Method and system for automatically controlling and correcting concentration of liquid medicine

Abstract

The invention relates to a liquid medicine concentration automatic control correction method and system in the technical field of semiconductor manufacturing, and the method comprises the steps: if the concentration deviation exceeds a preset threshold value, comparing real-time monitoring data with a chemical equilibrium model, obtaining the offset of a reaction rate, and judging the specific parameters of the liquid supplementing amount and the adjustment time; a proportional-integral control algorithm is combined with a trigger condition and a liquid supplementing parameter, the liquid medicine concentration is dynamically adjusted, a compensation factor is introduced for temperature fluctuation, and an adjusted concentration control instruction is obtained; liquid level distribution data in the cleaning tank are synchronously collected through a liquid level detection device, and whether the uniformity of the liquid medicine in the tank reaches a preset standard or not is judged in combination with the adjusted concentration control instruction; and continuously acquiring a stable distribution result, comparing the stable distribution result with an initial data set, judging the optimization degree of the control precision and the response speed, and obtaining a concentration fluctuation range and a uniformity index after dynamic adjustment.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor manufacturing technology, specifically to the field of semiconductor cleaning processes, and particularly to a method and system for automatic control and correction of liquid medicine concentration. Background Art

[0002] The field of semiconductor manufacturing technology occupies a core position in the global science and technology and economic pattern. Especially with the rapid growth of semiconductor demand, the stability of its production process has become a key factor driving the development of the industry. In the important process of wet process cleaning in mass production, the wide application of chemical solutions makes the precise control of liquid medicine concentration directly affect the cleaning effect and product quality. However, the existing technologies still face significant challenges in maintaining the stability of liquid medicine concentration.

[0003] Currently, the commonly used time mode or batch concentration compensation method can adjust the liquid medicine concentration to a certain extent, but its main defect is that the concentration value fluctuates greatly. This fluctuation stems from the lack of a real-time dynamic adjustment mechanism, resulting in the difficulty of meeting the requirements of high-precision processes in terms of concentration control accuracy, especially in terms of reaction rate and chemical equilibrium stability. Therefore, how to effectively reduce the concentration deviation and improve the control response speed has become an urgent technical problem to be solved.

[0004] Specifically, the core challenges faced in this field are concentrated in the following aspects: First, the liquid medicine concentration is greatly affected by temperature fluctuations and cannot achieve precise stability in a dynamic environment; second, the existing liquid addition mechanism lacks the ability to quickly respond to real-time concentration data, resulting in adjustment lags; third, the collaborative detection of liquid level and concentration is insufficient, making it difficult to ensure the uniformity of liquid medicine in the cleaning tank. These technical factors have not been fully resolved, resulting in limited stability of concentration control in the process, thereby affecting the reliability and consistency of semiconductor cleaning.

[0005] Therefore, how to achieve rapid stability of liquid medicine concentration through real-time monitoring and dynamic adjustment while ensuring the accuracy of concentration detection, and ensure the uniformity of temperature and concentration in the cleaning tank has become a key issue in improving the stability of semiconductor wet processes. The solution to this problem not only requires overcoming the deviation of concentration fluctuations but also achieving efficient automatic control under complex process conditions to provide reliable guarantees for mass production processes. Summary of the Invention

[0006] The present invention provides a method for automatic control and correction of liquid medicine concentration, including the following steps:

[0007] Collect liquid medicine concentration and temperature fluctuation data at multiple positions in the cleaning tank through a sensor array, and record the real-time monitoring results using high-frequency sampling technology to obtain the initial data set of concentration distribution and temperature change;

[0008] Calculate the spatial distribution characteristics of the liquid medicine concentration and the time series change trend of the temperature fluctuation based on the initial data set, conduct a quantitative analysis on the correlation between the concentration deviation and the temperature fluctuation, and determine the triggering conditions for dynamic adjustment;

[0009] If the concentration deviation exceeds the preset threshold, compare the real-time monitoring data with the chemical equilibrium model to obtain the offset of the reaction rate, and judge the specific parameters of the liquid supplement volume and the adjustment timing;

[0010] Adopt the proportional-integral control algorithm, combine the triggering conditions and the liquid supplement parameters, dynamically adjust the liquid medicine concentration, introduce a compensation factor for the temperature fluctuation, and obtain the adjusted concentration control instruction;

[0011] Synchronously collect the liquid level distribution data in the cleaning tank through the liquid level detection device, and combine the adjusted concentration control instruction to judge whether the uniformity of the liquid medicine in the tank reaches the preset standard;

[0012] If the uniformity is lower than the preset standard, obtain the distribution ratio of the supplementary liquid medicine according to the liquid level detection result and the concentration distribution characteristics through the multi-point injection system, and determine the flow control scheme for each injection point;

[0013] Drive the automatic control module to perform the liquid medicine supply and stirring operations according to the flow control scheme, update the concentration deviation and the temperature fluctuation state for the real-time monitoring data, and obtain the stable distribution result of the liquid medicine in the cleaning tank;

[0014] Judge the optimization degree of the control accuracy and the response speed by comparing the continuously collected stable distribution result with the initial data set, and obtain the concentration fluctuation range and the uniformity index after dynamic adjustment;

[0015] Adopt the feedback loop mechanism to input the optimization degree data into the proportional-integral control algorithm, adjust the compensation factor and the triggering conditions according to the concentration fluctuation range and the uniformity index, and determine the adjustment parameters for the next cycle.

[0016] The present invention provides an automatic control and correction system for the liquid medicine concentration, mainly including:

[0017] A data acquisition module, which is used to collect the liquid medicine concentration and temperature fluctuation data at multiple positions in the cleaning tank through a sensor array, record the real-time monitoring results by using the high-frequency sampling technology, and obtain the initial data set of the concentration distribution and the temperature change;

[0018] A feature analysis module, which is used to calculate the spatial distribution characteristics of the liquid medicine concentration and the time series change trend of the temperature fluctuation based on the initial data set, conduct a quantitative analysis on the correlation between the concentration deviation and the temperature fluctuation, and determine the triggering conditions for dynamic adjustment;

[0019] A trigger judgment module, which is used to, if the concentration deviation exceeds a preset threshold, obtain the offset of the reaction rate by comparing the real-time monitoring data with the chemical equilibrium model, and judge the specific parameters of the liquid supplement volume and the adjustment timing;

[0020] A dynamic adjustment module, which is used to adopt a proportional-integral control algorithm to combine the trigger condition and the liquid supplement parameters to dynamically adjust the liquid medicine concentration, introduce a compensation factor for temperature fluctuations, and obtain an adjusted concentration control instruction;

[0021] A uniformity detection module, which is used to synchronously collect the liquid level distribution data in the cleaning tank through a liquid level detection device, and combine the adjusted concentration control instruction to judge whether the uniformity of the liquid medicine in the tank reaches the preset standard;

[0022] A flow distribution module, which is used to, if the uniformity is lower than the preset standard, obtain the distribution ratio of the supplementary liquid medicine according to the liquid level detection result and the concentration distribution characteristics through a multi-point injection system, and determine the flow control scheme for each injection point;

[0023] An execution control module, which is used to drive the automatic control module to execute the liquid medicine supply and stirring operations according to the flow control scheme, update the concentration deviation and the temperature fluctuation state for the real-time monitoring data, and obtain a stable distribution result of the liquid medicine in the cleaning tank;

[0024] A performance evaluation module, which is used to judge the optimization degree of the control accuracy and the response speed by continuously collecting the comparison between the stable distribution result and the initial data set, and obtain the concentration fluctuation range and the uniformity index after dynamic adjustment;

[0025] A parameter optimization module, which is used to adopt a feedback loop mechanism to input the optimization degree data into the proportional-integral control algorithm, adjust the compensation factor and the trigger condition for the concentration fluctuation range and the uniformity index, and determine the adjustment parameters for the next cycle.

[0026] The technical solution provided by the embodiment of the present invention may include the following beneficial effects:

[0027] The present invention discloses an automatic control and correction method for the concentration of liquid medicine. This method collects multi-point concentration and temperature data through a sensor array, and uses high-frequency sampling technology to record real-time monitoring results. Analyze the concentration distribution characteristics and temperature change trends according to the initial data set to determine the dynamic adjustment trigger condition. When the concentration deviation exceeds the threshold, judge the liquid supplement parameters in combination with the chemical equilibrium model. Adopt a proportional-integral control algorithm for dynamic concentration adjustment, and introduce a temperature fluctuation compensation factor. At the same time, evaluate the uniformity through a liquid level detection device, and optimize the distribution by using a multi-point injection system when necessary. The present invention continuously optimizes the control accuracy and the response speed through a feedback loop mechanism, effectively realizes the precise control and uniform distribution of the liquid medicine concentration in the cleaning tank, and improves the cleaning efficiency and quality. Specific embodiments

[0028] The technical solutions in the embodiments of the present invention will be described clearly and in detail below. The described embodiments are only a part of the embodiments of the present invention.

[0029] A specific method for automatically controlling and correcting the concentration of a liquid medicine in this embodiment may specifically include:

[0030] S1. Collect data on the concentration and temperature fluctuations at multiple positions in the cleaning tank through a sensor array, and record the real-time monitoring results using high-frequency sampling technology to obtain an initial data set of the concentration distribution and temperature changes.

[0031] Collect multi-point data in the cleaning tank through a sensor array to obtain the initial values of the liquid medicine concentration and temperature data, and obtain the original data set of real-time monitoring. Process the original data set using high-frequency sampling technology to obtain the fluctuation characteristics of the concentration distribution and the time series of temperature changes. If the fluctuation characteristics of the concentration distribution exceed the preset threshold, smooth the data points through an interpolation algorithm to obtain an adjusted concentration distribution data set. According to the adjusted concentration distribution data set, partition the multi-point data in the cleaning tank using a clustering algorithm to determine the distribution range of the concentration anomaly area. Process the time series of the temperature changes through time series analysis technology to obtain the time trend of the temperature fluctuations and determine whether there are abnormally high temperature points. If the time trend of the temperature fluctuations shows abnormally high temperature points, combine the distribution range of the concentration anomaly area and use a regression algorithm to predict the abnormal distribution law in the cleaning tank. Remap the multi-point data in the cleaning tank according to the abnormal distribution law to obtain an optimized concentration distribution and temperature change data set.

[0032] S2. Calculate the spatial distribution characteristics of the liquid medicine concentration and the time series change trend of the temperature fluctuations according to the initial data set, perform a quantitative analysis on the correlation between the concentration deviation and the temperature fluctuations, and determine the trigger condition for dynamic adjustment.

[0033] Obtain the liquid medicine concentration sensor data and the temperature sensor data, and construct a three-dimensional space coordinate system; use an interpolation algorithm to interpolate the discrete sampling points in the three-dimensional space coordinate system to obtain a continuous concentration distribution function; analyze the spectral characteristics of the temperature sensor data through Fourier transform to determine the main fluctuation period; calculate the concentration gradient vector field according to the concentration distribution function, and judge the area and direction where the concentration changes most violently; establish a multiple regression model according to the concentration distribution function and the main fluctuation period; process the multiple regression model using the sliding time window method to obtain the real-time concentration-temperature correlation coefficient; if the real-time concentration-temperature correlation coefficient exceeds the preset threshold, trigger the automatic adjustment mechanism for the liquid medicine ratio.

[0034] Specifically, in the initial dataset, first, the spatial distribution characteristics of the liquid medicine concentration are calculated through a spatial interpolation algorithm (such as Kriging interpolation). Assume that the sampling points of the liquid medicine concentration at a certain moment are (x1, y1, c1), (x2, y2, c2), …, (xn, yn, cn), where (xi, yi) are spatial coordinates and ci is the concentration value. Through the Kriging interpolation algorithm, a continuous concentration distribution map can be generated. For example, the concentration distribution range in a certain area is from 0.5 mg / L to 2.3 mg / L, and the concentration gradient is higher in the central area and lower in the edge area. Then, a trend analysis of the temperature fluctuation is carried out through a time series analysis method (such as Fourier transform). Assume that the temperature data are T1, T2, …, Tn, and the sampling interval is 1 minute. Through Fourier transform, the main frequency components of the temperature fluctuation can be extracted. For example, there are significant periodic fluctuations in the range of 0.01 Hz to 0.1 Hz. Then, the Pearson correlation coefficient is used to quantitatively analyze the correlation between the concentration deviation and the temperature fluctuation. Assume that the concentration deviation sequence is ΔC1, ΔC2, …, ΔCn, and the temperature fluctuation sequence is ΔT1, ΔT2, …, ΔTn. The calculated correlation coefficient is 0.78, indicating a strong positive correlation between the two. Finally, based on the results of the correlation analysis, the trigger conditions for dynamic adjustment are determined.

[0035] For example, when the temperature fluctuation amplitude exceeds ±0.5 °C and the concentration deviation exceeds ±0.2 mg / L, the dynamic adjustment mechanism is triggered, and the liquid medicine injection rate is adjusted through the PID control algorithm to keep the concentration and temperature within the set range.

[0036] S3. If the concentration deviation exceeds the preset threshold, then by comparing the real-time monitoring data with the chemical equilibrium model, the offset of the reaction rate is obtained, and the specific parameters of the liquid supplement volume and the adjustment timing are judged.

[0037] Obtain the real-time concentration data of each component in the reaction kettle, collect multi-point concentration information through the sensor network. If it is detected that the concentration deviation exceeds the preset threshold, trigger the data analysis module for in-depth analysis. Fit the real-time monitoring data with the chemical equilibrium model to calculate the theoretical equilibrium concentration. Quantify the deviation degree of the reaction rate by comparing the actual concentration with the theoretical concentration. According to the deviation degree and the reaction kinetics model, calculate the required liquid supplement volume and the optimal liquid supplement timing. Call the automatic control system to accurately execute the liquid supplement operation and adjust the concentration of each component to the target range. Continuously monitor the concentration change trend and dynamically optimize the control parameters to maintain the stable balance of the reaction system.

[0038] Specifically, in a real-time monitoring system, when the concentration deviation of a certain reactant in the reactor is detected to exceed the preset threshold (for example, the set threshold is ±5%, the current detected value is 12.3% while the target value is 10%), the system automatically triggers the chemical equilibrium model comparison process. First, a dynamic equilibrium model constructed based on the law of conservation of mass and the Arrhenius equation (such as the rate constant k = 0.15 s-1 and the activation energy Ea = 45 kJ / mol) calculates the theoretical reaction rate (the expected value is 0.18 mol / L·s), and performs a difference analysis with the real-time rate (0.22 mol / L·s) collected by the actual sensor, obtaining an offset Δr = 0.04 mol / L·s. Then, the amount of raw liquid to be supplemented is calculated as 6 L through a material balance algorithm (such as the replenishment volume Q = Δr × V × Δt, where the reaction volume V = 500 L and the time window Δt = 30 s). At the same time, in combination with the PID control module (the proportional coefficient Kp = 1.2 and the integral time Ti = 60 s), the opening of the replenishment valve is dynamically adjusted to 68%, and the parameter update is completed before the next sampling period (with a 5 s interval). During the process, the system continuously checks the concentration change rate (such as whether the current slope -0.2% / s approaches zero) to judge the adjustment effect. If the sampling data for 3 consecutive times all enter the threshold range (9.8% - 10.2%), it is determined to be in a steady state and the correction is stopped.

[0039] S4. Adopt a proportional-integral control algorithm combined with trigger conditions and replenishment parameters to dynamically adjust the liquid medicine concentration, introduce a compensation factor for temperature fluctuations, and obtain an adjusted concentration control instruction.

[0040] Obtain the liquid medicine concentration and temperature data, which are obtained by real-time monitoring of the sensor; determine the current system state information according to the liquid medicine concentration and temperature data; judge whether the current system state information triggers a preset concentration adjustment threshold; if the preset concentration adjustment threshold is triggered, calculate an initial concentration adjustment amount using a proportional-integral control algorithm; determine a basic control instruction according to the initial concentration adjustment amount; calculate a temperature compensation coefficient for the temperature data; adjust the basic control instruction according to the temperature compensation coefficient to obtain an optimized control instruction; use an actuator to execute the optimized control instruction to achieve liquid medicine concentration adjustment; monitor the effect of the liquid medicine concentration adjustment through a feedback loop; update the control parameters according to the effect of the liquid medicine concentration adjustment; use the updated control parameters for the next round of liquid medicine concentration adjustment.

[0041] Specifically, in the liquid medicine concentration control system, the proportional-integral control algorithm calculates the difference between the concentration deviation e(t) and the set value in real time by setting the proportional gain Kp = 0.8 and the integral time Ti = 120 seconds. When it is detected that the current concentration deviates from the target value by 5%, a liquid replenishment instruction is triggered, and the liquid replenishment flow rate Q is dynamically adjusted according to the formula Q = Kp×e(t)+(1 / Ti)∫e(t)dt. For example, when the instantaneous deviation is 3%, the system outputs a liquid replenishment flow rate of 12 mL / min. For the concentration change caused by temperature fluctuations, a temperature compensation factor α = 0.05% / °C is introduced. When the temperature sensor detects that the ambient temperature rises by 2°C, the system automatically increases the target concentration by 0.1% to offset the evaporation effect. The compensated concentration instruction C_adj = C_set + α×ΔT, where C_set is the original set value. All parameters are updated in real time through the PID controller, the sampling period is set to 1 second, the control signal is output to the actuator through the D / A conversion module, and at the same time, a sliding window algorithm is used to perform weighted average filtering on the most recent 10 sampling data, and the weight coefficients are [0.1, 0.15, 0.2, 0.25, 0.3] to suppress measurement noise. When the concentration fluctuation exceeds ±1.5% within three consecutive cycles, the system automatically triggers the parameter self-tuning program and re-tunes Kp and Ti using the Ziegler-Nichols method to ensure that the control accuracy is always maintained within the range of ±0.8%.

[0042] S5. Synchronously collect the liquid level distribution data in the cleaning tank through the liquid level detection device, and combine the adjusted concentration control instruction to judge whether the uniformity of the liquid medicine in the tank reaches the preset standard.

[0043] Obtain the liquid level distribution data in the cleaning tank, and the liquid level distribution data is collected by arranging multi-point liquid level sensors. Use the data acquisition module to collect the liquid level sensor data in real time to obtain the liquid level distribution map of the cleaning tank. Calculate the current liquid medicine concentration distribution according to the preset concentration control algorithm, and judge whether the liquid medicine concentration distribution is uniform. If the liquid medicine concentration distribution is not uniform, start the stirring device to mix the liquid medicine. Measure the liquid medicine concentration at multiple sampling points using a spectral analyzer to obtain the actual concentration distribution data. Compare the calculation result of the preset concentration control algorithm with the actual concentration distribution data through the fuzzy control algorithm to determine the correction value of the concentration control instruction. Adjust the dosing device according to the correction value of the concentration control instruction to achieve precise control of the liquid medicine concentration.

[0044] Specifically, the liquid level detection device uses a high-precision ultrasonic sensor to collect the liquid level data of 8 monitoring points in the cleaning tank in real time at a sampling frequency of 10 times per second. The data of each monitoring point is processed by the Kalman filter algorithm to eliminate the fluctuating noise, and then a liquid level distribution matrix L = [L1, L2,..., L8] is generated. Among them, the liquid level value at point L5 has a periodic fluctuation of ±2 mm due to its proximity to the liquid inlet. The concentration control system dynamically adjusts the chemical dosing amount according to the real-time pH value detection result. When the pH sensor detects a deviation exceeding 0.5, the PID controller outputs a concentration adjustment command with parameters of Kp = 1.2 / Ki = 0.05 / Kd = 0.3. The system performs a multivariable coupling analysis on the liquid level matrix L and the concentration command C, and evaluates the uniformity by calculating the coefficient of variation CV = σ / μ×100% (where σ is the standard deviation of the liquid levels at 8 monitoring points, and μ is the average liquid level). When the CV value is continuously lower than 5% for 30 seconds and the maximum liquid level difference ΔLmax ≤ 15 mm, a compliance signal is triggered. For abnormal situations such as the CV suddenly rising above 8%, the system automatically starts a compensation algorithm based on the hydrodynamic model. This model uses the Reynolds number Re = ρvd / μ (where ρ is the liquid medicine density of 1020 kg / m 3 , v is the flow velocity of 0.3 m / s, d is the characteristic size of the tank of 1.2 m, and μ is the dynamic viscosity of 0.89 Pa·s) to judge the flow regime. When Re > 4000, the stirring motor speed is increased to 120% of the preset value for turbulent enhancement. All data is uploaded to the MES system through the OPCUA protocol and matched with 500 groups of compliance parameters in the historical process database. When the matching similarity reaches more than 90%, the process knowledge graph is updated.

[0045] S6. If the uniformity is lower than the preset standard, the multi-point injection system is used to obtain the distribution ratio of the supplementary liquid medicine according to the liquid level detection result and the concentration distribution characteristics, and determine the flow control scheme for each injection point.

[0046] Obtain the detection result of the liquid level detection system, and the detection result includes the concentration distribution data. If the uniformity of the concentration distribution data is lower than the preset standard, determine the distribution characteristics of the concentration distribution according to the detection result. According to the distribution characteristics and the detection result, use a preset algorithm to calculate the distribution ratio of the supplementary liquid medicine to obtain the demand of each region. Through the multi-point injection system, determine the flow control parameters of the injection points according to the distribution ratio to obtain the initial control scheme. If the concentration distribution under the initial control scheme does not reach the preset standard, adjust the flow control parameters through an iterative optimization algorithm to determine the final control scheme. Execute the flow control of the final control scheme through the multi-point injection system to obtain the adjusted uniformity data. Judge whether the preset standard is reached according to the adjusted uniformity data, and obtain the optimized distribution characteristics. Store the final control scheme and the optimized distribution characteristics through a data recording tool to determine the basis for subsequent operations.

[0047] Specifically, when the liquid level sensor detects that the height of the liquid medicine in the reaction tank is lower than the set threshold of 1200 mm, the system automatically triggers the uniformity analysis module. The reaction tank is divided into 25 detection units of 5×5 using the grid sampling method. The concentration values of each unit are obtained through a spectrometer and the coefficient of variation is calculated. If the current coefficient of variation exceeds the preset standard of 0.15, the control core starts a compensation program based on the fuzzy PID algorithm. First, a concentration distribution matrix [0.82, 0.85,..., 0.91] (25-dimensional data) is established, and 3 low-concentration regions are identified using K-means clustering (the cluster center values are 0.83, 0.84, and 0.85 respectively). According to the Euclidean distance weight distribution formula Q = K×(1 - d / D)×ΔC (where K is the system constant 0.8, d is the distance from the unit to the injection point, D is the maximum action radius of 2 m, and ΔC is the target concentration difference of 0.12), it is calculated that the injection flow rate in the northeast area needs to be increased to 12 L / min, the southwest area is adjusted to 8 L / min, and the central area maintains the reference flow rate of 5 L / min. The flow controller drives the solenoid valve to execute through PID closed-loop regulation (proportional coefficient 0.6, integral time 2 s, derivative time 0.5 s), and at the same time, the concentration change rate is monitored in real time. When the change rate is lower than 0.01 g / L·s, it is determined that the compensation is completed. During the whole process, the data acquisition period is 10 s, and the control instruction delay is controlled within 200 ms to ensure that the system response time does not exceed the process requirement threshold of 30 s.

[0048] S7. Drive the automatic control module to perform liquid medicine replenishment and stirring operations according to the flow control scheme, update the concentration deviation and temperature fluctuation status for the real-time monitoring data, and obtain the stable distribution result of the liquid medicine in the cleaning tank.

[0049] Obtain real-time monitoring data through the flow control scheme, use the preset threshold to judge the concentration deviation and temperature fluctuation status, and obtain the adjustment requirements. Extract the liquid medicine replenishment amount from the adjustment requirements, use the automatic control module to perform the replenishment operation, and obtain the initial state after the liquid medicine is replenished. Perform a stirring operation on the initial state, update the data in the cleaning tank through real-time monitoring, and obtain the distribution state after mixing. If the distribution state after mixing exceeds the stable distribution range, adjust the flow control parameters through the control module to obtain an optimized replenishment scheme. Re-perform the liquid medicine replenishment and stirring operations according to the optimized replenishment scheme, obtain the updated monitoring data, and determine the change trend of the concentration deviation. Analyze the change trend of the concentration deviation through the support vector machine algorithm, judge whether the temperature fluctuation converges, and obtain the stable distribution state of the liquid medicine in the cleaning tank. Use the random forest algorithm to verify the stable distribution state, obtain the deviation distribution characteristics, and judge the final distribution consistency.

[0050] Specifically, the flow control scheme adjusts the opening of the liquid supply valve in real time through the PID algorithm. The set flow value is set to 5 L / min. When the sensor detects that the actual flow deviates by ±0.3 L / min, the controller outputs adjustment parameters of P = 0.8, I = 0.2, and D = 0.1, driving the stepper motor to adjust the valve opening with 0.5° as the minimum adjustment unit. The stirring control module adopts the fuzzy logic algorithm, taking the temperature fluctuation value ΔT as the input variable. When ΔT exceeds ±2°C, the system starts a three-stage speed regulation strategy: when ΔT is in the range of 2 - 3°C, it starts low-speed stirring at 800 rpm; when it is in the range of 3 - 5°C, it switches to medium-speed at 1200 rpm; when it exceeds 5°C, it enables the high-speed mode at 1500 rpm. The concentration monitoring system collects conductivity data every 10 seconds, fits the concentration curve through the least squares method. When the concentration deviation exceeds ±0.15% within 3 consecutive sampling periods, it triggers the compensation algorithm Q = K × ΔC × V (where K = 1.2 is the compensation coefficient and V is the tank volume of 50 L), and automatically injects the compensation liquid. The temperature control system uses the Kalman filter algorithm to process the raw data of the PT100 sensor, establishing the state equation X(k) = 0.95X(k - 1) + 0.1U(k) + W(k), and the observation equation Z(k) = X(k) + V(k), where the covariance of the process noise W(k) and the observation noise V(k) are set to 0.01 and 0.05 respectively, to achieve

[0051] a steady-state control accuracy of ±0.3°C. All real-time data is uploaded to the central controller through the Modbus protocol. The system performs multi-parameter coupling analysis every 30 seconds. When there are simultaneously flow fluctuations > 8%, temperature changes > 1.5°C, and concentration deviations > 0.2%, it automatically starts the emergency balancing program, first adjusting the temperature to within ±0.5°C of the set value, and then gradually correcting other parameters.

[0052] S8. By continuously collecting the comparison between the stable distribution result and the initial data set, judge the optimization degree of the control accuracy and response speed, and obtain the concentration fluctuation range and uniformity index after dynamic adjustment.

[0053] Obtain the initial data set and the stable distribution data generated by continuous acquisition; use the comparative analysis method to process the initial data set and the stable distribution data to obtain the optimization degree of control accuracy and response speed; process the stable distribution data through the time series analysis algorithm to obtain the concentration fluctuation trend after dynamic adjustment; calculate the fluctuation range according to the concentration fluctuation trend to determine the preliminary value of the uniformity index; if the uniformity index exceeds the preset threshold, adjust the control accuracy through the dynamic adjustment algorithm to obtain the optimized concentration fluctuation data; for the optimized concentration fluctuation data, use the mean square error calculation method to judge the stability of the fluctuation range to obtain the stable fluctuation range and the final value of the uniformity index; judge whether the optimization degree meets the requirements by comparing and analyzing the stable fluctuation range and the final value of the uniformity index; extract the response speed of dynamic adjustment from the results of the comparative analysis to determine the adjustment efficiency of the overall system.

[0054] Specifically, during the dynamic adjustment process, the concentration data of the target substance in the reactor is continuously collected by a high-frequency sensor at a sampling frequency of 100 times per second. The initial data set shows that the concentration fluctuation range is

[0055] ±0.15 mol / L, and the uniformity index (standard deviation) is 0.08. The PID control algorithm is used to adjust the actuator, where the proportional coefficient Kp is set to 1.2, the integral time Ti is 5 seconds, and the derivative time Td is 0.3 seconds. By calculating the error value (the deviation between the set value and the actual value) in real time and outputting a control signal, the opening of the actuator (such as a control valve) is dynamically adjusted within the range of 0-100%. After 30 minutes of closed-loop control, the collected stable distribution results show that the concentration fluctuation range is reduced to ±0.05 mol / L, and the uniformity index is increased to 0.03. To quantify the optimization effect, the data in the initial and stable stages are compared using analysis of variance (ANOVA), and the F value is 25.6 (P < 0.01), indicating a significant improvement in control accuracy. At the same time, by calculating the system response time (the time required for the output value to reach 90% of the steady state from the change of the set value), it is shortened from the initial 12 seconds to 3 seconds, proving that the response speed has been improved. Further, the fast Fourier transform (FFT) is used to analyze the spectral characteristics of the concentration signal, and it is found that the amplitude of the high-frequency noise component is reduced from 0.02 mol / L to 0.005 mol / L, indicating that the dynamic adjustment effectively suppresses external interference. Finally, by fitting the concentration distribution curve, the kurtosis coefficient is reduced from 2.1 to 1.8, verifying the improvement of uniformity.

[0056] S9. Input the optimization degree data into the proportional-integral control algorithm using the feedback loop mechanism, adjust the compensation factor and trigger conditions for the concentration fluctuation range and the uniformity index, and determine the adjustment parameters for the next cycle.

[0057] Obtain real-time concentration data using a data acquisition system; calculate the current concentration fluctuation range and uniformity index based on the real-time concentration data; process the concentration fluctuation range and the uniformity index through a proportional-integral control algorithm to obtain a concentration adjustment parameter; if the concentration fluctuation range exceeds a preset threshold, adjust the compensation factor; if the uniformity index is lower than a preset standard, modify the trigger condition parameter; use the adjusted compensation factor and the modified trigger condition parameter for the next cycle of concentration adjustment; generate a concentration control instruction according to the concentration adjustment parameter; output the concentration control instruction to an execution system to obtain an optimized concentration adjustment result.

[0058] Specifically, in the concentration control system, by collecting the solute concentration data in the reaction kettle in real time (for example, the current concentration value is 2.35 mol / L and the sampling period is 500 ms), the standard deviation (0.08 mol / L) of the last 20 samplings is calculated using a sliding window as the fluctuation range index, and at the same time, the coefficient of variation (0.12) of the last 100 data is used as the basis for uniformity evaluation. When the standard deviation of the fluctuation exceeds the threshold of 0.1 mol / L, the proportional-integral (PI) algorithm is triggered for adjustment, and its transfer function is Gc(s) = 2.5 + 0.6 / s, where the proportional coefficient Kp is initially set to 2.5 and the integral time Ti is 1.67 seconds. The algorithm outputs the increment of the compensation flow valve opening Δu(t) = Kp[e(t) + 1 / Ti∫e(t)dt] according to the deviation e(t) = set value (2.5 mol / L) - measured value. For example, when the cumulative deviation integral term reaches 0.15, Δu(t) = 2.5×0.07 + 0.6×0.15 = 0.265 is calculated, corresponding to an increase in the feed pump speed of 3.2%. After each adjustment, the regression equation α (range 0.8 - 1.2) of the compensation factor α is established by fitting the last 50 sets of adjustment parameter and effect data using the least squares method, α = 0.05×standard deviation + 1.1×coefficient of variation. When the coefficient of determination R 2 > 0.9, the PI parameter is automatically updated to Kp_new = α×Kp_old. The trigger condition adopts a dual judgment mechanism. In addition to abnormal standard deviation of the fluctuation, when the decrease amplitude of the coefficient of variation < 5% after 5 consecutive adjustments, the integral time Ti is shortened by 20% to enhance the dynamic response.

[0059] The present invention provides an automatic control and correction system for the liquid medicine concentration, mainly including:

[0060] A data acquisition module, which is used to collect the liquid medicine concentration and temperature fluctuation data at multiple positions in the cleaning tank through a sensor array, record the real-time monitoring results using high-frequency sampling technology, and obtain the initial data set of the concentration distribution and temperature change.

[0061] A feature analysis module, which is used to calculate the spatial distribution characteristics of the liquid medicine concentration and the time-series change trend of the temperature fluctuation according to the initial data set, conduct a quantitative analysis on the correlation between the concentration deviation and the temperature fluctuation, and determine the trigger conditions for dynamic adjustment;

[0062] A trigger judgment module, which is used to, if the concentration deviation exceeds a preset threshold, obtain the offset of the reaction rate by comparing the real-time monitoring data with the chemical equilibrium model, and judge the specific parameters of the liquid supplement volume and the adjustment timing;

[0063] A dynamic adjustment module, which is used to adopt a proportional-integral control algorithm to combine the trigger conditions and the liquid supplement parameters to dynamically adjust the liquid medicine concentration, introduce a compensation factor for the temperature fluctuation, and obtain an adjusted concentration control instruction;

[0064] A uniformity detection module, which is used to synchronously collect the liquid level distribution data in the cleaning tank through a liquid level detection device, and combine the adjusted concentration control instruction to judge whether the uniformity of the liquid medicine in the tank meets the preset standard;

[0065] A flow distribution module, which is used to, if the uniformity is lower than the preset standard, obtain the distribution ratio of the supplementary liquid medicine according to the liquid level detection result and the concentration distribution characteristics through a multi-point injection system, and determine the flow control scheme for each injection point;

[0066] An execution control module, which is used to drive an automatic control module to execute the liquid medicine supply and stirring operations according to the flow control scheme, update the concentration deviation and the temperature fluctuation state according to the real-time monitoring data, and obtain a stable distribution result of the liquid medicine in the cleaning tank;

[0067] A performance evaluation module, which is used to judge the optimization degree of the control accuracy and the response speed by continuously collecting the comparison between the stable distribution result and the initial data set, and obtain the concentration fluctuation range and the uniformity index after dynamic adjustment;

[0068] A parameter optimization module, which is used to adopt a feedback loop mechanism to input the optimization degree data into the proportional-integral control algorithm, adjust the compensation factor and the trigger conditions according to the concentration fluctuation range and the uniformity index, and determine the adjustment parameters for the next cycle.

[0069] Inspired by the above embodiments of the present invention, through the above description, relevant staff can completely make various changes and modifications without departing from the technical idea of this invention. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An automatic control and calibration method for liquid medicine concentration, characterized in that, The method includes the following steps: S1. Collect the data of the liquid medicine concentration and temperature fluctuation at multiple positions in the cleaning tank through a sensor array, and record the real-time monitoring results by using high-frequency sampling technology to obtain the initial data set of the concentration distribution and temperature change; S2. Calculate the spatial distribution characteristics of the liquid medicine concentration and the time series change trend of the temperature fluctuation according to the initial data set, conduct a quantitative analysis on the correlation between the concentration deviation and the temperature fluctuation, and determine the trigger condition for dynamic adjustment; S3. If the concentration deviation exceeds the preset threshold, compare the real-time monitoring data with the chemical equilibrium model to obtain the offset of the reaction rate, and judge the specific parameters of the liquid supplement amount and the adjustment time; S4. Adopt the proportional-integral control algorithm to combine the trigger condition and the liquid supplement parameters to dynamically adjust the liquid medicine concentration, introduce a compensation factor for the temperature fluctuation, and obtain the adjusted concentration control instruction; S5. Synchronously collect the liquid level distribution data in the cleaning tank through the liquid level detection device, and combine the adjusted concentration control instruction to judge whether the uniformity of the liquid medicine in the tank reaches the preset standard; S6. If the uniformity is lower than the preset standard, obtain the distribution ratio of the supplementary liquid medicine according to the liquid level detection result and the concentration distribution characteristics through the multi-point injection system, and determine the flow control scheme for each injection point; S7. Drive the automatic control module to perform the liquid medicine supply and stirring operations according to the flow control scheme, update the concentration deviation and the temperature fluctuation state according to the real-time monitoring data, and obtain the stable distribution result of the liquid medicine in the cleaning tank; S8. Judge the optimization degree of the control accuracy and the response speed by comparing the continuously collected stable distribution result with the initial data set, and obtain the concentration fluctuation range and the uniformity index after dynamic adjustment; S9. Adopt the feedback loop mechanism to input the optimization degree data into the proportional-integral control algorithm, adjust the compensation factor and the trigger condition according to the concentration fluctuation range and the uniformity index, and determine the adjustment parameters for the next cycle.

2. The automatic control and calibration method for the concentration of liquid medicine according to claim 1, wherein, The S1 includes: Collect multi-point data in the cleaning tank through a sensor array, obtain the initial values of the liquid medicine concentration and temperature data, and obtain the original data set of real-time monitoring; Process the original data set by using high-frequency sampling technology to obtain the fluctuation characteristics of the concentration distribution and the time series of the temperature change; If the fluctuation characteristics of the concentration distribution exceed the preset threshold, smooth the data points by using the interpolation algorithm to obtain the adjusted concentration distribution data set; According to the adjusted concentration distribution data set, partition the multi-point data in the cleaning tank by using the clustering algorithm to determine the distribution range of the concentration abnormal area; Process the time series of the temperature change by using the time series analysis technology to obtain the time trend of the temperature fluctuation and judge whether there is an abnormally high temperature point; If the time trend of the temperature fluctuation shows an abnormally high temperature point, combine the distribution range of the concentration abnormal area, and predict the abnormal distribution law in the cleaning tank by using the regression algorithm; Remap the multi-point data in the cleaning tank according to the abnormal distribution law to obtain the optimized concentration distribution and temperature change data set.

3. The automatic control and calibration method for liquid medicine concentration according to claim 1, wherein The S2 includes: Obtain the liquid medicine concentration sensor data and the temperature sensor data, and construct a three-dimensional space coordinate system; Interpolate the discrete sampling points in the three-dimensional space coordinate system using an interpolation algorithm to obtain a continuous concentration distribution function; Analyze the spectral characteristics of the temperature sensor data through Fourier transform to determine the main fluctuation period; Calculate the concentration gradient vector field according to the concentration distribution function, and judge the region and direction where the concentration changes most violently; Establish a multiple regression model according to the concentration distribution function and the main fluctuation period; Process the multiple regression model using a sliding time window method to obtain the real-time concentration-temperature correlation coefficient; If the real-time concentration-temperature correlation coefficient exceeds a preset threshold, trigger the automatic adjustment mechanism of the liquid medicine ratio; 4. The automatic control and calibration method for the concentration of a liquid medicine according to any one of claims 1 to 3, characterized in that, The S3 includes: Obtain the real-time concentration data of each component in the reaction kettle, collect multi-point concentration information through the sensor network. If the detected concentration deviation exceeds the preset threshold, trigger the data analysis module for in-depth analysis. Fit the real-time monitoring data using a chemical equilibrium model, calculate the theoretical equilibrium concentration. By comparing the actual concentration with the theoretical concentration, quantify the deviation degree of the reaction rate. According to the deviation degree and the reaction kinetics model, calculate the required liquid supplement amount and the optimal liquid supplement timing. Call the automatic control system to accurately execute the liquid supplement operation, adjust the concentration of each component to the target range. Continuously monitor the concentration change trend, dynamically optimize the control parameters, and maintain the stable balance of the reaction system.

5. The automatic control and calibration method for the concentration of a liquid medicine according to any one of claims 1 to 3, characterized in that, The S4 includes: Obtain the liquid medicine concentration and temperature data, which are real-time monitored by sensors; Determine the current system state information according to the liquid medicine concentration and temperature data; Judge whether the current system state information triggers the preset concentration adjustment threshold; If the preset concentration adjustment threshold is triggered, calculate the initial concentration adjustment amount using a proportional-integral control algorithm; Determine the basic control instruction according to the initial concentration adjustment amount; Calculate the temperature compensation coefficient for the temperature data; Adjust the basic control instruction according to the temperature compensation coefficient to obtain an optimized control instruction; Use an actuator to execute the optimized control instruction to achieve liquid medicine concentration adjustment; Monitor the effect of the liquid medicine concentration adjustment through a feedback loop; Update the control parameters according to the effect of the liquid medicine concentration adjustment; Use the updated control parameters for the next round of liquid medicine concentration adjustment.

6. The automatic control and calibration method for the concentration of a liquid medicine according to any one of claims 1 to 3, characterized in that, The S5 includes: Obtain the liquid level distribution data in the cleaning tank, which is collected by arranging multi-point liquid level sensors; Use a data acquisition module to collect the liquid level sensor data in real time to obtain the liquid level distribution map of the cleaning tank; Calculate the current liquid medicine concentration distribution according to the preset concentration control algorithm, and judge whether the liquid medicine concentration distribution is uniform; If the liquid medicine concentration distribution is not uniform, start the stirring device to mix the liquid medicine; Measure the liquid medicine concentration at multiple sampling points using a spectral analyzer to obtain the actual concentration distribution data; Compare the calculation result of the preset concentration control algorithm with the actual concentration distribution data through a fuzzy control algorithm to determine the correction value of the concentration control instruction; Adjust the dosing device according to the correction value of the concentration control instruction to achieve precise control of the liquid medicine concentration.

7. The automatic control and calibration method for the concentration of liquid medicine according to any one of claims 1 to 3, characterized in that, The S6 includes: Obtain the detection result of the liquid level detection system, and the detection result includes concentration distribution data; If the uniformity of the concentration distribution data is lower than the preset standard, determine the distribution characteristics of the concentration distribution according to the detection result; According to the distribution characteristics and the detection result, calculate the distribution ratio of the supplementary liquid medicine by using a preset algorithm to obtain the demand of each region; Through a multi-point injection system, determine the flow control parameters of the injection points according to the distribution ratio to obtain an initial control scheme; If the concentration distribution under the initial control scheme does not reach the preset standard, adjust the flow control parameters through an iterative optimization algorithm to determine the final control scheme; Execute the flow control of the final control scheme through the multi-point injection system to obtain the adjusted uniformity data; Judge whether the preset standard is reached according to the adjusted uniformity data, and obtain the optimized distribution characteristics; ​ 8. The automatic control and calibration method for the concentration of a liquid medicine according to any one of claims 1 to 3, characterized in that, ​ ​ ​ ​ ​ ​ ​ ​ 9. The automatic control and calibration method for the concentration of a liquid medicine according to any one of claims 1 to 3, characterized in that, ​ ​ ​ ​ ​ ​ ​ ​ ​ 10. An automatic control and calibration system for liquid medicine concentration, characterized in that, ​ The data acquisition module is used to collect data on the concentration and temperature fluctuations of the liquid medicine at multiple positions in the cleaning tank through a sensor array, and record the real-time monitoring results using high-frequency sampling technology to obtain the initial data set of the concentration distribution and temperature change; The feature analysis module is used to calculate the spatial distribution characteristics of the liquid medicine concentration and the time series change trend of the temperature fluctuation according to the initial data set, conduct a quantitative analysis on the correlation between the concentration deviation and the temperature fluctuation, and determine the trigger condition for dynamic adjustment; The trigger judgment module is used to, if the concentration deviation exceeds the preset threshold, obtain the offset of the reaction rate by comparing the real-time monitoring data with the chemical equilibrium model, and judge the specific parameters of the liquid supplement amount and the adjustment timing; The dynamic adjustment module is used to adopt a proportional-integral control algorithm to combine the trigger condition and the liquid supplement parameters to dynamically adjust the liquid medicine concentration, introduce a compensation factor for the temperature fluctuation, and obtain the adjusted concentration control instruction; The uniformity detection module is used to synchronously collect the liquid level distribution data in the cleaning tank through a liquid level detection device, and combine the adjusted concentration control instruction to judge whether the uniformity of the liquid medicine in the tank reaches the preset standard; The flow distribution module is used to, if the uniformity is lower than the preset standard, obtain the distribution ratio of the supplementary liquid medicine according to the liquid level detection result and the concentration distribution characteristics through a multi-point injection system, and determine the flow control scheme for each injection point; The execution control module is used to drive the automatic control module to perform the liquid medicine supply and stirring operations according to the flow control scheme, update the concentration deviation and the temperature fluctuation state according to the real-time monitoring data, and obtain the stable distribution result of the liquid medicine in the cleaning tank; The performance evaluation module is used to judge the optimization degree of the control accuracy and response speed by comparing the continuously collected stable distribution result with the initial data set, and obtain the concentration fluctuation range and uniformity index after dynamic adjustment; The parameter optimization module is used to adopt a feedback loop mechanism to input the optimization degree data into the proportional-integral control algorithm, adjust the compensation factor and the trigger condition according to the concentration fluctuation range and the uniformity index, and determine the adjustment parameters for the next cycle.