Wafer loading correction method and apparatus

Abstract

The application discloses a wafer loading deviation rectification method and device, belongs to the wafer transmission field, and is applied to simf, efem and sort wafer transmission equipment. The method first collects the level attitude data and tab signal of a wafer loading groove in real time through a multi-source sensing assembly, determines whether the wafer installation is abnormal through a three-dimensional position sensing algorithm of a wafer channel, calculates the inclination of the equipment by using a multi-source sensing dynamic deviation fitting algorithm if the wafer installation is abnormal, and then converts the inclination into motor pulse numbers through a loading dynamic twin calibration model. The wafer position is adjusted by driving the motor according to the pulse numbers, real-time collection and update data are carried out on the SMIF loading precision control data analysis platform during the adjustment, the abnormality determination, deviation calculation and adjustment steps are repeated until the wafer installation reaches the standard. The method fuses multi-source data, realizes dynamic deviation rectification through real-time feedback and repeated verification, effectively avoids subsequent false alarms and wafer damage, and guarantees stable operation of the equipment and wafer processing quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wafer manufacturing technology, and in particular to a wafer loading correction method and device. BACKGROUND

[0002] In wafer manufacturing, SIMF, EFEM, Sort and other wafer transmission devices are key devices for realizing automatic transmission and processing of wafers, and accurate loading of wafers in the transmission process directly affects the subsequent processing quality and wafer yield. Currently, when such devices are running, wafers need to be stably placed in wafer loading slots, but their loading state is easily affected by factors such as device running vibration, environmental temperature change and mechanical wear, resulting in abnormal wafer installation. If the abnormality cannot be sensed and corrected in time, it may not only cause false alarms in subsequent device operation, but also cause wafer collision, scratching and other damage due to improper wafer loading, resulting in economic losses. Therefore, the industry urgently needs a method that can sense the wafer loading state in real time, accurately calculate the device deviation and efficiently complete the correction, in order to ensure the stable operation of the wafer transmission device and the wafer processing quality. A wafer loading correction method and device are proposed based on this demand, which realizes dynamic correction of wafer loading through multi-algorithm cooperation and multi-unit cooperation.

[0003] The prior art has obvious deficiencies in wafer loading correction. On the one hand, existing methods mostly rely on single sensing data for abnormality determination and deviation calculation, without effective fusion analysis of the attitude data collected by the level meter and the signal data collected by the tab recognition module, resulting in low accuracy of wafer installation abnormality determination and large error between the calculated device inclination and the actual deviation, which cannot provide accurate data support for subsequent correction. On the other hand, the existing correction process lacks dynamic calibration and real-time feedback mechanism. After converting the motor pulse number according to the deviation, only single fixed adjustment can be performed, which cannot update the data in real time and repeatedly check the adjustment effect during the adjustment process, and is prone to under-adjustment or over-adjustment, making it difficult to ensure that the final wafer loading state meets the preset standard and effectively avoiding subsequent false alarms and wafer damage. SUMMARY

[0004] In order to overcome the shortcomings and deficiencies of the prior art, the present application provides a wafer loading correction method and device.

[0005] The application provides a wafer loading deviation correction method, comprising the following steps: S1, calling a multi-source sensing component through an SMIF loading precision control data analysis platform to collect three-dimensional posture data output by a level meter arranged on a wafer loading slot in a wafer conveying device and wafer tab signal generated by a corresponding wafer tab recognition module in real time, wherein the three-dimensional posture data comprises inclination angle data of the loading slot in three directions of X axis, Y axis and Z axis, and the wafer tab signal comprises wafer tab position coordinates and wafer tab profile feature data; S2, inputting the wafer tab signal collected in S1 into a wafer channel three-dimensional pose perception algorithm, comparing the wafer tab position coordinates with preset standard wafer tab coordinates through the algorithm, combining the edge fitting result of the wafer tab profile feature data, and determining whether the wafer on the wafer loading slot is abnormally installed; S3, when it is determined in S2 that there is an installation abnormality, inputting the three-dimensional posture data of the level meter collected in S1 into a multi-source sensing dynamic deviation fitting algorithm, performing time sequence correlation analysis on the inclination angle data at different collection moments through the algorithm, calculating the equipment inclination of the wafer loading slot under the current working state after removing abnormal fluctuation data, and the equipment inclination comprises the overall inclination angle of the loading slot and an inclination direction vector; S4, based on the equipment inclination obtained in S3, calling a loading dynamic twin calibration model, calculating the deviation between the equipment inclination parameters and preset standard posture parameters of the loading slot in the model, converting the inclination deviation value into motor pulse numbers for driving the adjustment motor of the loading slot through the posture-pulse mapping logic built in the model, and the motor pulse numbers comprise X axis direction adjustment pulse numbers, Y axis direction adjustment pulse numbers and Z axis direction adjustment pulse numbers; S5, transmitting the motor pulse numbers in each direction obtained through S4 to the motor control unit of the SMIF machine, driving the corresponding adjustment motor to operate according to the pulse numbers in each direction, driving the wafer loading slot to perform multi-dimensional posture adjustment, so as to change the position of the abnormally installed wafer in the slot; and S6, in the adjustment process in S5, collecting the updated three-dimensional posture data of the level meter and the updated wafer tab signal of the wafer tab recognition module in real time through the SMIF loading precision control data analysis platform, repeating the operations of S2-S5, and until the wafer installation state is determined to meet the preset standard by the wafer channel three-dimensional pose perception algorithm.

[0006] Further, the expression of the wafer channel three-dimensional pose perception algorithm is as follows: wherein, is a wafer installation deviation value, is a weight coefficient of the i th group of wafer tab signals, is the i th group of wafer tab position coordinates, is a preset standard wafer tab coordinate, is an actual fitting length of the wafer tab profile feature data, is a preset standard wafer tab profile length, and n is the number of groups of collected wafer tab signals.

[0007] Further, the expression of the multi-source sensing dynamic deviation fitting algorithm is as follows: wherein, is the device inclination, m is the number of collected level data sets, is the inclination angle of the jth set of level data in the kth direction, is the inclination angle of the X-axis, Y-axis, Z-axis, is the inclination weight coefficient of the X-axis, Y-axis, Z-axis direction, is the time sequence decay coefficient, is the collection time of the jth set of data, is the average of all data collection times.

[0008] Further, the expression of the loading dynamic twin calibration model is: wherein, is the number of pulses of the kth adjustment motor, k corresponds to the X-axis, Y-axis, Z-axis direction motor respectively, is the pulse coefficient of the kth motor, is the standard inclination angle of the kth direction, is the standard deviation of the inclination data of the kth direction, is the preset standard deviation, is the dynamic calibration coefficient of the kth direction.

[0009] Further, the SMIF loading precision control data analysis platform adopts a data fusion formula in the data collection process: wherein, is the fused data value, is the data fusion weight, is the level collection data, is the tab signal collection data, is the error value of the level data, is the error value of the tab signal data.

[0010] Further, in the S4, the motor pulse number conversion process also adopts a compensation formula: wherein, is the compensated motor pulse number, is the inclination compensation coefficient of the kth motor, is the historical data compensation coefficient of the kth motor, is the inclination angle of the kth direction in the jth set of level data.

[0011] Further, step S3 includes the following sub-steps: S31 Extracting the three-dimensional attitude data of the level instrument collected in S1 from the database of the SMIF-loaded precision control data analysis platform, and sorting the data according to the acquisition time sequence to obtain a time-series data sequence; S32 Inputting the time-series data sequence into the preprocessing module of the multi-source sensor dynamic deviation fitting algorithm, and the module performs outlier detection on the data, removing outliers that exceed the preset tilt angle range. Data is marked as outlier and removed, while valid data is retained. S33 assigns corresponding weight coefficients to the X, Y, and Z axes based on the importance of the tilt angle in each direction for the retained valid data. S34 calculates the weighted skew value corresponding to each valid data point; S34 introduces a time-series decay coefficient based on the difference between the data acquisition time and the average acquisition time. The weighted tilt values ​​are corrected, and the final tilt angle is calculated by averaging the results. .

[0012] Further, S4 includes the following sub-steps: S41 retrieves preset standard attitude parameters of the loading slot from the parameter library of the loading dynamic twin calibration model, which include standard tilt angles in the X-axis, Y-axis, and Z-axis directions. and standard deviation S42 will obtain the equipment tilt angle from S3. The difference between the tilt angle and the standard tilt angle in each direction is calculated to obtain the tilt deviation value in each direction; S43 adjusts the motor model and performance parameters according to each direction and calls the corresponding pulse coefficient. and dynamic calibration coefficient Simultaneously calculate the standard deviation of the current tilt data. S44 substitutes the tilt deviation value, pulse coefficient, standard deviation, and dynamic calibration coefficient into the expression of the loading dynamic twin calibration model to calculate the initial pulse number of the adjustment motor in each direction. .

[0013] Further, S5 includes the following sub-steps: S51 The initial pulse count of each direction motor obtained in S4 is transmitted to the motor control unit of the SMIF machine. The control unit performs format conversion on the pulse count, converting the digital signal into a control signal recognizable by the motor; S52 The motor control unit sends the converted control signal to the corresponding motor... axis, axis, The shaft adjusting motor sends a driving instruction, which includes pulse number, operation speed and operation direction parameters; S53, after the adjusting motor receives the driving instruction, starts operation according to the instruction parameters, drives the wafer loading slot to adjust the posture in the corresponding direction through the transmission mechanism, and feeds back the operation state signal in real time during the adjustment; S54, the motor control unit receives the operation state signal fed back by the motor, confirms whether the motor completes the operation according to the preset pulse number, and if not, continues to send the driving instruction until the pulse number of the motor operation reaches the preset value.

[0014] A wafer loading deviation correction device comprises: a multi-source sensing data acquisition unit connected with a level meter and a tab identification module on a wafer loading slot, for collecting level meter three-dimensional posture data and tab signals in real time, and transmitting the collected data to a SMIF loading precision control data analysis unit; the SMIF loading precision control data analysis unit is connected with the multi-source sensing data acquisition unit and a wafer channel three-dimensional position perception algorithm processing unit, for storing, sorting and preprocessing the collected data, and sending the processed data to the corresponding algorithm processing unit; the wafer channel three-dimensional position perception algorithm processing unit is connected with the SMIF loading precision control data analysis unit and a multi-source sensing dynamic deviation fitting algorithm processing unit, for receiving the processed tab signal and determining the wafer installation state, and triggering the multi-source sensing dynamic deviation fitting algorithm processing unit to work if the determination is abnormal; the multi-source sensing dynamic deviation fitting algorithm processing unit is connected with the wafer channel three-dimensional position perception algorithm processing unit and a loading dynamic twin calibration model calculation unit, for receiving the level meter data and calculating the device inclination, and transmitting the inclination data to the loading dynamic twin calibration model calculation unit; the loading dynamic twin calibration model calculation unit is connected with the multi-source sensing dynamic deviation fitting algorithm processing unit and a motor control driving unit, for converting the motor pulse number according to the device inclination, and transmitting the pulse number to the motor control driving unit; the motor control driving unit is connected with the loading dynamic twin calibration model calculation unit and the adjusting motor of the wafer loading slot, for receiving the motor pulse number and driving the adjusting motor to operate, and adjusting the posture of the wafer loading slot.

[0015] Beneficial effects: the present application collects the level attitude data and the tab signal in real time through the multi-source sensing assembly, accurately determines the wafer mounting abnormality through the crystal channel three-dimensional position sensing algorithm, calculates the device inclination through the multi-source sensing dynamic deviation fitting algorithm, converts the motor pulse number combined with the loading dynamic twin calibration model, drives the motor to adjust the wafer position, simultaneously relies on the SMIF loading precision control data analysis platform to realize dynamic deviation correction. The method combines the level and tab signal two types of data, improves the abnormality determination accuracy and inclination calculation accuracy through algorithm processing, provides accurate data support for deviation correction; in view of the problems that the prior art lacks dynamic calibration and real-time feedback, single adjustment is easy to be not in place or excessive, the method collects and updates data in real time in the adjustment process, repeatedly executes the abnormality determination, deviation calculation and motor adjustment steps, until the wafer mounting meets the standard, avoids false alarm and wafer damage, and guarantees the stable operation of the equipment and the processing quality. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The method steps flowchart of the present application is shown in the figure.

[0017] Figure 2 The device unit composition diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0018] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict, and the present application will be further described in detail in combination with the drawings and specific embodiments.

[0019] As shown in the figure, a wafer loading deviation correction method comprises the following steps: Figure 1

[0020] S1, calling the multi-source sensing assembly through the SMIF loading precision control data analysis platform, collecting the three-dimensional attitude data output by the level configured on the wafer loading slot in the wafer transmission device and the tab signal generated by the corresponding wafer tab recognition module in real time, wherein the three-dimensional attitude data includes the inclination angle data of the loading slot X axis, Y axis and Z axis, and the tab signal includes the tab position coordinates and tab contour feature data;

[0021] ​Specifically, the implementation process of S1 first completes the connection debugging of the multi-source sensing component and the SMIF loading precision control data analysis platform, ensures that the data transmission rate is stable at above 100 Mbps, and the delay is controlled within 50 ms. The collected level meter three-dimensional attitude data includes inclination angle data in X, Y, and Z axes, each of which is measured in the range of -5° to +5°, with a measurement accuracy of 0.001° and a collection frequency of 10 Hz to ensure real-time capture of the small attitude changes of the loading slot. The tab signal is obtained by an industrial camera with a resolution of 1280x720 and a laser positioning component, including tab position coordinates (coordinate accuracy controlled at 0.01 mm) and tab profile feature data (profile extraction error not exceeding 0.005 mm). When collecting, the vertical distance between the camera and the loading slot is fixed at 300 mm, and the light intensity is maintained at 500-800 lux to avoid environmental light interference with data accuracy. This step provides comprehensive and accurate raw data support for subsequent abnormality judgment and deviation calculation by synchronously collecting two types of key data, avoiding errors in subsequent processing due to data loss or insufficient accuracy.

[0022] S2, input the tab signal collected by S1 into the wafer channel three-dimensional pose perception algorithm, compare the tab position coordinates with the preset standard tab coordinates through the algorithm, and combine the edge fitting result of the tab profile feature data to determine whether the wafer on the wafer loading slot is installed abnormally;

[0023] Specifically, S2 first imports the tab signal collected by S1 into the processing module of the wafer channel three-dimensional pose perception algorithm in time sequence. The algorithm first performs coordinate conversion on the tab position coordinates, converting the coordinates in the camera coordinate system to the coordinates in the loading slot coordinate system. During the conversion process, the preset coordinate system conversion parameters (including translation and rotation angle, with a translation error of 0.002 mm and a rotation angle error of not more than 0.001°) need to be introduced. Then, the algorithm compares the converted tab position coordinates with the preset standard tab coordinates (the standard coordinates need to be recorded into the system in advance according to the specifications of different types of wafers, such as a wafer with a diameter of 300 mm, whose standard tab coordinates are set as (150 mm, 0 mm, 0 mm)), calculates the coordinate deviation value; at the same time, the edge fitting of the tab profile feature data is performed, the least squares method is used to fit the profile curve, and the ratio of the actual fitting length to the preset standard tab profile length (the standard length is set according to the design specifications of the wafer tab, such as 10 mm) is calculated. The algorithm combines the coordinate deviation value and the profile length ratio value, sets the judgment threshold (the coordinate deviation threshold is 0.05 mm, and the profile length ratio threshold is 0.95-1.05), and when the coordinate deviation exceeds the threshold or the profile length ratio exceeds the set range, it is determined that the wafer has installation abnormalities. This step ensures the accuracy of abnormality judgment by multi-dimensional data comparison and precise threshold setting, avoiding misjudgment or omission.

[0024] S3, when S2 determines that there is an installation abnormality, inputting the level three-dimensional posture data collected by S1 into a multi-source sensing dynamic deviation fitting algorithm, performing time sequence correlation analysis on the tilt angle data at different collection time points through the algorithm, calculating the equipment tilt of the wafer loading slot under the current working state after eliminating abnormal fluctuation data, the equipment tilt including the overall tilt angle of the loading slot and the tilt direction vector;

[0025] Specifically, S3 is started after determining that there is an installation abnormality. First, the level three-dimensional posture data collected by S1 is extracted from the database of the SMIF loading precision control data analysis platform. The amount of extracted data needs to meet at least 20 groups of continuous data (corresponding to a collection time length of 2 seconds) to ensure the representativeness of the data. Then, the data is input into the multi-source sensing dynamic deviation fitting algorithm. The algorithm first detects abnormal values of the data, and uses the 3σ criterion (σ is the standard deviation of the data, which is determined based on at least 50 groups of historical normal data when calculating, and σ is usually not more than 0.002°) to mark and eliminate the data outside the range of [mean-3σ, mean+3σ] as abnormal data and retain valid data. Then, according to the force characteristics of each direction of the loading slot and the wafer installation requirements, weight coefficients are assigned to the X-axis, Y-axis and Z-axis respectively (for example, the X-axis weight is 0.4, the Y-axis weight is 0.3, and the Z-axis weight is 0.3. The weight coefficients need to be determined through more than 100 calibration experiments). The weighted tilt values of each valid data are calculated. Then, combined with the difference between the data collection time and the average collection time, a time sequence decay coefficient (set to 0.8, to ensure that the data closer to the current time has a larger weight) is introduced to correct the weighted tilt values. Finally, the mean value of all corrected weighted tilt values is obtained to obtain the equipment tilt (the calculation result is kept to 4 decimal places, with an accuracy of 0.0001°). This step effectively reduces the influence of environmental interference and data fluctuation on the tilt calculation and improves the calculation accuracy through data screening, weighted calculation and time sequence correction.

[0026] S4, based on the equipment tilt obtained by S3, calling the loading dynamic twin calibration model, calculating the deviation between the equipment tilt parameters and the preset standard posture parameters of the loading slot in the model, and converting the tilt deviation value into the motor pulse number for driving the loading slot adjustment motor through the built-in posture-pulse mapping logic in the model. The motor pulse number includes the X-axis direction adjustment pulse number, the Y-axis direction adjustment pulse number and the Z-axis direction adjustment pulse number.

[0027] Specifically, S4 first calls the parameter library loaded with the dynamic twin calibration model, extracts the preset loading slot standard posture parameters, wherein the standard inclination angles in each direction are set to 0° (allowable error ± 0.001°), and the standard deviation is set to 0.001° (determined based on the factory calibration data of the equipment). Then, the inclination of the equipment obtained by S3 is subtracted from the standard inclination angles in each direction to obtain the inclination deviation values in each direction (the deviation values are rounded to four decimal places); at the same time, the pulse coefficients of the motors in each direction are determined according to the model of the adjustment motor (such as a stepper motor, the step angle is set to 1.8°, and the subdivision multiple is 16) (for example, the pulse coefficient of the X-axis motor is 500 pulses / °, which is determined by the technical parameters provided by the motor manufacturer and actual debugging), combined with the running conditions of the equipment (such as temperature 25±2℃, humidity 40%-60%) to adjust the dynamic calibration coefficient (set to 1.02, used to compensate for the influence of environmental factors on the operation of the motor), and calculate the standard deviation of the current inclination data (based on the valid data retained in S3). Finally, the inclination deviation values, pulse coefficients, standard deviations, and dynamic calibration coefficients are substituted into the model to calculate the pulse number of the adjustment motor in each direction (the pulse number is an integer to ensure that the motor can be accurately executed), which introduces multiple sets of calibration parameters and working condition compensation coefficients to ensure the accuracy of the motor pulse number conversion and provides a scientific basis for subsequent adjustment.

[0028] S5, transmit the pulse number of the motor in each direction calculated by S4 to the motor control unit of the SMIF machine, and drive the corresponding adjustment motor to operate according to the pulse number in each direction, so as to drive the wafer loading slot to adjust the multi-dimensional posture and change the position of the abnormally mounted wafer in the slot;

[0029] Specifically, S5 starts, the pulse number of each direction motor (such as 1500 pulses of X axis, 800 pulses of Y axis, and 500 pulses of Z axis) obtained in S4 is transmitted to the motor control unit of the SMIF machine. The control unit receives data using RS485 communication protocol, and the transmission baud rate is set to 115200bps to ensure stable data transmission. The control unit first converts the pulse number, converts the decimal digital signal to a binary control signal recognizable by the motor, and sets the running speed (such as 500 pulses / second to avoid excessive speed causing the loading slot to vibrate) and running direction (based on the inclination deviation direction, such as setting the motor to positive rotation if the X axis deviation is positive) according to the motor operation requirements. Then the control unit sends a driving instruction to the corresponding adjustment motor, and the instruction sending interval is controlled within 10ms to prevent synchronization errors caused by instruction delay; after receiving the instruction, the motor drives the wafer loading slot to move in the corresponding direction through the ball screw transmission mechanism. During the movement, the motor feeds back the running state signal (including the current pulse number and running state) to the control unit every 100 pulses. The control unit receives the feedback signal in real time, and when it detects that the actual running pulse number of the motor reaches the preset value, it sends a stop instruction. If it does not reach, it continues to send a driving instruction until the pulse number matches. Through precise signal conversion and real-time feedback control, this step ensures that the loading slot completes the adjustment according to the preset parameters, avoiding excessive or insufficient adjustment.

[0030] S6, during the adjustment process of S5, the SMIF loading precision control data analysis platform collects the updated three-dimensional attitude data of the level meter and the updated tab signal of the tab recognition module in real time, and repeats the operations of S2-S5 until the wafer installation state meets the preset standard.

[0031] Specifically, S6 is started synchronously during the adjustment process of S5. The SMIF loading precision control data analysis platform collects the updated three-dimensional attitude data of the level meter (the collection parameters are the same as in S1 to ensure data comparability) and the updated tab signal of the tab recognition module (the same collection accuracy and environmental parameters are maintained) at a frequency of 10Hz. The new data collected needs to be transmitted to the corresponding algorithm processing module within 100ms, and the abnormality judgment process of S2 is repeated (the same judgment threshold and coordinate conversion parameters are used). If the wafer installation is still abnormal, the inclination calculation of S3 (the data processing logic is the same as in S3) and the pulse number conversion of S4 (the parameter library remains unchanged) are continued, and then the adjustment operation of S5 is performed according to the new pulse number. This cycle continues until the wafer installation state meets the preset standard (the coordinate deviation is ≤0.05mm and the contour length ratio is within the range of 0.95-1.05), at which point the platform records the final adjustment parameters and the time when the standard is met, and the cycle is stopped. Through real-time data collection and cyclic verification mechanism, this step ensures that the wafer loading state finally meets the standard and completely solves the installation abnormality problem.

[0032] Preferably, the expression for the crystal channel three-dimensional pose sensing algorithm is: ,in, This represents the wafer mounting tolerance value. Let be the weighting coefficient of the i-th group of convex signals. Let be the coordinates of the position of the i-th group of convex pieces. To preset standard convex coordinates, This represents the actual fitted length of the convex profile feature data. The preset standard convex profile length is given by n, where n is the number of convex signal groups collected.

[0033] Specifically, the 3D pose sensing algorithm for wafers calculates wafer mounting deviation values ​​using multi-dimensional data, providing a quantitative basis for anomaly detection. During implementation, the number of convex signal groups participating in the calculation is first determined, typically set to 10 continuously acquired signals to balance computational efficiency and result stability. The weighting coefficient of each convex signal group needs to be adjusted based on the environmental stability during signal acquisition. If the illumination intensity is stable at 500-800 lux and there is no vibration interference, the weighting coefficient is set to 0.1; if slight interference exists, the weighting coefficient is lowered to 0.08 to ensure that interference signals have less impact on the results. The convex position coordinates need to undergo coordinate system transformation first. The difference between the transformed actual coordinates and the preset standard coordinates (determined according to the wafer model, e.g., the standard coordinates for a 200mm diameter wafer are (100mm, 0mm, 0mm)) is calculated, and then combined with the ratio of the actual fitted length of the convex contour to the standard length (e.g., 8mm), and substituted into the algorithm for calculation. The least squares method is used for convex profile fitting, with the fitting error controlled within 0.005mm. The standard length needs to be confirmed through wafer design drawings and entered into the system in advance. This algorithm, through weight allocation and multi-parameter fusion, enables the calculated wafer mounting deviation value to reach an accuracy of 0.001mm, accurately reflecting the actual wafer mounting status and avoiding errors caused by single-parameter judgments. This provides reliable data support for whether to initiate correction in the future.

[0034] Preferably, the expression for the multi-source sensing dynamic deviation fitting algorithm is: ,in, The value is the equipment tilt angle, and m is the number of data sets collected from the level gauge. For the j-th group of level data Tilt angles of the X, Y, and Z axes. These are the tilt weighting coefficients in the X, Y, and Z axis directions. This is the time-series decay coefficient. Let j be the time of data collection for the j-th group. This is the average of all data collection times.

[0035] Specifically, the multi-source sensing dynamic deviation fitting algorithm is used to accurately extract the device inclination from the level data and exclude abnormal data interference. When implemented, first determine the number of collected level data sets, usually select 30 consecutive data sets (corresponding to a 3-second collection time), to ensure that the data is statistically representative. The weight coefficient of the inclination angle in each direction needs to be set according to the structural characteristics of the loading slot. The weight coefficient of the X-axis direction is set to 0.45 because the X-axis direction mainly bears the weight of the wafer. The weight coefficients of the Y-axis and Z-axis are 0.3 and 0.25, respectively. These values are determined through more than 150 empty and full load calibration experiments of the device to ensure that the influence of inclination in each direction on wafer installation can be accurately reflected. The time series decay coefficient is set to 0.85. This value is based on the data fluctuation law during device operation. The closer the collection time is to the current time, the greater the weight after decay, which can reduce the error influence of early data due to environmental changes. The 3σ criterion is used for abnormal value detection, and the standard deviation is calculated based on 100 consecutive level data collected during normal operation of the device. The standard deviation is usually controlled within 0.002°. Data outside the range of [mean-3σ, mean+3σ] is determined as abnormal and excluded. Through the above parameter setting and data processing process, the accuracy of the device inclination calculated by the algorithm can reach 0.0001°, which can truly reflect the current posture of the loading slot and provide accurate deviation data for subsequent pulse number conversion.

[0036] Preferably, the expression of the loading dynamic twin calibration model is: wherein, is the pulse number of the kth adjustment motor, k corresponds to the X-axis, Y-axis, and Z-axis direction motors, is the pulse coefficient of the kth motor, is the standard inclination angle in the kth direction, is the standard deviation of the inclination data in the kth direction, is the preset standard deviation, is the dynamic calibration coefficient in the kth direction.

[0037] Specifically, the dynamic twin calibration model is loaded to convert the device inclination into the number of pulses executable by the motor, so as to realize accurate correspondence between the deviation and the adjustment action. Before implementation, the standard inclination angle and the standard deviation in each direction need to be preset in the model parameter library. The standard inclination angle is set to 0° (with a factory error of ±0.001°), and the standard deviation is determined to be 0.0015° based on 200 calibration experiment data before the device leaves the factory, so as to ensure that the parameters meet the initial state of the device. The motor pulse coefficient needs to be determined according to the motor model. For example, if a stepper motor with a step angle of 1.8° and a subdivision multiple of 32 is selected, the X-axis pulse coefficient is set to 600 pulses / °, and the pulse coefficient of the Y-axis and the Z-axis is set to 550 pulses / ° due to smaller load. The value needs to be verified through motor locked-rotor experiment and actual load test to ensure that the motor can generate accurate displacement for each received pulse number. The dynamic calibration coefficient needs to be adjusted in combination with the real-time working condition. When the environmental temperature is 23-27℃ and the humidity is 40%-60%, the coefficient is set to 1.03; if the temperature exceeds this range, the coefficient is adjusted to 1.05 to compensate for the influence of temperature change on the running accuracy of the motor. The standard deviation of the current inclination data is calculated based on the effective data retained in the right 3, so as to ensure that the dispersion degree of the current data can be reflected. The motor pulse number error calculated by the model is controlled within ±1 pulse, which can ensure the accuracy of the motor driving the loading groove adjustment and avoid excessive or insufficient adjustment due to inaccurate pulse number.

[0038] Preferably, the SMIF loading precision control data analysis platform adopts a data fusion formula in the data acquisition process: wherein, is the fused data value, is the data fusion weight, is the level meter acquisition data, is the tab signal acquisition data, is the error value of the level meter data, is the error value of the tab signal data.

[0039] Specifically, the SMIF loading precision control data analysis platform uses a data fusion formula to integrate the two types of data, level meter and tab signal, to improve data reliability and subsequent processing efficiency. In implementation, the data fusion weight needs to be dynamically adjusted according to the collection accuracy of the two types of data. If the level meter measurement accuracy is 0.001° and the tab signal coordinate accuracy is 0.01mm, the weight is set to 0.6 to ensure that the level meter data with higher accuracy accounts for a larger proportion. If the tab signal accuracy is improved to 0.005mm through laser assisted positioning, the weight is adjusted to 0.5 to achieve balanced fusion of the two types of data. The level meter collected data needs to be subjected to outlier rejection and smoothing processing, and the error of the processed data is controlled within 0.0005°. The tab signal needs to complete contour fitting and coordinate conversion to ensure uniform data format. The error value of the level meter data is determined based on the equipment calibration report, which is usually 0.001°. The error value of the tab signal is calculated through multiple repeated collection and is set to 0.005mm. During the fusion process, the platform needs to monitor the collection state of the two types of data in real time. If a type of data is interrupted or the error exceeds the threshold value, the weight is automatically adjusted. For example, when the level meter data is abnormal, the weight is reduced to 0.3 and the tab signal weight is increased to 0.7 to ensure that the fused data can still reflect the actual state of the equipment. Through data fusion, the stability of the data output by the platform is improved by more than 40%, providing more reliable input for subsequent algorithm processing and reducing the impact of single data abnormality on the overall process.

[0040] Preferably, in the S4, the motor pulse number conversion process also uses a compensation formula: wherein, is the compensated motor pulse number, is the tilt compensation coefficient of the kth motor, is the historical data compensation coefficient of the kth motor, is the tilt angle of the kth direction in the jth group of level meter data.

[0041] Specifically, the motor pulse number compensation formula is used to correct the basic pulse number and eliminate adjustment errors caused by various factors during equipment operation. When implementing, the tilt compensation coefficient needs to be set according to the tilt range of the equipment. When the tilt is 0°-0.5°, the compensation coefficients of X-axis, Y-axis and Z-axis are all set to 0.02; when the tilt exceeds 0.5°, the compensation coefficient is adjusted to 0.03. This value is obtained through adjustment experiments of the equipment under different tilt conditions, to ensure that the influence of the increase of the tilt angle on the motor running accuracy can be offset. The historical data compensation coefficient needs to be calculated based on the adjustment records of the equipment in the past 30 days. If the average deviation between the actual displacement and the theoretical displacement of the motor in the historical adjustment is 0.002 mm, the compensation coefficient is set to 0.015; if the deviation exceeds 0.003 mm, the coefficient is adjusted to 0.02, to correct the mechanical wear error caused by long-term operation of the motor. When calculating, the historical tilt angle data in each direction in right 3 needs to be extracted, usually 50 groups of effective data are selected to ensure that the historical data can reflect the recent running state of the equipment. The motor pulse number after compensation needs to be verified by comparing the change of the tilt after the actual operation of the motor with the theoretical change, to ensure that the error after compensation is controlled within 0.001 mm. The application of the compensation formula improves the motor adjustment accuracy by 30%, effectively solves the problem of inaccurate basic pulse number caused by equipment wear and environmental changes, and ensures that the wafer can accurately reach the standard position after adjustment.

[0042] Preferably, S3 comprises the following steps: S31 extracts the three-dimensional attitude data of the level collected in S1 from the database of the SMIF loading precision control data analysis platform, sorts the data in time sequence to obtain a time series data sequence; S32 inputs the time series data sequence into the pretreatment module of the multi-source sensing dynamic deviation fitting algorithm, detects the abnormal values of the data through the module, marks the data exceeding the preset tilt angle range as abnormal data and eliminates them, and retains valid data; S33 assigns corresponding weight coefficients to the X-axis, Y-axis and Z-axis respectively according to the importance of the tilt angle in each direction , and calculates the weighted tilt values corresponding to each valid data; S34 combines the difference between the data collection time and the average collection time, introduces a time sequence attenuation coefficient to correct each weighted tilt value, and finally obtains the equipment tilt through mean value calculation.

[0043] Specifically, the step-by-step implementation process of S3 involves: S31 extracting the three-dimensional attitude data of the level instrument collected in S1 from the database of the precision control data analysis platform loaded in SMIF. During extraction, 20 consecutive data sets (corresponding to a 2-second acquisition duration and a 10Hz acquisition frequency) need to be selected to ensure the data covers the complete attitude change cycle. After extraction, the data is sorted in ascending order according to the acquisition timestamp to form a time-series data sequence. The sorting error is controlled within 10ms to prevent time sequence errors from causing subsequent analysis biases. S32 inputting the time-series data sequence into the multi-source sensor dynamic deviation fitting algorithm preprocessing module uses the 3σ criterion for outlier detection. The standard deviation is calculated based on 100 consecutive historical data sets during normal equipment operation, typically 0.002°. The preset tilt angle range is set to [-0.5°, 0.5°]. Data exceeding this range or deviating from the mean by more than 3σ are marked as outliers and removed. The amount of valid data retained needs to be... At least 15 sets of data are used to avoid abnormal data interfering with the calculation results. S33 assigns weight coefficients to the X, Y, and Z axes based on the structural characteristics of the loading slot and the wafer load requirements. The X-axis, which directly bears the wafer weight, is set to 0.4, while the Y and Z axes are each set to 0.3. These values ​​have been verified through 100 no-load and full-load calibration experiments to ensure accurate reflection of the impact of tilt in each direction on wafer mounting. The weighted tilt value for each set of valid data is then calculated according to these weights. S34 introduces a time-lapse coefficient of 0.8 (set based on the timeliness of data during equipment operation; data closer to the current moment has a higher weight). The weighted tilt value is corrected for time-lapse by combining the difference between the data acquisition time of each set and the average acquisition time of all data. Finally, the arithmetic mean of all corrected weighted tilt values ​​is taken to obtain the equipment tilt angle with an accuracy of 0.0001°, providing accurate deviation data for subsequent pulse count conversion.

[0044] Preferably, step S4 includes the following sub-steps: S41 retrieves preset standard attitude parameters of the loading slot from the parameter library of the loading dynamic twin calibration model, which include standard tilt angles in the X-axis, Y-axis, and Z-axis directions. and standard deviation S42 will obtain the equipment tilt angle from S3. The difference between the tilt angle and the standard tilt angle in each direction is calculated to obtain the tilt deviation value in each direction; S43 adjusts the motor model and performance parameters according to each direction and calls the corresponding pulse coefficient. and dynamic calibration coefficient Simultaneously calculate the standard deviation of the current tilt data. S44 substitutes the tilt deviation value, pulse coefficient, standard deviation, and dynamic calibration coefficient into the expression of the loading dynamic twin calibration model to calculate the initial pulse number of the adjustment motor in each direction. .

[0045] Specifically, the sub-step implementation process of S4, S41 calls the preset loading slot standard posture parameter from the dynamic twin calibration model parameter library, which includes the standard inclination angle and standard deviation of the X-axis, Y-axis and Z-axis direction, wherein the standard inclination angle is set to 0° (allowing ±0.001° factory calibration error), and the standard deviation is 0.0015° (determined based on 200 empty load calibration experimental data before the device leaves the factory), and the parameter integrity and validity need to be confirmed by the parameter verification mechanism when calling, if there is a missing or error, it will automatically call the supplement from the backup parameter library to ensure the accuracy of the parameter; S42 respectively calculates the difference between the device inclination calculated by S3 and the standard inclination angle of each axis to obtain the inclination deviation value of each X-axis, Y-axis and Z-axis, and the calculation process retains 4 decimal places, the accuracy is controlled within 0.0001°, to avoid the influence of difference calculation error on the subsequent pulse number conversion accuracy; S43 determines the key parameters according to the model specification of the adjustment motor, such as selecting a stepping motor with a step angle of 1.8° and a subdivision multiple of 16, the X-axis motor pulse coefficient is set to 500 pulses / °, and the Y-axis and Z-axis motors are set to 450 pulses / ° due to smaller load, the value is verified by motor locked-rotor experiment and actual load test to ensure that the pulse number and the actual displacement amount of the motor are accurately corresponding, and at the same time, the standard deviation of the current inclination data is calculated based on the effective data retained by S3, the sample size is not less than 15 groups, and the dynamic calibration coefficient 1.02 is determined according to the real-time environmental conditions (temperature 25±2℃, humidity 40%-60%), and when the temperature exceeds the range, it is adjusted to 1.05, to compensate for the influence of environmental factors on the motor running accuracy; S44 substitutes the inclination deviation value of each axis, the corresponding pulse coefficient, the current standard deviation and the dynamic calibration coefficient into the loading dynamic twin calibration model to obtain the initial pulse number of each axis adjustment motor through the built-in calculation logic of the model, and the calculation result is rounded to an integer, the error is controlled within ±1 pulse, and the motor can execute the adjustment action according to the accurate pulse number.

[0046] Preferably, S5 includes the following sub-steps: S51 transmits the initial pulse number of each direction motor obtained by S4 to the motor control unit of the SMIF machine, and the control unit converts the pulse number into a control signal recognizable by the motor through format conversion; S52 the motor control unit sends a driving instruction to the corresponding axis, axis, axis adjustment motor according to the converted control signal, and the instruction includes pulse number, running speed and running direction parameters; S53 after receiving the driving instruction, the adjustment motor starts running according to the instruction parameters, drives the wafer loading slot to adjust the posture along the corresponding direction through the transmission mechanism, and feedbacks the running state signal in real time during the adjustment process; S54 the motor control unit receives the running state signal feedback by the motor, confirms whether the motor completes the running according to the preset pulse number, if not, continues to send the driving instruction until the motor running pulse number reaches the preset value.

[0047] Specifically, the sub-step implementation process of S5, S51 transmits the initial pulse number of each axis motor obtained in S4 to the motor control unit of the SMIF machine through an industrial communication interface. The transmission adopts the RS485 communication protocol, the baud rate is set to 115200bps, and the transmission delay is controlled within 50ms to prevent data transmission lag from causing multi-axis adjustment out of sync. After receiving the data, the motor control unit first checks the reasonableness of the pulse number, confirms that the pulse number is within the maximum allowed pulse range of the motor (100-5000 pulses), and after the verification passes, converts the decimal digital pulse signal into a binary control signal recognizable by the motor, with a conversion error less than 0.1% to ensure the accuracy of the control signal. S52, the motor control unit generates a driving instruction according to the converted control signal, which contains the pulse number required for motor operation, operation speed (set to 500 pulses / second to avoid excessive speed causing the loading slot to vibrate and affect wafer stability), and operation direction (determined based on the direction of the deviation of the inclination of each axis, such as positive deviation for the X-axis, the motor rotates in the positive direction, and negative deviation for the motor to reverse), the instruction is stored in the local cache after generation, which is convenient for subsequent fault tracing and data review; then send the driving instruction to the corresponding motor at an interval of 10ms to prevent instruction congestion causing motor response delay; S53, after receiving the driving instruction, the adjustment motor of each axis drives the wafer loading slot to move in the corresponding direction through the ball screw transmission mechanism. During the movement, the motor feeds back the operation state signal to the motor control unit once every 100 pulses, which contains the current number of pulses operated, motor working voltage, working current and other parameters. The feedback frequency is consistent with the instruction sending frequency to ensure that the control unit can grasp the motor running state in real time; S54, the motor control unit compares the actual number of pulses fed back with the preset pulse number, if the difference between the two is within ±1 pulse, it is determined that the adjustment is completed, and a stop instruction is sent; if the difference exceeds 5 pulses, it is determined that the adjustment is abnormal, the driving instruction is re-sent for correction until the actual number of pulses operated is consistent with the preset value, ensuring that the loading slot completes the posture adjustment according to the preset parameters, avoiding excessive or insufficient adjustment affecting the installation precision of the wafer.

[0048] The wafer three-dimensional pose perception algorithm is the core algorithm for accurately determining the wafer mounting state in the present application. Its essence is to identify the abnormal mounting of the wafer in the loading slot through multi-dimensional data fusion and quantitative calculation. To implement this algorithm, the industrial camera and the laser positioning assembly are first used to collect the tab signal, including the tab position coordinates (the precision is controlled within 0.01 mm) and the tab profile feature data (the profile extraction error is not more than 0.005 mm). Then, the collected tab position coordinates are converted from the camera coordinate system to the loading slot coordinate system (the conversion parameter error is controlled within 0.002 mm and 0.001°). Subsequently, the algorithm compares and calculates the deviation value of the converted coordinates with the preset standard tab coordinates (set according to the wafer model, such as the standard coordinates of a 300 mm diameter wafer being (150 mm, 0 mm, 0 mm)), and performs least squares fitting on the tab profile data to calculate the ratio of the actual fitting length to the standard profile length (such as 10 mm). Finally, the coordinate deviation value and the profile length ratio are combined to determine whether the installation is abnormal through the preset threshold (the coordinate deviation threshold is 0.05 mm, and the profile length ratio range is 0.95-1.05). The role of this algorithm is to replace the traditional single-dimensional abnormal detection method and achieve accurate identification of the wafer mounting state. Its significance lies in avoiding invalid correction or non-correction due to misjudgment or missed judgment, providing reliable basis for the subsequent correction process, and ensuring the stable operation of the wafer transmission equipment.

[0049] The multi-source sensing dynamic deviation fitting algorithm is a key algorithm for extracting the precise device inclination from the level data, aiming to eliminate the influence of data fluctuations and environmental interference on the inclination calculation. In the implementation process, first, the level three-dimensional attitude data (including X-axis, Y-axis, and Z-axis inclination angles, with a measurement accuracy of 0.001° and a collection frequency of 10 Hz) is obtained from the data acquisition end. Then, 20 groups of valid data (corresponding to a 2-second duration) are selected and sorted by time. Next, the 3σ rule (based on 100 groups of historical normal data to calculate the standard deviation, usually 0.002°) is used to remove abnormal data outside the range of [mean-3σ, mean+3σ], and at least 15 groups of valid data are retained. Then, according to the force characteristics of the loading slot, the weight coefficients of each axis are assigned (X-axis 0.4, Y-axis 0.3, Z-axis 0.3, determined by 100 calibration experiments), the weighted inclination value of each group of data is calculated, and the time decay coefficient 0.8 (set according to the data timeliness rule) is introduced to correct the weighted inclination value. Finally, the mean value is taken to obtain the device inclination with a precision of 0.0001°. The role of this algorithm is to convert the original sensing data into quantitative deviation values that truly reflect the device attitude. Its significance lies in providing accurate data support for subsequent motor pulse number conversion, avoiding insufficient correction accuracy due to inclination calculation errors, and reducing the risk of wafer damage.

[0050] The loading dynamic twin calibration model is the core model connecting the device inclination and the motor adjustment action. Through establishing the mapping relationship between the deviation and the execution parameter, the accurate deviation correction control is realized. To realize the model, the standard posture parameters are preset in the parameter library, including the standard inclination angle of each axis (0°, with an error of ±0.001°) and the standard deviation (0.0015°, determined based on 200 factory calibration data), and the pulse coefficients of different motor models (for example, for a motor with a step angle of 1.8° and a subdivision multiple of 16, the X-axis has 500 pulses / ° and the Y-axis has 450 pulses / °) and the dynamic calibration coefficients (normal 1.02 and 1.05 in the case of temperature abnormality) are stored. During the model operation, the deviation value of the device inclination from the standard inclination angle is calculated first, and the standard deviation of the current inclination data (calculated based on 15 groups of effective data) is combined to put the deviation value, the pulse coefficient, the standard deviation, and the calibration coefficient into the built-in logic to convert the initial pulse number of each axis motor (with an error of ±1 pulse) and the pulse number can also be corrected through the compensation formula (introducing the inclination compensation coefficient of 0.02-0.03 and the historical data compensation coefficient of 0.015-0.02). The role of the model is to convert the abstract inclination deviation into specific pulse parameters executable by the motor, and the significance lies in realizing the accurate matching of the deviation and the adjustment action, avoiding over-adjustment or insufficient adjustment, ensuring the wafer to quickly return to the standard position, and improving the efficiency of the device operation.

[0051] The SMIF loading precision control data analysis platform is the core data processing and scheduling hub of the present application, which bears the functions of data acquisition, storage, transmission and collaborative scheduling. To realize the platform, a multi-source data access interface is built to connect with the level meter (transmission rate 100 Mbps, delay within 50 ms), the tab identification module (1280×720 resolution camera) and the motor control unit (RS485 protocol, 115200 bps baud rate) to collect three-dimensional posture data, tab signals and motor state signals in real time. The built-in database of the platform stores historical data and standard parameters, supports data sorting by time, preliminary screening of abnormal values (based on the preset range [-0.5°, 0.5°]), and has data distribution function to send the processed tab signals to the wafer three-dimensional position sensing algorithm, the level meter data to the multi-source sensing dynamic deviation fitting algorithm, and the motor pulse number to the control unit. The platform can also collect and update data in real time during the adjustment process to drive the process cycle until the standard is reached. The role of the platform is to realize the data flow and collaborative work of each technical module, and the significance lies in integrating the scattered sensing and control links to form a closed-loop deviation correction system, avoiding the low efficiency or data discontinuity caused by independent operation of each module, ensuring the smoothness and efficiency of the entire deviation correction process, and improving the automation and intelligence level of the wafer transport equipment.

[0052] As Figure 2As shown, a wafer loading correction device comprises: a multi-source sensing data acquisition unit connected with a level gauge and a tab recognition module on a wafer loading slot, for real-time acquisition of level gauge three-dimensional attitude data and tab signals, and transmission of the acquired data to an SMIF loading precision control data analysis unit; the SMIF loading precision control data analysis unit is connected with the multi-source sensing data acquisition unit and a wafer channel three-dimensional position perception algorithm processing unit, for storage, sorting and preprocessing of the acquired data, and sending of the processed data to the corresponding algorithm processing unit; the wafer channel three-dimensional position perception algorithm processing unit is connected with the SMIF loading precision control data analysis unit and a multi-source sensing dynamic deviation fitting algorithm processing unit, for receiving the processed tab signals and determining the wafer installation state, and triggering the multi-source sensing dynamic deviation fitting algorithm processing unit to work if the determination is abnormal; the multi-source sensing dynamic deviation fitting algorithm processing unit is connected with the wafer channel three-dimensional position perception algorithm processing unit and a loading dynamic twin calibration model calculation unit, for receiving the level gauge data and calculating the device inclination, and transmitting the inclination data to the loading dynamic twin calibration model calculation unit; the loading dynamic twin calibration model calculation unit is connected with the multi-source sensing dynamic deviation fitting algorithm processing unit and a motor control driving unit, for converting the motor pulse number according to the device inclination, and transmitting the pulse number to the motor control driving unit; the motor control driving unit is connected with the loading dynamic twin calibration model calculation unit and the adjusting motor of the wafer loading slot, for receiving the motor pulse number and driving the adjusting motor to operate, and adjusting the attitude of the wafer loading slot.

[0053] A wafer loading correction method and device can comprehensively optimize the wafer loading correction effect and specifically overcome the defects in the prior art. The advantages are: first, the data acquisition is comprehensive and real-time, relying on multi-source sensing components to synchronously obtain level gauge attitude data and tab signals, providing a rich data basis for subsequent processing; second, the abnormality determination and deviation calculation are accurate, through wafer channel three-dimensional position perception algorithm and multi-source sensing dynamic deviation fitting algorithm to respectively realize accurate identification of wafer installation abnormalities and accurate calculation of device inclination; third, the adjustment process is scientific and controllable, relying on the loading dynamic twin calibration model to convert the motor pulse number, ensuring reasonable adjustment parameters, and relying on the SMIF loading precision control data analysis platform to realize real-time feedback and repeated verification. In view of the defect in the prior art that relying on single sensing data leads to large determination and calculation errors, this method fuses two types of key data, processes them through professional algorithms, greatly improves the determination accuracy and calculation precision, and provides reliable data support for correction; for the problems of lack of dynamic calibration and real-time feedback in the prior art, and single adjustment prone to not in place or excessive, this method continuously acquires and updates data during the adjustment process, repeatedly executes the abnormality determination, deviation calculation and motor adjustment process until the wafer installation meets the standard, effectively avoids false positives and wafer damage, and ensures stable operation of the equipment and wafer processing quality.

[0054] In the description of the application, it is necessary to point out that, unless otherwise explicitly specified and limited, the terms "arrange", "install", "connect", "join", "fix" should be understood in a broad sense, for example, can be fixedly connected, can also be detachably connected, or integrally connected; can be mechanically connected, can also be electrically connected; can be directly connected, can also be indirectly connected through an intermediate medium, can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the application can be understood according to the specific circumstances.

[0055] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various equivalent changes, modifications, replacements and variations of the embodiments can be made without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalent scope.

Claims

1. A wafer loading correction method, characterized by, The method comprises the following steps: S1, calling a multi-source sensing component through an SMIF loading precision control data analysis platform to collect three-dimensional attitude data output by a level meter arranged on a wafer loading slot in a wafer transfer device and wafer tab signal generated by a corresponding wafer tab recognition module in real time, wherein the three-dimensional attitude data comprises inclination angle data of the loading slot in X, Y and Z axes, and the wafer tab signal comprises wafer tab position coordinates and wafer tab profile feature data; S2, inputting the wafer tab signal collected in S1 into a wafer channel three-dimensional pose perception algorithm, comparing the wafer tab position coordinates with preset standard wafer tab coordinates through the algorithm, combining edge fitting results of the wafer tab profile feature data, and determining whether the wafer on the wafer loading slot is installed abnormally; S3, when it is determined in S2 that there is an installation abnormality, inputting the level meter three-dimensional attitude data collected in S1 into a multi-source sensing dynamic deviation fitting algorithm, performing time sequence correlation analysis on the inclination angle data at different collection time points through the algorithm, calculating the equipment inclination of the wafer loading slot under the current working state after removing abnormal fluctuation data, and the equipment inclination comprises the overall inclination angle and inclination direction vector of the loading slot; S4, based on the equipment inclination obtained in S3, calling a loading dynamic twin calibration model, calculating the deviation between the equipment inclination parameters and preset standard posture parameters of the loading slot in the model, converting the inclination deviation value into motor pulse numbers for driving the adjustment motor of the loading slot through the posture-pulse mapping logic built in the model, and the motor pulse numbers comprise X-axis direction adjustment pulse numbers, Y-axis direction adjustment pulse numbers and Z-axis direction adjustment pulse numbers.

2. The wafer load correction method of claim 1, wherein, The method further comprises: S5, transmitting the motor pulse numbers in each direction converted in S4 to a motor control unit of the SMIF machine, driving the corresponding adjustment motor to operate according to the pulse numbers in each direction, driving the wafer loading slot to perform multi-dimensional posture adjustment, and changing the position of the wafer installed abnormally in the slot; S6, in the adjustment process in S5, collecting the updated three-dimensional attitude data of the level meter and the updated wafer tab signal of the wafer tab recognition module in real time through the SMIF loading precision control data analysis platform, repeating the operations of S2-S5, and until the wafer channel three-dimensional pose perception algorithm determines that the wafer installation state meets the preset standard; The expression of the crystal channel three-dimensional pose perception algorithm is: wherein, is a wafer installation deviation value, is a weight coefficient of the i-th group of tab signals, is the i-th group of tab position coordinates, is a preset standard tab coordinate, is an actual fitting length of the tab profile feature data, is a preset standard tab profile length, and n is the number of collected tab signal groups.

3. The wafer load correction method of claim 1, wherein, The expression of the multi-source sensing dynamic deviation fitting algorithm is: wherein, is the device inclination, m is the number of collected level data groups, is the inclination of the X-axis, Y-axis and Z-axis in the jth group of level data, is the inclination angle of the X-axis, Y-axis and Z-axis, is the inclination weight coefficient of the X-axis, Y-axis and Z-axis direction, is the time sequence attenuation coefficient, is the collection time of the jth group of data, is the average value of all data collection times.

4. The wafer load correction method of claim 1, wherein, The expression of the loading dynamic twin calibration model is: Wherein, is the pulse number of the kth adjusting motor, k corresponds to the X-axis, Y-axis, and Z-axis direction motor respectively, is the pulse coefficient of the kth motor, is the standard tilt angle of the kth direction, is the standard deviation of the tilt data of the kth direction, is the preset standard deviation, is the dynamic calibration coefficient of the kth direction.

5. The wafer load correction method of claim 1, wherein, The SMIF loading precision control data analysis platform adopts a data fusion formula in the data acquisition process: wherein, is the fused data value, is the data fusion weight, is the level meter acquisition data, is the tab signal acquisition data, is the error value of the level meter data, is the error value of the tab signal data.

6. The wafer load correction method of claim 1, wherein, In the S4, the motor pulse number conversion process also adopts a compensation formula: wherein, is the compensated motor pulse number, is the tilt compensation coefficient of the kth motor, is the historical data compensation coefficient of the kth motor, is the tilt angle of the kth direction in the jth group of level data.

7. The wafer load correction method of claim 1, wherein, S3 includes the following steps: S31 Extract the three-dimensional attitude data of the level instrument collected in S1 from the database of the SMIF-loaded precision control data analysis platform, and sort the data according to the acquisition time sequence to obtain a time-series data sequence; S32 Input the time-series data sequence into the preprocessing module of the multi-source sensor dynamic deviation fitting algorithm, and the module performs outlier detection on the data, removing outliers that exceed the preset tilt angle range. Data is marked as outlier and removed, while valid data is retained. S33 assigns corresponding weight coefficients to the X, Y, and Z axes based on the importance of the tilt angle in each direction for the retained valid data. S34 calculates the weighted skew value corresponding to each valid data point; S34 introduces a time-series decay coefficient based on the difference between the data acquisition time and the average acquisition time. The weighted tilt values ​​are corrected, and the final tilt angle is calculated by averaging the results. .

8. The wafer loading and correction method of claim 1, wherein, The S4 comprises the following sub-steps: S41, calling preset loading slot standard posture parameters from a parameter library loaded with a dynamic twin calibration model, the parameters comprising standard inclination angles in X-axis, Y-axis and Z-axis directions and standard deviations ; S42 obtains the device inclination degree from S3 The inclination degree deviation value of each direction is obtained by calculating the difference between the standard inclination angle of each direction. S43, according to the model and performance parameters of the adjusting motor in each direction, call the corresponding pulse coefficient and dynamic calibration coefficient , at the same time, calculate the standard deviation of the current inclination data ; S44, put the inclination deviation value, pulse coefficient, standard deviation, dynamic calibration coefficient into the expression of the loading dynamic twin calibration model, calculate the initial pulse number of the adjusting motor in each direction .

9. The wafer load correction method of claim 2, wherein, The S5 comprises the following steps: S51 transmitting the initial pulse number of each direction motor obtained in S4 to a motor control unit of the SMIF machine, and the control unit converts the pulse number into a control signal recognizable by the motor; S52 the motor control unit sends a driving instruction to the corresponding shaft, shaft, shaft adjustment motor according to the converted control signal, and the driving instruction comprises the pulse number, the operation speed and the operation direction parameters; S53 the adjustment motor starts operation according to the instruction parameters after receiving the driving instruction, drives the wafer loading slot to adjust the posture along the corresponding direction through the transmission mechanism, and feeds back the operation state signal in real time during the adjustment process; S54 the motor control unit receives the operation state signal fed back by the motor, confirms whether the motor completes the operation according to the preset pulse number, and continues to send the driving instruction until the pulse number of the motor operation reaches the preset value if the operation is not completed.

10. A wafer loading correction device, characterized by, It comprises: a multi-source sensing data collection unit connected with the level meter and the wafer tab recognition module on the wafer loading slot, used for collecting the three-dimensional attitude data of the level meter and the wafer tab signal in real time, and transmitting the collected data to the SMIF loading precision control data analysis unit; an SMIF loading precision control data analysis unit connected with the multi-source sensing data collection unit and the wafer channel three-dimensional pose perception algorithm processing unit, used for storing, sorting and preprocessing the collected data, and sending the processed data to the corresponding algorithm processing unit; a wafer channel three-dimensional pose perception algorithm processing unit connected with the SMIF loading precision control data analysis unit and the multi-source sensing dynamic deviation fitting algorithm processing unit, used for receiving the processed wafer tab signal and determining the wafer installation state, and triggering the multi-source sensing dynamic deviation fitting algorithm processing unit to work if the wafer installation state is determined to be abnormal; The multi-source sensing dynamic deviation fitting algorithm processing unit is connected with the wafer three-dimensional position and posture sensing algorithm processing unit and the loading dynamic twin calibration model calculation unit, is used for receiving the level data and calculating the device inclination, and transmits the inclination data to the loading dynamic twin calibration model calculation unit; The loading dynamic twin calibration model calculation unit is connected with the multi-source sensing dynamic deviation fitting algorithm processing unit and the motor control driving unit, is used for converting the motor pulse number according to the device inclination, and transmits the pulse number to the motor control driving unit; The motor control driving unit is connected with the loading dynamic twin calibration model calculation unit and the adjusting motor of the wafer loading slot, is used for receiving the motor pulse number and driving the adjusting motor to run, and adjusts the posture of the wafer loading slot.

Images