UPA pressure automatic compensation system and method for mask substrate polishing

Abstract

The invention discloses a UPA pressure automatic compensation system and method for mask substrate polishing. The UPA pressure automatic compensation system comprises a pressure execution module, a distributed sensing feedback network and an intelligent control unit. Wherein the pressure execution module is used for directly polishing a mask base plate and comprises a force application shaft, a polishing head, an air source, a UPA air bag and an electric proportional valve, the distributed sensing feedback network comprises an air path pressure sensor and a tail end contact pressure sensor, and the intelligent control unit receives a collection signal of the distributed sensing feedback network and sends the collection signal to the polishing head. On the basis of a current process formula and process parameter coupling model, a data fusion algorithm, a feed-forward compensation algorithm and a fuzzy PID algorithm are operated to obtain a feed-forward compensation amount and a fuzzy PID output amount, the feed-forward compensation amount and the fuzzy PID output amount are superposed, a total control voltage is generated and output to an electric proportional valve, and then the input air pressure of a UPA air bag is adjusted. Therefore, closed-loop control over the polishing pressure of the mask substrate is achieved. Through multi-sensor data fusion, fuzzy PID self-adaptive control and intelligent feed-forward compensation based on a process parameter coupling model, real-time, accurate and constant control of the polishing pressure is realized, so that the surface precision and yield of the mask substrate are improved.

Description

Technical Field

[0001] This invention relates to the precision manufacturing and processing of semiconductors, specifically to an automatic compensation system and method for UPA pressure in mask substrate polishing, belonging to the technical field of semiconductor processing equipment. Background Technology

[0002] In the manufacturing process of semiconductor mask substrates, the nanoscale flatness and sub-nanometer roughness of the substrate surface directly determine the quality of subsequent photolithography imaging. Chemical mechanical polishing (CMP) is the core process for achieving ultra-smooth surfaces, and the precision of pressure control applied by the polishing head to the substrate directly determines the uniformity of the removal rate, thus affecting the surface accuracy and yield of the polished mask substrate.

[0003] Currently, mainstream mask polishing equipment generally adopts a polishing head structure driven by ultra-purified air (UPA) airbags, and uses an electric proportional valve to adjust the air pressure to set the target load. Please refer to [link / reference]. Figure 1 During the polishing operation, the mask substrate 1 is placed on the polishing pad 2, which is attached to the upper surface of the rotating polishing disk 3. The force shaft fixed at the center of the polishing head 4 is connected to the rotary motor 6 through the coupling 5 to achieve rotation. The UPA (ultra-pure clean air) airbag 7 is placed in the cavity inside the polishing head 4. The electric proportional valve 8 is connected to the air path of the UPA airbag 7 and precisely controls the input positive pressure to inflate the UPA airbag 7, thereby applying uniform pressure downward to the mask substrate 1. Under the flexible pressure of the UPA airbag 7, the mask substrate 1 is in close contact with the polishing pad 2 below. Through the relative rotational movement of the polishing head 4 and the polishing disk 3, the polishing operation of the surface of the mask substrate 1 is completed.

[0004] However, existing UPA pressure control systems generally have the following drawbacks: 1. Open-loop or simple closed-loop control: Most devices only set the air pressure through an electro-proportional valve (precision pressure regulating valve) and cannot sense the actual pressure of the UPA airbag at the end of the polishing head contacting the workpiece on the mask substrate.

[0005] 2. Weak resistance to disturbances: During the polishing process, as the polishing pad wears and thins, the polishing fluid flows, and the spindle of the equipment vibrates mechanically, the actual contact pressure between the UPA airbag and the mask substrate will fluctuate nonlinearly.

[0006] 3. Lack of process adaptability: Traditional PID (proportional-integral-derivative) control parameters used to control polishing pressure are fixed, making it difficult to take into account the different requirements for response speed and stability under different process formulations. This results in poor consistency between batches of workpieces polished from mask substrates, and easily leads to defects such as surface collapse or insufficient polishing removal.

[0007] Therefore, there is an urgent need for an intelligent pressure compensation system that can sense the end contact force in real time, actively predict process disturbances, and adaptively adjust the control strategy to break through the current bottleneck of pressure control technology in ultra-precision polishing of mask substrates. Summary of the Invention

[0008] To address the shortcomings of the prior art, this invention aims to solve the problem of pressure instability caused by consumable wear, mechanical vibration, and changes in process parameters during mask substrate polishing. It provides an automatic compensation system and method for UPA pressure in mask substrate polishing. Through multi-sensor data fusion, fuzzy PID adaptive control, and intelligent feedforward compensation based on process model, it achieves real-time, accurate, and constant control of polishing pressure, thereby improving the surface accuracy and yield of the mask substrate.

[0009] The technical solution adopted by this invention to solve its technical problem is as follows: An automatic compensation system for UPA pressure in mask substrate polishing, wherein the mask substrate is disposed on a polishing pad attached to the upper surface of a rotating polishing disk, characterized in that the automatic compensation system comprises: The pressure actuation module, used to directly polish a mask substrate, includes a force-applying shaft, a polishing head, an air source, a UPA airbag, and an electro-proportional valve. The UPA airbag, connected to the air source, is disposed in the inner cavity of the polishing head and pressed onto the mask substrate. The rotating force-applying shaft is fixedly connected to the top center of the polishing head and drives the UPA airbag to apply a downward, uniform, flexible pressure to the mask substrate to achieve polishing of the mask substrate surface. The electro-proportional valve is connected in the air path between the UPA airbag and the air source and is used to adjust the input air pressure of the UPA airbag to regulate the polishing pressure on the mask substrate. A distributed sensing feedback network includes an air path pressure sensor and an end contact pressure sensor. The air path pressure sensor is installed in the air path between the output end of the electro-proportional valve and the air inlet of the UPA airbag, and is used to detect in real time the actual pipeline air pressure output by the electro-proportional valve and supplied to the UPA airbag. The end contact pressure sensor is installed at the end of the force application shaft and is used to directly detect the actual normal contact force of the UPA airbag pressing against the mask substrate. The intelligent control unit, connected to the electro-proportional valve, the pneumatic pressure sensor, and the end-contact pressure sensor, incorporates a data fusion algorithm, a process parameter coupling model trained on historical data, a feedforward compensation algorithm, and a fuzzy PID algorithm. This intelligent control unit receives signals from the pneumatic pressure sensor and the end-contact pressure sensor. Based on the current process formulation and the process parameter coupling model, it runs the data fusion algorithm, the feedforward compensation algorithm, and the fuzzy PID algorithm to obtain the feedforward compensation amount and the output of the fuzzy PID algorithm, which are then superimposed to generate the final total control voltage, which is output to the electro-proportional valve. This adjusts the input air pressure of the UPA airbag, thereby achieving closed-loop control of the polishing pressure of the mask substrate.

[0010] Furthermore, the force-applying shaft is connected to a rotary motor via a coupling or universal joint.

[0011] Furthermore, the current process formula refers to a set of pre-set process parameters for controlling the polishing process in the current processing, including target set pressure, polishing disc spindle speed, polishing fluid flow rate, and polishing time.

[0012] Another technical solution of the present invention is as follows: An automatic compensation method for UPA pressure in mask substrate polishing, implemented through the above-mentioned automatic compensation system, is characterized by upgrading the pressure control from "passive response" to "active prediction" through multi-sensor data fusion, fuzzy PID adaptive control, and intelligent feedforward compensation based on a process parameter coupling model, thereby achieving real-time, precise, and constant control of polishing pressure.

[0013] Furthermore, the automatic compensation method includes the following specific steps: Step 1) System initialization and loading of current process recipe — The intelligent control unit reads the current process formula to be executed, which includes the target set pressure, polishing disc spindle speed, polishing fluid flow rate and polishing time. At the same time, it reads the current actual thickness of the polishing pad as the basis for subsequent intelligent feedforward compensation calculation. Step 2) Intelligent feedforward compensation calculation — The intelligent control unit calculates the estimated pressure loss caused by the wear and thinning of the polishing pad and the deformation of the UPA airbag due to the rotation of the polishing disc, based on the process parameter coupling model, and uses this estimate as the feedforward compensation pressure value. In the formula, Feedforward compensation pressure value Initial standard thickness of polishing pad The current actual thickness of the polishing pad. Polishing disc spindle speed, : Based on the experimentally calibrated proportionality coefficient; Before the system is pressurized, the feedforward compensation pressure value is added to the target set pressure of the current process formula to calculate the initial total set pressure: In the formula, Initial total pressure setting Feedforward compensation pressure value Setting targets creates pressure; The calculation results are then converted into a feedforward compensation control voltage and sent to the pressure execution module. Step 3) Multi-source data acquisition and fusion — After the system starts polishing, the intelligent control unit collects the actual pipeline air pressure signal from the air pressure sensor and the actual normal contact force signal from the end contact pressure sensor in real time. At the same time, it uses Kalman filtering or moving average filtering algorithm to remove mechanical vibration noise, and converts the obtained actual normal contact force into an equivalent actual contact pressure for comparison with the initial total set pressure obtained in step 2). Step 4) Fuzzy PID Adaptive Control — The intelligent control unit calculates the pressure error. and its rate of change , In the formula, Initial total pressure setting Actual contact pressure; Then calculate the pressure error. and pressure error change rate The fuzzy logic controller of the intelligent control unit performs fuzzy inference and simultaneously performs parameter self-tuning based on the fuzzy rule table, dynamically adjusting the proportional coefficient of the fuzzy logic controller online. Integral coefficient and differential coefficients ; Step 5) Closed-loop output execution — The intelligent control unit converts the calculation result obtained by the fuzzy logic controller in step 4) into a fuzzy PID feedback control voltage, and superimposes it with the feedforward compensation control voltage obtained in step 2). In the formula, Total control voltage Feedforward compensation control voltage Fuzzy PID feedback control voltage; Finally, the generated total control voltage is sent to the electro-proportional valve, which linearly adjusts the input air pressure of the UPA airbag according to the magnitude of the total control voltage, thereby realizing closed-loop control of the actual contact pressure of the UPA airbag against the mask substrate.

[0014] Furthermore, in step 3), the actual contact pressure is obtained by dividing the actual normal contact force by the calculated contact area between the UPA airbag and the mask substrate.

[0015] Furthermore, in step 4), when the pressure error... When the value is large, increase the scaling factor. To accelerate response; when the pressure error is small and the actual contact pressure Approaching the initial total set pressure When, increase the integral coefficient To eliminate steady-state error and reduce the proportionality coefficient To prevent overshoot.

[0016] Furthermore, in step 5), the total control voltage is a DC voltage analog quantity of 0-10V.

[0017] Furthermore, steps 3) to 5) are continuously cycled at a high frequency with a period of 1ms until the polishing operation is completed.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: 1) Ultra-high precision constant control: Combining end-point direct sensing and data fusion technology, it eliminates the interference of air path delay and mechanical friction, realizes true contact pressure closed-loop control, and significantly reduces pressure fluctuation rate.

[0019] 2) Excellent dynamic response characteristics: The fuzzy PID algorithm solves the problem of parameter matching difficulties in traditional PID under nonlinear conditions (such as airbag deformation and rapid pressure change), and achieves control effects with fast response, small overshoot and zero steady-state error.

[0020] 3) Improved process consistency: An intelligent feedforward compensation model based on process parameters (polishing pad thickness, rotation speed) was introduced, upgrading pressure control from "passive response" to "active prediction". This effectively compensates for system errors caused by consumable wear, significantly improves the surface uniformity of different batches of mask substrates, and reduces the scrap rate. Attached Figure Description

[0021] Figure 1 This is a diagram showing the polishing process of the mask substrate.

[0022] Figure 2 This is a schematic diagram of the system hardware structure of the present invention.

[0023] Figure 3 This is a flowchart of the system control logic of the present invention.

[0024] Figure 4 This is a block diagram of the fuzzy PID control principle.

[0025] In the picture, 1—Mask substrate, 2—Polishing pad, 3—Polishing disc, 4—Polishing head, 5—Coupling, 6—Rotary motor, 7—UPA airbag, 8—Electrical proportional valve. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0027] This invention is used for the automatic control and compensation of the pressure exerted on the mask substrate by ultrapure clean air (UPA) during mask substrate polishing.

[0028] The automatic compensation system described in this invention mainly consists of three parts: a pressure execution module, a distributed sensor feedback network, and an intelligent control unit. (See attached diagram) Figure 2 .

[0029] 1. Pressure Execution Module: Please refer to the following document. Figure 1 and Figure 2 It includes a force-applying shaft, polishing head 4, air source, electric proportional valve 8, air pipeline and UPA airbag 7.

[0030] The UPA airbag 7 is connected to an ultrapure clean air (UPA) source, is disposed in the inner cavity of the polishing head 4, and is pressed onto the mask substrate 1. A rotating force shaft is fixedly connected to the top center of the polishing head 4 and drives the UPA airbag 7 to apply a downward uniform flexible pressure to the mask substrate 1 through the polishing head 4, so as to polish the surface of the mask substrate 1. An electro-proportional valve 8 is connected in the air path between the UPA airbag 7 and the air source, and is used to adjust the input air pressure of the UPA airbag according to the signal of the intelligent control unit, so as to regulate the polishing pressure on the mask substrate.

[0031] 2. Distributed sensor feedback network: Please refer to Figure 1 ,include: Air pressure sensor: Installed on the air pipeline between the output of the electro-proportional valve 8 and the air inlet of the UPA airbag 7, its function is to detect the actual output of the electro-proportional valve 8 and the actual pipeline air pressure supplied to the UPA airbag 7 in real time. This forms an inner loop monitoring system for the gas path itself; End contact pressure sensor: Embedded at the end of the force-applying shaft or at the universal joint connection of the polishing head 4, its function is to directly collect the actual normal contact force of the UPA airbag 7 pressing against the mask substrate 1. This signal is input to the intelligent control unit as the system's main feedback signal.

[0032] 3. Intelligent control unit (PLC): It has a built-in process parameter coupling model trained based on historical data, receives the acquisition signals from the air pressure sensor and the end contact pressure sensor, runs the data fusion algorithm, feedforward compensation algorithm and fuzzy PID algorithm, generates the total control voltage and outputs it to the electro-proportional valve 8, thereby adjusting the input air pressure of the UPA airbag 7.

[0033] Explanation of control and compensation methods: Step 1: System Initialization and Recipe Loading — Please refer to the following: Figure 2 and Figure 3 The intelligent control unit reads the current process recipe to be executed. Here, the process recipe refers to a pre-set set of parameters used to control the chemical mechanical polishing process, mainly including: target set pressure. Polishing disc spindle speed The system also monitors the polishing fluid flow rate and polishing time; simultaneously, it reads the current lifespan data of polishing pad 2 (i.e., the current actual thickness of polishing pad 2) from the database or storage area. This serves as the input basis for subsequent feedforward compensation calculations.

[0034] Step 2: Intelligent Feedforward Compensation Calculation — The intelligent control unit uses a process parameter coupling model trained based on historical data to calculate the estimated pressure loss caused by the wear and thinning of the polishing pad 2 and the deformation of the UPA airbag 7 due to the high-speed rotation of the polishing disk 3, and uses this as the feedforward compensation pressure value.

[0035] The model formula for defining the feedforward compensation pressure value is as follows: The symbols in the formula have the following meanings: : Feedforward compensation pressure value (i.e., the pressure that the system predicts needs to be added).

[0036] Target pressure (derived from the process formulation, i.e., ideal pressure).

[0037] Initial standard thickness of polishing pad.

[0038] : Current actual thickness of the polishing pad.

[0039] : Polishing disc spindle speed (from process formula).

[0040] : Proportional coefficient based on experimental calibration.

[0041] Before pressurization, the system pre-sets the pressure according to the process formulation target. The initial total set pressure is calculated by adding this compensation amount to the base pressure. : And will The initial control signal is output and sent to the pressure execution module.

[0042] Step 3: Multi-source data acquisition and fusion — During the polishing process, the intelligent control unit (PLC) acquires two signals in real time through the analog input module: 1) Actual pipeline air pressure from the air pressure sensor Signal; 2) Actual normal contact force from the end contact pressure sensor Signal.

[0043] The intelligent control unit (PLC) internally uses Kalman filtering or moving average filtering algorithms to remove mechanical vibration noise and collects the actual normal contact force. (Unit: N) Divide by the contact area S between the UPA airbag 7 and the mask substrate 1 to convert to the equivalent actual contact pressure. (Pressure), used for comparison with the set value.

[0044] Step 4: Fuzzy PID Adaptive Control — Please refer to the following: Figure 3 and Figure 4 The intelligent control unit calculates the pressure error. and its rate of change ; Fuzzy reasoning: and Input fuzzy logic controller; Parameter self-tuning: Based on a fuzzy rule table, the proportional gain of the PID controller is dynamically adjusted online. Integral coefficient and differential coefficients ; For example: when the error is large, increase... To speed up the response; when the error is small but close to the target value, increase Eliminate steady-state error and reduce Prevent overshoot.

[0045] Step 5: Closed-loop output execution — Please see Figure 4 The feedforward compensation amount is superimposed with the output of the fuzzy PID controller, which is the feedforward compensation control voltage calculated in step 2. The fuzzy PID feedback control voltage calculated in step 4 The voltages are superimposed to generate the final total control voltage. (0-10V DC analog signal): The intelligent control unit will (Corresponding to a 0-10V DC analog voltage signal) is sent to the electro-proportional valve 8, which then adjusts the voltage signal accordingly. The inflation pressure of the UPA airbag 7 is linearly adjusted to achieve the actual contact pressure between the UPA airbag 7 and the mask substrate 1. Closed-loop control.

[0046] This cycle runs continuously in the PLC at a high frequency (e.g., a 1ms cycle).

[0047] Example Reference Figure 1 As shown, the automatic compensation system of the present invention mainly consists of three parts: a pressure execution module, a distributed sensor feedback network, and an intelligent control unit.

[0048] 1. Pressure Execution Module: This module includes an air source, an electric proportional valve 8, air piping, and a UPA airbag 7. The electric proportional valve 8 adjusts the output air pressure according to the signal from the intelligent control unit, driving the polishing head 4 to apply pressure to the mask substrate 1. A digital pressure gauge is installed at the output of the electric proportional valve 8.

[0049] 2. Distributed sensor feedback network: including: Air pressure sensor: Installed in the air path between the electro-proportional valve 8 and the UPA airbag 7, used to detect the actual air pressure supplied by the system to the pipeline. .

[0050] End contact pressure sensor: A miniature spoke-type load cell (range 0-500N, accuracy 0.1%FS) is embedded in the universal joint or the end of the force transmission shaft of the polishing head 4 to directly sense the actual normal contact force of the UPA airbag 7 on the mask substrate 1. .

[0051] 3. Intelligent Control Unit (PLC): Beckhoff or equivalent high-performance PLC is selected as the main controller; sensor signals are acquired through a high-speed analog input module (such as an EtherCAT bus module), with the sampling frequency set to 1kHz. Then, the data fusion algorithm, process compensation model and fuzzy PID algorithm are run to output the total control voltage to the electrical proportional valve 8.

[0052] The operation process of the automatic compensation system is as follows (according to...) Figure 3 and Figure 4 (Explanation required) 1. Start: The PLC program starts and enters the initialization state of the CASE statement.

[0053] 2. Feedforward Calculation: Based on the current usage time of the polishing pad (converted to thickness), the feedforward compensation pressure value is calculated using a formula. For example: if the target is set with pressure The initial set pressure is 200 mbar. The model predicts that an additional 10 mbar is needed due to the thinning of the subbase. Therefore, the initial total set pressure output is... .

[0054] 3. Acquisition and Filtering: The TON timer triggers sampling, and the moving average filtering algorithm is used to process the end contact pressure sensor data of the most recent 10 cycles to filter out high-frequency noise caused by the spindle rotation.

[0055] 4. Fuzzy logic processing: Define error The fuzzy subset is {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}; If a sharp drop in pressure is detected (with a positive error), the fuzzy rule output will be larger. and smaller This causes the electric proportional valve to open rapidly; If the pressure has stabilized near the target value (error is zero), the fuzzy rule enhances the integral coefficient. Its function is to lock in pressure.

[0056] 5. Output: The calculation results drive the electric proportional valve 8 through the analog output module to physically adjust the airbag pressure.

[0057] 6. Loop: Return to step 3 until the process timer ends.

[0058] This embodiment incorporates a safety monitoring mechanism into the control logic to achieve anomaly protection. If the sensor feedback pressure exceeds 120% of the set value for 500ms continuously (hard limit protection), the system will forcibly reset the output of the electro-proportional valve 8 to zero and trigger an alarm to prevent crushing of the expensive mask substrate 1.

[0059] The above are merely preferred embodiments of the present invention. It must be noted that all equivalent modifications, variations and alterations made by those skilled in the art based on the contents of this application should be within the scope of protection of this invention.

Claims

1. An automatic compensation system for UPA pressure in polishing a mask substrate, wherein the mask substrate is disposed on a polishing pad attached to the upper surface of a rotating polishing disc, characterized in that, The automatic compensation system includes: The pressure actuation module, used to directly polish a mask substrate, includes a force-applying shaft, a polishing head, an air source, a UPA airbag, and an electro-proportional valve. The UPA airbag, connected to the air source, is disposed in the inner cavity of the polishing head and pressed onto the mask substrate. The rotating force-applying shaft is fixedly connected to the top center of the polishing head and drives the UPA airbag to apply a downward, uniform, flexible pressure to the mask substrate to achieve polishing of the mask substrate surface. The electro-proportional valve is connected in the air path between the UPA airbag and the air source and is used to adjust the input air pressure of the UPA airbag to regulate the polishing pressure on the mask substrate. A distributed sensing feedback network includes an air path pressure sensor and an end contact pressure sensor. The air path pressure sensor is installed in the air path between the output end of the electro-proportional valve and the air inlet of the UPA airbag, and is used to detect in real time the actual pipeline air pressure output by the electro-proportional valve and supplied to the UPA airbag. The end contact pressure sensor is installed at the end of the force application shaft and is used to directly detect the actual normal contact force of the UPA airbag pressing against the mask substrate. The intelligent control unit, connected to the electro-proportional valve, the pneumatic pressure sensor, and the end-contact pressure sensor, incorporates a data fusion algorithm, a process parameter coupling model trained on historical data, a feedforward compensation algorithm, and a fuzzy PID algorithm. This intelligent control unit receives signals from the pneumatic pressure sensor and the end-contact pressure sensor. Based on the current process formulation and the process parameter coupling model, it runs the data fusion algorithm, the feedforward compensation algorithm, and the fuzzy PID algorithm to obtain the feedforward compensation amount and the output of the fuzzy PID algorithm, which are then superimposed to generate the final total control voltage, which is output to the electro-proportional valve. This adjusts the input air pressure of the UPA airbag, thereby achieving closed-loop control of the polishing pressure of the mask substrate.

2. The automatic compensation system for UPA pressure in mask substrate polishing according to claim 1, characterized in that, The force-applying shaft is connected to the rotary motor via a coupling or universal joint.

3. The automatic compensation system for UPA pressure in mask substrate polishing according to claim 1, characterized in that, The current process formula refers to a set of pre-set process parameters for controlling the polishing process in the current processing, including target set pressure, polishing disc spindle speed, polishing fluid flow rate, and polishing time.

4. An automatic compensation method for UPA pressure in mask substrate polishing implemented by the automatic compensation system of claim 1, characterized in that, By using multi-sensor data fusion, fuzzy PID adaptive control, and intelligent feedforward compensation based on a process parameter coupling model, pressure control is upgraded from "passive response" to "active prediction," achieving real-time, precise, and constant control of polishing pressure.

5. The automatic compensation method for UPA pressure in mask substrate polishing according to claim 4, characterized in that, The automatic compensation method includes the following specific steps: Step 1) System initialization and loading of current process recipe — The intelligent control unit reads the current process formula to be executed, which includes the target set pressure, polishing disc spindle speed, polishing fluid flow rate and polishing time. At the same time, it reads the current actual thickness of the polishing pad as the basis for subsequent intelligent feedforward compensation calculation. Step 2) Intelligent feedforward compensation calculation — The intelligent control unit calculates the estimated pressure loss caused by the wear and thinning of the polishing pad and the deformation of the UPA airbag due to the rotation of the polishing disc, based on the process parameter coupling model, and uses this estimate as the feedforward compensation pressure value. In the formula, Feedforward compensation pressure value Initial standard thickness of polishing pad The current actual thickness of the polishing pad. Polishing disc spindle speed, : Based on experimentally calibrated proportionality coefficients; Before the system is pressurized, the feedforward compensation pressure value is added to the target set pressure of the current process formula to calculate the initial total set pressure: In the formula, Initial total pressure setting Feedforward compensation pressure value Setting targets creates pressure; The calculation results are then converted into a feedforward compensation control voltage and sent to the pressure execution module. Step 3) Multi-source data acquisition and fusion — After the system starts polishing, the intelligent control unit collects the actual pipeline air pressure signal from the air pressure sensor and the actual normal contact force signal from the end contact pressure sensor in real time. At the same time, it uses Kalman filtering or moving average filtering algorithm to remove mechanical vibration noise, and converts the obtained actual normal contact force into an equivalent actual contact pressure for comparison with the initial total set pressure obtained in step 2). Step 4) Fuzzy PID Adaptive Control — The intelligent control unit calculates the pressure error. and its rate of change , In the formula, Initial total pressure setting Actual contact pressure; Then calculate the pressure error. and pressure error change rate The fuzzy logic controller of the intelligent control unit performs fuzzy inference and simultaneously performs parameter self-tuning based on the fuzzy rule table, dynamically adjusting the proportional coefficient of the fuzzy logic controller online. Integral coefficient and differential coefficients ; Step 5) Closed-loop output execution — The intelligent control unit converts the calculation result obtained by the fuzzy logic controller in step 4) into a fuzzy PID feedback control voltage, and superimposes it with the feedforward compensation control voltage obtained in step 2). In the formula, Total control voltage Feedforward compensation control voltage Fuzzy PID feedback control voltage; Finally, the generated total control voltage is sent to the electro-proportional valve, which linearly adjusts the input air pressure of the UPA airbag according to the magnitude of the total control voltage, thereby realizing closed-loop control of the actual contact pressure of the UPA airbag against the mask substrate.

6. The automatic pressure compensation method for UPA in a mask substrate polishing machine according to claim 5, characterized in that, In step 3), the actual contact pressure is obtained by dividing the actual normal contact force by the calculated contact area between the UPA airbag and the mask substrate.

7. The automatic compensation method for UPA pressure in mask substrate polishing according to claim 5, characterized in that, In step 4), when the pressure error... When the value is large, increase the scaling factor. To accelerate response; when the pressure error is small and the actual contact pressure Approaching the initial total set pressure When, increase the integral coefficient To eliminate steady-state error and reduce the proportionality coefficient To prevent overshoot.

8. The automatic compensation method for UPA pressure in mask substrate polishing according to claim 5, characterized in that, In step 5), the total control voltage is a DC voltage analog quantity of 0-10V.

9. The automatic pressure compensation method for UPA in a mask substrate polishing machine according to claim 5, characterized in that, Steps 3) to 5) are continuously cycled at a high frequency with a period of 1ms until the polishing operation is completed.

Images