A method, system and medium for controlling growth of silica sol seed for wafer grinding

Abstract

The present invention relates to a method, system and medium for controlling the growth of silica sol crystal seeds for wafer grinding, the method comprising: U1. In a reactor, a silicate metering pump is turned on to add a certain amount of silicate at a set speed, and a stabilizer metering pump is turned on to add an alkaline stabilizer to adjust the pH value, a stirrer is stirred at a certain speed, and heated to a certain temperature for reaction, and an initial silica sol crystal seed product is obtained after being kept warm for a certain time; U2. A laser particle size analyzer is used to detect the particle size of the initial silica sol crystal seed product to obtain particle size data information of the initial silica sol crystal seed product, a silicate metering pump is used to obtain data information of the silicate addition speed in real time, a stabilizer metering pump is used to obtain data information of the stabilizer addition speed in real time, and a pH value sensor is used to obtain pH value data information of the reactant in real time. The present invention can not only accurately control the growth of silica sol crystal seeds, but also improve the production efficiency of silica sol.

Description

Technical Field

[0001] The invention relates to the technical field of silica sol preparation, and in particular to a silica sol seed growth control method, system and medium for wafer grinding. Background Art

[0002] With the rapid development of industry and economic society, the high-purity silica sol industry has also ushered in an opportunity for great changes. The products of the high-purity silica sol industry are widely used in electronics, aerospace, heavy machinery, construction, aviation and other industries, and its market demand is in sustained growth. With the development of science and technology and the progress of society, the high-purity silica sol industry is growing day by day.

[0003] In the wafer grinding process, silica sol is an important grinding medium, and its performance directly affects the grinding effect and product quality. The growth of seeds in silica sol is a complex process, which is affected by many factors, such as temperature, pressure, stirring speed, silica acid concentration, etc. Therefore, precise control of the growth of silica sol seeds becomes the key to improving the quality of silica sol and production efficiency. However, in the existing technology, the control method of silica sol seed growth usually relies on experience or simple process parameter control, which cannot achieve precise control, resulting in low production efficiency and unstable product quality.

[0004] Therefore, how to improve the production efficiency of silica sol and how to continuously improve the production process of silica sol have become issues that we urgently need to solve.

[0005] In the prior art, a patent (application number: 200910054211.2) discloses a method for preparing silica sol seeds, wherein an alkaline silicate aqueous solution is mixed with an acidic silicate aqueous solution to obtain a weakly alkaline mixed silicate aqueous solution, and the obtained silicate aqueous solution is subjected to a hydrothermal reaction to obtain silica sol seeds. Although the preparation of silica sol seeds is provided, the production process is not improved, resulting in low production efficiency and poor control of the seed growth method.

[0006] In the prior art, the patent (application number: 201610382474.6) discloses a polydisperse large-particle silica sol and a preparation method thereof, wherein a monodisperse spherical silica sol with a particle size of 20nm-30nm is used as a seed, stirred and heated, and a monodisperse spherical silica sol seed with a particle size of 20nm-30nm and active silicic acid are continuously added to the reaction system, and a heating concentration method is used to maintain a constant liquid level during the overall reaction process. During this period, a dilute inorganic alkali solution is added to keep the pH value of the system at 9.5-10.5, and the system is cooled after being kept warm. However, the preparation process is not very precise, resulting in the inability to accurately control the production of silica sol seeds. Summary of the invention

[0007] In view of the above deficiencies in the prior art, the present invention provides a method, system and medium for controlling the growth of silica sol seeds for wafer grinding, which can not only accurately control the growth of silica sol seeds, but also improve the production efficiency of silica sol.

[0008] In order to achieve the above-mentioned purpose and other related purposes, the technical solution provided by the present invention is as follows: a method for controlling the growth of silica sol seed crystals for wafer grinding, the method comprising:

[0009] U1. Open the silicic acid metering pump on the reactor, add silicic acid to the reactor, open the stabilizer metering pump on the reactor, add alkaline stabilizer to adjust the pH value, stir the reactor at a certain speed, heat to a certain temperature for reaction, and keep warm for a certain period of time to obtain the initial silica sol seed product;

[0010] U2. Use an online laser particle size analyzer to detect the particle size of the initial silica sol seed crystal product to obtain data information on the particle size of the initial silica sol seed crystal product, use a silicate metering pump to obtain data information on the silicate addition rate in real time, use a stabilizer metering pump to obtain data information on the stabilizer addition rate in real time, use a pH value sensor to obtain data information on the pH value of the reactant in real time, obtain data information on the speed of the stirrer in real time by detecting the speed of the stirring motor, and use a temperature sensor in the reactor to obtain data information on the temperature of the reactor in real time;

[0011] U3. Input the reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information, silicic acid addition rate data information and silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning, and obtain the optimal solution, and output the optimized reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information and silicic acid addition rate data information;

[0012] U4. Input the optimized reactor temperature data information, agitator speed data information, reactant PH data information, stabilizer addition speed data information and silicate addition speed data information into ADRC algorithm to adjust and control the reactor, silicate metering pump, stabilizer metering pump and agitator, output the reactor temperature control data information, silicate metering pump control data information, stabilizer metering pump control data information and agitator control data information to each control mechanism, keep the reactor warm for 50-60 minutes, and obtain silica sol seed crystals for wafer grinding.

[0013] Further, in step U3, the step of inputting the reactor temperature data information, the stirrer speed data information, the pH value data information of the reactants, the stabilizer addition speed data information, the silicic acid addition speed data information and the silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning includes:

[0014] U31. Based on the reactor temperature data information, the stirrer speed data information, the reactant pH value data information, the stabilizer addition speed data information, the silicic acid addition speed data information and the silica sol seed initial product particle size data information, establish a relationship mapping J, J (x1, x2, x3, x4, x5) = α0 + α1f1 (x1) + α2f2 (x2) + α3f3 (x3) + α4f4 (x4) + α5f5 (x5), where x1 is the reactor temperature data information information, f1(x1) is the temperature function of the reactor, x2 is the speed data of the stirrer, f2(x2) is the speed function of the stirrer, x3 is the pH value data of the reactants, f3(x3) is the pH value function of the reactants, x4 is the data of the stabilizer addition speed, f4(x4) is the stabilizer addition speed function, x5 is the data of the silica acid addition speed, f5(x5) is the silica acid addition speed function, and J is the particle size data of the initial silica sol seed crystal product;

[0015] U32. Based on the relationship mapping function J (x1, x2, x3, x4, x5), a relationship matrix G of the reactor temperature, the speed of the stirrer, the pH value of the reactants, the rate of adding the stabilizer and the rate of adding the silicic acid is established.

[0016]

[0017] Among them, x 1n is the temperature data information of the reactor sampled at the nth time, x 2n is the speed data information of the nth sampling stirrer, x 3n is the pH value data information of the nth sampling reactant, x 4n The data information of the speed of adding the stabilizer for the nth sampling, x 5n is the data information of the silicate addition rate of the nth sampling, J n The particle size data information of the initial silica sol seed sampled for the nth time;

[0018] U33. Based on the relationship matrix G of the reactor temperature, the speed of the stirrer, the pH value of the reactants, the rate of addition of the stabilizer and the rate of addition of silicate, a weight function F of the improved Hopfer neural network model is established.

[0019]

[0020] Where i = 1, 2, 3, ... n, j = 1, 2, 3, ... n, G i s and G j s are the i-th and j-th elements of the s-th sequence, k is the k-th sequence to be stored, G is the relationship matrix, α and β are weight analysis factors;

[0021] U34. Based on the weight function F of the improved Hopfield neural network model, output the optimized reactor temperature data information, agitator speed data information, reactant pH value data information, stabilizer addition rate data information and silicate addition rate data information.

[0022] Furthermore, in step U31, the temperature relationship function f1(x1) of the reactor, the speed relationship function f2(x2) of the stirrer and the pH value relationship function f3(x3) of the reactant are respectively:

[0023] f1(x1)=λ2x1 2 +λ1x1 1 +λ0,

[0024] f2(x2)=ω1x2+ω0,

[0025] f3(x3)=σlnx3,

[0026] Among them, x1 is the data information of the reactor temperature, x2 is the data information of the agitator speed, x3 is the data information of the pH value of the reactant, λ0, λ1 and λ2 are constant parameters of the reactor temperature relationship function, ω1 and ω0 are constant parameters of the agitator speed relationship function, and σ is a constant parameter of the pH value relationship function of the reactant.

[0027] Further, in step U31, the stabilizer addition rate relationship function f4(x4) and the silicate addition rate relationship function f5(x5) are respectively:

[0028] f4(x4)=δ3x4 3 +δ2x4 2 +δ1x4+δ0,

[0029] f5(x5)=τ5x5 5 +τ4x5 4 +τ3x5 3 +τ2x5 2 +τ1x5+τ0,

[0030] Among them, x4 is the data information of the stabilizer addition rate, x5 is the data information of the silicate addition rate, δ0, δ1, δ2, δ3 and δ4 are constant parameters of the stabilizer addition rate relationship function, τ0, τ1, τ2, τ3, τ4 and τ5 are constant parameters of the silicate addition rate relationship function.

[0031] Further, in step U4, the step of inputting the optimized reactor temperature data information, agitator speed data information, reactant pH data information, stabilizer addition speed data information and silicic acid addition speed data information into the ADRC algorithm to adjust and control the reactor, silicic acid metering pump, stabilizer metering pump and agitator includes:

[0032] U41. Based on the optimized reactor temperature data information, stirrer speed data information, reactant pH data information, stabilizer addition speed data information and silicic acid addition speed data information, establish a state tracking differential control function R,

[0033]

[0034] Among them, x1′ is the optimized data information of reactor temperature, x′2 is the optimized data information of stirrer speed, x3′ is the optimized data information of reactant pH, x′4 is the optimized data information of stabilizer addition speed, x5′ is the optimized data information of silicic acid addition speed, h and z are state tracking adjustment factors, and the state change data information of each control mechanism is obtained;

[0035] U42. Based on the state change data information of each control mechanism, establish the real-time dynamic disturbance function H of each control mechanism,

[0036] H(x1,x2,x3,x4,x5,Δt)=H0(x1,x2,x3,x4,x5,Δt)+H1(x1,x2,x3,x4,x5,Δt), wherein H0 is the real-time dynamic external disturbance function of each control mechanism, H1 is the real-time dynamic internal disturbance function of each control mechanism, x1 is the data information of reactor temperature, x2 is the data information of agitator speed, x3 is the data information of pH value of reactant, x4 is the data information of stabilizer addition speed, x5 is the data information of silicate addition speed, and the state disturbance change data information of each control mechanism is obtained, and Δt is the state change time of each control mechanism;

[0037] U43. Based on the state disturbance variation data information of each control mechanism and the state variation data information of each control mechanism, establish a real-time control law function L of each control mechanism,

[0038] L=bfal(n1,n2,φ),

[0039] Among them, n1 is the state disturbance change data information of each control mechanism, n2 is the state change data information of each control mechanism, φ is the control change factor, b is the control constant parameter, and fal is the nonlinear combination function of each control mechanism;

[0040] U44. Based on the real-time control law function L of each control mechanism, output the control quantity data information of each control mechanism.

[0041] Furthermore, in step U43, the nonlinear combination function fal of each control mechanism,

[0042]

[0043] And φ is greater than 0, where n1 is the state disturbance change data information of each control mechanism, n2 is the state change data information of each control mechanism, and φ is the control change factor.

[0044] Furthermore, the control mechanisms are a temperature control mechanism of the reactor, a stirrer, a silicate metering pump and a stabilizer metering pump.

[0045] In order to achieve the above-mentioned purpose and other related purposes, the present invention also provides a silica sol seed crystal growth control system for wafer grinding, including a computer device, which is programmed or configured to execute any step of the silica sol seed crystal growth control method for wafer grinding.

[0046] In order to achieve the above-mentioned purpose and other related purposes, the present invention also provides a computer-readable storage medium, which stores a computer program programmed or configured to execute any one of the silica sol seed growth control methods for wafer grinding.

[0047] The present invention has the following positive effects:

[0048] 1. The present invention inputs the reactor temperature data information, the stirrer speed data information, the pH value data information of the reactants, the stabilizer addition speed data information, the silicic acid addition speed data information and the particle size data information of the initial silica sol seed crystal into the improved Hopfield neural network model for training and learning, and obtains the optimal solution, which can not only effectively control the particle size of the silica sol, but also reasonably distribute the raw materials during the control process, thereby reducing unnecessary consumption in the production process and reducing production costs.

[0049] 2. The present invention inputs the optimized reactor temperature data information, agitator speed data information, reactant PH data information, stabilizer addition speed data information and silicic acid addition speed data information into the ADRC algorithm to adjust and control the reactor, silicic acid metering pump, stabilizer metering pump and agitator, which not only accurately controls the reactor, silicic acid metering pump, stabilizer metering pump and agitator, but also further improves the control accuracy of the growth of silica sol seed crystals, and reduces the situation of unqualified finished products caused by insufficient or excessive feeding.

[0050] 3. The silica sol seed growth control method, system and medium for wafer grinding of the present invention can not only accurately control the growth of silica sol seeds and improve the production efficiency of silica sol by real-time monitoring and adjusting the growth process of silica sol seeds, but also ensure the stability and consistency of the product, providing new possibilities for industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0052] Figure 2 It is a schematic diagram of the structure of the reactor of the present invention.

[0053] Explanation of the numbers in the figure: 1-motor, 2-silicic acid metering pump, 3-stabilizer metering pump, 4-PH value sensor, 5-temperature sensor, 6-agitator, 7-online laser particle size analyzer. DETAILED DESCRIPTION

[0054] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0055] Example 1: Figure 1 or Figure 2 As shown, a method for controlling the growth of silica sol seed crystals for wafer grinding is applied to a reactor equipped with an online laser particle size analyzer, and the method comprises:

[0056] U1. Open the silicic acid metering pump on the reactor, add silicic acid to the reactor, and at the same time open the stabilizer metering pump on the reactor, add ammonia water as an alkaline stabilizer, adjust the pH value to 8.5-9.5, stir the reactor at a certain speed, heat to 97-99 degrees and keep warm for 25-35 minutes to grow the silicon dioxide particles, and obtain the initial silica sol seed product, at this time the particle size is about 5nm;

[0057] U2. Use an online laser particle size analyzer to detect the particle size of the initial silica sol seed crystal product to obtain data information on the particle size of the initial silica sol seed crystal product, use a silicate metering pump to obtain data information on the silicate addition rate in real time, use a stabilizer metering pump to obtain data information on the stabilizer addition rate in real time, use a pH value sensor to obtain data information on the pH value of the reactant in real time, obtain data information on the speed of the stirrer in real time by detecting the speed of the stirring motor, and use a temperature sensor in the reactor to obtain data information on the temperature of the reactor in real time;

[0058] U3. Input the reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information, silicic acid addition rate data information and silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning, and obtain the optimal solution, and output the optimized reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information and silicic acid addition rate data information;

[0059] U4. The optimized reactor temperature data information, agitator speed data information, reactant PH data information, stabilizer addition speed data information and silicate addition speed data information are input into ADRC algorithm to adjust and control the reactor, silicate metering pump, stabilizer metering pump and agitator, and the reactor temperature control data information, silicate metering pump control data information, stabilizer metering pump control data information and agitator control data information are output to each control mechanism. The reactor is kept warm for 50-70 minutes, and the silica sol seeds grow in a perfect form. The obtained silica sol seeds for wafer grinding have a controllable particle size according to demand during the reaction process, in order to achieve the general industry requirement of 20-80nm.

[0060] In this embodiment, data preprocessing: First, we need to preprocess the input data, including data cleaning, filling missing values, data standardization and other operations. These operations can make data of different dimensions and types comparable, so as to better carry out subsequent analysis and modeling.

[0061] In this embodiment, in step U3, the step of inputting the reactor temperature data information, the stirrer speed data information, the pH value data information of the reactants, the stabilizer addition speed data information, the silicic acid addition speed data information and the silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning includes:

[0062] U31. Based on the reactor temperature data information, the stirrer speed data information, the reactant pH value data information, the stabilizer addition speed data information, the silicic acid addition speed data information and the silica sol seed initial product particle size data information, establish a relationship mapping J, J (x1, x2, x3, x4, x5) = α0 + α1f1 (x1) + α2f2 (x2) + α3f3 (x3) + α4f4 (x4) + α5f5 (x5), where x1 is the reactor temperature data information information, f1(x1) is the temperature function of the reactor, x2 is the speed data of the stirrer, f2(x2) is the speed function of the stirrer, x3 is the pH value data of the reactants, f3(x3) is the pH value function of the reactants, x4 is the data of the stabilizer addition speed, f4(x4) is the stabilizer addition speed function, x5 is the data of the silica acid addition speed, f5(x5) is the silica acid addition speed function, and J is the particle size data of the initial silica sol seed crystal product;

[0063] U32. Based on the relationship mapping function J (x1, x2, x3, x4, x5), a relationship matrix G of the reactor temperature, the speed of the stirrer, the pH value of the reactants, the rate of adding the stabilizer and the rate of adding the silicic acid is established.

[0064]

[0065] Among them, x 1n is the temperature data information of the reactor sampled at the nth time, x 2n is the speed data information of the nth sampling stirrer, x 3n is the pH value data information of the nth sampling reactant, x 4n The data information of the speed of adding the stabilizer for the nth sampling, x 5n is the data information of the silicate addition rate of the nth sampling, J n The particle size data information of the initial silica sol seed sampled for the nth time;

[0066] U33. Based on the relationship matrix G of the reactor temperature, the speed of the stirrer, the pH value of the reactants, the rate of addition of the stabilizer and the rate of addition of silicate, a weight function F of the improved Hopfer neural network model is established.

[0067]

[0068] Where i = 1, 2, 3, ... n, j = 1, 2, 3, ... n, G i s and G j s are the i-th and j-th elements of the s-th sequence, k is the k-th sequence to be stored, G is the relationship matrix, α and β are weight analysis factors;

[0069] U34. Based on the weight function F of the improved Hopfield neural network model, output the optimized reactor temperature data information, agitator speed data information, reactant pH value data information, stabilizer addition rate data information and silicate addition rate data information.

[0070] In this embodiment, in step U31, the temperature relationship function f1(x1) of the reactor, the speed relationship function f2(x2) of the stirrer and the pH value relationship function f3(x3) of the reactant are respectively:

[0071] f1(x1)=λ2x1 2 +λ1x1 1 +λ0,

[0072] f2(x2)=ω1x2+ω0,

[0073] f3(x3)=σlnx3,

[0074] Among them, x1 is the data information of the reactor temperature, x2 is the data information of the agitator speed, x3 is the data information of the pH value of the reactant, λ0, λ1 and λ2 are constant parameters of the reactor temperature relationship function, ω1 and ω0 are constant parameters of the agitator speed relationship function, and σ is a constant parameter of the pH value relationship function of the reactant.

[0075] Determination of the relationship function: In the formula

[0076] J(x1,x2,x3,x4,x5)=α0+α1f1(x1)+α2f2(x2)+α3f3(x3)+α4f4(x4)+α

[0077] 5f5(x5), we need to determine the various relationship functions f1(x1), f2(x2), f3(x3), f4(x4) and f5(x5).

[0078] These relationship functions can be determined by fitting experimental data or using prior knowledge. For example, we can establish a suitable model to describe this relationship by observing the relationship between the reactor temperature and the initial particle size of the silica sol seed crystal.

[0079] Model training: Use known data to train the model, that is, determine the coefficients α0, α1, α2, α3, α4, and α5. This can be done through machine learning algorithms such as least squares and gradient descent. During the training process, we need to choose a suitable loss function and optimization algorithm so that the model can better fit the training data.

[0080] Model validation and evaluation: Use a portion of independent data to verify the performance of the model and evaluate the model's prediction accuracy and generalization ability. We can use indicators such as mean square error (MSE) and root mean square error (RMSE) to measure the model's prediction accuracy, and use methods such as cross-validation to evaluate the model's generalization ability.

[0081] Model optimization: Based on the results of validation and evaluation, the model is optimized. This may include adjusting the model's parameters, selecting different features, modifying the model structure, etc. In addition, we can also use regularization techniques to prevent overfitting and improve the generalization ability of the model.

[0082] Model application: After the model is optimized and verified, it can be applied to actual production. By real-time monitoring of the reactor temperature, agitator speed, reactant pH value, stabilizer addition rate, and silicic acid addition rate, the model is used to calculate the particle size data of the initial silica sol seed product, thereby achieving precise control of silica sol production.

[0083] In this embodiment, in step U31, the stabilizer addition rate relationship function f4(x4) and the silicate addition rate relationship function f5(x5) are respectively:

[0084] f4(x4)=δ3x4 3 +δ2x4 2 +δ1x4+δ0,

[0085] f5(x5)=τ5x5 5 +τ4x5 4 +τ3x5 3 +τ2x5 2 +τ1x5+τ0,

[0086] Among them, x4 is the data information of the stabilizer addition rate, x5 is the data information of the silicate addition rate, δ0, δ1, δ2, δ3 and δ4 are constant parameters of the stabilizer addition rate relationship function, τ0, τ1, τ2, τ3, τ4 and τ5 are constant parameters of the silicate addition rate relationship function.

[0087] Example 2: Figure 1 or Figure 2 , a method for controlling the growth of silica sol seed crystals for wafer grinding, the method comprising:

[0088] U1. In the reactor, open the silicic acid metering pump on the reactor, add silicic acid to the reactor, and at the same time open the stabilizer metering pump on the reactor, add ammonia water as an alkaline stabilizer, adjust the pH value to 8.5-9.5, stir the reactor at a certain speed, heat to 97-99 degrees and keep warm for 25-35 minutes to grow the silicon dioxide particles, and obtain the initial silica sol seed crystal product, at this time the particle size is about 5nm;

[0089] U2. Use a laser particle size analyzer to detect the particle size of the initial silica sol seed crystal product to obtain data information on the particle size of the initial silica sol seed crystal product, use a silicate metering pump to obtain data information on the silicate addition rate in real time, use a stabilizer metering pump to obtain data information on the stabilizer addition rate in real time, use a pH value sensor to obtain data information on the pH value of the reactant in real time, obtain data information on the speed of the stirrer in real time by detecting the speed of the stirring motor, and use a temperature sensor to obtain data information on the temperature of the reactor in real time;

[0090] U3. Input the reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information, silicic acid addition rate data information and silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning, and obtain the optimal solution, and output the optimized reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information and silicic acid addition rate data information;

[0091] U4. Input the optimized reactor temperature data information, agitator speed data information, reactant PH data information, stabilizer addition speed data information and silicate addition speed data information into ADRC algorithm to adjust and control the reactor, silicate metering pump, stabilizer metering pump and agitator, output the reactor temperature control data information, silicate metering pump control data information, stabilizer metering pump control data information and agitator control data information to each control mechanism, keep the reactor warm for 50-70 minutes, and obtain silica sol seed crystals for wafer grinding.

[0092] In this embodiment, in step U4, the step of inputting the optimized reactor temperature data information, agitator speed data information, reactant pH data information, stabilizer addition speed data information and silicic acid addition speed data information into the ADRC algorithm to adjust and control the reactor, silicic acid metering pump, stabilizer metering pump and agitator includes:

[0093] U41. Based on the optimized reactor temperature data information, stirrer speed data information, reactant pH data information, stabilizer addition speed data information and silicic acid addition speed data information, establish a state tracking differential control function R,

[0094]

[0095] Among them, x1′ is the optimized data information of reactor temperature, x′2 is the optimized data information of stirrer speed, x3′ is the optimized data information of reactant pH, x′4 is the optimized data information of stabilizer addition speed, x5′ is the optimized data information of silicic acid addition speed, h and z are state tracking adjustment factors, and the state change data information of each control mechanism is obtained;

[0096] U42. Based on the state change data information of each control mechanism, establish the real-time dynamic disturbance function H of each control mechanism,

[0097] H(x1,x2,x3,x4,x5,Δt)=H0(x1,x2,x3,x4,x5,Δt)+H1(x1,x2,x3,x4,x5,Δt), wherein H0 is the real-time dynamic external disturbance function of each control mechanism, H1 is the real-time dynamic internal disturbance function of each control mechanism, x1 is the data information of reactor temperature, x2 is the data information of agitator speed, x3 is the data information of pH value of reactant, x4 is the data information of stabilizer addition speed, x5 is the data information of silicate addition speed, and the state disturbance change data information of each control mechanism is obtained, and Δt is the state change time of each control mechanism;

[0098] U43. Based on the state disturbance variation data information of each control mechanism and the state variation data information of each control mechanism, establish a real-time control law function L of each control mechanism,

[0099] L=bfal(n1,n2,φ),

[0100] Among them, n1 is the state disturbance change data information of each control mechanism, n2 is the state change data information of each control mechanism, φ is the control change factor, b is the control constant parameter, and fal is the nonlinear combination function of each control mechanism;

[0101] U44. Based on the real-time control law function L of each control mechanism, output the control quantity data information of each control mechanism.

[0102] In this embodiment, in step U43, the nonlinear combination function fal of each control mechanism is:

[0103]

[0104] And φ is greater than 0, where n1 is the state disturbance change data information of each control mechanism, n2 is the state change data information of each control mechanism, and φ is the control change factor.

[0105] In this embodiment, the control mechanisms are a temperature control mechanism of the reactor, a stirrer, a silicate metering pump and a stabilizer metering pump.

[0106] The present invention provides a control system for the growth of silica sol seed crystals for wafer grinding, comprising a computer device which is programmed or configured to execute any one of the steps of the control method for the growth of silica sol seed crystals for wafer grinding.

[0107] The silica sol seed crystal growth control system for wafer grinding includes: a data acquisition module for real-time acquisition of data information on the particle size of the initial silica sol seed crystal, data information on the rate of addition of silicic acid, data information on the rate of addition of a stabilizer, and data information on the pH value of the reactants; and a data processing module for processing and analyzing the acquired data information, so as to accurately control the growth of the silica sol seed crystal.

[0108] The data processing module includes: a data analysis unit and a feedback control unit. The data analysis unit is used to analyze data information and identify the growth state of the silica sol seed crystals. The feedback control unit is used to regulate the silicate metering pump, stabilizer metering pump, agitator and temperature control system according to the analysis results of the data analysis unit.

[0109] The present invention provides a computer-readable storage medium, on which is stored a computer program programmed or configured to execute any one of the methods for controlling the growth of silica sol seeds for wafer grinding.

[0110] The control mechanism used in the embodiments provided in the present application can be any combination of components such as a frequency converter, a D / A conversion module, a stepper motor, a digital switch, etc. that controls the heating temperature of the reactor, the delivery volume of the silicate metering pump, the delivery volume of the stabilizer metering pump and the speed of the agitator.

[0111] Any reference to memory, storage, database, or other medium may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0112] In summary, the present invention can not only accurately control the growth of silica sol seed crystals, but also improve the production efficiency of silica sol.

[0113] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for controlling the growth of silica sol seed crystals for wafer grinding, characterized in that: The method comprises: U1. Open the silicic acid metering pump on the reactor, add silicic acid to the reactor, open the stabilizer metering pump on the reactor, add alkaline stabilizer to adjust the pH value, stir the reactor at a certain speed, heat to a certain temperature for reaction, and keep warm for a certain period of time to obtain the initial silica sol seed product; U2. Use an online laser particle size analyzer to detect the particle size of the initial silica sol seed crystal product to obtain data information on the particle size of the initial silica sol seed crystal product, use a silicate metering pump to obtain data information on the silicate addition rate in real time, use a stabilizer metering pump to obtain data information on the stabilizer addition rate in real time, use a pH value sensor to obtain data information on the pH value of the reactant in real time, obtain data information on the speed of the stirrer in real time by detecting the speed of the stirring motor, and use a temperature sensor in the reactor to obtain data information on the temperature of the reactor in real time; U3. Input the reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information, silicic acid addition rate data information and silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning, and obtain the optimal solution, and output the optimized reactor temperature data information, stirrer speed data information, reactant pH value data information, stabilizer addition rate data information and silicic acid addition rate data information; U4. Input the optimized reactor temperature data information, stirrer speed data information, reactant PH data information, stabilizer addition speed data information and silicic acid addition speed data information into ADRC algorithm to adjust and control the reactor, silicic acid metering pump, stabilizer metering pump and stirrer, output reactor temperature control data information, silicic acid metering pump control data information, stabilizer metering pump control data information and stirrer control data information to each control mechanism, and obtain silica sol seed crystals for wafer grinding after the reactor is kept warm for a certain period of time; In step U3, the step of inputting the reactor temperature data information, the stirrer speed data information, the pH value data information of the reactants, the stabilizer addition speed data information, the silicic acid addition speed data information and the silica sol seed initial product particle size data information into the improved Hopfield neural network model for training and learning includes: U31. Based on the reactor temperature data information, the stirrer speed data information, the reactant pH value data information, the stabilizer addition speed data information, the silicic acid addition speed data information and the silica sol seed initial product particle size data information, establish a relationship mapping J, J(x1,x2,x3,x4,x5)=α0+α1f1(x1)+α2f2(x2)+α3f3(x3)+α4f4(x4)+α5f5(x5), Wherein, x1 is the data information of reactor temperature, f1(x1) is the relationship function of reactor temperature, x2 is the data information of stirrer speed, f2(x2) is the relationship function of stirrer speed, x3 is the data information of pH value of reactant, f3(x3) is the relationship function of pH value of reactant, x4 is the data information of stabilizer addition speed, f4(x4) is the relationship function of stabilizer addition speed, x5 is the data information of silicic acid addition speed, f5(x5) is the relationship function of silicic acid addition speed, and J is the data information of particle size of initial silica sol seed crystal; U32. Based on the relationship mapping function J (x1, x2, x3, x4, x5), a relationship matrix G of the reactor temperature, the speed of the stirrer, the pH value of the reactants, the rate of adding the stabilizer and the rate of adding the silicic acid is established. ， Among them, x 1n is the temperature data information of the reactor sampled at the nth time, x 2n is the speed data information of the nth sampling stirrer, x 3n is the pH value data information of the nth sampling reactant, x 4n The data information of the speed of adding the stabilizer for the nth sampling, x 5n is the data information of the silicate addition rate of the nth sampling, J n The particle size data information of the initial silica sol seed sampled for the nth time; U33. Based on the relationship matrix G of the reactor temperature, the speed of the stirrer, the pH value of the reactants, the rate of addition of the stabilizer and the rate of addition of silicate, a weight function F of the improved Hopfer neural network model is established. ， Where i=1,2,3,...n, j=1,2,3,...n, G i s and G j s are the i-th and j-th elements of the s-th sequence, k is the k-th sequence to be stored, G is the relationship matrix, α and β are weight analysis factors; U34. Based on the weight function F of the improved Hopfield neural network model, output the optimized reactor temperature data information, agitator speed data information, reactant pH value data information, stabilizer addition rate data information and silicate addition rate data information.

2. The method for controlling the growth of silica sol seed crystals for wafer polishing according to claim 1, characterized in that: In step U31, the temperature relationship function f1(x1) of the reactor, the speed relationship function f2(x2) of the stirrer and the pH value relationship function f3(x3) of the reactant are respectively: f1(x1)=λ2x1 2 +λ1x1 1 +λ0, f2(x2)=ω1x2+ω0, f3(x3)=σlnx3, Among them, x1 is the data information of the reactor temperature, x2 is the data information of the agitator speed, x3 is the data information of the pH value of the reactant, λ0, λ1 and λ2 are constant parameters of the reactor temperature relationship function, ω1 and ω0 are constant parameters of the agitator speed relationship function, and σ is a constant parameter of the pH value relationship function of the reactant.

3. The method for controlling the growth of silica sol seed crystals for wafer polishing according to claim 1, characterized in that: In step U31, the stabilizer addition rate relationship function f4(x4) and the silicate addition rate relationship function f5(x5) are respectively: f4(x4)=δ3x4 3 +δ2x4 2 +δ1x4+δ0, f5(x5)=τ5x5 5 +τ4x5 4 +τ3x5 3 +τ2x5 2 +τ1x5+τ0, Among them, x4 is the data information of the stabilizer addition rate, x5 is the data information of the silicate addition rate, δ0, δ1, δ2, δ3 and δ4 are constant parameters of the stabilizer addition rate relationship function, τ0, τ1, τ2, τ3, τ4 and τ5 are constant parameters of the silicate addition rate relationship function.

4. The method for controlling the growth of silica sol seed crystals for wafer polishing according to claim 1, characterized in that: In step U4, the step of inputting the optimized reactor temperature data information, agitator speed data information, reactant pH data information, stabilizer addition speed data information and silicic acid addition speed data information into the ADRC algorithm to adjust and control the reactor, silicic acid metering pump, stabilizer metering pump and agitator includes: U41. Based on the optimized reactor temperature data information, stirrer speed data information, reactant pH data information, stabilizer addition speed data information and silicic acid addition speed data information, establish a state tracking differential control function R, ， in, This is the optimized reactor temperature data information. Speed ​​data information for optimized agitator, is the pH data information of the optimized reactants, This is the data information of the optimized stabilizer addition speed. is the data information of the optimized silicate addition rate, h and z are the state tracking adjustment factors, and the state change data information of each control mechanism is obtained; U42. Based on the state change data information of each control mechanism, establish the real-time dynamic disturbance function H of each control mechanism, H(x1,x2,x3,x4,x5,Δt)=H0(x1,x2,x3,x4,x5,Δt)+H1(x1,x2,x3,x4,x5,Δt), Among them, H0 is the real-time dynamic external disturbance function of each control mechanism, H1 is the real-time dynamic internal disturbance function of each control mechanism, x1 is the data information of the reactor temperature, x2 is the data information of the speed of the stirrer, x3 is the data information of the pH value of the reactant, x4 is the data information of the stabilizer addition speed, x5 is the data information of the silicate addition speed, and the state disturbance change data information of each control mechanism is obtained, and Δt is the state change time of each control mechanism; U43. Based on the state disturbance variation data information of each control mechanism and the state variation data information of each control mechanism, establish a real-time control law function L of each control mechanism, L = bfal(n1,n2,φ), Among them, n1 is the state disturbance change data information of each control mechanism, n2 is the state change data information of each control mechanism, φ is the control change factor, b is the control constant parameter, and fal is the nonlinear combination function of each control mechanism; U44. Based on the real-time control law function L of each control mechanism, output the control quantity data information of each control mechanism.

5. The method for controlling the growth of silica sol seed crystals for wafer polishing according to claim 4, characterized in that: In step U43, the nonlinear combination function fal of the control mechanisms is: ， And φ is greater than 0, where n1 is the state disturbance change data information of each control mechanism, n2 is the state change data information of each control mechanism, and φ is the control change factor.

6. The method for controlling the growth of silica sol seed crystals for wafer polishing according to claim 5, characterized in that: The control mechanisms are a temperature control mechanism of the reactor, a stirrer, a silicate metering pump and a stabilizer metering pump.

7. A silica sol seed growth control system for wafer grinding, characterized in that: The invention comprises a computer device, characterized in that the computer device is programmed or configured to execute the steps of the method for controlling the growth of silica sol seed crystals for wafer grinding according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program programmed or configured to execute the method for controlling the growth of silica sol seed crystals for wafer grinding according to any one of claims 1 to 6.

Images