Intelligent control method for wafer cleaning and conveying process

Abstract

The invention relates to the technical field of information, in particular to an intelligent control method for a wafer cleaning and conveying process of a semiconductor wafer cleaning technology, which comprises the following steps of: in the wafer conveying process, detecting vibration and displacement signals of a conveying device in real time by adopting a gyroscope sensor; the collected signals are input into a pre-trained support vector machine model to be classified, whether current vibration and displacement exceed preset threshold values or not is judged, the support vector machine model is obtained through historical data training, and normal and abnormal vibration and displacement states can be accurately distinguished; when vibration or displacement exceeding a threshold value is detected, a servo motor is triggered to carry out compensation control, vibration and displacement are restrained by adjusting the conveying speed and acceleration parameters, it is ensured that the wafer is stably conveyed to the rotary cleaning device, and a compensation control algorithm dynamically adjusts control parameters according to the amplitude of vibration and displacement, so that quick response and accurate compensation are achieved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, in particular to semiconductor wafer cleaning technology, and in particular to an intelligent control method for a wafer cleaning and conveying process. Background Art

[0002] In the semiconductor wafer manufacturing process, wafer cleaning is a key step. The commonly used wafer conveying device is prone to vibration and displacement during the conveying process, which affects the cleaning effect. The position of the wafer on the conveyor belt is not accurate enough, and dead corners are prone to appear during cleaning, resulting in inadequate cleaning.

[0003] In addition, wafers of different sizes and weights require different conveying speeds and cleaning parameters. Existing conveying devices cannot be dynamically adjusted according to the characteristics of the wafers, which affects the efficiency and quality of cleaning.

[0004] The rotation angle and speed of the wafer during transportation are also important factors affecting the cleaning effect. An inappropriate rotation angle will result in the cleaning liquid not being able to evenly cover the wafer surface, and too fast or too slow rotation speed will affect the thoroughness of cleaning.

[0005] Therefore, it is necessary to design a wafer conveying device that can accurately control the wafer position, dynamically adjust the conveying speed, optimize the rotation angle and speed, and a matching cleaning method to improve the stability and quality of wafer cleaning and ensure the reliability and yield of semiconductor devices. Summary of the invention

[0006] The present invention provides an intelligent control method for a wafer cleaning and conveying process, the method comprising the following steps:

[0007] The machine vision system is used to detect the position coordinates of the wafer on the conveyor belt in real time. Based on the acquired coordinate data, the servo motor accurately controls the movement of the conveyor belt to adjust the wafer to the preset optimal cleaning position. The optimal cleaning position is defined as the position where the center of the wafer is aligned with the center of the cleaning nozzle, ensuring that the cleaning liquid evenly covers the wafer surface and eliminating cleaning dead corners caused by wafer position deviation.

[0008] Based on the wafer image obtained by the machine vision system, the size information of the wafer is calculated. At the same time, the wafer weight is measured using a high-precision weighing sensor, and a mapping table of wafer size and weight and optimal conveying speed is established. The mapping table is constructed based on a large amount of experimental data and expert experience. The corresponding optimal conveying speed parameters are queried from the mapping table, and the servo motor is controlled to adjust the conveyor belt speed to achieve dynamic optimization of the conveying speed.

[0009] During the wafer conveying process, a gyroscope sensor is used to detect the vibration and displacement signals of the conveying device in real time. The collected signals are input into a pre-trained support vector machine model for classification to determine whether the current vibration and displacement exceed the preset threshold. The support vector machine model is trained using historical data and can accurately distinguish between normal and abnormal vibration and displacement states.

[0010] When vibration or displacement exceeding the threshold is detected, the servo motor is triggered to perform compensation control. By adjusting the conveying speed and acceleration parameters, vibration and displacement are suppressed to ensure stable conveyance of the wafer to the rotary cleaning device. The compensation control algorithm dynamically adjusts the control parameters according to the amplitude of vibration and displacement to achieve fast response and precise compensation.

[0011] For the spin cleaning process, before the cleaning liquid is sprayed, a computer vision algorithm is used to analyze the static wafer surface image. The initial state of the wafer surface is determined through image segmentation and feature extraction. During the cleaning process, a high-speed camera is used to capture the instantaneous image of the rotating wafer. Combined with the initial state analysis results, the uniformity of the cleaning liquid coverage on the wafer surface is evaluated to obtain a quantitative coverage score.

[0012] Compare the coverage score with the preset threshold. When it is lower than the threshold, dynamically adjust the rotation angle and speed of the wafer. Optimize the rotation parameters through the fuzzy control algorithm. The algorithm outputs the optimal angle and speed adjustment based on the three input variables of coverage score, rotation angle and speed. The fuzzy rules are set according to expert knowledge. The effectiveness of the algorithm is ensured through repeated tuning. The coverage trend is continuously monitored and the cleaning parameters are optimized in real time.

[0013] According to the changing trend of coverage and the rotation time of the wafer, a reinforcement learning algorithm is used to continuously optimize the control strategy of rotation angle and speed. The reinforcement learning algorithm aims to maximize the thoroughness of cleaning and uses the coverage score as a reward signal. The optimal combination of rotation parameters is learned through multiple iterations. The characteristics of different types of wafers are considered during the learning process to form a highly adaptable control strategy and improve the stability of the cleaning effect.

[0014] After each wafer is cleaned, defect detection is performed on the cleaned wafer surface. The surface defects are identified and located through a deep learning algorithm, and a defect distribution map is generated. The defect information is correlated with the previous cleaning parameters for analysis to find out the factors that may lead to incomplete cleaning. The analysis results are fed back to the upstream process parameter optimization system to adjust the cleaning fluid formula, injection pressure and other parameters, continuously improve the wafer cleaning process, and form a closed-loop optimization.

[0015] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0016] The present invention discloses an intelligent control method for a wafer cleaning and conveying process, which uses machine vision to detect the position of the wafer in real time, combines high-precision weighing and vibration detection, and dynamically optimizes the conveying speed and position. In the rotation cleaning stage, computer vision is used to analyze the uniformity of cleaning liquid coverage, and fuzzy control and reinforcement learning algorithms are used to adjust the rotation parameters in real time to ensure the thoroughness of cleaning. Defect detection is performed after cleaning, and the results are fed back to the process parameter optimization system to form a closed-loop improvement. The present invention realizes intelligent control of the entire wafer cleaning process, significantly improves the cleaning quality and efficiency, reduces manual intervention, and provides an innovative solution for the wafer cleaning process in semiconductor manufacturing. DETAILED DESCRIPTION

[0017] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the embodiments. The present application is further described in detail below in conjunction with the embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention.

[0018] The intelligent control method of a wafer cleaning and conveying process in this embodiment specifically includes the following steps:

[0019] S1. Use a machine vision system to detect the position coordinates of the wafer on the conveyor belt in real time. Based on the acquired coordinate data, the servo motor accurately controls the movement of the conveyor belt to adjust the wafer to the preset optimal cleaning position. The optimal cleaning position is defined as the position where the center of the wafer is aligned with the center of the cleaning nozzle, ensuring that the cleaning liquid evenly covers the wafer surface and eliminates cleaning dead corners caused by wafer position deviation.

[0020] S1 includes: acquiring image data of wafers on the conveyor belt, the image data being acquired by a machine vision system through real-time image acquisition; preprocessing the image data, denoising and enhancing the image data using an image preprocessing algorithm, and acquiring a preprocessed wafer image; extracting contour features of the wafer from the preprocessed wafer image, calculating the geometric center coordinates of the wafer based on the contour features, and acquiring real-time position coordinate data of the wafer on the conveyor belt; acquiring preset optimal cleaning position coordinates, comparing the real-time position coordinate data of the wafer on the conveyor belt with the optimal cleaning position coordinates, and calculating the deviation between the center of the wafer and the optimal cleaning position; if the deviation exceeds a preset threshold, triggering a servo motor to adjust the conveyor belt movement; based on the deviation, using The PID control algorithm controls the rotation speed and movement direction of the servo motor, and moves the wafer to the optimal cleaning position by adjusting the movement of the conveyor belt so that the center of the wafer is aligned with the center of the cleaning nozzle; when the wafer reaches the optimal cleaning position, the cleaning nozzle is triggered to start, the flow rate and spraying time of the cleaning liquid are controlled, and the cleaning liquid is evenly sprayed on the center of the wafer so that the cleaning liquid completely covers the wafer surface; after cleaning, the machine vision system is used to collect and analyze images of the wafer surface to detect the cleanliness of the wafer surface; if the cleanliness does not meet the preset standard, the cleaning process is triggered again until the wafer surface is completely clean; the qualified wafers are transmitted to the next process, the cleaning process data are recorded and statistically analyzed, the cleaning effect is evaluated, and the cleaning parameters are dynamically optimized.

[0021] Specifically, the machine vision system is a key technology in the wafer cleaning process. It uses a high-resolution camera to collect images of wafers on the conveyor belt in real time to obtain the position and surface state information of the wafer. For example, a 10-megapixel industrial camera can be used to collect images at a speed of 30 frames per second to ensure that the subtle features of the wafer are captured. Image preprocessing is an important step to improve the quality of wafer images. Common preprocessing methods include Gaussian filtering denoising and histogram equalization to enhance contrast. Gaussian filtering can effectively remove random noise in the image, while histogram equalization can enhance the overall contrast of the image and make the wafer edge clearer. The edge detection algorithm is used to extract the contour features of the wafer. The Canny algorithm is a commonly used edge detection method. It can accurately extract the edge contour of the wafer by calculating the gradient amplitude and direction of the image and performing non-maximum suppression and double threshold detection. According to the extracted contour, the geometric center coordinates of the wafer can be calculated to obtain the real-time position of the wafer on the conveyor belt. The PID control algorithm is used to accurately control the servo motor and adjust the movement of the conveyor belt. The PID controller calculates the appropriate control signal based on the deviation between the center of the wafer and the optimal cleaning position. For example, if the deviation is 5mm, the PID controller may output a corresponding voltage signal to make the servo motor run at the appropriate speed and direction to move the wafer to the optimal cleaning position. Control of the cleaning nozzle is the key to ensuring uniform cleaning of the wafer surface. A piezoelectric nozzle can be used to control the flow rate and spray range of the cleaning liquid by adjusting the vibration frequency of the piezoelectric crystal. For example, the nozzle can be set to vibrate at a frequency of 300Hz and spray 0.5ml of cleaning liquid per second for 10 seconds to ensure that the wafer surface is completely covered. Cleanliness detection is an important part of evaluating the cleaning effect. Image analysis techniques such as grayscale threshold segmentation or texture analysis can be used to detect residues on the wafer surface. For example, a cleanliness threshold, such as 99.9% of the surface area without visible residues, can be set as a standard for qualified cleaning. Data recording and statistical analysis help to continuously optimize the cleaning process. Data such as cleaning parameters, cleaning time and cleanliness of each wafer can be recorded, and the relationship between the cleaning effect and each parameter can be discovered through big data analysis techniques such as machine learning algorithms, so as to dynamically adjust the cleaning parameters and improve cleaning efficiency and quality. This data-driven optimization approach can significantly improve the overall efficiency of wafer cleaning, reduce unnecessary re-cleaning, and save time and resources.

[0022] S2. Calculate the size information of the wafer based on the wafer image obtained by the machine vision system. At the same time, use a high-precision weighing sensor to measure the weight of the wafer. Establish a mapping table of wafer size and weight and optimal conveying speed, which is built based on a large amount of experimental data and expert experience. Query the corresponding optimal conveying speed parameters from the mapping table, control the servo motor to adjust the conveyor belt speed, and realize dynamic optimization of the conveying speed.

[0023] S2 includes: acquiring a wafer image, and using an image processing algorithm to determine the diameter and thickness information of the wafer in the wafer image; acquiring wafer weight measurement data collected by a high-precision weighing sensor to obtain the weight information of the wafer; querying a pre-constructed wafer parameter and conveying speed mapping table based on the diameter, thickness and weight information of the wafer to obtain the optimal conveying speed parameters corresponding to the current wafer parameters; if the optimal conveying speed parameters are different from the current conveyor belt speed, controlling the servo motor to adjust the conveyor belt speed to the optimal conveying speed parameters; during the wafer conveying process, continuously acquiring wafer images and weighing sensor data, and updating the optimal conveying speed parameters in real time to achieve dynamic optimization of the conveying speed; using a machine learning algorithm, based on historical wafer parameters and conveying speed data, continuously optimizing the wafer parameter and conveying speed mapping table to improve the accuracy of conveying speed optimization; converting the experience of wafer production experts into optimization rules, supplementing and improving the wafer parameter and conveying speed mapping table, to cope with conveying speed optimization under special process requirements.

[0024] Specifically, wafer image processing is a key step in optimizing the conveying speed. The wafer image is captured by a high-resolution camera, the wafer outline is extracted using an edge detection algorithm, and the wafer diameter is calculated using methods such as the Hough transform. For example, for an 8-inch wafer, the diameter may be measured to be 200.3 mm. The thickness can be measured using laser triangulation with an accuracy of microns. A typical 8-inch wafer is about 725 microns thick. High-precision weighing sensors are key equipment for obtaining wafer weight information. Strain gauge weighing sensors are used with an accuracy of 0.01 grams. For example, the typical weight of an 8-inch wafer is about 53 grams. Combining size and weight information, the material and processing status of the wafer can be inferred, which is crucial to determining the optimal conveying speed. The pre-built mapping table is a bridge between wafer parameters and conveying speed. The table may contain the following correspondence: 8-inch silicon wafer, 725 microns thick, 53 grams, corresponding to an optimal conveying speed of 0.5 m / s. If the detected wafer parameters match an item in the table, the system will use the corresponding speed parameters. Servo motor control is the core of achieving precise speed regulation. A closed-loop control system is used to feedback the real-time speed through the encoder, and PID regulation is performed after comparing it with the target speed. For example, if the current speed is 0.4 m / s and the target speed is 0.5 m / s, the system will gradually increase the motor speed until the target value is reached. Dynamic optimization is the key to improving the adaptability of the system. During the conveying process, the system continuously collects wafer data. If minor defects are found on the surface of the wafer, the conveying speed may be appropriately reduced to ensure safety. For example, if a minor crack is detected on the edge of the wafer, the system may reduce the speed from 0.5 m / s to 0.4 m / s. The application of machine learning algorithms enables the system to have self-optimization capabilities. Through algorithms such as support vector machines (SVM), the system can learn the complex relationship between wafer parameters and optimal conveying speed. For example, the system may find that the optimal conveying speed of a certain type of wafer is slightly different under specific temperature and humidity conditions, and update the mapping table accordingly. The introduction of expert systems makes up for the shortcomings of pure data-driven methods. By encoding engineers' experience into a rule base, the system can handle some special cases. For example, for wafers that have just completed the polishing process, the expert system may recommend reducing the conveying speed by 20% to prevent possible surface scratches. This approach is particularly suitable for processing new or small-batch wafers, where historical data may not be sufficient to support machine learning algorithms to make accurate decisions.

[0025] S3. During the wafer conveying process, a gyroscope sensor is used to detect the vibration and displacement signals of the conveying device in real time. The collected signals are input into a pre-trained support vector machine model for classification to determine whether the current vibration and displacement exceed the preset threshold. The support vector machine model is trained using historical data and can accurately distinguish between normal and abnormal vibration and displacement states.

[0026] S3 includes: obtaining real-time vibration signals and displacement signals of the wafer conveying device, wherein the vibration signals and displacement signals are collected by a gyroscope sensor arranged on the conveying device; preprocessing the vibration signals and displacement signals to obtain standardized feature data, wherein the preprocessing includes denoising and normalization; inputting the standardized feature data into a pre-trained support vector machine model, and using the support vector machine model to classify and judge the standardized feature data to determine whether the vibration signals and displacement signals are within a normal range; if the result of the classification judgment shows that the vibration signals and displacement signals exceed a preset threshold, it is determined that the current state is abnormal, and an early warning signal is output; if the result of the classification judgment shows that the vibration signals and displacement signals do not exceed the preset threshold, it is determined that the current state is normal, and the vibration signals and displacement signals at the next moment are continuously obtained and monitored; the vibration signals, displacement signals and corresponding classification judgment results at each moment are continuously recorded to form a historical data set; the support vector machine model is regularly retrained using the historical data set to obtain an optimized support vector machine model for subsequent classification judgment.

[0027] Specifically, gyroscope sensors play an important role in the wafer conveying process. They can capture the vibration and displacement information of the conveying device in real time. For example, a high-precision gyroscope can detect angular velocity changes as small as 0.01 degrees per second, which is crucial for monitoring subtle vibrations during wafer conveying. The acquired raw signal usually contains noise and needs to be filtered. Common filtering methods include Kalman filtering and wavelet transform, which can effectively remove high-frequency noise and retain useful signal features. In the data preprocessing stage, the normalization operation unifies data of different dimensions to the same scale. For example, the minimum-maximum normalization method can be used to map both the vibration amplitude and the displacement to the range of 0 to 1. The processed data is more suitable for input into the machine learning model. The support vector machine (SVM) model performs well in anomaly detection. It can find the optimal classification hyperplane in the high-dimensional feature space. In wafer conveying monitoring, the radial basis function (RBF) kernel can be used. This kernel function can handle nonlinear classification problems and is suitable for capturing complex patterns of vibration and displacement data. The design of the early warning system requires a trade-off between sensitivity and false alarm rate. For example, you can set multiple levels of warning thresholds: a mild warning is issued when the vibration amplitude exceeds 1.5 times the normal range, and a severe warning is issued when it exceeds 2 times. This hierarchical warning mechanism can help engineers better assess the status of the equipment and take appropriate maintenance measures in a timely manner. The historical data that is continuously recorded is not only used for model training, but also for equipment life prediction. By analyzing the long-term trend of vibration and displacement data, it is possible to predict when the equipment may need to be overhauled or replaced, thereby achieving preventive maintenance. Regular retraining of the model is key to maintaining system performance. An incremental learning method can be used to update the model every time a certain amount of new data is accumulated (such as weekly or monthly). This method enables the model to adapt to small changes in the equipment over time, such as changes in vibration characteristics caused by component wear. In actual applications, it may also be necessary to consider the differences in vibration characteristics under different process conditions. For example, when processing wafers of different sizes or materials, the normal vibration range may be different. Therefore, multiple SVM models can be established for different process conditions, and the model used can be automatically switched according to the current production task to improve the accuracy of anomaly detection. In addition, the reliability of the monitoring system can be further improved by combining other sensor data. For example, it can monitor motor current, temperature and other parameters at the same time, and integrate multiple data sources for abnormal detection. This multimodal data fusion approach can provide more comprehensive equipment status information, reduce false alarm rates, and improve the accuracy and credibility of early warnings. Through this comprehensive and sophisticated monitoring and early warning system, the stability and reliability of the wafer conveying process can be greatly improved, and product losses and downtime caused by equipment failures can be reduced, thereby improving overall production efficiency.

[0028] S4. When vibration or displacement exceeding the threshold is detected, the servo motor is triggered to perform compensation control. By adjusting the conveying speed and acceleration parameters, vibration and displacement are suppressed to ensure stable conveyance of the wafer to the rotary cleaning device. The compensation control algorithm dynamically adjusts the control parameters according to the amplitude of vibration and displacement to achieve fast response and precise compensation.

[0029] S4 includes: obtaining real-time vibration and displacement data during wafer conveying, wherein the real-time vibration and displacement data are obtained by collecting vibration and displacement signals through sensors, and converting the vibration and displacement signals into digital quantities; judging whether the real-time collected vibration and displacement data exceed the preset vibration threshold and displacement threshold range; if the real-time collected vibration and displacement data exceed the preset vibration threshold and displacement threshold range, triggering the servo motor to perform compensation control; filtering the collected vibration and displacement data using a Kalman filter algorithm to obtain filtered vibration and displacement data; calculating compensation control parameters of the servo motor using a PID control algorithm based on the filtered vibration and displacement data, wherein the compensation control parameters include adjustment values ​​of conveying speed and acceleration; transmitting the compensation control parameters to a servo motor controller, and achieving vibration and displacement suppression during wafer conveying by adjusting the rotation speed and acceleration of the servo motor; during the compensation control process, continuously monitoring the vibration and displacement state of the wafer, and dynamically optimizing the compensation control parameters through an adaptive control algorithm; when the wafer is stably conveyed to the rotary cleaning device, stopping the compensation control, and recording relevant data during the compensation control process.

[0030] Specifically, vibration and displacement control during wafer conveying is a key link to ensure wafer quality. Real-time data acquisition usually uses high-precision accelerometers and displacement sensors, such as MEMS accelerometers and laser displacement sensors. These sensors can capture tiny vibration and displacement changes with an accuracy of up to micron level. For example, the accelerometer used in a certain wafer production line has a sensitivity of 0.1g, a displacement sensor resolution of 0.1μm, and a sampling frequency of 1kHz, which can monitor subtle changes in the conveying process in real time. Threshold setting is the key to judging abnormal conditions. It is usually determined according to the wafer material characteristics and process requirements. For example, a 300mm wafer production line sets the vibration threshold to 0.5g and the displacement threshold to ±50μm. When the threshold is exceeded, the system immediately triggers compensation control to prevent wafer damage. The Kalman filter algorithm plays an important role in signal processing and can effectively remove random noise and improve data reliability. For example, after a wafer fab applied Kalman filtering, the signal-to-noise ratio of the vibration signal increased by 30%, the accuracy of the displacement data increased by 20%, and the accuracy of subsequent control was greatly improved. The PID control algorithm is the core of achieving precise compensation. By adjusting the three parameters of proportion, integration and differentiation, vibration and displacement can be quickly responded and stably suppressed. In practical applications, a wafer conveying system adopts adaptive PID control to automatically adjust parameters according to different working conditions. For example, in the startup phase, the proportional coefficient is increased for rapid response; in the stable phase, the integral coefficient is increased to eliminate static errors; in the case of sudden disturbances, the differential coefficient is increased to improve system stability. Servo motor control is a key link in execution compensation. High-precision servo motors can achieve micron-level position adjustment and millisecond-level response speed. For example, a wafer conveying system uses a servo motor with a resolution of 0.01°, which is matched with a reducer to achieve a minimum adjustment of 0.1μm, and can accurately execute the control instructions output by the PID algorithm. The adaptive control algorithm can dynamically optimize the control parameters according to real-time monitoring data. For example, a wafer fab adopts a fuzzy neural network adaptive algorithm to continuously adjust the control strategy by learning historical data and real-time feedback. After one month of operation, the vibration suppression effect of the system has been improved by 15% and the displacement control accuracy has been improved by 10%. Data recording and analysis are essential for system optimization. For example, a wafer production line has established a complete database to record the conveying parameters, vibration displacement data and control effects of each batch of wafers. Through big data analysis, the correlation between equipment aging and increased vibration was found, maintenance needs were predicted in advance, and unplanned downtime was reduced by 30%. The comprehensive application of these technologies not only ensures the stability of the wafer conveying process, but also improves production efficiency and yield. After a large wafer fab introduced this system, the wafer breakage rate was reduced by 50%, the product yield was increased by 3 percentage points, and the average annual cost savings were tens of millions of yuan.

[0031] S5. For the rotating cleaning process, before the cleaning liquid is sprayed, a computer vision algorithm is used to analyze the static wafer surface image, and the initial state of the wafer surface is determined through image segmentation and feature extraction. During the cleaning process, a high-speed camera is used to capture the instantaneous image of the rotating wafer, and combined with the initial state analysis results, the uniformity of the cleaning liquid coverage on the wafer surface is evaluated to obtain a quantitative coverage score.

[0032] S5 includes: using a high-resolution camera to obtain a still wafer surface image, graying it through an image preprocessing algorithm to enhance the image contrast, and providing high-quality input for subsequent image segmentation and feature extraction. Using an edge detection algorithm to segment the preprocessed image, accurately extract the wafer edge contour, and using a region growing algorithm to segment the wafer surface area to obtain the feature data of the initial state of the wafer surface. Extracting the feature information of surface defects and stains from the segmented image, using a feature matching algorithm to compare with the preset defect library to determine the initial state classification label of the wafer surface. During the cleaning process, a high-speed camera is used to capture the instantaneous image of the rotating wafer in real time, and the instantaneous image is aligned with the initial state image through an image registration algorithm to ensure the temporal and spatial consistency of the image analysis. The registered instantaneous image is converted into color space to extract the color characteristics of the cleaning solution on the wafer surface, and the regional coverage analysis algorithm is used to evaluate the coverage area and uniformity of the cleaning solution. According to the coverage area and uniformity analysis results, combined with the preset coverage scoring model, a quantitative coverage score is calculated. The scoring model considers the indicators of coverage area ratio and distribution uniformity. If the coverage score is lower than the preset threshold, feedback is provided to adjust the cleaning liquid injection parameters and the cleaning process is repeated until the score reaches the preset standard to ensure that the wafer surface cleaning effect meets the process requirements.

[0033] Specifically, high-resolution cameras that acquire wafer surface images are the basis for cleaning quality assessment. Industrial cameras with 4K or 8K resolutions are usually used, combined with high-quality lenses, to ensure that the tiny details of the wafer surface are captured. Grayscale processing in image preprocessing can convert RGB color images into single-channel grayscale images, simplifying subsequent processing steps. Contrast enhancement commonly uses histogram equalization methods to redistribute pixel values ​​to the range of 0-255 to improve the overall contrast of the image. Edge detection is a key step in image segmentation. The Canny algorithm is a commonly used edge detection method that accurately extracts wafer edges through steps such as Gaussian filtering, non-maximum suppression, and dual threshold detection. The region growing algorithm starts from a preset seed point and gradually expands the similar pixel area to eventually obtain a complete wafer surface area. Feature extraction and matching are the core of identifying the surface state of the wafer. The extractable features include texture features (such as grayscale co-occurrence matrix), shape features (such as area, perimeter, roundness), etc. Feature matching can use the KNN (K nearest neighbor) algorithm to compare the extracted features with the pre-established defect library to determine the initial state classification of the wafer surface. High-speed cameras capture rotating wafer images in real time, usually at an acquisition rate of more than 1,000 frames per second to ensure that clear instantaneous images are captured. Image registration algorithms such as SIFT (Scale Invariant Feature Transform) can be used to align the instantaneous image with the initial state image to compensate for the differences caused by rotation and displacement. Color space conversion often uses RGB to HSV conversion, and HSV space is more suitable for describing the color characteristics of the cleaning fluid. Regional coverage analysis can extract the cleaning fluid coverage area by setting the HSV threshold, and calculate the coverage area ratio and distribution uniformity. The coverage scoring model can consider the coverage area ratio (such as greater than 95%) and uniformity indicators (such as standard deviation less than 0.1) to comprehensively evaluate the cleaning effect. If the coverage score does not meet the standard, the system will automatically adjust the cleaning fluid injection parameters. For example, increase the injection pressure (such as from 2 bar to 2.5 bar), adjust the nozzle angle (such as from 60° to 75°), or extend the cleaning time (such as from 30 seconds to 35 seconds). This closed-loop control ensures the adaptive optimization of the cleaning process and ultimately achieves the preset cleaning quality standard. The entire process reflects the application of computer vision technology in semiconductor manufacturing. Through real-time image analysis and feedback control, the accuracy and efficiency of wafer cleaning have been greatly improved. This not only reduces the subjectivity and labor intensity of manual inspection, but also realizes real-time monitoring and optimization of the cleaning process, which is of great significance to improving product yield and production efficiency.

[0034] S6. Compare the coverage score with the preset threshold. When it is below the threshold, dynamically adjust the rotation angle and speed of the wafer. Optimize the rotation parameters through the fuzzy control algorithm, which outputs the optimal angle and speed adjustment based on the three input variables of coverage score, rotation angle and speed. The fuzzy rules are set according to expert knowledge, and the effectiveness of the algorithm is ensured through repeated tuning. Continuously monitor the changing trend of coverage and optimize the cleaning parameters in real time.

[0035] S6 includes: obtaining real-time coverage score data, comparing the real-time coverage score data with a preset threshold, and triggering a fuzzy control algorithm if the real-time coverage score data is lower than the preset threshold; inputting the current coverage score, rotation angle and speed into the fuzzy control algorithm, calculating according to preset fuzzy rules, and obtaining an optimal angle adjustment amount and an optimal speed adjustment amount; feeding back the optimal angle adjustment amount and the optimal speed adjustment amount to the cleaning equipment, and dynamically adjusting the wafer rotation parameters; continuously monitoring coverage changes, and if the adjusted coverage score is still lower than the preset threshold, restarting the fuzzy control algorithm for optimization; recording the coverage score and the corresponding rotation angle and speed parameters after each adjustment, and constructing an optimized data set; iteratively improving the fuzzy rules in the fuzzy control algorithm according to the optimized data set, and obtaining an improved fuzzy control algorithm; using the improved fuzzy control algorithm to optimize and control the subsequent wafer cleaning process, and improve the adaptability and accuracy of the cleaning effect.

[0036] Specifically, obtaining real-time coverage score data is a key step in the cleaning process. For example, assuming that the preset coverage threshold is 90%, the real-time monitoring system shows that the current coverage is 85%. At this point, the system immediately triggers the fuzzy control algorithm to ensure that the cleaning effect meets the standard. The input of the fuzzy control algorithm includes the current coverage score, rotation angle and speed. Assume that the current rotation angle is 30 degrees and the speed is 100 revolutions per minute. According to the preset fuzzy rules, the algorithm will perform the following calculations: if the coverage is low and the rotation speed is fast, the speed will be appropriately reduced; if the coverage is low and the rotation angle is small, the angle will be appropriately increased. Through fuzzy logic reasoning, the algorithm outputs the optimal angle adjustment of 5 degrees and the speed adjustment of 10 revolutions per minute. These adjustments are fed back to the cleaning equipment, and the equipment dynamically adjusts the wafer rotation parameters. The new rotation angle becomes 35 degrees and the speed becomes 90 revolutions per minute. The coverage changes are continuously monitored. If the score is still below the threshold, the system will start the fuzzy control algorithm again for optimization. The coverage score and parameter changes after each adjustment are recorded to construct an optimized data set. For example, the coverage is 88% after the first adjustment and 92% after the second adjustment. These data are used to iterate and improve the algorithm. Through multiple iterations, the accuracy and adaptability of the fuzzy rules are improved. Improving the accuracy of the fuzzy rules means that the algorithm can more accurately predict the impact of the adjustment amount on the coverage. For example, the initial algorithm may overestimate the impact of speed adjustment on the coverage. After iteration, the algorithm can more accurately balance the adjustment of speed and angle to ensure that each adjustment can effectively improve the coverage. Improved adaptability is reflected in the algorithm's ability to adapt to different cleaning conditions. For example, different types of wafers may respond differently to the cleaning solution. After multiple iterations, the algorithm can adjust the fuzzy rules according to the characteristics of different wafers to ensure that the ideal cleaning effect can be achieved under various conditions. Doing so can not only improve cleaning efficiency, but also reduce the waste of cleaning solution and reduce production costs. For example, through optimization and adjustment, the amount of cleaning solution used was reduced by 10%, and the cleaning time was shortened by 5%, significantly improving the overall efficiency of the production line. In addition, recording and analyzing the data of each adjustment can help to find potential problems and improvement points. For example, the data may show that the cleaning effect is always unsatisfactory at certain specific angles and speeds. Through in-depth analysis, the reasons can be found and targeted improvements can be made. Continuous optimization of cleaning effects can also improve product quality. For example, if tiny defects and stains on the wafer surface are not thoroughly cleaned, they may lead to amplified defects in subsequent processes and affect the performance of the final product. By precisely controlling the cleaning process, the wafer surface is ensured to be highly clean, thereby improving the yield and reliability of the product. In short, through real-time monitoring, dynamic adjustment of fuzzy control algorithms, and data-driven iterative optimization, the cleaning process can be continuously improved to ensure that the wafer surface cleaning effect meets high-standard process requirements and improve production efficiency and product quality.

[0037] S7. Based on the coverage trend and the wafer rotation time, the reinforcement learning algorithm is used to continuously optimize the rotation angle and speed control strategy. The reinforcement learning algorithm aims to maximize the thoroughness of cleaning, uses the coverage score as a reward signal, and learns the optimal rotation parameter combination through multiple iterations. The characteristics of different types of wafers are considered during the learning process to form a highly adaptable control strategy and improve the stability of the cleaning effect.

[0038] The type information of the wafer is obtained, and the initial control parameters of the rotation cleaning process of the type of wafer are determined according to the pre-established wafer characteristic database, including the rotation angle range and speed range. The wafer is placed in the cleaning equipment, the cleaning process is started, the rotation angle and speed are set according to the current control parameters, and the wafer is rotationally cleaned. The surface image of the wafer is collected in real time by the visual sensor, the image is processed, and the coverage change curve of the wafer surface during the cleaning process is calculated. The coverage change curve is compared with the preset ideal coverage change curve, the similarity of the two curves is calculated, and the similarity value is used as the instant reward value of the reinforcement learning algorithm. According to the reward value, the reinforcement learning algorithm updates the parameters of the control strategy network through the policy gradient method to obtain new rotation angle and speed control values. The updated control parameters are applied to the wafer cleaning equipment, the rotation angle and speed are adjusted, and multiple rounds of iterative learning optimization are performed. When the similarity between the coverage change curve and the ideal curve reaches the preset threshold, or the number of iterations reaches the upper limit, the learning process is stopped, and the optimized rotation angle and speed control strategy is output to guide the subsequent cleaning process of the same type of wafer.

[0039] Specifically, the wafer characteristic database is established based on the physical and chemical properties, surface structure and other information of different types of wafers. For example, for an 8-inch silicon wafer, the initial rotation angle range may be set to 0-180 degrees and the speed range may be 500-2000rpm. The selection of these parameters is based on factors such as the size of the wafer, material strength and cleaning requirements. When the visual sensor collects images of the wafer surface, a high-resolution CCD camera can be used with an appropriate light source system. The image processing process includes preprocessing steps such as denoising and contrast enhancement, and then the cleaning area is identified through algorithms such as edge detection and region segmentation. The coverage rate can be calculated using pixel statistics methods, such as dividing the number of pixels in the cleaning area by the total number of pixels. The ideal coverage rate change curve is usually S-shaped, with a slow growth in the early stage, a rapid rise in the middle stage, and a flattening in the later stage. The similarity between the actual curve and the ideal curve can be quantified by calculating the Euclidean distance or correlation coefficient of the two curves. For example, if the correlation coefficient of the two curves is 0.95, 0.95 can be input into the reinforcement learning algorithm as the immediate reward value. In the reinforcement learning algorithm, a deep Q network (DQN) or a policy gradient method can be used. Taking DQN as an example, the input state may include the current coverage, rotation angle and speed, and the output is the Q value of different angle and speed combinations. According to the ε-greedy strategy, the action is selected to gradually optimize the Q network parameters. For example, the initial ε value is set to 0.9, and gradually reduced to 0.1 as the training progresses, balancing exploration and utilization. In the process of updating the control parameters, a progressive adjustment strategy can be adopted. For example, if the current angle is 90 degrees and the speed is 1000rpm, if the optimal action output by DQN is to increase the angle and reduce the speed, the angle can be adjusted to 100 degrees and the speed can be adjusted to 950rpm. This progressive adjustment helps to maintain the stability of the cleaning process. In the iterative learning optimization process, reasonable stop conditions need to be set. For example, the similarity threshold can be set to 0.98 and the maximum number of iterations can be set to 100. Stop the optimization when any condition is met. The optimization result can be represented as a policy network, which inputs the wafer type and cleaning stage and outputs the optimal angle and speed control values. Compared with traditional fixed parameter control, this optimization method based on reinforcement learning can better adapt to the changes of different types of wafers and cleaning environments. Through continuous learning and optimization, the system can gradually improve cleaning efficiency and quality, reduce manual intervention, and achieve intelligent control. At the same time, the accumulated optimization experience can be used to guide the cleaning parameter settings of new wafers and improve overall production efficiency.

[0040] S8. After each wafer is cleaned, the surface of the cleaned wafer is inspected for defects. The surface defects are identified and located through a deep learning algorithm to generate a defect distribution map. Defects include contamination and scratches. The defect information is correlated with the previous cleaning parameters to find out the factors that may lead to incomplete cleaning. The analysis results are fed back to the upstream process parameter optimization system to adjust the cleaning solution formula, injection pressure and other parameters, continuously improve the wafer cleaning process, and form a closed-loop optimization.

[0041] Acquire an image of the wafer surface after cleaning, the image being a defect image; preprocess the defect image to obtain a preprocessed image, the preprocessing including denoising and contrast enhancement; use a convolutional neural network to perform defect recognition on the preprocessed image to obtain a defect distribution map; acquire cleaning parameter data, the cleaning parameter data including a cleaning liquid formula and a spray pressure, and construct a parameter database; perform correlation analysis on the defect distribution map and the parameter database, use a random forest algorithm to analyze the correlation between defects and cleaning parameters, and obtain correlation analysis results; determine the cleaning parameters that need to be adjusted based on the correlation analysis results, and generate optimization suggestions; transmit the optimization suggestions to a process parameter optimization system to adjust the cleaning liquid formula and the spray pressure; monitor the cleaning effect in real time through a closed-loop control system, feed back the adjustment results, and continuously optimize the cleaning process.

[0042] Specifically, deep learning algorithms play an important role in wafer surface defect detection. Taking convolutional neural networks as an example, image features can be extracted through multi-layer convolution and pooling operations to effectively identify defects such as tiny scratches and particle contamination. In practical applications, advanced network structures such as ResNet can be used to improve detection accuracy. For example, a semiconductor factory uses an improved YOLOv5 model to increase the defect detection rate to 99.5%, significantly surpassing traditional machine vision methods. Image preprocessing is crucial to improving defect detection results. Gaussian filtering can effectively remove image noise, while histogram equalization can enhance image contrast. In one case, the problem of reflective interference on the wafer surface was successfully solved by introducing an adaptive threshold segmentation algorithm, making the defect edge more clearly identifiable. The generation of defect distribution maps lays the foundation for subsequent analysis. The heat map can intuitively display the defect-intensive area, assisting engineers to quickly locate the problem. A factory used this method to find that the defect rate in the edge area was high, and then optimized the layout of the cleaning nozzle, significantly improving the cleaning uniformity. The construction of a cleaning parameter database helps to comprehensively analyze the factors affecting the process. In addition to the cleaning fluid formula and injection pressure, key parameters such as temperature and time should also be included. A study shows that by accurately controlling the pH value of the cleaning fluid, the particle removal rate can be increased by 15%. Machine learning algorithms play an important role in parameter optimization. The random forest algorithm has good feature importance evaluation capabilities and can quickly identify key factors that affect defect generation. In a study, random forest analysis found that the concentration of the cleaning fluid was strongly correlated with micro-scratch defects. After adjusting the formula accordingly, the defect rate was reduced by 30%. The closed-loop control system ensures continuous process optimization. By real-time monitoring of the cleaning effect, parameters can be adjusted in time to avoid quality fluctuations between batches. A production line introduced an adaptive system based on fuzzy control to dynamically adjust the cleaning pressure according to the defect detection results, so that the yield rate is stable at more than 99%. This data-driven cleaning process optimization method not only improves product quality, but also reduces production costs. By reducing chemical consumption and improving equipment utilization, a factory saves more than 1 million yuan per year. In addition, the method has a high degree of automation, reduces human intervention, improves process consistency and repeatability, and provides strong support for the intelligent transformation of the semiconductor manufacturing industry.

[0043] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent control method for a wafer cleaning and conveying process, characterized in that: The method comprises the following steps: S1. Use a machine vision system to detect the position coordinates of the wafer on the conveyor belt in real time. Based on the acquired coordinate data, the servo motor accurately controls the movement of the conveyor belt to adjust the wafer to the preset optimal cleaning position. The optimal cleaning position is defined as the position where the center of the wafer is aligned with the center of the cleaning nozzle to ensure that the cleaning liquid evenly covers the wafer surface and eliminates cleaning dead corners caused by wafer position deviation. S2. Calculate the size information of the wafer based on the wafer image obtained by the machine vision system. At the same time, use a high-precision weighing sensor to measure the weight of the wafer, and establish a mapping table of wafer size and weight and optimal conveying speed. The mapping table is constructed based on a large amount of experimental data and expert experience. The corresponding optimal conveying speed parameters are obtained from the mapping table, and the servo motor is controlled to adjust the conveyor belt speed to achieve dynamic optimization of the conveying speed. S3. During wafer conveying, a gyroscope sensor is used to detect the vibration and displacement signals of the conveying device in real time, and the collected signals are input into a pre-trained support vector machine model for classification to determine whether the current vibration and displacement exceed a preset threshold. The support vector machine model is trained using historical data and can accurately distinguish between normal and abnormal vibration and displacement states; S4. When vibration or displacement exceeding the threshold is detected, the servo motor is triggered to perform compensation control. By adjusting the conveying speed and acceleration parameters, vibration and displacement are suppressed to ensure that the wafer is stably conveyed to the rotary cleaning device. The compensation control algorithm dynamically adjusts the control parameters according to the amplitude of vibration and displacement to achieve rapid response and accurate compensation. S5. For the rotation cleaning process, before the cleaning liquid is sprayed, a computer vision algorithm is used to analyze the static wafer surface image, and the initial state of the wafer surface is determined through image segmentation and feature extraction. During the cleaning process, a high-speed camera is used to capture the instantaneous image of the rotating wafer. Combined with the initial state analysis results, the uniformity of the cleaning liquid coverage on the wafer surface is evaluated to obtain a quantitative coverage score; S6. Compare the coverage score with a preset threshold. When it is lower than the threshold, dynamically adjust the rotation angle and speed of the wafer. Optimize the rotation parameters through a fuzzy control algorithm. The algorithm outputs the optimal angle and speed adjustment based on the three input variables of coverage score, rotation angle and speed. The fuzzy rules are set according to expert knowledge. The effectiveness of the algorithm is ensured through repeated tuning. The coverage trend is continuously monitored and the cleaning parameters are optimized in real time. S7. Based on the coverage trend and the rotation time of the wafer, the reinforcement learning algorithm is used to continuously optimize the control strategy of the rotation angle and speed. The reinforcement learning algorithm aims to maximize the thoroughness of cleaning and uses the coverage score as a reward signal. The optimal combination of rotation parameters is learned through multiple iterations. The characteristics of different types of wafers are considered during the learning process to form a highly adaptable control strategy and improve the stability of the cleaning effect. S8. After each wafer is cleaned, defect detection is performed on the cleaned wafer surface. The surface defects are identified and located through a deep learning algorithm, and a defect distribution map is generated. The defect information is correlated with the previous cleaning parameters for analysis to find out the factors that may lead to incomplete cleaning. The analysis results are fed back to the upstream process parameter optimization system to continuously improve the wafer cleaning process and form a closed-loop optimization.

2. According to claim 1, an intelligent control method for a wafer cleaning and conveying process is characterized in that: The S1 includes: Acquire image data of wafers on a conveyor belt, wherein the image data is acquired by a machine vision system through real-time image acquisition; Preprocessing the image data, denoising and enhancing the image data using an image preprocessing algorithm to obtain a preprocessed wafer image; Extracting the contour features of the wafer from the preprocessed wafer image, calculating the geometric center coordinates of the wafer according to the contour features, and obtaining the real-time position coordinate data of the wafer on the conveyor belt; Obtaining preset optimal cleaning position coordinates, comparing the real-time position coordinate data of the wafer on the conveyor belt with the optimal cleaning position coordinates, and calculating the deviation between the center of the wafer and the optimal cleaning position; If the deviation exceeds a preset threshold, the servo motor is triggered to adjust the conveyor belt movement; According to the deviation, a PID control algorithm is used to control the rotation speed and movement direction of the servo motor, and the wafer is moved to the optimal cleaning position by adjusting the movement of the conveyor belt so that the center of the wafer is aligned with the center of the cleaning nozzle; When the wafer reaches the optimal cleaning position, the cleaning nozzle is triggered to start, the flow rate and spraying time of the cleaning liquid are controlled, and the cleaning liquid is evenly sprayed at the center of the wafer so that the cleaning liquid completely covers the surface of the wafer; After cleaning is completed, the machine vision system is used to collect and analyze images of the wafer surface to detect the cleanliness of the wafer surface; If the cleanliness does not meet the preset standard, a re-cleaning process is triggered until the wafer surface is completely clean; The wafers that pass the cleaning process are transferred to the next process, the cleaning process data is recorded and statistically analyzed, the cleaning effect is evaluated, and the cleaning parameters are dynamically optimized.

3. The intelligent control method for wafer cleaning and conveying process according to claim 1, characterized in that: The S2 comprises: Acquire a wafer image, and use an image processing algorithm to determine the diameter and thickness information of the wafer in the wafer image; Obtaining wafer weight measurement data collected by a high-precision weighing sensor to obtain weight information of the wafer; According to the diameter, thickness and weight information of the wafer, a pre-constructed wafer parameter and conveying speed mapping table is searched to obtain the optimal conveying speed parameters corresponding to the current wafer parameters; If the optimal conveying speed parameter is different from the current conveyor belt speed, controlling the servo motor to adjust the conveyor belt speed to the optimal conveying speed parameter; During the wafer conveying process, the wafer image and weighing sensor data are continuously acquired, and the optimal conveying speed parameters are updated in real time to achieve dynamic optimization of the conveying speed; Using a machine learning algorithm, based on historical wafer parameter and conveying speed data, the wafer parameter and conveying speed mapping table is continuously optimized to improve the accuracy of conveying speed optimization; The experience of wafer production experts is converted into optimization rules, and the wafer parameter and conveying speed mapping table is supplemented and improved to meet the conveying speed optimization under special process requirements.

4. The intelligent control method for wafer cleaning and conveying process according to any one of claims 1 to 3, characterized in that: The S3 includes: Acquiring a real-time vibration signal and a displacement signal of a wafer conveying device, wherein the vibration signal and the displacement signal are acquired by a gyroscope sensor disposed on the conveying device; Preprocessing the vibration signal and the displacement signal to obtain standardized feature data, wherein the preprocessing includes denoising and normalization; Inputting the standardized feature data into a pre-trained support vector machine model, and using the support vector machine model to classify and judge the standardized feature data to determine whether the vibration signal and the displacement signal are within a normal range; If the result of the classification judgment indicates that the vibration signal and the displacement signal exceed the preset threshold value, it is determined that the current state is abnormal and an early warning signal is output; If the result of the classification judgment indicates that the vibration signal and the displacement signal do not exceed the preset threshold, it is determined that the current state is normal, and the vibration signal and the displacement signal at the next moment are continuously acquired and monitored; Continuously record the vibration signal and displacement signal at each moment and the corresponding classification and judgment results to form a historical data set; The support vector machine model is regularly retrained using the historical data set to obtain an optimized support vector machine model for subsequent classification judgment.

5. The intelligent control method for wafer cleaning and conveying process according to any one of claims 1 to 3, characterized in that: The S4 includes: Acquire real-time vibration and displacement data during wafer conveying, wherein the real-time vibration and displacement data are obtained by collecting vibration and displacement signals through sensors, and converting the vibration and displacement signals into digital quantities; Determining whether the vibration and displacement data collected in real time exceed a preset vibration threshold and displacement threshold range; If the vibration and displacement data collected in real time exceed the preset vibration threshold and displacement threshold range, the servo motor is triggered to perform compensation control; Using a Kalman filter algorithm to filter the collected vibration and displacement data to obtain filtered vibration and displacement data; According to the filtered vibration and displacement data, the compensation control parameters of the servo motor are calculated by a PID control algorithm, wherein the compensation control parameters include adjustment values ​​of conveying speed and acceleration; The compensation control parameters are transmitted to the servo motor controller, and the vibration and displacement suppression of the wafer conveying process is achieved by adjusting the rotation speed and acceleration of the servo motor; During the compensation control process, the vibration and displacement state of the wafer are continuously monitored, and the compensation control parameters are dynamically optimized through an adaptive control algorithm; When the wafer is stably transported to the spin cleaning device, the compensation control is stopped and relevant data during the compensation control process is recorded.

6. The intelligent control method for wafer cleaning and conveying process according to any one of claims 1 to 3, characterized in that: The S5 includes: Use a high-resolution camera to obtain a still wafer surface image, perform grayscale processing through an image preprocessing algorithm, enhance image contrast, and provide high-quality input for subsequent image segmentation and feature extraction; The preprocessed image is segmented using the edge detection algorithm to accurately extract the wafer edge contour, and the wafer surface area is segmented using the region growing algorithm to obtain the characteristic data of the initial state of the wafer surface; Extract feature information of surface defects and stains from the segmented image, use feature matching algorithm to compare with the preset defect library, and determine the initial state classification label of the wafer surface; During the cleaning process, a high-speed camera is used to capture the instantaneous image of the rotating wafer in real time. The instantaneous image is aligned with the initial state image through an image registration algorithm to ensure the temporal and spatial consistency of image analysis. Perform color space conversion on the registered instantaneous image to extract the color characteristics of the cleaning liquid on the wafer surface, and use the regional coverage analysis algorithm to evaluate the coverage area and uniformity of the cleaning liquid; Based on the coverage area and uniformity analysis results, combined with the preset coverage scoring model, a quantitative coverage score is calculated. The scoring model takes into account the indicators of coverage area ratio and distribution uniformity; If the coverage score is lower than the preset threshold, feedback is provided to adjust the cleaning liquid injection parameters and the cleaning process is repeated until the score reaches the preset standard to ensure that the wafer surface cleaning effect meets the process requirements.

7. An intelligent control method for a wafer cleaning and conveying process according to any one of claims 1 to 3, characterized in that: The S6 comprises: Acquire real-time coverage score data, compare the real-time coverage score data with a preset threshold, and trigger a fuzzy control algorithm if the real-time coverage score data is lower than the preset threshold; The current coverage score, rotation angle and speed are input into the fuzzy control algorithm, and the optimal angle adjustment amount and the optimal speed adjustment amount are calculated according to the preset fuzzy rules; Feeding back the optimal angle adjustment amount and the optimal speed adjustment amount to the cleaning equipment to dynamically adjust the wafer rotation parameters; Continuously monitoring the coverage change, if the adjusted coverage score is still lower than the preset threshold, restarting the fuzzy control algorithm for optimization; Record the coverage score and corresponding rotation angle and speed parameters after each adjustment to build an optimized data set; Iteratively improving the fuzzy rules in the fuzzy control algorithm according to the optimized data set to obtain an improved fuzzy control algorithm; The improved fuzzy control algorithm is used to optimize the control of the subsequent wafer cleaning process, thereby improving the adaptability and accuracy of the cleaning effect.