A method and system for concentrate thickener reagent control

Abstract

The application discloses a thickener dosing control method and system, relates to mineral processing, and comprises the following steps: obtaining a concentration value measured by an i-th concentration sensor at different depths at time t; obtaining overflow water turbidity T, effective viscosity of overflow water or underflow, volume fraction of fine particles and volume fraction of coarse particles in overflow water, Zeta potential at a feeding end, Zeta potential at an overflow end and time derivative of the Zeta potential at the overflow end, and particle equivalent diameter d, particle density, fluid density and fluid viscosity; calculating an effective settling velocity of a particle group; performing online correction on the effective settling velocity by using a least square method or a Kalman filtering algorithm; performing time lag correction on the concentration values at different depths by using the corrected effective settling velocity; correcting the overflow water turbidity T according to the corrected concentration profile; and respectively calculating a flocculant dosing rate and a coagulant dosing rate; and the application improves the accurate and collaborative control of the flocculant dosing rate and the coagulant dosing rate by using an artificial intelligence model.

Description

Technical Field

[0001] This application relates to the field of mineral processing, and in particular to a method and system for controlling the dosing of chemicals in a concentrator. Background Technology

[0002] In mineral processing, thickening is a crucial step in achieving solid-liquid separation and is widely used in industries such as mineral processing, metallurgy, and chemicals. Thickeners separate solid particles from water in mineral slurry using the principle of gravity sedimentation, obtaining a high-concentration underflow and clarified overflow. To improve sedimentation efficiency and separation effect, flocculants and coagulants are typically added to the thickener. Flocculants primarily work by bridging coarse particles into larger agglomerates through the bridging effect of polymer molecules, accelerating sedimentation; coagulants, on the other hand, compress the electric double layer on the surface of fine particles through charge neutralization, promoting particle aggregation. The synergistic effect of these two agents is essential for achieving efficient solid-liquid separation.

[0003] Current technologies typically treat flocculants and coagulants as single agents for control, failing to fully consider the differences in their mechanisms of action: flocculants primarily act as bridging agents for coarse particles, while coagulants primarily act as charge neutralizing agents for fine particles. Due to the lack of real-time monitoring and differentiated response to key parameters such as particle size distribution (volume fraction of fine to coarse particles) and particle surface electrical properties (Zeta potential and its dynamic changes), precise and coordinated control of flocculant and coagulant dosing rates cannot be achieved, leading to unreasonable agent ratios and ineffective treatment of some particles.

[0004] Furthermore, existing technologies fail to fully consider the dynamic characteristics of particle settling, particularly the time differences required for particles to settle from the feed point to different depths. The concentration values ​​measured by the concentration sensors at different depths actually correspond to particle groups entering the thickener at different times. This time lag is not taken into account, resulting in distorted concentration profiles and making it difficult to correct the dosing strategy in a timely and accurate manner.

[0005] Furthermore, during the operation of the thickener, the properties of the feed (particle size distribution, mineral composition, etc.) and operating conditions (feed concentration, feed flow rate, etc.) are constantly changing, causing a deviation between the actual settling velocity of the particles and the theoretically calculated value. If the settling velocity cannot be corrected in real time, the control strategy based on the theoretical settling velocity will be inaccurate, affecting the dosing effect.

[0006] The aforementioned problems lead to waste of reagents, increased operating costs, unstable concentration effects, decreased quality of clarified water, and insufficient automation, making it difficult to meet the requirements of modern mineral processing for efficient, energy-saving, and intelligent control. Summary of the Invention

[0007] To address the lack of precise and coordinated control over the addition of flocculants and coagulants in existing technologies, this application provides a method and system for controlling the addition of flocculants and coagulants in a concentrator, thereby improving the precise and coordinated control of the addition rates of flocculants and coagulants.

[0008] One aspect of this application provides a method for controlling the dosing of chemicals in a concentrator, comprising: S1, obtaining different depths along the vertical direction of the concentrator. The concentration value measured by the i-th concentration sensor at time t And obtain the overflow turbidity T and the effective viscosity of the overflow or bottom flow. Volume fraction of fine particles in overflow water With coarse particle volume fraction Zeta potential at the feed end Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential And the equivalent diameter d of the particles, particle density Fluid density and fluid viscosity ;

[0009] S2, based on the equivalent particle diameter d and particle density Fluid density and fluid viscosity Calculate the effective settling velocity of the particle group. S3, using the least squares method or Kalman filtering algorithm, based on the concentration distribution curves measured by multiple concentration sensors over time, the effective sedimentation velocity is determined. Perform online correction to obtain the corrected effective settlement velocity. S4, utilizing the corrected effective settlement velocity For different depths Concentration value at Perform time lag correction to obtain the corrected concentration profile; correct the overflow turbidity T based on the corrected concentration profile;

[0010] S5, based on the corrected overflow turbidity Effective viscosity of overflow or underflow Volume fraction of fine particles in overflow water With coarse particle volume fraction Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential Calculate the flocculant dosing rate separately. and coagulant dosing rate ;

[0011] S6, based on the flocculant dosage rate and coagulant dosing rate The dosing of flocculants and coagulants is controlled by a dosing pump and an electric valve.

[0012] The overflow turbidity (T) refers to the content of suspended solid particles in the clarified water discharged from the overflow outlet at the top of the thickener. It is measured by a turbidity sensor and is expressed in NTU (turbidity unit) or mg / L. Overflow turbidity is an important indicator for evaluating the solid-liquid separation effect of the thickener. The lower the turbidity, the clearer the overflow water and the better the solid-liquid separation effect. Increased turbidity usually indicates that fine particles have not settled effectively or that the reagent dosage is insufficient.

[0013] Effective viscosity Effective viscosity refers to the apparent viscosity of overflow or underflow slurry, reflecting the flow resistance of the slurry under shear stress, and is measured in mPa·s or Pa·s. Effective viscosity is affected by factors such as particle concentration, particle size distribution, slurry temperature, and reagent addition. High-concentration slurry and the addition of polymeric flocculants significantly increase effective viscosity, thereby affecting particle settling velocity and the separation performance of the thickener. It is measured using an online viscometer.

[0014] Fine particle volume fraction This refers to the volume fraction of fine particles smaller than a certain critical value (usually 10 to 20 μm) in overflow water. It is dimensionless and ranges from 0 to 1. Fine particles, due to their small size, large specific surface area, and slow settling velocity, are easily discharged with overflow water and are a major cause of increased overflow water turbidity. The volume fraction of fine particles is measured using a laser particle size analyzer or an online particle size analyzer.

[0015] coarse particle volume fraction This refers to the volume fraction of coarse particles larger than a certain critical value (usually 20 to 50 μm) in the overflow water. It is dimensionless and ranges from 0 to 1. Due to their large size and fast settling velocity, coarse particles should normally settle to the bottom of the thickener and be discharged with the underflow. An abnormally high volume fraction of coarse particles in the overflow water indicates insufficient flocculant dosage or poor flocculation, causing coarse particles to fail to effectively aggregate into large particle clusters and be carried out by the overflow. The volume fraction of coarse particles is measured using a laser particle size analyzer or an online particle size analyzer.

[0016] Zeta potential at the feed end The zeta potential refers to the electrostatic potential on the surface of particles in the slurry at the thickener inlet. It reflects the surface charge state and double-layer thickness of the particles, and is measured in mV. A larger absolute value of the zeta potential indicates more surface charge on the particles, stronger electrostatic repulsion between particles, and greater difficulty in particle aggregation. A smaller absolute value of the zeta potential (closer to zero) indicates that the surface charge on the particles is neutralized, the electrostatic repulsion between particles is weakened, and particle aggregation is facilitated. It is measured using a zeta potential analyzer.

[0017] Overflow terminal Zeta potential The zeta potential refers to the electrokinetic potential of residual particles on the surface of the overflow water at the overflow port of the thickener, measured in mV. The zeta potential at the overflow end reflects the neutralization effect of the agent (especially the coagulant) on the surface charge of the particles. If the absolute value of the zeta potential at the overflow end is still large, it indicates insufficient coagulant dosage, and the surface charge of the fine particles is not sufficiently neutralized, causing the fine particles to be difficult to coagulate and settle, and thus discharged with the overflow. It is measured using a zeta potential analyzer.

[0018] Time derivative of the overflow terminal Zeta potential The rate of change of the overflow terminal Zeta potential over time is expressed in mV / s or mV / min. This reflects the dynamic trend of changes in the electrical properties of the particle surface. A positive value indicates an increase in negative charge (or a decrease in positive charge) on the particle surface, which may be due to a weakening of the coagulant effect or a change in the properties of the feed. A negative value indicates a decrease in negative charge (or an increase in positive charge) on the particle surface, which may be due to the coagulant playing a role. The magnitude of the electrical properties reflects the rate of change and can predict the dynamic trend of the drug's effect.

[0019] The equivalent diameter is assumed to be spherical and is expressed in μm or mm. For non-spherical particles, the equivalent diameter can be either the volume equivalent diameter (the diameter of a sphere with the same volume as the particle) or the sedimentation equivalent diameter (the diameter of a sphere with the same sedimentation velocity). The particle equivalent diameter is a fundamental parameter for calculating particle sedimentation velocity. It is measured using methods such as laser particle size analyzer, sieve analysis, or microscopic image analysis, and is usually represented by the median particle size d50 or the average particle size of the particle group.

[0020] Particle density Particle density refers to the density of solid particles, measured in kg / m³ or g / cm³. It is an inherent physical property of minerals, and different minerals have significantly different densities. For example, quartz has a density of approximately 2650 kg / m³, magnetite approximately 5150 kg / m³, and chalcopyrite approximately 4200 kg / m³. Particle density is measured using a density bottle method, a specific gravity bottle method, or a hydrometer, or it can be calculated based on the chemical analysis results of the mineral's composition.

[0021] fluid density This refers to the density of the liquid phase (usually water or an aqueous solution containing dissolved salts) in a slurry, expressed in kg / m³ or g / cm³. The density of pure water at room temperature and pressure is approximately 1000 kg / m³. Fluid density is affected by temperature and dissolved substances and is measured using a densitometer or online density sensor. The difference between fluid density and particle density is also considered. It is the power source that drives the gravitational settling of particles.

[0022] fluid viscosity Dynamic viscosity refers to the dynamic viscosity of the liquid phase in a slurry, reflecting the frictional resistance between molecules within the fluid, and is measured in mPa·s or Pa·s. The viscosity of pure water at 20°C is approximately 1.0 mPa·s. Fluid viscosity decreases with increasing temperature and is also affected by dissolved substances. Fluid viscosity is measured using a viscometer and is an important parameter for calculating particle settling resistance. Note the distinction between fluid viscosity and dynamic viscosity. (Referring only to liquid phase viscosity) and effective viscosity (Refers to the overall viscosity of a slurry containing solid particles).

[0023] A particle cluster refers to an aggregate of solid particles within a thickener that possesses a specific particle size distribution, density distribution, and physicochemical properties. A particle cluster is not a single particle, but rather the totality of a batch of particles entering the thickener from the feed inlet at a given moment. These particles maintain relative independence during the settling process. The particle cluster concept in this application describes the time lag characteristic of the settling process: particles entering in the same batch reach different depths at different times; therefore, concentration values ​​measured at different depths correspond to different settling stages of the same particle cluster.

[0024] Effective Settlement Velocity Effective settling velocity refers to the settling velocity of a particle group in a slurry, after considering factors such as particle shape and interparticle interactions (impeding settling effect). The unit is m / s or mm / s. Effective settling velocity differs from the free settling velocity of a single spherical particle in an infinitely diluted fluid (Stokes settling velocity); it comprehensively reflects the overall settling characteristics of the particle group during actual concentration. Effective settling velocity decreases with increasing particle concentration (impeding settling effect) and is also affected by factors such as particle shape, particle size distribution, and reagent action.

[0025] Flocculants are a class of high-molecular-weight polymer agents, primarily composed of polyacrylamide (PAM) and its derivatives, with molecular weights typically ranging from millions to tens of millions. Flocculants work by adsorbing and bridging long polymer chains onto the surfaces of multiple particles, linking dispersed coarse particles into larger flocs (flocs), thereby significantly increasing the effective particle size and settling velocity of the particle group. Flocculants primarily act as bridging agents for larger coarse particles, with a weaker effect on fine particles. Excessive flocculant dosage leads to excessively high slurry viscosity, which is detrimental to settling; insufficient dosage results in weak flocs that are easily broken.

[0026] Coagulants are a class of low molecular weight or inorganic salt agents, such as aluminum sulfate, ferric chloride, and polyaluminum chloride (PAC), with molecular weights typically ranging from hundreds to thousands. Coagulants neutralize the surface charge of particles by providing counter-charged ions, compressing the electric double layer thickness of the particles, and lowering the absolute value of the zeta potential on the particle surface (approaching zero). This weakens the electrostatic repulsion between particles, promoting collisions and aggregation into larger particles. Coagulants primarily exert their charge-neutralizing and agglomerating effect on fine particles with smaller diameters; their effect on coarse particles is weaker. Excessive coagulant addition can lead to charge reversal, which can actually increase the repulsion between particles; insufficient addition results in inadequate charge neutralization, making it difficult for fine particles to aggregate.

[0027] Furthermore, S2, calculate the effective settling velocity of the particle group. ,include: Where g is the acceleration due to gravity. To inhibit sedimentation factors, This is the particle shape correction factor; Where C is the current concentration. Where n is the maximum packing concentration and n is the resistance coefficient.

[0028] Furthermore, S3 yields the corrected effective settlement velocity. This includes: effective settlement velocity Using the impediment coefficient n as the parameter to be estimated; collecting concentration values ​​C measured by multiple concentration sensors over a continuous time period; and based on the current effective settling velocity... Calculations at different depths using one-dimensional convection-diffusion equations Theoretical predicted concentration value at time t ; Calculate the measured concentration value Compared with theoretically predicted concentration values Deviation between According to the deviation The effective settlement velocity is determined using the least squares method or Kalman filtering algorithm. The effective settlement velocity was obtained by performing optimization and correction. ;

[0029] Furthermore, S4 utilizes the corrected effective settlement velocity. For different depths Concentration value at Perform time lag correction to obtain the corrected concentration profile, including setting the current measurement time t as a unified reference time. Based on the corrected effective settlement velocity Calculate the settling depth of the particle group from the feed point. Time required ; at each depth The concentration value at that location was corrected for time shift: The corrected concentration value was obtained. ,in, Indicates depth At time Measured concentration values; corrected concentration values ​​for all depths. By depth Arranged from smallest to largest, the corrected concentration profile is formed by combining them.

[0030] Furthermore, the overflow turbidity T is corrected based on the corrected concentration profile, including: based on the corrected concentration profile, according to the formula... Calculate the concentration gradient ; concentration gradient With preset threshold Compare; when Exceeding the threshold When this is determined to be an abnormal settling condition, the overflow turbidity T is normalized and compensated for, according to the formula... Calculate the corrected overflow turbidity ,in, The overflow turbidity reference value is given, and k is the compensation coefficient; threshold value. The value range is 0.05~0.15 kg / (m³·m); the compensation coefficient k ranges from 0.1 to 0.5. When Not exceeding the threshold At that time, according to the formula The overflow turbidity T is normalized to obtain the corrected overflow turbidity. ;

[0031] Further, in step S5, the flocculant dosing rate is calculated. ,include: ;in, This is the minimum dosing rate for the flocculant. This represents the maximum dosing rate of the flocculant. is the normalized dosing intensity function for flocculants, with a value range of [0, 1]. The corrected overflow turbidity. Effective viscosity of overflow or underflow The normalized value, The volume fraction of coarse particles in the overflow water , This is the normalized value of the overflow terminal Zeta potential ζ. This is the time derivative of the Zeta potential at the overflow terminal.

[0032] Furthermore, the coagulant dosing rate was calculated. ,include: ;in, This is the minimum dosing rate for the coagulant. This represents the maximum dosing rate of the coagulant. This is the normalized dosing intensity function for the coagulant, with a value range of [0, 1]. The corrected overflow turbidity. The volume fraction of fine particles in the overflow water , This is the normalized value of the overflow terminal Zeta potential ζ.

[0033] Furthermore, Where σ(*) is the Sigmoid function;

[0034] Furthermore, Where Ash is based on the corrected overflow turbidity. The calculated normalized turbidity index, .

[0035] Another aspect of this application provides a dosing control system for a concentrator, comprising: a data acquisition module for acquiring data at different depths along the vertical direction of the concentrator. The concentration value measured by the i-th concentration sensor at time t And obtain the overflow turbidity T and the effective viscosity of the overflow or bottom flow. Volume fraction of fine particles in overflow water With coarse particle volume fraction Zeta potential at the feed end Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential And the equivalent diameter d of the particles, particle density Fluid density and fluid viscosity The settling velocity calculation module calculates the settling velocity based on the equivalent particle diameter d and particle density. Fluid density and fluid viscosity Calculate the effective settling velocity of the particle group. The settling velocity correction module uses the least squares method or Kalman filter algorithm to adjust the effective settling velocity based on the concentration distribution curves measured by multiple concentration sensors over time. Perform online correction to obtain the corrected effective settlement velocity. ;

[0036] The concentration profile correction module utilizes the corrected effective sedimentation velocity. For different depths Concentration value at Time lag correction is performed to obtain the corrected concentration profile; the turbidity correction module corrects the overflow turbidity T based on the corrected concentration profile to obtain the corrected overflow turbidity. The dosing rate calculation module calculates the dosing rate based on the corrected overflow turbidity. Effective viscosity of overflow or underflow Volume fraction of fine particles in overflow water With coarse particle volume fraction Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential Calculate the flocculant dosing rate separately. and coagulant dosing rate The dosing execution module, based on the flocculant dosing rate... and coagulant dosing rate The dosing of flocculants and coagulants is controlled by a dosing pump and an electric valve.

[0037] Compared to existing technologies, the advantages of this application are:

[0038] To address the shortcomings of existing thickener dosing control methods, such as insufficient consideration of the dynamic characteristics of particle settling, time lag between concentration measurements at different depths, and lack of precise coordinated control of flocculant and coagulant dosing, this application provides a thickener dosing control method. This method employs the least squares method or Kalman filter algorithm to online correct the effective settling velocity of the particle group based on measured data from multiple concentration sensors, improving the accuracy of settling velocity prediction. Furthermore, the corrected effective settling velocity is used to correct for time lag in concentration measurements at different depths, eliminating measurement errors caused by differences in particle settling time. Deviation is identified to obtain a concentration profile that accurately reflects the particle group of the same batch. The concentration gradient is calculated based on the corrected concentration profile. When the concentration gradient exceeds a preset threshold, it is judged as an abnormal sedimentation and the overflow turbidity is compensated and corrected. A normalized dosing intensity function for flocculants and coagulants is established. Taking into account multiple physicochemical parameters such as the corrected overflow turbidity, effective viscosity, volume fraction of fine and coarse particles, Zeta potential and its time derivative, the dosing rate of flocculants and coagulants is precisely and synergistically controlled to ensure that the synergistic effect of flocculants in bridging and flocculating coarse particles and coagulants in neutralizing and flocculating fine particles is fully utilized. Attached Figure Description

[0039] This application will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:

[0040] Figure 1This is an exemplary flowchart of a dosing control method for a concentrator according to some embodiments of this application;

[0041] Figure 2 This is a schematic diagram of a drug dispensing mechanism according to some embodiments of this application. Detailed Implementation

[0042] The methods and systems provided in the embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0043] A vertical sampling frame is suspended on the rake frame of the thickener, with multiple concentration sensors evenly spaced along the vertical direction to achieve real-time detection of the concentration gradient within the pool. Eight sensors are used, spaced 0.5 meters apart, covering the main settling area. The sensor housings are made of corrosion-resistant and wear-resistant materials and feature a waterproof sealing structure, meeting the requirements for long-term immersion operation. To prevent damage from large particles in the slurry and from the movement of the rake frame, a protective sleeve is added to the outside of the sampling frame. All sensors are connected to a signal collection module at the top of the sampling frame via an internal communication bus. The measured data is collected and transmitted to the central control unit for multi-point information fusion and analysis.

[0044] Because particle settling is time-consuming, sensor measurements at different depths do not reflect the true state at the same moment, and direct comparison can lead to time errors. This application estimates the equivalent settling velocity of the particle group, retrospectively estimating concentration data at each depth to a unified reference time, thereby obtaining a real-time concentration profile. The settling velocity estimation is based on particle size, fluid properties, and the effect hindering settling, and is improved in terms of accuracy through online data adaptive correction.

[0045] like Figure 1 As shown, different depths are obtained along the vertical direction of the concentrator. The concentration value measured by the i-th concentration sensor at time t And obtain the overflow turbidity T and the effective viscosity of the overflow or bottom flow. Volume fraction of fine particles in overflow water With coarse particle volume fraction Zeta potential at the feed end Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential And the equivalent diameter d of the particles, particle density Fluid density and fluid viscosity Based on the equivalent diameter d of the particles and the particle density Fluid density and fluid viscosity Calculate the effective settling velocity of the particle group. The effective sedimentation velocity is determined using the least squares method or Kalman filtering algorithm based on the concentration distribution curves measured by multiple concentration sensors over time. Perform online correction to obtain the corrected effective settlement velocity. ; Utilizing the corrected effective settlement velocity For different depths Concentration value at Perform time lag correction to obtain the corrected concentration profile; correct the overflow turbidity T based on the corrected concentration profile; and correct the overflow turbidity T based on the corrected overflow turbidity. Effective viscosity of overflow or underflow Volume fraction of fine particles in overflow water With coarse particle volume fraction Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential Calculate the flocculant dosing rate separately. and coagulant dosing rate According to the flocculant dosage rate and coagulant dosing rate The dosing of flocculants and coagulants is controlled by a dosing pump and an electric valve.

[0046] Specifically, a mathematical model is introduced after data acquisition to correct for time lag and settlement error. Let the i-th sensor be located at depth... The concentration was measured at time t. This signal actually reflects an earlier time. The particle group that enters and settles to that depth. This is determined by estimating the effective settling velocity. This allows it to be traced back to a unified reference time. This allows us to obtain the corrected concentration value. The basic relationship is as follows: The corrected concentration can be expressed as .

[0047] Settlement velocity The estimation is based on Stokes' law and modified with sedimentation inhibition factors and particle shape correction factors to better reflect the complex fluid conditions inside the thickener. Specifically, it is in the form of:

[0048] Where d is the equivalent diameter of the particle, obtained from particle size analysis. and These represent particle and fluid densities, respectively, with μ being the fluid viscosity and temperature compensation considered, along with the sedimentation insufficiency factor. According to the Richardson-Zaki empirical formula Calculate, where C is the current concentration. The maximum packing concentration is given by n, which is the hindering coefficient. The initial value of the hindering coefficient n is determined based on the particle Reynolds number Re. When Re < 0.2, n takes a value of 4.65, and when 0.2 ≤ Re < 1, n takes a value of... When 1 ≤ Re < 500, n takes the following value: During online operation, the impedance coefficient n is used as an estimable parameter, based on the concentration values ​​measured by multiple concentration sensors. The deviation from the theoretical prediction is corrected in real time by using the least squares method or Kalman filtering algorithm; particle shape correction coefficient. The shape factor SF is determined based on the particle shape factor, where SF is defined as SF = Sactual / Ssphere, where Sactual is the actual surface area of ​​the particle, and Ssphere is the surface area of ​​spherical particles of the same volume; when the particle is spherical... When the particles are ellipsoidal The specific value is determined by analyzing microscopic images or measuring the particle shape factor SF using a laser particle size analyzer, and then calculated according to the formula. Calculated.

[0049] Specifically, a state-space model is established, and the effective settlement velocity is... Using the impedance coefficient n as an estimable parameter; collecting concentration values ​​measured by multiple concentration sensors over a continuous time period. This generates a concentration distribution curve that varies with time and depth; based on the concentration distribution curve and the current effective settling velocity... According to the one-dimensional convection-diffusion equation Calculate the theoretically predicted concentration value Where z is the depth coordinate and D is the diffusion coefficient, the equation is numerically solved using the finite difference method or the finite element method to obtain the values ​​at different depths. Theoretical predicted concentration value at time t ; Calculate the measured concentration value Compared with theoretically predicted concentration values Deviation between When using the least squares method, the objective function is minimized. For effective settlement velocity The effective settlement velocity is obtained by optimizing the solution with the drag coefficient n. Alternatively, when using the Kalman filter algorithm, the effective settlement velocity will be... As a state variable, the measured concentration value As an observation, the corrected effective settlement velocity is obtained through prediction-update iterative calculation. Corrected effective settlement velocity It updates in real time during each sampling period to achieve adaptive dynamic correction.

[0050] In actual operation, the initial value of vsetir is obtained through theoretical calculation, and then online correction is performed using least squares or Kalman filtering methods based on the concentration distribution curves of multiple sensors over time. By establishing a state-space model and using vsetir and the impediment coefficient n as estimable parameters, the system can continuously update the settling velocity, thereby achieving adaptive dynamic correction. This method ensures the temporal consistency of concentration signals at different depths and provides more accurate real-time data support for subsequent automatic dosing decisions.

[0051] like Figure 2 As shown, online Zeta potential measuring devices are installed in the feed slurry pipeline and the overflow water pipeline, respectively. Zeta potential reflects the surface electrical properties of particles and the adsorption of reagents, and is a key parameter for judging the effectiveness of reagents. The sensor adopts a flow-through cell structure, enabling continuous monitoring during slurry flow. It is equipped with automatic flushing and self-calibration functions to ensure measurement stability and long-term reliability. The Zeta measurement at the feed end reflects the initial characteristics of the feed particles, while the Zeta measurement at the overflow end evaluates the effect of the reagent in the clarified water, thus achieving before-and-after comparison and feedback correction.

[0052] The specific control logic is as follows: Let T represent the overflow turbidity (NTU); let μeff represent the effective viscosity of the overflow or underflow; let fffine and fcoarse represent the volume fraction or relative proportion of fine and coarse particles in the overflow, respectively (satisfying fffine + fcoarse = 1); let ζ represent the Zeta potential (mV). Its time derivative (mV / s) is used to measure the adsorption rate of the reagent; let This represents the depth concentration profile characteristics (upper layer concentration) after hysteresis correction; let Qf and Qc be the instantaneous dosing rates (g / h) of flocculant and coagulant, respectively; let u = [Qf, Qc] be the control input vector. To eliminate dimensions and facilitate nonlinear mapping, the main quantities are normalized, and the reference values ​​can be obtained from historical operation or field calibration.

[0053] ; ; ; ;in, This is a reference value for effective viscosity, ranging from 1 to 10 mPa·s;

[0054] First, offline identification is performed through static and semi-dynamic experiments (rod sinking / column sinking experiments and online Zeta curves under different dosages) to obtain initial coefficients; then, data is collected on-site with a small perturbation strategy to train a data-driven lightweight neural network prediction model; finally, the learned model is embedded into the prediction kernel of MPC and online parameter identification is enabled to achieve adaptation.

[0055] Leveraging the smoothing properties of fuzzy / nonlinear mapping and sigmoid / hyperbolic tangent, we propose a real-time control mapping of the following form, which nonlinearly fuses multiple sensor signals into a dosing decision:

[0056] Among them, the minimum dosing rate of flocculant The value range is 0.1~5 mL / min; the maximum dosing rate of the flocculant. The value range is 10~100mL / min.

[0057] Among them, the minimum dosing rate of the coagulant The value range is 0.1~5 mL / min; the maximum dosing rate of the coagulant. The value range is 10~100 mL / min.

[0058] The decision function takes the form of a weighted sigmoid / tanh composite to reflect nonlinear coupling and saturation constraints.

[0059] ,in, These are weighting coefficients, with values ​​ranging from 0.1 to 5. To adjust the parameters; These are exponential parameters, and their values ​​range from 0.5 to 3.

[0060] ;here For the logistic function, ensure output ; These are weighting coefficients, with values ​​ranging from 0.1 to 5. The index parameter has a value range of 0.5 to 3; Ash is the normalized turbidity index calculated based on the corrected overflow turbidity T'. Ashref is the normalized turbidity reference value, ranging from 0.1 to 1; coefficients ai, bi, ci and exponents pi, qi are controller parameters, whose initial values ​​can be obtained offline / experimentally and then adaptively adjusted online. This structure achieves: turbidity increase ( This strongly encourages increased flocculant dosage, leading to increased viscosity. Inhibiting flocculant dosage to avoid excessive flocculation; high coarse particle ratio ( Larger sizes can also promote the addition of flocculants; while The sign and amplitude determine the adsorption rate and direction of the reagent. A significantly positive (or negative, according to the definition of reference polarity) indicates that the agent produces strong adsorption in a short period of time, and the dosage should be reduced immediately by adding more drug (through negative inhibition), and vice versa. If the flow is smooth or tends towards equilibrium, drug administration is permitted. This fusion strategy is used in data-driven fuzzy inference to handle process nonlinearity and measurement noise.

[0061] Turbidity and particle size detection devices are installed in the overflow water. Turbidity detection reflects the degree of clarity of the overflow, while particle size detection provides information on the particle size distribution, especially distinguishing the ratio of coarse to fine particles. These data together determine the settling effect and the quality of the overflow water. For signal acquisition and transmission, data from various sensors undergo time synchronization and filtering processing before being transmitted to the central control unit via an industrial communication bus. The control system uses a unified clock to ensure that measurement data from different depths and with different parameters are fused under the same time reference.

[0062] The entire system consists of a sensor layer, a signal aggregation module, a central controller, and actuators. The central controller fuses and analyzes sensor data, executes a hybrid control strategy combining rules and models, and outputs dosing commands. The actuators are an adjustable-speed dosing pump and a feedback-equipped electric valve, capable of precisely adjusting the dosage and interval of the pesticide according to the commands. The system also features automatic cleaning, self-calibration, and historical data recording functions to ensure long-term stable operation.

[0063] In addition to its chemical dosing function, the actuator includes an automatic cleaning unit that periodically or on-demand cleans the sampling pipeline, detection probe, and related fluid channels to prevent the adhesion of sediment, algae, and contaminants, ensuring detection accuracy and system stability. To improve the intelligence and efficiency of the self-cleaning process, the system introduces an artificial intelligence-based self-cleaning prediction and execution method. The central controller embeds a machine learning model, preferably using the random forest algorithm, which automatically determines the degree of contamination and cleaning needs of the sampling components by analyzing historical operating data and real-time monitoring data.

[0064] The model's input features include the stability of each sensor signal (fluctuation amplitude and drift trend of turbidity and particle size signals), historical operating time and cumulative working cycles, environmental parameters (temperature, flow rate, pH value, solid content), and signal recovery characteristics and duration after the most recent cleaning. The model performs multidimensional analysis on these features and outputs a cleaning necessity index. When this index exceeds a set threshold, the system automatically triggers a cleaning task; when the index is within a critical range, the system enters predictive monitoring mode, continuously tracking signal change trends to determine the optimal cleaning time.

[0065] During the execution phase, the central controller drives the cleaning valves, backwash pumps, or spray devices to perform cleaning operations based on the instructions output by the artificial intelligence model. After cleaning is completed, the system automatically enters the verification mode, compares the signal stability before and after cleaning, and feeds the results back to the artificial intelligence module to correct the prediction model, achieving continuous self-learning and parameter optimization.

[0066] This method avoids the waste of resources caused by timed cleaning and can clean in advance before the contamination seriously affects the measurement accuracy, thus maintaining the long-term accuracy and reliability of the detection system.

[0067] The foregoing illustrative description of the present application and its embodiments is not restrictive and can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. The accompanying drawings are only one embodiment of the present application, and the actual structure is not limited thereto. Therefore, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the present application, such designs should fall within the scope of protection of this application. Furthermore, the word "comprising" does not exclude other elements or steps, and the word "a" preceding an element does not exclude the inclusion of "a plurality" of that element. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

Claims

1. A method for controlling the dosing of chemicals in a concentrator, characterized in that, include: S1, obtaining different depths along the vertical direction of the concentrator. The concentration value measured by the i-th concentration sensor at time t And obtain the overflow turbidity T and the effective viscosity of the overflow or bottom flow. Volume fraction of fine particles in overflow water With coarse particle volume fraction Zeta potential at the feed end Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential And the equivalent diameter d of the particles, particle density Fluid density and fluid viscosity ; S2, based on the equivalent particle diameter d and particle density Fluid density and fluid viscosity Calculate the effective settling velocity of the particle group. ; S3, using the least squares method or Kalman filter algorithm, based on the concentration value distribution curves measured by multiple concentration sensors over time, determines the effective sedimentation velocity. Perform online correction to obtain the corrected effective settlement velocity. ; S4, utilizing the corrected effective settlement velocity For different depths Concentration value at Perform time lag correction to obtain the corrected concentration profile; The overflow turbidity T was corrected based on the corrected concentration profile. S5, based on the corrected overflow turbidity Effective viscosity of overflow or underflow Volume fraction of fine particles in overflow water With coarse particle volume fraction Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential Calculate the flocculant dosing rate separately. and coagulant dosing rate ; S6, based on the flocculant dosage rate and coagulant dosing rate The dosing of flocculants and coagulants is controlled by a dosing pump and an electric valve.

2. The dosing control method for a concentrator according to claim 1, characterized in that: S2, Calculate the effective settling velocity of the particle group. ,include: Where g is the acceleration due to gravity. To inhibit sedimentation factors, This is the particle shape correction factor; Where C is the current concentration. denoted as the maximum packing concentration, and n as the impediment coefficient.

3. The dosing control method for a concentrator according to claim 2, characterized in that: S3, using the least squares method or Kalman filter algorithm, based on the concentration value distribution curves measured by multiple concentration sensors over time, determines the effective sedimentation velocity. Perform online correction to obtain the corrected effective settlement velocity. ,include: Effective settlement velocity The obstacle coefficient n is used as the parameter to be estimated; Collect concentration values ​​C measured by multiple concentration sensors over a continuous time period; Based on the current effective settlement rate Calculations at different depths using one-dimensional convection-diffusion equations Theoretical predicted concentration value at time t ; Calculate the measured concentration value Compared with theoretically predicted concentration values Deviation between ; According to deviation The effective settlement velocity is determined using the least squares method or Kalman filtering algorithm. The effective settlement velocity was obtained by performing optimization and correction. .

4. The dosing control method for a concentrator according to claim 3, characterized in that: S4, utilizing the corrected effective settlement velocity For different depths Concentration value at Time lag correction is performed to obtain the corrected concentration profile, including: Set the current measurement time t as the unified reference time. ; Based on the corrected effective settlement velocity Calculate the settling depth of the particle group from the feed point. Time required ; Each depth The concentration value at that location was corrected for time shift: The corrected concentration value was obtained. ,in, Indicates depth At time The measured concentration value; Corrected concentration values ​​for all depths By depth Arranged from smallest to largest, they are combined to form a corrected concentration profile.

5. The dosing control method for a concentrator according to claim 4, characterized in that: The overflow turbidity T is corrected based on the corrected concentration profile, including: Based on the corrected concentration profile, according to the formula Calculate the concentration gradient ; Concentration gradient With preset threshold Compare; when Exceeding the threshold When this is determined to be an abnormal settling condition, the overflow turbidity T is normalized and compensated for, according to the formula... Calculate the corrected overflow turbidity ,in, The overflow turbidity reference value is given, and k is the compensation coefficient; threshold value. The value range is 0.05~0.15 kg / (m³·m); the value range of the compensation coefficient k is 0.1~0.5; when Not exceeding the threshold At that time, according to the formula The overflow turbidity T is normalized to obtain the corrected overflow turbidity. .

6. The dosing control method for a concentrator according to claim 1, characterized in that: Calculate the dosing rate of coagulant ,include: ;in, This is the minimum dosing rate for the coagulant. This represents the maximum dosing rate of the coagulant. This is the normalized dosing intensity function for the coagulant, with a value range of [0, 1]. The corrected overflow turbidity. The volume fraction of fine particles in the overflow water , This is the normalized value of the overflow terminal Zeta potential ζ.

7. The dosing control method for a concentrator according to claim 6, characterized in that: Where Ash is based on the corrected overflow turbidity. The calculated normalized turbidity index, Ash ref This is the reference value for normalized turbidity.

8. A dosing control system for a concentrator, characterized in that, include: The data acquisition module acquires data at different depths along the vertical direction of the concentrator. The concentration value measured by the i-th concentration sensor at time t And obtain the overflow turbidity T and the effective viscosity of the overflow or bottom flow. Volume fraction of fine particles in overflow water With coarse particle volume fraction Zeta potential at the feed end Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential And the equivalent diameter d of the particles, particle density Fluid density and fluid viscosity ; The settling velocity calculation module calculates the settling velocity based on the equivalent particle diameter d and particle density. Fluid density and fluid viscosity Calculate the effective settling velocity of the particle group. ; The settling velocity correction module uses the least squares method or Kalman filter algorithm to adjust the effective settling velocity based on the concentration distribution curves measured by multiple concentration sensors over time. Perform online correction to obtain the corrected effective settlement velocity. ; The concentration profile correction module utilizes the corrected effective sedimentation velocity. For different depths Concentration value at Perform time lag correction to obtain the corrected concentration profile; The turbidity correction module corrects the overflow turbidity T based on the corrected concentration profile to obtain the corrected overflow turbidity. ; The dosing rate calculation module calculates the dosing rate based on the corrected overflow turbidity. Effective viscosity of overflow or underflow Volume fraction of fine particles in overflow water With coarse particle volume fraction Overflow terminal Zeta potential and the time derivative of the overflow terminal Zeta potential Calculate the flocculant dosing rate separately. and coagulant dosing rate ; The dosing execution module, based on the flocculant dosing rate and coagulant dosing rate The dosing of flocculants and coagulants is controlled by a dosing pump and an electric valve.

Images