CN118331071B - Method for monitoring and controlling concentration of particles in reaction solution - Google Patents
Method for monitoring and controlling concentration of particles in reaction solution Download PDFInfo
- Publication number
- CN118331071B CN118331071B CN202410725165.9A CN202410725165A CN118331071B CN 118331071 B CN118331071 B CN 118331071B CN 202410725165 A CN202410725165 A CN 202410725165A CN 118331071 B CN118331071 B CN 118331071B
- Authority
- CN
- China
- Prior art keywords
- temperature
- concentration
- model
- data
- solution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000002245 particle Substances 0.000 title claims abstract description 112
- 238000006243 chemical reaction Methods 0.000 title claims abstract description 86
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000012544 monitoring process Methods 0.000 title claims abstract description 30
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 111
- 238000004088 simulation Methods 0.000 claims abstract description 30
- 230000008569 process Effects 0.000 claims abstract description 26
- 238000005457 optimization Methods 0.000 claims abstract description 23
- 238000011217 control strategy Methods 0.000 claims abstract description 22
- 230000036961 partial effect Effects 0.000 claims abstract description 21
- 230000002787 reinforcement Effects 0.000 claims abstract description 15
- 230000009471 action Effects 0.000 claims description 43
- 230000006870 function Effects 0.000 claims description 35
- 238000012549 training Methods 0.000 claims description 26
- 230000000694 effects Effects 0.000 claims description 19
- 230000008859 change Effects 0.000 claims description 13
- 238000009792 diffusion process Methods 0.000 claims description 13
- 238000005259 measurement Methods 0.000 claims description 13
- 238000009826 distribution Methods 0.000 claims description 11
- 230000004907 flux Effects 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 8
- 238000012546 transfer Methods 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 5
- 238000005507 spraying Methods 0.000 claims description 5
- 230000006399 behavior Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 238000007865 diluting Methods 0.000 claims description 4
- 238000004134 energy conservation Methods 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 4
- 239000003153 chemical reaction reagent Substances 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 230000005674 electromagnetic induction Effects 0.000 claims description 3
- 239000004973 liquid crystal related substance Substances 0.000 claims description 3
- 230000036962 time dependent Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 3
- 238000007499 fusion processing Methods 0.000 claims description 2
- 239000004065 semiconductor Substances 0.000 abstract description 5
- 238000011161 development Methods 0.000 description 7
- 238000004519 manufacturing process Methods 0.000 description 5
- 229920002120 photoresistant polymer Polymers 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 239000013043 chemical agent Substances 0.000 description 3
- 238000013178 mathematical model Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 239000000047 product Substances 0.000 description 3
- 229910000679 solder Inorganic materials 0.000 description 3
- 239000007795 chemical reaction product Substances 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 239000000758 substrate Substances 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012993 chemical processing Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000011859 microparticle Substances 0.000 description 1
- 238000005375 photometry Methods 0.000 description 1
- 230000036632 reaction speed Effects 0.000 description 1
- 238000009991 scouring Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000002277 temperature effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention provides a method for monitoring and controlling the concentration of particles in a reaction solution, and belongs to the technical field of particle concentration monitoring and controlling in a semiconductor process. Real-time particle concentration data, temperature data and flow rate data of pure water in a water inlet in an inner circulation system pipeline are collected, a particle concentration model and a temperature model in a reaction chamber are constructed based on the preprocessed data, and a partial differential equation is utilized to describe the dynamic evolution process of concentration and temperature in a solution. Secondly, the constructed concentration model and temperature model are digitalized into simulation models, optimization is carried out in a virtual environment, and a control strategy which best meets the control target of pure water flow rate and temperature compensation quantity is searched. Based on the simulation environment, a controller based on reinforcement learning is established, and the controller is trained in the simulation environment to learn an optimal control strategy. And finally, deploying the trained controller into an actual concentration monitoring system to realize efficient closed-loop monitoring and control.
Description
Technical Field
The invention belongs to the technical field of particle concentration monitoring and control in a semiconductor process, and particularly relates to a method for monitoring and controlling the particle concentration of a reaction solution.
Background
In recent years, as semiconductor device manufacturing processes continue to increase, so too has the requirements for controlling micro particles or debris on the surface of a PCB or IC substrate. These small particles and residues can lead to increased chip defect rates, affecting product quality and yield. Specifically, the process steps such as dry/wet film development, solder resist development or dry/wet film stripping require photoresist treatment, and the particles (chips) have acceptable effects on the product in the prior production of the product, but the negative problems caused by the chips are considered in the new generation of technology. The control of these particulates (debris) is increasingly important in new generation technology for the negative impact of the process itself, overall yield, and the lifetime of the chemical processing chamber.
The current stage is to control the above-mentioned particle (chip) management manner in the process of producing high-end products such as IC substrates, more in such a manner that the whole particle (chip) amount is kept below a threshold value by injecting new chemical agents as much as possible and by hedging the increasing particle (chip), and when the chip concentration is detected to be out of a reasonable range, diluting the chip concentration by supplementing an appropriate amount of water into the system; when the concentration of the chemical agent in the solution is detected to be lower than a reasonable range in real time, a proper amount of chemical agent is added into the system, so that the production requirement is met.
In the prior art, the particle concentration is usually measured by adopting a conductivity measurement mode, the measurement mode covers most of process equipment in the market, besides conductivity measurement, a photometric measurement mode is adopted for relatively high-end products, but the sampling precision and the sampling rate of the two measurement modes cannot meet the process monitoring requirements more and more along with the improvement of the control requirements, and a high-precision and high-efficiency control strategy based on the monitored concentration is lacking.
Disclosure of Invention
In view of the above problems, the present invention provides a method for monitoring and controlling the concentration of particles in a reaction solution, comprising the steps of:
The temperature sensor and the ultrasonic sensor are simultaneously arranged at the adjacent positions of the inner circulation pipeline wall based on the pipeline inflow system or the spraying system, the flow rate of the chemical reagent is fixed, and the flow rate of pure water is controlled while the temperature data of the reaction solution and the concentration data of solution particles are monitored; taking the temperature and the concentration of solution particles as optimal control targets, and ensuring the detection precision of the ultrasonic sensor through temperature compensation;
establishing a controller model based on reinforcement learning according to a control strategy meeting an optimal control target, namely an optimal set value of the opening degree of the water valve and the temperature compensation quantity;
The controller model is deployed into a controller of the whole system, and training and application are carried out on the controller model, so that the particle concentration monitoring and the pure water flow rate control of the high-efficiency closed loop are realized.
Preferably, the method further comprises the following steps:
Collecting real-time particle concentration data of inner wall of inner circulation pipeline Solution temperature data at ultrasonic sensorTemperature data at water inletAnd flow rate data of pure water in water inletPreprocessing the acquired data;
constructing a particle concentration model and a temperature model in a reaction chamber, and observing dynamic changes in the reaction process;
Constructing a simulation environment based on the particle concentration model and the temperature model to obtain a control strategy which optimally meets a control target, namely, an optimal set value of the opening degree of the water valve and the temperature compensation quantity;
training the controller model in a simulation environment based on the reinforcement-learning controller model to obtain an initialization model;
The initialization model is deployed into a controller of the system, and further training and application are performed in a real environment, so that efficient closed-loop particle concentration monitoring and pure water flow rate control are realized.
Preferably, the collecting data specifically includes:
Particle concentration data acquisition: the ultrasonic sensor is contacted with the solution and emits ultrasonic waves, the concentration of particles is calculated according to the reflection condition of the ultrasonic waves, and the concentration data is recorded and updated with time, and is recorded as ；
And (3) temperature data acquisition: in the reaction process, certain temperature change occurs to influence the measurement accuracy of the ultrasonic sensor, so that temperature compensation is needed, the first temperature sensor is arranged on the inner wall of the inner circulation pipeline to measure the temperature change of the ultrasonic sensor, and the temperature data are updated in real time along with the time and recorded as; Meanwhile, a second temperature sensor is used for collecting water inlet temperature data, which is recorded as；
And (3) collecting pure water flow rate data: the first electromagnetic flowmeter is used for measuring the flow rate of pure water in the water inlet; by electromagnetic induction principle, measuring the flow rate data of pure water flowing through the water inlet, and recording asAt the same time, the time is acquired by the second electromagnetic flowmeterThe velocity of the flow field at the moment, i.e. the flow velocity of the solution in the whole reaction chamber, is recorded as。
Preferably, the specific process of preprocessing the collected data is as follows:
Particle concentration data pretreatment of ultrasonic collection: denoising the particle concentration data by adopting wavelet transformation to eliminate high-frequency noise interference introduced by measuring equipment or environment; the wavelet transform decomposes the signal into components of different frequencies, and then retains useful signal information by selectively removing high frequency noise, as the formula:
（1）
Wherein, Representing the wavelet transform denoising algorithm,Is the filtered concentration data;
Temperature data pretreatment: and carrying out fusion processing on the temperature data by adopting Kalman filtering to inhibit random noise and drift introduced by measurement errors or system drift, wherein the formula is as follows:
（2）
Wherein, Representing the kalman filter algorithm,For the filtered temperature data of the reaction chamber,The temperature of the water inlet after filtering.
Preferably, the construction of the particle concentration model and the temperature model in the reaction chamber is specifically that:
real-time concentration data after pretreatment Representing at timeConcentration distribution of small particles in the solution at the moment; adopting an infinite element modeling method based on data to disperse the whole solution flow field into a limited number of small units, and dispersing the solution flow field into each small unitThe concentration evolution of small particles satisfies the partial differential equation:
（3）
Wherein, As an operator of the gradient,Is time ofTime unitThe concentration distribution of the medium and small particles,Is time ofTime unitThe particle flux in (a);
Particle flux Comprising both diffusion and convection effects, thus, particle fluxCan be expressed as a superposition of these two partial effects:
（4）
Wherein, For the diffusion coefficient, the diffusion rate of small particles in solution is described,Is the flow field velocity;
Will be The flux summation of all small units at the moment can be obtainedOverall rate of change of concentration throughout the flow field at time:
（5）
By combining real-time concentration data Substituting the above equation and combining flow field velocitySolving by a numerical method to obtain the system at any timeConcentration profile field of (2)。
Preferably, the particle concentration model and the temperature model in the reaction chamber are constructed, wherein the temperature model is specifically built by:
real-time concentration data after pretreatment Representing at timeTemperature distribution of the solution at the moment; the solution temperature is affected by two thermal effects, including the volumetric flow rate of pure waterThe convection heat transfer and the heat conduction of certain chemical reaction inside the solution are brought about; establishing an energy conservation equation describing the dynamic change of the temperature field based on the effects of these two thermal effects:
（6）
Wherein, In order to achieve the density of the solution,Is the specific heat capacity of the solution,For the thermal conductivity of the solution,Is thatAn external heat source item at a moment;
External heat source items come from two parts: chemical reaction heat conduction inside the solution at the moment AndPure water volume flow at momentConvective heat transfer by：
（7）
Wherein,For the volume flow rate in the reaction chamber,The inlet temperature of the water inlet; adding the above two parts to obtainExternal heat source item of moment of time：
（8）
Temperature data in the reaction chamberAnd、Substituting the above equation and combining flow field velocitySolving by a numerical method to obtain the system at any timeTemperature distribution field of (2)。
Preferably, a simulation environment is constructed based on a particle concentration model and a temperature model, and a control strategy which best meets a control target is obtained, specifically:
taking the established particle concentration model and temperature model as the basis, and digitizing the particle concentration model and the temperature model into a partial differential equation set to serve as a simulation model:
（9）
Wherein, For the amount of temperature compensation,AndThe diffusion coefficients corresponding to the concentration field and the temperature field respectively,AndRespectively as control variablesAndAn influence function on concentration and temperature; the partial differential equation is solved on a computer through a numerical method;
Determining control variables and optimization targets: for control requirements, the control variables are naturally selected as two main parameters affecting the concentration and temperature of the solution, namely the pure water flow rate And temperature compensation amount; Meanwhile, constraint conditions are set:
setting a maximum concentration threshold: setting an allowable maximum concentration threshold ; The current concentration field is greater than the maximum concentration thresholdBy controlling the flow rate of pure water at the timeDiluting the particle concentration by adding an appropriate amount of pure water into the reaction chamber so that the temperature field is maintained at a maximum concentration thresholdUnder that, the stability of the chemical solution in the reaction chamber is ensured;
Setting a temperature control range: setting a temperature control range according to the working temperature range of the ultrasonic sensor Temperature compensation is adopted to avoid the influence of temperature on the measurement of the ultrasonic sensor;
On the basis of the above, searching the optimal control strategy in the simulation environment.
Preferably, the partial differential equation is solved on a computer by a numerical method, and the specific flow is as follows:
discretizing: discretizing the continuous variable in the partial differential equation into discrete points on grid points, including discretizing the concentration and temperature fields in time and space;
numerical approximation: according to the discretized equation, converting the partial differential equation into an algebraic equation set by adopting a numerical approximation method;
Solving algebraic equation sets: solving the algebraic equation set obtained after discretization by using a numerical method for solving the algebraic equation set to obtain numerical solutions of solution concentration and temperature at different time and space positions;
Based on the result of the numerical simulation, determining a control variable and an optimization target, and searching for an optimal pure water flow rate by using an optimization algorithm And temperature compensation amountThe combination ensures that the concentration and the temperature of the solution in the reaction process can meet the requirements and achieve the optimal reaction effect.
Preferably, the searching for the optimal control strategy in the simulation environment specifically includes:
Initializing pure water flow rate And temperature compensation amountIs a value of (2);
Flow rate of pure water to be initialized And temperature compensation amountSubstituting into a simulation model (9), and solving by a numerical method to obtain a solution concentration fieldAnd a temperature field；
From the resulting solution concentration fieldAnd a temperature fieldChecking whether the constraint condition is satisfied;
if the constraint condition is not satisfied, updating the pure water flow rate according to the optimization algorithm And temperature compensation amount；
Repeating the steps until the constraint conditions are met and the optimization target reaches an acceptable minimum value;
Outputting an optimal control strategy: pure water flow rate And temperature compensation amount。
Preferably, the building of the controller model based on reinforcement learning specifically includes:
in order to obtain an optimal result, a state space and an action space are established, and a state transition process of the environment is determined; by using Representing time stepsIn the state of time, the time-dependent state,Representing time stepsThe action of the time is that,Is shown in the stateExecuting an actionThe instant rewards obtained later; state spaceFrom the concentration field in the reaction chamberAnd a temperature fieldThe constitution is expressed as a vector:
（10）
The action space represents a collection of actions taken by the controller, typically expressed as either a continuous space or a discrete space; action The opening degree of a valve for controlling the flow rate of pure water and the power of a temperature controller for controlling the temperature compensation amount; the opening of the valve controlling the flow of pure water and the power of the temperature controller are represented by a continuous value ranging from 0 to 1, wherein 0 means that the valve is completely closed or the temperature controller is completely inoperative, and 1 means that the valve is completely opened or the temperature controller is operated at the maximum power;
according to the set state space and action space, the controller environment is modeled as follows:
（11）
Wherein, As a state transfer function of the environment,Is a random disturbance in the environment.
Preferably, the training process of the controller model includes:
Designing a reward function to guide the behavior of the reinforcement learning controller; reward function Consistent with the optimization target; the optimization objective is to maintain the optimal pure water flow rateAnd temperature compensation amountThus, the bonus function is defined as:
（12）
Wherein, To balance the flow rate of pure waterAnd optimal pure water flow rateIs used for the control of the temperature of the liquid crystal display device,To balance the temperature compensation in the reaction chamberAnd an optimal temperature compensation amountParameters of (2);
constructing a Q value function to learn an action value function so as to guide the selection of actions; the Q-value function is expressed in state Execute action downwardsIs based on the expected jackpot, i.e. the Q-value functionAnd updated; controller environment model based on Q value functionTraining is performed by defining a loss function through each experience tupleIs calculated from the expected value of the samples of (a),Is the next state; the loss function is defined as:
（13）
Wherein, The desired operator is indicated as being a function of the user,As a discount factor, the number of times the discount is calculated,For parameters of the target network, for calculating a target Q value, i.e. a future expected jackpot,Is a parameter of the primary network for evaluating the Q value of taking different actions in a given state.
Preferably, the controller model is trained based on a loss function, and the specific steps are as follows:
s1, initializing experience tuples Parameters of the primary networkAnd target network parametersState space；
S2, selecting actions according to the current Q network；
S3, executing actions in the environmentObserving the next stateObtain instant rewards;
S4, willStoring the experience tuples;
s5, sampling batch data from the experience tuples, and calculating the loss of the batch data according to a formula (13);
S6, updating the Q value network parameters by using random gradient descent, and carrying out state Is updated according to the update of (a);
s7, repeating the steps S2-S7, and when the total rewards of the model are not obviously improved, proving that the model tends to converge and finishing training;
after training, a controller model is obtained ：
（14）
Wherein,Is the current state, comprising a concentration field and a temperature field in the reaction chamber; To be in the current state The action of lower selection, namely the opening degree of the valve and the power of the temperature controller; the controller model parameters obtained for training, namely the trained deep Q network parameters.
Compared with the prior art, the invention has the following beneficial effects:
1. The limitations of the traditional method are overcome through the data-driven modeling and reinforcement learning technology, and the automatic and efficient monitoring and control of the concentration of the solution particles on various photoresist processes and production line platforms are realized. By optimally controlling the flow rate of pure water And temperature compensation amountResidual particles are removed to the maximum extent, and mechanical damage and chemical pollution are avoided.
2. The reaction quality is stable and controllable. By monitoring particle concentration data in real timeAnd temperature dataThe control parameters are quickly adjusted to ensure that the reaction temperature is controlled in a reasonable range on various processes and platformsAnd the particles are thoroughly removed, so that the chip yield is improved.
3. Flexible adaptation to different process conditions. The mathematical model constructed by the invention can describe the dynamic change of the concentration and the temperature of particles in the reaction process, and is not limited by specific process parameters. The reinforcement learning controller is trained in a simulation environment, so that the reinforcement learning controller can learn and adapt to optimal control strategies under working conditions of different initial concentrations, temperatures, flow rates and the like, has strong universality and robustness, and can be widely applied to various process links and machine platforms, wherein the process links comprise dry film development, liquid photoresist development, solder resist development, dry film stripping and liquid photoresist stripping, and the machine platforms comprise vertical continuous lines, horizontal continuous lines and single-plate multi-process treatment.
Drawings
FIG. 1 is a schematic block diagram of a method for monitoring and controlling the concentration of particles in a reaction solution according to the present invention.
FIG. 2 is a schematic diagram of two system environments to which the present invention is applicable.
Detailed Description
The invention will be further illustrated with reference to examples. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The specific principle of the invention is shown in figure 1:
The invention relates to a process links of dry/wet film development, solder resist development or dry/wet film stripping and the like for treating photoresist, which generally adopts two structures of a pipeline inflow system and a spraying system, as shown in fig. 2, the internal circulation system can be used for carrying out disturbance to accelerate the reaction speed; taking the temperature and the concentration of solution particles as optimal control targets, and ensuring the detection precision of the ultrasonic sensor through temperature compensation; the method specifically comprises the following steps:
Collecting real-time particle concentration data of inner wall of inner circulation pipeline Solution temperature data at ultrasonic sensorTemperature data at water inletAnd flow rate data of pure water in water inletPreprocessing the acquired data;
constructing a particle concentration model and a temperature model in a reaction chamber, and observing dynamic changes in the reaction process;
Constructing a simulation environment based on the particle concentration model and the temperature model to obtain a control strategy which optimally meets a control target, namely, an optimal set value of the opening degree of the water valve and the temperature compensation quantity;
based on the simulation environment, establishing a controller model based on reinforcement learning, and training the controller model in the simulation environment to obtain an initialization model;
The initialization model is deployed into a controller of the system, and further training and application are performed in a real environment, so that efficient closed-loop particle concentration monitoring and pure water flow rate control are realized.
1. Data acquisition
In the application scene of the semiconductor manufacturing process, the inner wall of the pipeline of the internal circulation system and the water inlet are provided with required sensor equipment, including an ultrasonic concentration sensor, a high-precision temperature sensor and an electromagnetic flowmeter. These sensors are used to monitor key parameters in the reaction process in real time. In the whole reaction chamber system, the solution in the reaction chamber, namely the solution for short, comprises pure water flowing into a water inlet and quantitative chemical reagent.
Particle concentration data acquisition: the ultrasonic sensor is contacted with the solution and emits ultrasonic waves, the concentration of particles is calculated according to the reflection condition of the ultrasonic waves, and the concentration data is recorded and updated with time, and is recorded as；
And (3) temperature data acquisition: in the reaction process, certain temperature change occurs to influence the measurement accuracy of the ultrasonic sensor, so that temperature compensation is needed, the first temperature sensor is arranged on the inner wall of the inner circulation pipeline to measure the temperature change of the ultrasonic sensor, and the temperature data are updated in real time along with the time and recorded as; Meanwhile, a second temperature sensor is used for collecting water inlet temperature data, which is recorded as; Wherein, in the pipeline inflow system, the second temperature sensor is installed at the inlet of pure water into the reaction chamber, in the spraying system, the second temperature sensor is installed at the pipeline of pure water into the water spraying shower;
And (3) collecting pure water flow rate data: the first electromagnetic flowmeter is used for measuring the flow rate of pure water in the water inlet, and the installation positions of the first electromagnetic flowmeter and the second electromagnetic flowmeter are the same; by electromagnetic induction principle, measuring the flow rate data of pure water flowing through the water inlet, and recording as At the same time, the time is acquired by the second electromagnetic flowmeterThe velocity of the flow field at the moment, i.e. the flow velocity of the solution in the whole reaction chamber, is recorded as。
2. Data preprocessing
Particle concentration data pretreatment of ultrasonic collection: denoising the particle concentration data by adopting wavelet transformation to eliminate high-frequency noise interference introduced by measuring equipment or environment; the wavelet transform decomposes the signal into components of different frequencies, and then retains useful signal information by selectively removing high frequency noise, as the formula:
（1）
Wherein, Representing the wavelet transform denoising algorithm,Is the filtered concentration data;
Temperature data pretreatment: the temperature data is fused by adopting Kalman filtering to inhibit random noise and drift caused by measurement errors or system drift, the Kalman filtering is a recursive filtering technology, and can provide optimal state estimation by carrying out joint estimation on a system dynamic model and a measured value, and the formula is as follows:
（2）
Wherein, Representing the kalman filter algorithm,For the filtered temperature data of the reaction chamber,The temperature of the water inlet after filtering.
3. Construction of particle concentration and temperature models in reaction chambers
The particle concentration model is established specifically as follows:
real-time concentration data after pretreatment Representing at timeConcentration profile of small particles in solution at the moment. Considering the complex dynamic behavior of small particles under the actions of water flow scouring, centrifugal force, gravity and the like, an analytical model cannot be simply established. Therefore, the invention adopts an infinite element modeling method based on data to disperse the whole solution flow field into a limited small unit. At each small unitThe concentration evolution of small particles satisfies the partial differential equation:
（3）
Wherein, As an operator of the gradient,Is time ofTime unitThe concentration distribution of the medium and small particles,Is time ofTime unitThe particle flux in (a);
Particle flux Comprises two parts:
diffusion effect: the diffusion of particles due to the concentration gradient, i.e. the particles diffuse along the concentration gradient from a high concentration region to a low concentration region;
convection effect: the convective motion of the particles due to the velocity of the flow field, i.e., the particles move with the fluid flow.
Thus, particle fluxCan be expressed as a superposition of these two partial effects:
（4）
Wherein, For the diffusion coefficient, the diffusion rate of small particles in solution is described,Is the flow field velocity;
Will be The flux summation of all small units at the moment can be obtainedOverall rate of change of concentration throughout the flow field at time:
（5）
By combining real-time concentration data Substituting the above equation and combining flow field velocitySolving by a numerical method to obtain the system at any timeConcentration profile field of (2). The infinite element modeling method based on the data can better consider the movement and distribution condition of small particles in a complex flow field, thereby more accurately describing the change of the concentration of the small particles in the solution.
The temperature model is established specifically as follows:
real-time concentration data after pretreatment Representing at timeTemperature profile of the solution at the moment. The solution temperature is affected by two thermal effects, including the volumetric flow rate of pure water(The volume of pure water passing through the water inlet per unit time can be controlled by the flow rate of pure water)Calculated by multiplying the cross-sectional area of the water inlet); the solution generates heat conduction with certain chemical reaction inside. Establishing an energy conservation equation describing the dynamic change of the temperature field based on the effects of these two thermal effects:
（6）
Wherein, In order to achieve the density of the solution,Is the specific heat capacity of the solution,For the thermal conductivity of the solution,Is thatAn external heat source item at a moment;
External heat source items come from two parts: chemical reaction heat conduction inside the solution at the moment AndPure water volume flow at momentConvective heat transfer by：
（7）
Wherein,For the volume flow rate in the reaction chamber,The inlet temperature of the water inlet; adding the above two parts to obtainExternal heat source item of moment of time：
（8）
Temperature data in the reaction chamberAnd、Substituting the above equation and combining flow field velocitySolving by a numerical method to obtain the system at any timeTemperature distribution field of (2). The model based on energy conservation can describe the dynamic change of the temperature in the solution more accurately, thereby providing reliable temperature information for an intelligent control system.
4. Obtaining a control strategy that best meets control objectives
(1) Constructing a simulation model of the reaction process: in order to evaluate and optimize the effects of different control strategies in a virtual environment, a simulated mathematical model of the semiconductor wafer reaction process needs to be built. Taking the established particle concentration model and temperature model as the basis, and digitizing the particle concentration model and the temperature model into a partial differential equation set to serve as a simulation model:
（9）
Wherein, For the amount of temperature compensation,AndThe diffusion coefficients corresponding to the concentration field and the temperature field respectively,AndRespectively as control variablesAndInfluence function on concentration and temperature. The partial differential equation can be solved on a computer by a numerical method, and the specific flow is as follows:
Discretizing: the continuous variable in the partial differential equation is discretized into discrete points on the grid points. This includes discretizing the concentration and temperature fields in time and space.
Numerical approximation: and according to the discretized equation, converting the partial differential equation into an algebraic equation set by adopting a numerical approximation method.
Solving algebraic equation sets: and solving the algebraic equation set obtained after discretization by using a numerical method for solving the algebraic equation set to obtain numerical solutions of solution concentration and temperature at different time and space positions.
Based on the result of the numerical simulation, determining a control variable and an optimization target, and searching for an optimal pure water flow rate by using an optimization algorithmAnd temperature compensation amountThe combination ensures that the concentration and the temperature of the solution in the reaction process can meet the requirements and achieve the optimal reaction effect.
(2) Determining control variables and optimization targets: for control requirements, the control variables are naturally selected as two main parameters affecting the concentration and temperature of the solution, namely the pure water flow rateAnd temperature compensation amount. Meanwhile, constraint conditions are set to ensure the reaction effect and the stability of the system:
setting a maximum concentration threshold: setting an allowable maximum concentration threshold . The current concentration field is greater than the maximum concentration thresholdBy controlling the flow rate of pure water at the timeDiluting the particle concentration by adding an appropriate amount of pure water into the reaction chamber so that the temperature field is maintained at a maximum concentration thresholdIn the meantime, stability of the chemical solution in the reaction chamber is ensured.
Setting a temperature control range: setting a temperature control range according to the working temperature range of the ultrasonic sensorTo avoid temperature effects on the sensor measurements. The ultrasonic sensor may introduce measurement errors at too high or too low a temperature, so that by setting a suitable temperature range, the accuracy and stability of the sensor can be ensured.
(3) Searching an optimal control strategy in a simulation environment:
Initializing pure water flow rate And temperature compensation amountIs a value of (2);
Flow rate of pure water to be initialized And temperature compensation amountSubstituting into a simulation model (9), and solving by a numerical method to obtain a solution concentration fieldAnd a temperature field；
From the resulting solution concentration fieldAnd a temperature fieldChecking whether the constraint condition is satisfied;
if the constraint condition is not satisfied, updating the pure water flow rate according to the optimization algorithm And temperature compensation amount；
Repeating the steps until the constraint conditions are met and the optimization target reaches an acceptable minimum value;
Outputting an optimal control strategy: pure water flow rate And temperature compensation amount。
Through the process based on mathematical model simulation and optimization, the pure water flow rate which can meet the reaction requirement and has optimal effect is obtained through searchingAnd temperature compensation amountAnd (5) combining.
5. Establishing a reinforcement learning-based controller model
And according to the obtained output optimal control strategy: pure water flow rateAnd temperature compensation amountA controller model for controlling the flow rate of pure water and the temperature compensation amount is established based on deep reinforcement learning. In deep reinforcement learning, in order to obtain an optimal result, it is necessary to establish a state space and an action space and determine a state transition process of the environment. By usingRepresenting time stepsIn the state of time, the time-dependent state,Representing time stepsThe action of the time is that,Is shown in the stateExecuting an actionThe instant rewards obtained later. State spaceFrom the concentration field in the reaction chamberAnd a temperature fieldThe constitution is expressed as a vector:
（10）
the action space represents a collection of actions taken by the controller, typically expressed as either a continuous space or a discrete space. In the invention, the actions The opening degree of the valve for controlling the flow rate of pure water and the power of the temperature controller for controlling the temperature compensation amount are expressed. In general, the opening of the valve controlling the flow of pure water and the power of the temperature controller can be represented by a continuous value ranging from 0 to 1, where 0 means that the valve is completely closed or the temperature controller is completely inoperative and 1 means that the valve is completely opened or the temperature controller is operated at maximum power.
According to the set state space and action space, the controller environment is modeled as follows:
（11）
Wherein, As a state transfer function of the environment,Is a random disturbance in the environment.
6. Training of controller models
Based on the established controller environment model, a reward function is designed to guide the behavior of the reinforcement learning controller. Reward functionShould be consistent with the optimization objective. In this invention, the optimization objective is to maintain an optimal pure water flow rateAnd temperature compensation amount. Thus, the bonus function is defined as:
（12）
Wherein, To balance the flow rate of pure waterAnd optimal pure water flow rateIs used for the control of the temperature of the liquid crystal display device,To balance the temperature compensation in the reaction chamberAnd an optimal temperature compensation amountIs a parameter of (a).
A Q-value function is constructed to learn the action value function to guide the selection of actions. The Q-value function is expressed in stateExecute action downwardsIs based on the expected jackpot, i.e. the Q-value functionAnd updated. Controller environment model based on Q value functionTraining is performed by defining a loss function through each experience tupleIs calculated from the expected value of the samples of (a),The next state. The loss function is defined as:
（13）
Wherein, The desired operator is indicated as being a function of the user,As a discount factor, the number of times the discount is calculated,For parameters of the target network, for calculating a target Q value, i.e. a future expected jackpot,Is a parameter of the primary network for evaluating the Q value of taking different actions in a given state.
Training a controller model based on the loss function, wherein the specific steps are as follows:
1) Initializing experience tuples Parameters of the primary networkAnd target network parametersState space；
2) Selecting actions based on current Q network；
3) Performing actions in an environmentObserving the next stateObtain instant rewards;
4) Will beStoring the experience tuples;
5) Sampling the batch data from the experience tuple and calculating a loss of the batch data according to equation (13);
6) Updating Q-value network parameters using random gradient descent and performing state Is updated according to the update of (a);
7) Repeating the steps 2) -6), when the total rewards of the model are not obviously improved, proving that the model tends to be converged, and finishing training.
After training, a controller model is obtained：
（14）
Wherein,Is the current state, comprising a concentration field and a temperature field in the reaction chamber; To be in the current state The action of lower selection, namely the opening degree of the valve and the power of the temperature controller; the controller model parameters obtained for training, namely the trained deep Q network parameters.
7. Overall control of the operating process
The trained controller is deployed into a reaction system to realize high-efficiency closed-loop particle concentration monitoring and temperature control, and the process comprises the following steps:
the controller deployment, namely deploying the intelligent controller subjected to reinforcement learning training into a total control unit of the whole reaction system;
in the reaction process, the controller needs to acquire current state data of the system in real time through a sensor, and obtains state information of a particle concentration field and a temperature field of the solution in the reaction chamber according to the state data and the modeling type;
The motion selection, namely obtaining the output optimal pure water flow rate and the temperature compensation quantity based on the simulation model according to the current state information, and selecting the optimal motion by using the trained model by the controller;
Executing control, namely sending the selected action value to a corresponding executing mechanism by the controller: the opening of the valve controlling the flow rate of pure water and the heating power of the heater controlling the temperature compensation amount, thereby implementing the control of the reaction process;
And (3) feeding back and adjusting, wherein in the control process, the sensor continuously feeds back state information of the concentration field and the temperature field of the solution particles. The controller evaluates the deviation between the current state and the expected target and performs strategy adjustment according to feedback so as to continuously optimize the control performance;
Continuous operation: the controller can continuously circulate the process, select actions according to the real-time state, execute control instructions, acquire feedback and adjust strategies, and therefore automatic closed-loop control is achieved. By the mode, the system can automatically adjust key parameters, keep the concentration and temperature of solution particles in a reasonable range, and improve monitoring precision and control efficiency.
The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.
While the foregoing describes the embodiments of the present invention, it should be understood that the present invention is not limited to the embodiments, and that various modifications and changes can be made by those skilled in the art without any inventive effort.
Claims (10)
1. A method for monitoring and controlling the concentration of particles in a reaction solution, comprising the steps of:
The temperature sensor and the ultrasonic sensor are simultaneously arranged at the adjacent positions of the inner circulation pipeline wall based on the pipeline inflow system or the spraying system, the flow rate of the chemical reagent is fixed, and the flow rate of pure water is controlled while the temperature data of the reaction solution and the concentration data of solution particles are monitored; taking the temperature and the concentration of solution particles as optimal control targets, and ensuring the detection precision of the ultrasonic sensor through temperature compensation;
According to a control strategy meeting an optimal control target, namely an optimal set value of the opening degree of the water valve and the temperature compensation quantity, a controller model based on reinforcement learning is established, and the controller model specifically comprises:
in order to obtain an optimal result, a state space and an action space are established, and a state transition process of the environment is determined; by using Representing time stepsIn the state of time, the time-dependent state,Representing time stepsThe action of the time is that,Is shown in the stateExecuting an actionThe instant rewards obtained later; state spaceFrom the concentration field in the reaction chamberAnd a temperature fieldThe constitution is expressed as a vector:
（10）
The action space represents a set of actions taken by the controller, expressed as either a continuous space or a discrete space; action The opening degree of a valve for controlling the flow rate of pure water and the power of a temperature controller for controlling the temperature compensation amount; the opening of the valve controlling the flow of pure water and the power of the temperature controller are represented by a continuous value ranging from 0 to 1, wherein 0 means that the valve is completely closed or the temperature controller is completely inoperative, and 1 means that the valve is completely opened or the temperature controller is operated at the maximum power;
according to the set state space and action space, the controller environment is modeled as follows:
（11）
Wherein, As a state transfer function of the environment,Is a random disturbance in the environment;
The training process of the controller model comprises the following steps:
Designing a reward function to guide the behavior of the reinforcement learning controller; reward function Consistent with the optimization target; the optimization objective is to maintain the optimal pure water flow rateAnd temperature compensation amountThus, the bonus function is defined as:
（12）
Wherein, To balance the flow rate of pure waterAnd optimal pure water flow rateIs used for the control of the temperature of the liquid crystal display device,To balance the temperature compensation in the reaction chamberAnd an optimal temperature compensation amountParameters of (2);
constructing a Q value function to learn an action value function so as to guide the selection of actions; the Q-value function is expressed in state Execute action downwardsIs based on the expected jackpot, i.e. the Q-value functionAnd updated; controller environment model based on Q value functionTraining is performed by defining a loss function through each experience tupleIs calculated from the expected value of the samples of (a),In order to be in the next state,Is the next action; the loss function is defined as:
（13）
Wherein, The desired operator is indicated as being a function of the user,As a discount factor, the number of times the discount is calculated,For parameters of the target network, for calculating a target Q value, i.e. a future expected jackpot,Parameters of the main network for evaluating the Q values of different actions taken in a given state;
The controller model is deployed into a controller of the whole system, and training and application are carried out on the controller model, so that the particle concentration monitoring and the pure water flow rate control of the high-efficiency closed loop are realized.
2. A method of monitoring and controlling the concentration of particles in a reaction solution as set forth in claim 1, further comprising the process of:
Collecting real-time particle concentration data of inner wall of inner circulation pipeline Solution temperature data at ultrasonic sensorTemperature data at water inletAnd flow rate data of pure water in water inletPreprocessing the acquired data;
constructing a particle concentration model and a temperature model in a reaction chamber, and observing dynamic changes in the reaction process;
Constructing a simulation environment based on the particle concentration model and the temperature model to obtain a control strategy which optimally meets a control target, namely, an optimal set value of the opening degree of the water valve and the temperature compensation quantity;
training the controller model in a simulation environment based on the reinforcement-learning controller model to obtain an initialization model;
The initialization model is deployed into a controller of the system, and further training and application are performed in a real environment, so that efficient closed-loop particle concentration monitoring and pure water flow rate control are realized.
3. A method of monitoring and controlling the concentration of particles in a reaction solution as claimed in claim 2 wherein the collecting data comprises:
Particle concentration data acquisition: the ultrasonic sensor is contacted with the solution and emits ultrasonic waves, the concentration of particles is calculated according to the reflection condition of the ultrasonic waves, and the concentration data is recorded and updated with time, and is recorded as ；
And (3) temperature data acquisition: a first temperature sensor for measuring temperature changes at the ultrasonic sensor, the temperature data being updated in real time over time and recorded as; Meanwhile, a second temperature sensor is used for collecting water inlet temperature data, which is recorded as；
And (3) collecting pure water flow rate data: the first electromagnetic flowmeter is used for measuring the flow rate of pure water in the water inlet; by electromagnetic induction principle, measuring the flow rate data of pure water flowing through the water inlet, and recording asAt the same time, the time is acquired by the second electromagnetic flowmeterThe velocity of the flow field at the moment, i.e. the flow velocity of the solution in the whole reaction chamber, is recorded as。
4. A method for monitoring and controlling the concentration of particles in a reaction solution according to claim 3, wherein the preprocessing of the collected data is performed by:
Particle concentration data pretreatment of ultrasonic collection: denoising the particle concentration data by adopting wavelet transformation to eliminate high-frequency noise interference introduced by measuring equipment or environment; the wavelet transform decomposes the signal into components of different frequencies, and then retains useful signal information by selectively removing high frequency noise, as the formula:
（1）
Wherein, Representing the wavelet transform denoising algorithm,Is the filtered concentration data;
Temperature data pretreatment: and carrying out fusion processing on the temperature data by adopting Kalman filtering to inhibit random noise and drift introduced by measurement errors or system drift, wherein the formula is as follows:
（2）
Wherein, Representing the kalman filter algorithm,For the filtered temperature data of the reaction chamber,The temperature of the water inlet after filtering.
5. The method for monitoring and controlling the concentration of particles in a reaction solution according to claim 4, wherein the construction of a particle concentration model and a temperature model in the reaction chamber is performed, and the construction of the particle concentration model is specifically as follows:
real-time concentration data after pretreatment Representing at timeConcentration distribution of small particles in the solution at the moment; adopting an infinite element modeling method based on data to disperse the whole solution flow field into a limited number of small units, and dispersing the solution flow field into each small unitThe concentration evolution of small particles satisfies the partial differential equation:
（3）
Wherein, As an operator of the gradient,Is time ofTime unitThe concentration distribution of the medium and small particles,Is time ofTime unitThe particle flux in (a);
Particle flux Comprising both diffusion and convection effects, thus, particle fluxCan be expressed as a superposition of these two partial effects:
（4）
Wherein, For the diffusion coefficient, the diffusion rate of small particles in solution is described,Is the flow field velocity;
Will be The flux summation of all small units at the moment can be obtainedOverall rate of change of concentration throughout the flow field at time:
（5）
By combining real-time concentration data Substituting the above equation and combining flow field velocitySolving by a numerical method to obtain the system at any timeConcentration profile field of (2)。
6. The method for monitoring and controlling the concentration of particles in a reaction solution according to claim 5, wherein a particle concentration model and a temperature model in the reaction chamber are constructed, and wherein the temperature model is constructed specifically as follows:
real-time concentration data after pretreatment Representing at timeTemperature distribution of the solution at the moment; the solution temperature is affected by two thermal effects, including the volumetric flow rate of pure waterThe convection heat transfer and the heat conduction of certain chemical reaction inside the solution are brought about; establishing an energy conservation equation describing the dynamic change of the temperature field based on the effects of these two thermal effects:
（6）
Wherein, In order to achieve the density of the solution,Is the specific heat capacity of the solution,For the thermal conductivity of the solution,Is thatAn external heat source item at a moment;
External heat source items come from two parts: chemical reaction heat conduction inside the solution at the moment AndPure water volume flow at momentConvective heat transfer by：
（7）
Wherein,For the volume flow rate in the reaction chamber,The inlet temperature of the water inlet; adding the above two parts to obtainExternal heat source item of moment of time：
（8）
Temperature data in the reaction chamberAnd、Substituting the above equation and combining flow field velocitySolving by a numerical method to obtain the system at any timeTemperature distribution field of (2)。
7. The method for monitoring and controlling the concentration of particles in a reaction solution according to claim 2, wherein a simulation environment is constructed based on a particle concentration model and a temperature model, and a control strategy which best meets a control target is obtained, specifically:
taking the established particle concentration model and temperature model as the basis, and digitizing the particle concentration model and the temperature model into a partial differential equation set to serve as a simulation model:
（9）
Wherein, For the amount of temperature compensation,AndThe diffusion coefficients corresponding to the concentration field and the temperature field respectively,AndRespectively as control variablesAndAn influence function on concentration and temperature; the partial differential equation is solved on a computer through a numerical method;
determining control variables and optimization targets: for the control requirements, the control variables are selected as two main parameters affecting the concentration and temperature of the solution, namely pure water flow rate And temperature compensation amount; Meanwhile, constraint conditions are set:
setting a maximum concentration threshold: setting an allowable maximum concentration threshold ; The current concentration field is greater than the maximum concentration thresholdBy controlling the flow rate of pure water at the timeDiluting the particle concentration by adding an appropriate amount of pure water into the reaction chamber so that the temperature field is maintained at a maximum concentration thresholdUnder that, the stability of the chemical solution in the reaction chamber is ensured;
Setting a temperature control range: setting a temperature control range according to the working temperature range of the ultrasonic sensor Temperature compensation is adopted to avoid the influence of temperature on the measurement of the ultrasonic sensor;
On the basis of the above, searching the optimal control strategy in the simulation environment.
8. The method for monitoring and controlling the concentration of particles in a reaction solution according to claim 7, wherein the partial differential equation is solved on a computer by a numerical method, and the specific flow is as follows:
discretizing: discretizing the continuous variable in the partial differential equation into discrete points on grid points, including discretizing the concentration and temperature fields in time and space;
numerical approximation: according to the discretized equation, converting the partial differential equation into an algebraic equation set by adopting a numerical approximation method;
Solving algebraic equation sets: solving the algebraic equation set obtained after discretization by using a numerical method for solving the algebraic equation set to obtain numerical solutions of solution concentration and temperature at different time and space positions;
Based on the result of the numerical simulation, determining a control variable and an optimization target, and searching for an optimal pure water flow rate by using an optimization algorithm And temperature compensation amountThe combination ensures that the concentration and the temperature of the solution in the reaction process can meet the requirements.
9. The method for monitoring and controlling the concentration of particles in a reaction solution according to claim 7, wherein the searching for an optimal control strategy in a simulation environment is specifically:
Initializing pure water flow rate And temperature compensation amountIs a value of (2);
Flow rate of pure water to be initialized And temperature compensation amountSubstituting into a simulation model (9), and solving by a numerical method to obtain a solution concentration fieldAnd a temperature field；
From the resulting solution concentration fieldAnd a temperature fieldChecking whether the constraint condition is satisfied;
if the constraint condition is not satisfied, updating the pure water flow rate according to the optimization algorithm And temperature compensation amount；
Repeating the steps until the constraint conditions are met and the optimization target reaches an acceptable minimum value;
Outputting an optimal control strategy: pure water flow rate And temperature compensation amount。
10. A method for monitoring and controlling the concentration of particles in a reaction solution according to claim 1, wherein the training of the controller model based on the loss function is performed by:
s1, initializing experience tuples Parameters of the primary networkAnd target network parametersState space；
S2, selecting actions according to the current Q network；
S3, executing actions in the environmentObserving the next stateObtain instant rewards;
S4, willStoring the experience tuples;
s5, sampling batch data from the experience tuples, and calculating the loss of the batch data according to a formula (13);
S6, updating the Q value network parameters by using random gradient descent, and carrying out state Is updated according to the update of (a);
s7, repeating the steps S2-S7, and when the total rewards of the model are not obviously improved, proving that the model tends to converge and finishing training;
after training, a controller model is obtained ：
（14）
Wherein,Is the current state, comprising a concentration field and a temperature field in the reaction chamber; To be in the current state The action of lower selection, namely the opening degree of the valve and the power of the temperature controller; the controller model parameters obtained for training, namely the trained deep Q network parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410725165.9A CN118331071B (en) | 2024-06-06 | 2024-06-06 | Method for monitoring and controlling concentration of particles in reaction solution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410725165.9A CN118331071B (en) | 2024-06-06 | 2024-06-06 | Method for monitoring and controlling concentration of particles in reaction solution |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118331071A CN118331071A (en) | 2024-07-12 |
| CN118331071B true CN118331071B (en) | 2024-08-09 |
Family
ID=91778037
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410725165.9A Active CN118331071B (en) | 2024-06-06 | 2024-06-06 | Method for monitoring and controlling concentration of particles in reaction solution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118331071B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1106359A (en) * | 1993-08-26 | 1995-08-09 | 皮奇尼铝公司 | Process for the production of trihydrate of alumina having controlled sodium content and grain size |
| CN110108610A (en) * | 2019-05-27 | 2019-08-09 | 三峡大学 | The measurement method of suspension middle particle concentration |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005091970A2 (en) * | 2004-03-06 | 2005-10-06 | Michael Trainer | Methods and apparatus for determining the size and shape of particles |
| CN116125814A (en) * | 2023-02-21 | 2023-05-16 | 浙江工业职业技术学院 | Optimal control method for conveying micro-channel colloid particles |
| CN116822346A (en) * | 2023-06-15 | 2023-09-29 | 北京工业大学 | Sewage treatment nitrate nitrogen concentration control method based on Q learning |
-
2024
- 2024-06-06 CN CN202410725165.9A patent/CN118331071B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1106359A (en) * | 1993-08-26 | 1995-08-09 | 皮奇尼铝公司 | Process for the production of trihydrate of alumina having controlled sodium content and grain size |
| CN110108610A (en) * | 2019-05-27 | 2019-08-09 | 三峡大学 | The measurement method of suspension middle particle concentration |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118331071A (en) | 2024-07-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | A practical approach to flow field reconstruction with sparse or incomplete data through physics informed neural network | |
| CN100476733C (en) | System and method for semiconductor simulation on tool | |
| CN1860487B (en) | System and method for using first-principles simulation to analyze a process performed by a semiconductor processing tool | |
| CN100568249C (en) | System and method with ultimate principle emulation semiconductor-assisted manufacture process | |
| TWI849667B (en) | Learned model generating method, abnormality factor estimating device, learned model, abnormality factor estimating method, learning method, learning device, and learning data generating method | |
| CN1867896A (en) | System and method for on-tool semiconductor simulation | |
| CN101976045A (en) | Panel quality virtual measurement method and system for TFT-LCD etching process | |
| CN117195747B (en) | Uniform heat distribution optimization method for magnetic material drying | |
| CN118331071B (en) | Method for monitoring and controlling concentration of particles in reaction solution | |
| CN114110824A (en) | Intelligent control method and device for constant humidity machine | |
| CN102129242A (en) | Product quality control method during batch processing production process based on two-layer hybrid intelligent optimization | |
| CN114912364B (en) | Natural gas well flow prediction method, device, equipment and computer readable medium | |
| Mao et al. | Coating thickness modeling and prediction for hot-dip galvanized steel strip based on GA-BP neural network | |
| Zhang et al. | Oil holdup prediction of oil–water two phase flow using thermal method based on multiwavelet transform and least squares support vector machine | |
| Ren et al. | Data-driven and mechanism-based hybrid model for semiconductor silicon monocrystalline quality prediction in the czochralski process | |
| Tang et al. | Gas flow-rate measurement using a transit-time multi-path ultrasonic flow meter based on PSO-SVM | |
| KR102393813B1 (en) | Deep learning-based semiconductor chemical liquid temperature control system | |
| Seraj et al. | Data-driven predictor and soft-sensor models of a cement grate cooler based on neural network and effective dynamics | |
| CN110705186B (en) | Real-time online instrument checksum diagnosis method through RBF particle swarm optimization algorithm | |
| CN116341399A (en) | Thermodynamic hydraulic heat exchange coefficient prediction method based on physical constraint neural network | |
| CN110750756B (en) | Real-time on-line instrument checksum diagnosis method through optimal support vector machine algorithm | |
| Yang et al. | Rid-noise: Towards robust inverse design under noisy environments | |
| Song et al. | Multivariate prediction of airflow and temperature distributions using artificial neural networks | |
| Dai et al. | A GODFIP control algorithm for an IRC grain dryer | |
| Xie et al. | Identification of dual‐rate sampled nonlinear systems based on the cycle reservoir with regular jumps network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |