeposition of the colloid particles to ensure that the water quality remains at the best state for a long time; This dynamic adjustment mechanism not only improves the quality of ultrapure water, but also effectively prevents the redeposition of colloid particles, ensuring the stable operation of the system and the continuous optimization of water quality. It also includes: obtaining real-time monitoring data from the high-precision sensor of the colloid agglomeration degree analyzer, using a pre-established algorithm model to analyze the relationship between the colloid agglomeration state and the pH value and ionic strength of water quality parameters, and adjusting the water quality according to the analysis results. If the adjusted water quality does not achieve the expected effect, the algorithm model re-analyzes the data and adjusts the control parameters until the optimal water quality state is obtained.

7. The method for cleaning organic glue residue on the surface of a carrier according to claim 6, characterized in that: Also includes: The real-time monitoring data of the agglomeration state of the colloid particles is obtained through high-precision sensors, and the monitoring data is transmitted to the data processing module; The data processing module pre-processes the received monitoring data, extracts key characteristic parameters, and inputs the processed data into the pre-built model; The pre-built model analyzes the relationship between the colloid agglomeration state and water quality parameters based on the input characteristic parameters, and obtains the water quality adjustment strategy; The water quality adjustment strategy is transmitted to the water quality control module, and the water quality control module adjusts the water quality parameters through the control device according to the adjustment strategy; After the adjustment is completed, the high-precision sensor continues to monitor the agglomeration state of the colloid particles in real time and transmits the monitoring data to the data processing module; The data processing module compares and analyzes the data of the agglomeration state of the colloid particles before and after the adjustment to determine whether the water quality adjustment effect meets the expectations; If expectations are not met, the analysis results are fed back into the pre-built model; The pre-built model retrains and optimizes model parameters based on the feedback analysis results and generates new water quality adjustment strategies.

8. The method for cleaning organic glue residue on the surface of a carrier according to claim 1, characterized in that: The S106 also includes: An evaluation device was used to collect data on the removal efficiency of particles of different sizes, and the correlation between the removal efficiency, filtration accuracy and water flow rate was obtained through algorithm analysis; Based on these data, the technical parameters of the filtration device are dynamically adjusted to improve the filtration efficiency; Obtain the adjusted parameters and optimize the horizontal arrangement spacing and staggered arrangement to adapt to the agglomeration degree of the rubber particles; Through this optimization, a new particle removal process is formed, which can effectively improve the removal efficiency; According to the implementation effect of the new process, the evaluation device is used again to test the effect, and the test results are analyzed by algorithms to determine whether it is necessary to further adjust the technical parameters or optimize the layout; If the test results are satisfactory, the process will be standardized and used for particle removal in actual applications.

Images