Flocculant addition amount real-time optimization method and system based on multi-flocculation jar testing

Abstract

Disclosed in the present invention are a flocculant addition amount real-time optimization method and system based on multi-flocculation jar testing. The method comprises: obtaining an addition amount of a flocculant and a coagulant aid during first operation; on the basis of a flocculant basic addition amount in an r-th round of flocculation jar testing, respectively obtaining three sets of floc image and water quality index data corresponding to three intelligent JarTest modules which carry out flocculation testing at the same time in the r-th round of flocculation jar testing; respectively inputting the three sets of floc image and water quality index data into a pre-trained deep learning model, to correspondingly obtain a supernatant turbidity predicted value Td1(r), a supernatant turbidity predicted value Td2(r), and a supernatant turbidity predicted value Td3(r) in the r-th round of flocculation jar testing; and on the basis of the supernatant turbidity predicted value Td1(r), the supernatant turbidity predicted value Td2(r), and the supernatant turbidity predicted value Td3(r), obtaining a new flocculant actual addition amount, and taking the new flocculant actual addition amount as a flocculant basic addition amount in the (r+1)-th round. The present invention can achieve real-time adjustment and optimization of the flocculant addition amount.

Description

Real-time optimization method and system for flocculant dosage based on multi-cup flocculation experiment Technical Field

[0001] The present invention belongs to the technical field of integration of artificial intelligence, automatic control and water treatment, and particularly relates to a method and system for real-time optimization of flocculant dosage based on a multi-cup flocculation experiment. Background Art

[0002] Coagulation treatment is widely used in tap water, municipal sewage and industrial wastewater, especially in tap water treatment, and is basically a standard unit in water plants.

[0003] The coagulation process requires the addition of flocculants. Under certain hydraulic conditions, whether the dosage of flocculants is appropriate determines whether the removal of pollutants in the water meets the standards after coagulation. For example, in tap water treatment, this refers to whether the turbidity removal meets the standards.

[0004] For the raw water to be treated, there is a complex nonlinear relationship between the amount of flocculant added and the amount of pollutants removed. This is because the coagulation effect depends on the combined effects of multiple water quality indicators such as turbidity, TOC\COD (TOC stands for total organic carbon, COD stands for chemical oxygen demand), pH value, temperature, etc.

[0005] During coagulation, water needs to flow slowly through a very long horizontal flow sedimentation tank, and water quality indicators can only be obtained through sensors at the tail end of the horizontal flow sedimentation tank. Therefore, there is a time delay of 1 to 2 hours between the addition of flocculants and the acquisition of coagulation results. This phenomenon is also called the large time lag phenomenon.

[0006] In order to quickly determine the appropriate flocculant dosage under the constraints of nonlinearity and large time lag, the current common practice is to conduct a six-beaker experiment in the laboratory every day, that is, to add different doses of flocculant to six groups of the same raw water. After the cup flocculation is completed, the turbidity of the supernatant of the water sample in the beaker is measured, that is, the turbidity after sedimentation, to obtain the corresponding relationship between different flocculant dosages and turbidity, and then determine the flocculant dosage based on the experience of the technicians.

[0007] Although the method of determining the flocculant dosage through the six-beaker test can basically cope with daily water treatment, it still has the following disadvantages:

[0008] 1. Unable to cope with sudden changes in water quality. In water treatment, sudden changes in water quality are inevitable. For example, reservoir discharge caused by summer rainstorms or reservoir overflow caused by water shortages will cause a sharp rise in turbidity. The turbidity of raw water will quickly rise from a few NTU (Nephelometric Turbidity Unit) to hundreds or even thousands of NTU within a few hours. In addition, when the output or process of the sewage discharge enterprise changes, the water quality of industrial wastewater will suddenly change. When the raw water suddenly changes, the flocculant dosage must be adjusted quickly to achieve the effluent water quality standard. At this time, the dosage determined by the last six-beaker experiment is no longer appropriate. In actual treatment, the dosage can only be determined entirely by manual experience;

[0009] 2. It is difficult to treat low-temperature and low-turbidity water in winter. At low temperatures, the flocculant is not fully dissolved, and the low turbidity results in insufficient floc growth. In this state, if the flocculant dosage is too small, the turbidity removal after flocculation is insufficient, and the effluent turbidity does not meet the standard. If the flocculant dosage is excessive, the inadequately dissolved flocculant itself will become a turbidity component, causing the effluent turbidity to not meet the standard. Therefore, it is usually necessary to repeatedly conduct six-beaker experiments and actual dosage adjustments to determine the flocculant dosage that meets the effluent turbidity requirements.

[0010] 3. Excessive flocculant addition. Because it is impossible to automatically adjust the flocculant dosage in real time according to changes in the incoming water, excessive flocculant addition is usually chosen to ensure that the effluent water quality meets the standards. This excessive addition will increase the company's water treatment costs. The increased costs are composed of two parts: one is the flocculant cost, and the other is the subsequent treatment costs of other units. For example, excessive flocculant will produce more sludge, which increases the sludge treatment fee; excessive flocculant PAC (polyaluminum chloride) will result in excessive residual aluminum, which requires additional cost to treat.

[0011] 4. Increased carbon emissions. In addition to increased costs, excessive flocculant addition also results in increased carbon emissions. Water treatment companies are not major contributors to direct carbon emissions, but they are major contributors to indirect carbon emissions. Among indirect carbon emissions, electricity consumption accounts for the largest proportion, followed by chemicals.

[0012] The above-mentioned disadvantages of the method of determining the flocculant dosage by the six-beaker experiment are mainly due to the inability of the six-beaker experiment to achieve real-time control of the flocculant dosage. Therefore, a real-time control scheme for flocculant dosage is needed to achieve real-time adjustment and optimization of the flocculant dosage.

[0013] Summary of the Invention

[0014] In order to solve the above problems existing in the prior art, the present invention provides a method and system for real-time optimization of flocculant dosage based on a multi-cup flocculation experiment.

[0015] A real-time optimization method for flocculant dosage based on a multi-cup flocculation experiment, the real-time optimization method for flocculant dosage comprising:

[0016] Step 1: Obtain the dosage of flocculant and coagulant aid during the first run, which will serve as the basic dosage of flocculant for the first round of cup flocculation experiment;

[0017] Step 2: Based on the basic flocculant dosage of the r-th round cup-jar flocculation experiment, respectively obtain the first group of flocs images and water quality index data, the second group of flocs images and water quality index data, and the third group of flocs images and water quality index data corresponding to the first intelligent Jar Test module, the second intelligent Jar Test module, and the third intelligent Jar Test module that simultaneously perform flocculation experiments in the r-th round cup-jar flocculation experiment, wherein the flocculant dosage Pd1(r) of the r-th round cup-jar flocculation experiment of the first intelligent Jar Test module is less than the current actual flocculant dosage, the flocculant dosage Pd2(r) of the r-th round cup-jar flocculation experiment of the second intelligent Jar Test module is equal to the current actual flocculant dosage, and the flocculant dosage Pd3(r) of the r-th round cup-jar flocculation experiment of the third intelligent Jar Test module is greater than the current actual flocculant dosage;

[0018] Step 3: Input the first set of floc images and water quality index data, the second set of floc images and water quality index data, and the third set of floc images and water quality index data corresponding to the rth round of cup flocculation experiment into a pre-trained deep learning model, and obtain the supernatant turbidity prediction value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity prediction value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity prediction value Td3(r) after the flocculation experiment of the third intelligent Jar Test module in the rth round of cup flocculation experiment;

[0019] Step 4: Obtain a new actual flocculant dosage based on the predicted value Td1(r) of the supernatant turbidity after the flocculation experiment of the first intelligent Jar Test module, the predicted value Td2(r) of the supernatant after the flocculation experiment of the second intelligent Jar Test module, and the predicted value Td3(r) of the supernatant after the flocculation experiment of the third intelligent Jar Test module, and use the new actual flocculant dosage as the basic flocculant dosage for the r+1th round. Repeat steps 2 to 4 to determine the new actual flocculant dosage for the r+1th round.

[0020] Optionally, the flocculant dosage Pd1(r) of the rth round of cup-jar flocculation experiment is Pd0(r-1)*L, the flocculant dosage Pd2(r) is Pd0(r-1)*K, and the flocculant dosage Pd3(r) is Pd0(r-1)*H, wherein Pd0(r-1) is the basic flocculant dosage of the r-1th round of cup-jar flocculation experiment, L is the dosage decrease coefficient, K is the dosage base coefficient, and H is the dosage increase coefficient, L<1, K=1, and H>1.

[0021] Optionally, step 4 includes:

[0022] Determine the relationship between the predicted turbidity value Td1(r) of the supernatant after the flocculation experiment of the first intelligent Jar Test module, the predicted turbidity value Td2(r) of the supernatant after the flocculation experiment of the second intelligent Jar Test module, the predicted turbidity value Td3(r) of the supernatant after the flocculation experiment of the third intelligent Jar Test module, and the supernatant turbidity standard Ts after each round of flocculation experiment;

[0023] If Td1(r)<Ts, the actual dosage of flocculant is adjusted to the dosage of flocculant Pd1(r) in the rth round of cup flocculation experiment;

[0024] If Td1(r)>Ts and Td2(r)<Ts, the actual dosage of flocculant is the dosage of flocculant Pd2(r) in the rth round of cup flocculation experiment;

[0025] If Td2(r)>Ts and Td3(r)<Ts, the actual flocculant dosage is adjusted to the flocculant dosage Pd3(r) of the rth round of cup flocculation experiment;

[0026] If Td1(r)>Ts, Td2(r)>Ts and Td3(r)≥Ts, the actual dosage of the new flocculant is determined based on the preset relationship between the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module and the supernatant turbidity standard Ts after each round of flocculation experiment.

[0027] Optionally, determining the new actual dosage of the flocculant according to a preset relationship between the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and the supernatant turbidity standard Ts after each round of flocculation experiment includes:

[0028] Judge the preset relationship among the predicted supernatant turbidity value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the predicted supernatant turbidity value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and the supernatant turbidity standard Ts after each round of flocculation experiment;

[0029] If the preset relationship is Td1(r)>Td2(r)>Td3(r)≥Ts, the turbidity removal rate is in the monotonically rising region. Then judge whether the state is in the preset turbidity state. If it is in the preset turbidity state, the actual dosage of the flocculant is adjusted to Pd3(r)*Q H , if it is not in the preset turbidity state, continue to judge whether it is in the preset temperature and turbidity state. If it is not in the preset temperature and turbidity state, the actual dosage of the flocculant is adjusted to Pd3(r)*Q. If it is in the preset temperature and turbidity state, then Pd3(r)*Q L , where Q is the dosing increase step, Q H is the first increase step of the preset turbidity state, Q L is the second increase step of the preset turbidity state;

[0030] If the preset relationship is Ts<Td1(r)<Td2(r)<Td3(r), the turbidity removal rate is in the monotonically decreasing region. Then the actual dosage of the flocculant is adjusted to Pd1(r) / H, and the values of H and L for the next round of beaker flocculation experiment are determined according to whether the state is in the preset temperature and turbidity state;

[0031] If the preset relationship is min(Td1(r),Td2(r),Td3(r))>Ts, Td1(r)>Td2(r), Td2(r)<Td3(r), or min(Td1(r),Td2(r),Td3(r))>Ts, Td1(r)<Td2(r), Td2(r)>Td3(r), the turbidity removal rate is in the non-monotonically rising region. At this time, the new actual dosage of the flocculant is determined according to the relationship among the predicted supernatant turbidity value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and the values of H and L for the next round of beaker flocculation experiment are determined according to whether the turbidity is in the preset temperature and turbidity state.

[0032] Optionally, judging whether the state is in the preset turbidity state includes:

[0033] Determine whether the climbing speed Z in two consecutive rounds is greater than the turbidity climbing rate b. If not, it is not in the preset turbidity state. If so, it is in the preset turbidity state, where Z = abs[Tini(r) - Tini(r - 1)] / Tini(r - 1)], abs is the absolute value operation, Tini(r) is the average influent turbidity of the r-th round of beaker flocculation experiment, and Tini(r - 1) is the average influent turbidity of the (r - 1)-th round of beaker flocculation experiment;

[0034] Among them, when the state is in the preset turbidity state, continue to determine whether to jump out of the preset turbidity state. If Tini(r - 3) > Tini(r), and at the same time Tini(r - 2) > Tini(r), and at the same time Tini(r) < c, then jump out of the preset turbidity state. Otherwise, it remains in the preset turbidity state, where Tini(r - 3) is the average influent turbidity of the (r - 3)-th round of beaker flocculation experiment, Tini(r - 2) is the average influent turbidity of the (r - 2)-th round of beaker flocculation experiment, and c is the turbidity when jumping out of the preset turbidity state.

[0035] Optionally, determining whether it is in the preset temperature and turbidity state includes:

[0036] Determine whether the monitored water temperature data Temp is less than the low temperature judgment standard Temp - L, and at the same time whether the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is less than the low turbidity judgment standard Tini_L. If the monitored water temperature data Temp is less than the low temperature judgment standard Temp - L, and at the same time the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is less than the low turbidity judgment standard Tini_L, then the state is in the preset temperature and turbidity state. Otherwise, it is not in the preset temperature and turbidity state.

[0037] Optionally, the dosing increase step Q = [(Tini(r) - Ts) / (Tini(r) - Td3(r)] * m, where m is the slope correction coefficient; the second increase step of the preset turbidity state is Q L = 1 + (Q cur - 1) / f, where f is the Q value adjustment coefficient in the preset temperature and turbidity state, and Q cur is the Q value of the current round of beaker flocculation experiment;

[0038] The value-taking method of the first increase step of the preset turbidity state includes:

[0039] Determine whether the average influent turbidity Tini(r - 1) of the (r - 1)-th round of beaker flocculation experiment is greater than the average influent turbidity Tini(r - 2) of the (r - 2)-th round of beaker flocculation experiment;

[0040] If the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-2) of the r-2th round of cup-jar flocculation experiment, then continue to judge whether the average turbidity of the influent water Tini(r) of the rth round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment. If so, then Q H =Qr, if not, then Q H =Qp;

[0041] If the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment is not greater than the average turbidity of the influent water Tini(r-2) of the r-2th round of cup-jar flocculation experiment, then continue to judge whether the average turbidity of the influent water Tini(r) of the rth round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment. If not, then Q H =Qd;

[0042] Among them, Qr=Qr1*n, Qp=Q, Qd=Qd1*n, Qr1=(Tpre1-Ts) / [Tini(r)-Ts], Tpre1=Tini(r)+kpre*yr, Qd1=(Tpre2-Ts) / [Tini(r)-Ts], Tpre2=Tini(r)+kpre*yd, kpre=k(r)+[k(r)-k(r-1)], k(r)=Tini(r)-Tini(r-1), k(r-1)=Tini(r-1)-Tini(r-2), yr is the first correction coefficient, yd is the second correction coefficient, and n is the correction coefficient of Q.

[0043] Optionally, the values ​​of H and L for the next round of cup flocculation experiment are determined based on whether the state is within the preset temperature and turbidity state, including:

[0044] Determine whether the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and whether the average turbidity of the inlet water Tini(r) of the rth round of cup flocculation experiment is less than the low turbidity judgment standard Tini_L;

[0045] If the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and the average turbidity of the inlet water Tini(r) of the rth round of cup flocculation experiment is less than the low turbidity judgment standard Tini_L, then the state is in the preset temperature and turbidity state. When entering the preset temperature and turbidity state for the first time, the value of L is updated to LL, and the value of H is updated to HL, and LL and HL are used as the values ​​of H and L in the next round, where LL=L cur -(1-L cur ) / s,HL=1+(H cur -1) / s,L curL value for the current round of jar flocculation experiment, H cur is the H value for the current round of jar flocculation experiment, s is the adjustment coefficient, f is the Q value adjustment coefficient under the preset temperature and turbidity conditions. When it is not the first time to enter the preset temperature and turbidity conditions, the values of L, H, and Q L remain unchanged;

[0046] If the monitored water temperature data Temp is greater than or equal to the low temperature judgment standard Temp-L, or the influent average turbidity Tini(r) of the r-th round of jar flocculation experiment is greater than or equal to the low turbidity judgment standard Tini_L, then L, H, and Q are set to the initialization data, and at the same time, the number of times of entering the preset temperature and turbidity conditions is set to 0.

[0047] Optionally, determine the new actual dosage of the flocculant according to the relationship between the predicted supernatant turbidity values Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity values Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity values Td3(r) after the flocculation experiment of the third intelligent Jar Test module, including:

[0048] Judge the relationship between the predicted supernatant turbidity values Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity values Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity values Td3(r) after the flocculation experiment of the third intelligent Jar Test module;

[0049] If Td1(r)>Td2(r) and Td2(r)<Td3(r), then the actual dosage of the flocculant is adjusted to [Pd2(r)+Pd3(r)] / 2;

[0050] If Td1(r)<Td2(r) and Td2(r)>Td3(r), then the actual dosage of the flocculant is adjusted to [Pd1(r)+Pd2(r)] / 2.

[0051] An embodiment of the present invention further provides a real-time optimization system for the dosage of flocculant based on multi-jar flocculation experiments. The real-time optimization system for the dosage of flocculant includes:

[0052] A basic dosage acquisition module, which is used to obtain the dosage of the flocculant and the coagulant aid during the first operation as the basic dosage of the flocculant for the first round of jar flocculation experiment;

[0053] a data acquisition module for respectively acquiring, based on the basic flocculant dosage of the r-th round cup-jar flocculation experiment, the first intelligent Jar Test module, the second intelligent Jar Test module, the first set of flocculent images and water quality index data, the second set of flocculent images and water quality index data, and the third set of flocculent images and water quality index data corresponding to the first intelligent Jar Test module, which simultaneously perform flocculation experiments in the r-th round cup-jar flocculation experiment, wherein the flocculant dosage Pd1(r) of the r-th round cup-jar flocculation experiment of the first intelligent Jar Test module is less than the current actual flocculant dosage, the flocculant dosage Pd2(r) of the r-th round cup-jar flocculation experiment of the second intelligent Jar Test module is equal to the current actual flocculant dosage, and the flocculant dosage Pd3(r) of the r-th round cup-jar flocculation experiment of the third intelligent Jar Test module is greater than the current actual flocculant dosage;

[0054] A prediction value acquisition module is used to input the first set of floc images and water quality index data, the second set of floc images and water quality index data, and the third set of floc images and water quality index data corresponding to the rth round of cup flocculation experiment into a pre-trained deep learning model, and correspondingly obtain the supernatant turbidity prediction value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity prediction value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity prediction value Td3(r) after the flocculation experiment of the third intelligent Jar Test module in the rth round of cup flocculation experiment;

[0055] The flocculant actual dosage acquisition module is used to obtain a new flocculant actual dosage based on the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and use the new flocculant actual dosage as the basic flocculant dosage for the r+1th round, and repeatedly execute the processing process from the data acquisition module to the flocculant actual dosage acquisition module to determine the new flocculant actual dosage for the r+1th round.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention proposes an iterative optimization method based on multiple cup-jar flocculation experiments. The method uses three intelligent Jar Test modules to conduct experiments simultaneously, obtaining a first set of floc images and water quality index data, a second set of floc images and water quality index data, and a third set of floc images and water quality index data. Then, based on the three sets of floc images and water quality index data obtained, a deep learning model is used to predict the turbidity after multiple cup-jar flocculation experiments, obtaining three predicted values ​​of supernatant turbidity after the simultaneous flocculation experiments. The three predicted values ​​of supernatant turbidity after the simultaneous flocculation experiments are then used to obtain the flocculant dosage for the next round of cup-jar flocculation experiments, thereby achieving iterative optimization of the flocculant dosage. The present invention uses a deep learning model to predict the flocculation experiment results, and achieves rapid optimization of the dosage through multiple cup-jar flocculation experiments. It can achieve real-time adjustment and optimization of the flocculant dosage, which can solve the difficulties of nonlinearity, time-varying and large time lag in the current flocculation dosing control, as well as the resulting problems of excessive flocculant addition, difficulty in coping with sudden water quality changes and special water quality. The method provided by the present invention shortens the prediction time of flocculation experimental results, optimizes effluent control, reduces the dosage of flocculant, saves costs for water treatment companies, and reduces indirect carbon emissions.

[0058] The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] 1 is a schematic flow chart of a method for real-time optimization of flocculant dosage based on a multi-cup flocculation experiment according to an embodiment of the present invention;

[0060] FIG2 is a schematic diagram of a model of a real-time optimization method for flocculant dosage based on a multi-cup flocculation experiment provided by an embodiment of the present invention;

[0061] 3 is a schematic flow chart of a method for determining the actual dosage of a flocculant provided in an embodiment of the present invention;

[0062] FIG4 is a graph showing the relationship between turbidity removal rate and flocculant dosage provided by an embodiment of the present invention;

[0063] 5 is a schematic flow chart of a method for determining a first increasing step value of a preset turbidity state according to an embodiment of the present invention;

[0064] 6 is a flow chart of a method for determining a preset temperature and turbidity state according to an embodiment of the present invention;

[0065] FIG7 is a schematic diagram of a real-time optimization system for flocculant dosage based on a multi-cup flocculation experiment provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The present invention will be further described in detail below with reference to specific examples, but the embodiments of the present invention are not limited thereto.

[0067] Example 1

[0068] There are two main types of existing flocculant dosage control methods. One is to predict turbidity through flocs generated during the coagulation process and then determine the flocculant dosage. In the paper "Dosage optimization of polyaluminum chloride by the application of convolutional neural network to the floc images captured injar tests" by Yamamura et al., the cup flocculation process is photographed, and the CNN neural network algorithm is used to train the model based on the floc images to achieve rapid prediction of the turbidity of the supernatant after sedimentation; in the patent "A method for establishing a coagulation intelligent monitoring linkage system" published by the China Patent Office on November 22, 2022, a camera is used to capture images of alum flowers in the coagulation tank, and combined with the influent parameters to jointly predict the flocculant dosage and the turbidity of the supernatant after sedimentation at the tail of the horizontal sedimentation tank. Another method is to use the historical operation data of the water plant to predict the flocculant dosage. In the paper "Coagulant dosage determination using deep learning-based graph attention multivariate time series forecasting model" by Lin et al., a graph attention neural network algorithm is used to train based on the historical data of the water plant to predict the flocculant dosage of the incoming water; in the patent "Coagulation and sedimentation control method, device, equipment and storage medium" published by the China Patent Office on March 14, 2023, the inlet parameters of the water inlet end of the sedimentation container in the preset water purification plant and the preset dosage of the preset coagulant are first obtained; according to the inlet parameters and dosage, the pre-trained turbidity time series prediction model is used for processing to obtain the predicted value of the effluent turbidity at the outlet end of the sedimentation container; according to the predicted value of the effluent turbidity, the preset dosage is adjusted to obtain the target dosage; according to the target dosage, the dosing equipment is controlled to add the preset coagulant of the target dosage into the sedimentation container.

[0069] However, the problem with the above method is that in the cup flocculation process, a group of continuously taken floc images have a clear time sequence between the images, while the CNN network predicts the samples as isolated individuals. Therefore, simply using the CNN neural network for turbidity prediction cannot effectively utilize the time sequence information of the floc images. At the same time, simply predicting the turbidity cannot achieve the regulation and optimization of the dosage. Using an underwater camera to photograph the alum flowers at the water inlet of the sedimentation tank, the quality of the alum flower photos is difficult to guarantee due to the dim light underwater. In addition, the humid environment in the water will also cause great damage to the service life of equipment such as cameras, resulting in more frequent equipment replacement and higher use costs. The model trained using the historical operation data of the water plant can only determine the historical dosage corresponding to the current water quality based on the inlet water parameters, but cannot fit a better flocculant dosage relationship, that is, it cannot achieve optimization of the dosage.

[0070] Based on the above reasons, the present invention proposes a real-time optimization method for flocculant dosage based on a multi-cup flocculation experiment. Please refer to Figures 1 and 2. Figure 1 is a flow chart of a real-time optimization method for flocculant dosage based on a multi-cup flocculation experiment provided by an embodiment of the present invention. Figure 2 is a model diagram of a real-time optimization method for flocculant dosage based on a multi-cup flocculation experiment provided by an embodiment of the present invention. The real-time optimization method for flocculant dosage includes:

[0071] Step 1: Obtain the dosage of flocculant and coagulant aid during the first run, and use it as the actual dosage of flocculant P-real. The actual dosage of flocculant P-real is used as the basic dosage of flocculant in the first round of the jar flocculation experiment (Jar Test).

[0072] Specifically, when the method of the present invention is run for the first time, the current dosage P-real of the flocculant and coagulant aid of the water plant is read as the basic dosage P0 of the flocculant in the cup flocculation experiment.

[0073] Step 2. Based on the basic flocculant dosage of the r-th round cup-jar flocculation experiment, obtain the first set of floc images and water quality index data, the second set of floc images and water quality index data, and the third set of floc images and water quality index data corresponding to the first intelligent Jar Test module, the second intelligent Jar Test module, and the third intelligent Jar Test module that simultaneously perform flocculation experiments in the r-th round cup-jar flocculation experiment, respectively. Among them, the flocculant dosage Pd1(r) of the r-th round cup-jar flocculation experiment of the first intelligent Jar Test module is less than the current actual flocculant dosage, the flocculant dosage Pd2(r) of the r-th round cup-jar flocculation experiment of the second intelligent Jar Test module is equal to the current actual flocculant dosage, and the flocculant dosage Pd3(r) of the r-th round cup-jar flocculation experiment of the third intelligent Jar Test module is greater than the current actual flocculant dosage.

[0074] Here, the initialization parameters are P0 = P-real, P1 = P0*L, P2 = P0*K, and P3 = P0*H, where L is the dosage reduction coefficient, K is the dosage base coefficient, and H is the dosage increase coefficient. L < 1, K = 1, and H > 1. Flocculants were then added to the three intelligent Jar Test modules at dosages of P1, P2, and P3, respectively, to conduct flocculation experiments.

[0075] Three smart Jar Test modules (i.e., the first smart Jar Test module, the second smart Jar Test module, and the third smart Jar Test module) simultaneously start flocculation and capture floc images and collect water quality index data. After M minutes, the first set of floc images and water quality index data obtained by the first smart Jar Test module, the second set of floc images and water quality index data obtained by the second smart Jar Test module, and the third set of floc images and water quality index data obtained by the third smart Jar Test module are obtained.

[0076] Furthermore, the flocculant dosage Pd1(r) of the rth round of cup-jar flocculation experiment is Pd0(r-1)*L, that is, the flocculant dosage is lower than the current dosage of the water plant, the flocculant dosage is Pd0(r-1)*K, that is, the flocculant dosage is the same as the current dosage of the water plant, and the flocculant dosage Pd3 is Pd0(r-1)*H, that is, the flocculant dosage is higher than the current dosage of the water plant, among which Pd0(r-1) is the basic flocculant dosage of the r-1th round of cup-jar flocculation experiment.

[0077] Here, each group of flocculent images has a number between 30 and 8,000, and the specific number of images depends on different deep learning models. Preferably, each group of flocculent images has a number of 900.

[0078] Here, water quality parameter data generally include turbidity, pH, temperature, TOC / COD, SS, sulfur and other indicator parameters, which can be set specifically according to the selected deep learning model.

[0079] Step 3. Input the first set of floc images and water quality index data, the second set of floc images and water quality index data, and the third set of floc images and water quality index data corresponding to the rth round of cup flocculation experiment into the pre-trained deep learning model, and obtain the supernatant turbidity prediction value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity prediction value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity prediction value Td3(r) after the flocculation experiment of the third intelligent Jar Test module of the rth round of cup flocculation experiment.

[0080] Here, during the flocculation process, the collected flocculation image data (ie, flocculent body image) and water quality index data will be stored in the server. When the collection time is M, the collection of the image and water quality index data will be stopped.

[0081] After collecting three sets of floc images and water quality index data, they are respectively input into a pre-trained deep learning model to obtain the corresponding supernatant turbidity prediction value. The deep learning model can be a CNN combined with an RNN network, or an autoencoder combined with an RNN network.

[0082] Optionally, the deep learning model is implemented using a convolutional neural network (CNN) combined with a recurrent neural network (RNN). Specifically, floc images are fed into the ResNet network within the CNN in chronological order. Image features are obtained from the output of the penultimate fully connected layer, or the previous flattening layer. These features are then combined with the water quality index data at the corresponding moment into a new tensor, which is then fed into the RNN. The RNN then captures the temporal sequence information between these tensors, ultimately outputting a predicted turbidity value for the supernatant after flocculation. Preferably, the RNN uses an LSTM.

[0083] Optionally, the deep learning model is implemented using an image dimensionality reduction algorithm combined with a recurrent neural network (RNN). Specifically, floc images are sequentially input into the image dimensionality reduction algorithm module in chronological order. The image dimensionality reduction algorithm extracts image features and outputs them in reduced dimensions. The reduced image features are then combined with the corresponding water quality index data to form a new tensor, which is then input into the RNN. The RNN then acquires temporal information between the tensors and finally outputs a predicted turbidity value for the supernatant after flocculation. Preferably, the image dimensionality reduction algorithm uses autoencoder technology, and the RNN uses LSTM technology.

[0084] In this example, three intelligent Jar Test modules simultaneously initiate flocculation, capture floc images, and collect water quality data. After M minutes, these three sets of floc images and water quality data are fed into a pre-trained deep learning model to generate predicted turbidity values ​​for the supernatant after flocculation. Based on these three predicted turbidity values, the new actual flocculant dosage, PE, and the baseline dosage, Pnew, for the next Jar Test round are calculated. The time M can range from 5 to 15 minutes, depending on the performance of the deep learning model.

[0085] Step 4: Obtain a new actual flocculant dosage based on the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and use the new actual flocculant dosage as the basic flocculant dosage for the r+1th round. Repeat steps 2 to 4 to determine the new actual flocculant dosage for the r+1th round.

[0086] In a specific embodiment, referring to FIG3 , step 4 may specifically include:

[0087] Determine the relationship between the predicted turbidity value Td1(r) of the supernatant after the flocculation experiment of the first intelligent Jar Test module, the predicted turbidity value Td2(r) of the supernatant after the flocculation experiment of the second intelligent Jar Test module, and the predicted turbidity value Td3(r) of the supernatant after the flocculation experiment of the third intelligent Jar Test module, and the supernatant turbidity standard Ts after each round of flocculation experiment;

[0088] If Td1(r)<Ts, the actual dosage of flocculant is adjusted to the dosage of flocculant Pd1(r) in the rth round of cup flocculation experiment;

[0089] If Td1(r)≥Ts and Td2(r)<Ts, the actual dosage of flocculant is the dosage of flocculant Pd2(r) in the rth round of cup flocculation experiment;

[0090] If Td2(r)≥Ts and Td3(r)<Ts, the actual flocculant dosage is adjusted to the flocculant dosage Pd3(r) of the rth round of cup flocculation experiment;

[0091] If Td1(r)>Ts, Td2(r)>Ts and Td3(r)≥Ts, the actual dosage of the new flocculant is determined based on the preset relationship between the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module and the supernatant turbidity standard Ts after each round of flocculation experiment.

[0092] Here, the supernatant turbidity standard Ts after each round of flocculation experiment, that is, the turbidity standard that the turbidity after sedimentation must meet, the value of Ts can be set by the water plant based on the national standard and the actual working conditions of the water plant. Usually the value is lower than the national standard, which means it is better than the national standard.

[0093] Optionally, Ts=0.8NTU.

[0094] In this embodiment, the actual dosage of the new flocculant is determined based on the preset relationship between the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and the supernatant turbidity standard Ts after each round of flocculation experiment, including:

[0095] Determine the preset relationship between the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and the supernatant turbidity standard Ts after each round of flocculation experiment;

[0096] (1) If the preset relationship is Td1(r)>Td2(r)>Td3(r)≥Ts, the turbidity removal rate is in a monotonically rising zone, and it is determined whether the state is in the preset turbidity state. If it is in the preset turbidity state, the actual dosage of the flocculant is adjusted to Pd3(r)*Q H If it is not in the preset turbidity state, continue to judge whether it is in the preset temperature and turbidity state. If it is not in the preset temperature and turbidity state, the actual dosage of flocculant is adjusted to Pd3(r)*Q. If it is in the preset temperature and turbidity state, then Pd3(r)*Q L , where Q is the dosage increase step, Q H The first increment step of the preset turbidity state, Q L This is the second increment step for the preset turbidity state.

[0097] Here, the turbidity removal rate represents the proportion of the decrease in the turbidity after sedimentation compared to the initial turbidity Tini of the influent water after the flocculation experiment. At the same initial turbidity, the higher the turbidity removal rate, the lower the turbidity after sedimentation of the water after the flocculation experiment. The initial turbidity Tini is the initial influent turbidity, and this influent water is the raw water to be treated in the water plant. Please refer to Figure 4. The turbidity removal rate and the flocculant dosage show a non-linear relationship. When the flocculant dosage is small, as the flocculant dosage increases, the turbidity removal rate increases accordingly, and the turbidity after sedimentation continuously decreases. As the flocculant dosage continues to increase, the turbidity removal rate will increase to a peak value, corresponding to the lowest turbidity after sedimentation at this time. After that, as the flocculant dosage continues to increase, the turbidity removal rate will start to decline, and the turbidity after sedimentation will start to rise.

[0098] When Td1(r) > Td2(r) > Td3(r), it indicates that the larger the flocculant dosage, the higher the turbidity removal rate. Because Pd1(r) < Pd2(r) < Pd3(r), this shows that at the maximum flocculant dosage, the dosage "P-peak" corresponding to the peak value of the turbidity removal rate has not been reached. As the dosage increases, the turbidity removal rate will increase monotonically, that is, at this time, the turbidity removal rate is in the monotonic climbing area.

[0099] Here, the preset turbidity state is a high-turbidity state, and the preset temperature and turbidity state is a low-temperature and low-turbidity state.

[0100] When the turbidity removal rate is in the removal rate monotonic climbing area and enters the high-turbidity state at the same time, then it is judged as the "high-turbidity mode", and the new flocculant dosage (i.e., the new actual flocculant dosage) PE = Pd3(r) * Q H , Q H The value of is determined by the "high-turbidity mode".

[0101] Furthermore, judging whether the state is in the preset turbidity state includes:

[0102] Judging whether the climbing speed Z of two consecutive rounds is greater than the turbidity climbing rate b. If not, it is not in the preset turbidity state. If so, it is in the preset turbidity state. Here, Z = abs[Tini(r) - Tini(r - 1)] / Tini(r - 1)], abs is the absolute value, Tini(r) is the average influent turbidity of the r-th round of beaker flocculation experiment, and Tini(r - 1) is the average influent turbidity of the (r - 1)-th round of beaker flocculation experiment;

[0103] Among them, when the state is in the preset turbidity state, it continues to judge whether to jump out of the preset turbidity state. If Tini(r - 3) > Tini(r), and at the same time Tini(r - 2) > Tini(r), and at the same time Tini(r) < c, then it jumps out of the preset turbidity state; otherwise, it remains in the preset turbidity state. Here, Tini(r - 3) is the average influent turbidity of the (r - 3)-th round of beaker flocculation experiment, Tini(r - 2) is the average influent turbidity of the (r - 2)-th round of beaker flocculation experiment, and c is the turbidity when jumping out of the preset turbidity state.

[0104] Preferably, c = 25 NTU.

[0105] Here, the turbidity climbing rate is b, and preferably b = 1.2 is set.

[0106] In Figure 3, HT is the identifier for entering the high turbidity mode. If HT = 1, it means it is already in the high turbidity state; if HT = 0, it means it is not in the high turbidity state.

[0107] In this embodiment, the dosing increase step is Q = [(Tini(r) - Ts) / (Tini(r) - Td3(r)] * m, where m is the slope correction coefficient; preferably, m = 1.1.

[0108] In this embodiment, the second increase step of the preset turbidity state is Q L = 1 + (Q cur - 1) / f, where f is the Q value adjustment coefficient under the preset temperature and turbidity state, and Q cur is the Q value of the current round of beaker flocculation experiment; preferably, f = 10.

[0109] In this embodiment, please refer to Figure ⑤. The method for obtaining the value of the first increase step of the preset turbidity state includes:

[0110] Judge whether the average influent turbidity Tini(r - 1) of the (r - 1)-th round of beaker flocculation experiment is greater than the average influent turbidity Tini(r - 2) of the (r - 2)-th round of beaker flocculation experiment;

[0111] If the average influent turbidity Tini(r - 1) of the (r - 1)-th round of beaker flocculation experiment is greater than the average influent turbidity Tini(r - 2) of the (r - 2)-th round of beaker flocculation experiment, then continue to judge whether the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is greater than the average influent turbidity Tini(r - 1) of the (r - 1)-th round of beaker flocculation experiment. If so, then this is the soaring section of the high turbidity state, and Q H = Qr; if not, then this is the peak area of the high turbidity state, and Q H = Qp; where Qr is the increase step in the soaring section, and Qp is the increase step in the peak area.

[0112] If the average turbidity of the influent water Tini(r-1) in the r-1th round of cup-jar flocculation experiment is not greater than the average turbidity of the influent water Tini(r-2) in the r-2th round of cup-jar flocculation experiment, then continue to judge whether the average turbidity of the influent water Tini(r) in the rth round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-1) in the r-1th round of cup-jar flocculation experiment. If not, then this is the fallback zone of the high turbidity state, then Q H =Qd, if so, then Dell, Dell refers to the empty state, that is, the state that will not appear in practice, so no operation is required, Qd is the increase step of the fallback segment;

[0113] Where, Qr = Qr1*n, Qp = Q, Qd = Qd1*n, Qr1 = (Tpre1-Ts) / [Tini(r)-Ts], Tpre1 = Tini(r)+kpre*yr, Qd1 = (Tpre2-Ts) / [Tini(r)-Ts], Tpre2 = Tini(r)+kpre*yd, Tpre is the predicted value of the initial turbidity of the raw water in the r+1th round, kpre is the predicted change value between the initial turbidity of the raw water in the r+1th round and the initial turbidity of the raw water in the rth round, kpre = k( r)+[k(r)-k(r-1)], K(r) is the change value between the initial turbidity of the raw water in the rth round and the initial turbidity of the raw water in the r-1th round, k(r)=Tini(r)-Tini(r-1), K(r-1) is the change value between the initial turbidity of the raw water in the r-1th round and the initial turbidity of the raw water in the r-2th round, k(r-1)=Tini(r-1)-Tini(r-2), yr is the first correction coefficient, that is, the correction coefficient for the prediction of the soaring segment, yd is the second correction coefficient, that is, the correction coefficient for the prediction of the falling segment, and n is the correction coefficient for Q.

[0114] Optionally, the value of yd is between 0.8 and 3, the value of yr is between 0.3 and 2, and the value of n is between 0.8 and 2.

[0115] Preferably, yd=1.2, yr=0.8, and n=1.1.

[0116] Further, referring to FIG6 , determining whether the temperature and turbidity are in the preset state includes:

[0117] Determine whether the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and whether the average inlet turbidity Tini(r) of the rth round cup flocculation experiment is less than the low turbidity judgment standard Tini_L. If the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and the average inlet turbidity Tini(r) of the rth round cup flocculation experiment is less than the low turbidity judgment standard Tini_L, then the state is in the preset temperature and turbidity state; otherwise, it is not in the preset temperature and turbidity state.

[0118] Here, when the inlet water temperature is relatively low and the turbidity is also relatively low, it enters the "low temperature and low turbidity state". When entering the low temperature and low turbidity state, the identifier DL of the low temperature and low turbidity mode is equal to 1.

[0119] That is to say, if the temperature Temp < Temp-L and the average inlet turbidity Tini(r) < Tini_L, it enters the low temperature and low turbidity state. If Temp > Temp_L or the average inlet turbidity Tini(r) > Tini_L, it jumps out of the low temperature and low turbidity state. When jumping out, the low temperature and low turbidity round variable Lr is set to 0, and L, H, and Q are set to the initialization data.

[0120] Preferably, Temp_L = 70°C and Tini_L = 4 NTU.

[0121] In this embodiment, the calculation method of the average inlet turbidity Tini(r) of the r-th round of beaker flocculation experiment is: Tini(r) = (Tini1(r) + Tini2(r) + Tini3(r)) / 3, Tini1(r) = Tur1(r), Tini2(r) = Tur2(r), Tini3(r) = Tur3(r), where Tini1(r) is the initial turbidity of the water sample in beaker No. 1 in the r-th round of beaker flocculation experiment, Tini2(r) is the initial turbidity of the water sample in beaker No. 2 in the r-th round of beaker flocculation experiment, Tini3(r) is the initial turbidity of the water sample in beaker No. 3 in the r-th round of beaker flocculation experiment, Tur1(r) is the turbidity of the water in beaker No. 1 measured by the turbidity probe in the r-th round of beaker flocculation experiment, Tur2(r) is the turbidity of the water in beaker No. 2 measured by the turbidity probe in the r-th round of beaker flocculation experiment, and Tur3(r) is the turbidity of the water in beaker No. 3 measured by the turbidity probe in the r-th round of beaker flocculation experiment.

[0122] (2) If the preset relationship is Ts < Td1(r) < Td2(r) < Td3(r), then the turbidity removal rate is in the monotonically decreasing region, and the actual flocculant dosage is adjusted to Pd1(r) / H, and the values of H and L for the next round of beaker flocculation experiment are determined according to whether the state is in the preset temperature and turbidity state.

[0123] In this embodiment, determining the values of H and L for the next round of beaker flocculation experiment according to whether the state is in the preset temperature and turbidity state includes:

[0124] Judging whether the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L and whether the average inlet turbidity Tini(r) of the r-th round of beaker flocculation experiment is less than the low turbidity judgment standard Tini_L;

[0125] If the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and at the same time the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is less than the low turbidity judgment standard Tini_L, the state is in the preset temperature and turbidity state. When first entering the preset temperature and turbidity state, the value of L is updated to LL, the value of H is updated to HL, and LL and HL are used as the values of H and L for the next round. Where LL = L cur -(1 - L cur ) / s, HL = 1 + (H cur -1) / s, L cur is the value of L in the current round of beaker flocculation experiment, H cur is the value of H in the current round of beaker flocculation experiment, s is the adjustment coefficient, specifically the adjustment coefficient of L and H in the preset temperature and turbidity state, f is the Q value adjustment coefficient in the preset temperature and turbidity state. When not first entering the preset temperature and turbidity state, the values of L, H, and Q L remain unchanged. LL is the reduction coefficient of the dosage in the preset temperature and turbidity state, and HL is the increase coefficient of the dosage in the preset temperature and turbidity state;

[0126] If the monitored water temperature data Temp is greater than the low temperature judgment standard Temp-L, or the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is greater than the low turbidity judgment standard Tini_L, jump out of the preset temperature and turbidity state, set L, H, and Q to the initial data, and at the same time set the number of times of entering the preset temperature and turbidity state to 0.

[0127] Preferably, s = 5, f = 10.

[0128] (3) If the preset relationship is min(Td1(r), Td2(r), Td3(r)) > Ts, Td1(r) > Td(r), Td2(r) < Td3(r), or min(Td1(r), Td2(r), Td3(r)) > Ts, Td1(r) < Td(r), Td2(r) > Td3(r), then the turbidity removal rate is in the non-monotonic climbing region. At this time, determine the new actual dosage of the flocculant according to the relationship between the predicted values of the supernatant turbidity Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted value of the supernatant turbidity Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted value of the supernatant turbidity Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and determine the values of H and L for the next round of beaker flocculation experiment according to whether the turbidity is in the preset temperature and turbidity state.

[0129] In this embodiment, the new actual dosage of the flocculant is determined according to the relationship among the predicted supernatant turbidity values Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity values Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity values Td3(r) after the flocculation experiment of the third intelligent Jar Test module, including:

[0130] Judge the relationship among the predicted supernatant turbidity values Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity values Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity values Td3(r) after the flocculation experiment of the third intelligent Jar Test module;

[0131] If Td1(r)>Td2(r) and Td2(r)<Td3(r), then the actual dosage of the flocculant is adjusted to [Pd₂(r)+Pd₃(r)] / 2;

[0132] If Td1(r)<Td2(r) and Td2(r)>Td3(r), then the actual dosage of the flocculant is adjusted to [Pd₁(r)+Pd₂(r)] / 2.

[0133] Finally, the new dosage of the flocculant PE (i.e., the new actual dosage of the flocculant) is determined by the method provided in this embodiment. Then, the parameter PE is passed to the Jar Test module to conduct a new round of experiments. Pnew(r)=PE, where Pnew(r) is the new dosage in the r-th round. At the same time, Pnew(r) is passed to the control system of the dosage added to the water plant. Specifically, the judgment conditions can be set according to the actual situation of the water plant to decide whether to change the actual dosage of the flocculant.

[0134] This paper proposes an iterative optimization method based on multiple jar flocculation experiments. This method uses three intelligent Jar Test modules to simultaneously conduct experiments, obtaining first, second, and third sets of floc images and water quality index data, respectively. Based on these three sets of floc images and water quality index data, a deep learning model is used to predict the turbidity of the multiple jar flocculation experiments, resulting in three predicted values ​​for the supernatant turbidity after the simultaneous flocculation experiments. These predicted values ​​are then used to determine the flocculant dosage for the next round of jar flocculation experiments, achieving iterative optimization of the flocculant dosage. This method uses a deep learning model to predict the flocculation effect and rapidly optimizes the dosage through multiple jar flocculation experiments. This method enables real-time adjustment and optimization of the flocculant dosage, addressing the current difficulties in flocculation dosing control, such as nonlinearity, time-varying behavior, and large time lags. This can also address the resulting problems of excessive flocculant dosage, making it difficult to address sudden changes in water quality and unusual water quality. The method provided by the present invention shortens the flocculation effect prediction time, optimizes effluent control, reduces the flocculant dosage, saves costs for water treatment companies, and reduces indirect carbon emissions.

[0135] Example 2

[0136] Please refer to FIG7 , which is a schematic diagram of a flocculant dosage real-time optimization system based on a multi-cup flocculation experiment provided by an embodiment of the present invention. In addition to the above embodiment, the present invention further provides a flocculant dosage real-time optimization system based on a multi-cup flocculation experiment. The flocculant dosage real-time optimization system includes:

[0137] The basic dosage acquisition module is used to obtain the dosage of flocculant and coagulant aid during the first run, which is used as the basic dosage of flocculant in the first round of cup flocculation experiment;

[0138] a data acquisition module for respectively acquiring, based on the basic flocculant dosage of the r-th round cup-jar flocculation experiment, the first intelligent Jar Test module, the second intelligent Jar Test module, the first set of flocculent images and water quality index data, the second set of flocculent images and water quality index data, and the third set of flocculent images and water quality index data corresponding to the first intelligent Jar Test module, which are simultaneously performing flocculation experiments in the r-th round cup-jar flocculation experiment, wherein the flocculant dosage Pd1(r) of the r-th round cup-jar flocculation experiment of the first intelligent Jar Test module is less than the current actual flocculant dosage, the flocculant dosage Pd2(r) of the r-th round cup-jar flocculation experiment of the second intelligent Jar Test module is equal to the current actual flocculant dosage, and the flocculant dosage Pd3(r) of the r-th round cup-jar flocculation experiment of the third intelligent Jar Test module is greater than the current actual flocculant dosage;

[0139] A prediction value acquisition module is used to input the first set of floc images and water quality index data, the second set of floc images and water quality index data, and the third set of floc images and water quality index data corresponding to the rth round of cup flocculation experiment into a pre-trained deep learning model, and correspondingly obtain the supernatant turbidity prediction value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity prediction value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity prediction value Td3(r) after the flocculation experiment of the third intelligent Jar Test module in the rth round of cup flocculation experiment;

[0140] The flocculant actual dosage acquisition module is used to obtain a new flocculant actual dosage based on the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and use the new flocculant actual dosage as the basic flocculant dosage for the r+1th round, and repeatedly execute the processing process from the data acquisition module to the flocculant actual dosage acquisition module to determine the new flocculant actual dosage for the r+1th round.

[0141] The flocculant dosage real-time optimization system provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effects are similar, which will not be described in detail here.

[0142] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0143] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described can be combined in any suitable manner in one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification.

[0144] Although the present invention has been described herein with reference to various embodiments, those skilled in the art may understand and implement other variations of the disclosed embodiments by reviewing the drawings and the disclosure in the process of implementing the claimed invention. In the specification, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple components or steps. The fact that certain measures are described in different embodiments does not mean that these measures cannot be combined to produce good results.

[0145] The above is a further detailed description of the present invention in conjunction with specific preferred embodiments, and the specific implementation of the present invention should not be considered to be limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be considered to fall within the scope of protection of the present invention.

Claims

1. A real-time optimization method for flocculant dosage based on multi-cup flocculation experiment, characterized in that: The real-time optimization method for flocculant dosage comprises: Step 1, obtaining the dosage of flocculant and coagulant aid during the first operation, and using it as the basic dosage of flocculant in the first round of cup flocculation experiment; Step 2: Based on the basic flocculant dosage of the r-th round cup-jar flocculation experiment, respectively obtain the first group of flocculent images and water quality index data, the second group of flocculent images and water quality index data, and the third group of flocculent images and water quality index data corresponding to the first intelligent Jar Test module, the second intelligent Jar Test module, and the third intelligent Jar Test module that simultaneously perform flocculation experiments in the r-th round cup-jar flocculation experiment, wherein the flocculant dosage Pd1(r) of the r-th round cup-jar flocculation experiment of the first intelligent Jar Test module is less than the current actual flocculant dosage, the flocculant dosage Pd2(r) of the r-th round cup-jar flocculation experiment of the second intelligent Jar Test module is equal to the current actual flocculant dosage, and the flocculant dosage Pd3(r) of the r-th round cup-jar flocculation experiment of the third intelligent Jar Test module is greater than the current actual flocculant dosage; Step 3, the first group of floc images and water quality index data, the second group of floc images and water quality index data, and the third group of floc images and water quality index data corresponding to the rth round of cup flocculation experiment are respectively input into the pre-trained deep learning model, and the turbidity prediction value Td1(r) of the supernatant after the flocculation experiment of the first intelligent Jar Test module, the turbidity prediction value Td2(r) of the supernatant after the flocculation experiment of the second intelligent Jar Test module, and the turbidity prediction value Td3(r) of the supernatant after the flocculation experiment of the third intelligent Jar Test module of the rth round of cup flocculation experiment are obtained; Step 4: Obtain a new actual flocculant dosage according to the predicted value Td1(r) of the supernatant turbidity after the flocculation experiment of the first intelligent Jar Test module, the predicted value Td2(r) of the supernatant turbidity after the flocculation experiment of the second intelligent Jar Test module, and the predicted value Td3(r) of the supernatant turbidity after the flocculation experiment of the third intelligent Jar Test module, and use the new actual flocculant dosage as the basic flocculant dosage for the r+1th round, and repeat steps 2 to 4 to determine the new actual flocculant dosage for the r+1th round.

2. The method for real-time optimization of flocculant dosage according to claim 1, characterized in that: The flocculant dosage Pd1(r) of the rth round of cup-jar flocculation experiment is Pd0(r-1)*L, the flocculant dosage Pd2(r) is Pd0(r-1)*K, and the flocculant dosage Pd3(r) is Pd0(r-1)*H, wherein Pd0(r-1) is the basic flocculant dosage of the r-1th round of cup-jar flocculation experiment, L is the dosage decrease coefficient, K is the dosage base coefficient, and H is the dosage increase coefficient, L<1, K=1, H>1.

3. The method for real-time optimization of flocculant dosage according to claim 2, characterized in that: The step 4 comprises: Judge the relationship between the predicted supernatant turbidity value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the predicted supernatant turbidity value Td3(r) after the flocculation experiment of the third intelligent Jar Test module and the supernatant turbidity standard Ts after each round of flocculation experiment; If Td1(r) < Ts, adjust the actual flocculant dosage to the flocculant dosage Pd1(r) in the r-th round of beaker flocculation experiment; If Td1(r) ≥ Ts and Td2(r) < Ts, the actual flocculant dosage is the flocculant dosage Pd2(r) in the r-th round of beaker flocculation experiment; If Td2(r) ≥ Ts and Td3(r) < Ts, adjust the actual flocculant dosage to the flocculant dosage Pd3(r) in the r-th round of beaker flocculation experiment; If Td1(r) > Ts, Td2(r) > Ts and Td3(r) ≥ Ts, determine the new actual flocculant dosage according to the preset relationship between the predicted supernatant turbidity value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the predicted supernatant turbidity value Td3(r) after the flocculation experiment of the third intelligent Jar Test module and the supernatant turbidity standard Ts after each round of flocculation experiment.

4. The method for real-time optimization of flocculant dosage according to claim 3, characterized in that: Determine the new actual flocculant dosage according to the preset relationship between the predicted supernatant turbidity value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the predicted supernatant turbidity value Td3(r) after the flocculation experiment of the third intelligent Jar Test module and the supernatant turbidity standard Ts after each round of flocculation experiment, including: Judge the preset relationship between the predicted supernatant turbidity value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, the predicted supernatant turbidity value Td3(r) after the flocculation experiment of the third intelligent Jar Test module and the supernatant turbidity standard Ts after each round of flocculation experiment; If the preset relationship is Td1(r)>Td2(r)>Td3(r)≥Ts, the turbidity removal rate is in a monotonically rising zone, and it is determined whether the state is in the preset turbidity state. If it is in the preset turbidity state, the actual dosage of the flocculant is adjusted to Pd3(r)*Q H If it is not in the preset turbidity state, continue to determine whether it is in the preset temperature and turbidity state. If it is not in the preset temperature and turbidity state, the actual dosage of the flocculant is adjusted to Pd3(r)*Q. If it is in the preset temperature and turbidity state, Pd3(r)*Q L , where Q is the dosage increase step, Q H The first increment step for the preset turbidity state, Q L A second increasing step for a preset turbidity state; If the preset relationship is Ts < Td1(r) < Td2(r) < Td3(r), the turbidity removal rate is in the monotonically decreasing region, then adjust the actual flocculant dosage to Pd1(r) / H, and determine the values of H and L for the next round of beaker flocculation experiment according to whether the state is in the preset temperature and turbidity state; If the preset relationship is min(Td1(r), Td2(r), Td3(r)) > Ts, Td1(r) > Td2(r), Td2(r) < Td3(r), or min(Td1(r), Td2(r), Td3(r)) > Ts, Td1(r) < Td2(r), Td2(r) > Td3(r), then the turbidity removal rate is in the non-monotonic climbing region. At this time, determine the new actual dosage of the flocculant according to the relationship between the predicted turbidity Td1(r) of the supernatant after the flocculation experiment of the first intelligent Jar Test module, the predicted turbidity Td2(r) of the supernatant after the flocculation experiment of the second intelligent Jar Test module, and the predicted turbidity Td3(r) of the supernatant after the flocculation experiment of the third intelligent Jar Test module, and determine the values of H and L for the next round of beaker flocculation experiment according to whether the turbidity is in the preset temperature and turbidity state.

5. The method for real-time optimization of flocculant dosage according to claim 4, characterized in that: Judging whether the state is in the preset turbidity state includes: Judging whether the climbing speed Z of two consecutive rounds is greater than the turbidity climbing rate b. If not, it is not in the preset turbidity state; if so, it is in the preset turbidity state, where Z = abs[Tini(r) - Tini(r - 1)] / Tini(r - 1)], abs is the absolute value operation, Tini(r) is the average influent turbidity of the r-th round of beaker flocculation experiment, and Tini(r - 1) is the average influent turbidity of the (r - 1)-th round of beaker flocculation experiment. Among them, when the state is in the preset turbidity state, continue to judge whether to jump out of the preset turbidity state. If Tini(r - 3) > Tini(r), and at the same time Tini(r - 2) > Tini(r), and at the same time Tini(r) < c, then jump out of the preset turbidity state; otherwise, it is still in the preset turbidity state, where Tini(r - 3) is the average influent turbidity of the (r - 3)-th round of beaker flocculation experiment, Tini(r - 2) is the average influent turbidity of the (r - 2)-th round of beaker flocculation experiment, and c is the turbidity when jumping out of the preset turbidity state.

6. The method for real-time optimization of flocculant dosage according to claim 4, characterized in that: Judging whether it is in the preset temperature and turbidity state includes: Judging whether the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and at the same time whether the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is less than the low turbidity judgment standard Tini_L. If the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and at the same time the average influent turbidity Tini(r) of the r-th round of beaker flocculation experiment is less than the low turbidity judgment standard Tini_L, then the state is in the preset temperature and turbidity state; otherwise, it is not in the preset temperature and turbidity state.

7. [Corrected 02.09.2024 according to Rule 26] The method for real-time optimization of flocculant dosage according to claim 4, characterized in that: The dosage increase step is Q = [(Tini (r) - Ts) / (Tini (r) - Td3 (r)] * m, m is the slope correction coefficient; the second increase step of the preset turbidity state is Q L =1+(Q cur -1) / f, f is the Q value adjustment coefficient under the preset temperature and turbidity conditions, Q cur is the Q value of the current round of cup flocculation experiment; The value-taking method of the first increasing step of the preset turbidity state includes: Judging whether the average influent turbidity Tini(r - 1) of the (r - 1)-th round of beaker flocculation experiment is greater than the average influent turbidity Tini(r - 2) of the (r - 2)-th round of beaker flocculation experiment. If the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-2) of the r-2th round of cup-jar flocculation experiment, then continue to judge whether the average turbidity of the influent water Tini(r) of the rth round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment. If so, then Q H =Qr, if not, then Q H =Qp; If the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment is not greater than the average turbidity of the influent water Tini(r-2) of the r-2th round of cup-jar flocculation experiment, then continue to judge whether the average turbidity of the influent water Tini(r) of the rth round of cup-jar flocculation experiment is greater than the average turbidity of the influent water Tini(r-1) of the r-1th round of cup-jar flocculation experiment. If not, then Q H =Qd; Wherein, Qr = Qr1 * n, Qp = Q, Qd = Qd1 * n, Qr1 = (Tpre1 - Ts) / [Tini(r) - Ts], Tpre1 = Tini(r) + kpre * yr, Qd1 = (Tpre2 - Ts) / [Tini(r) - Ts], Tpre2 = Tini(r) + kpre * yd, kpre = k(r) + [k(r) - k(r - 1)], k(r) = Tini(r) - Tini(r - 1), k(r - 1) = Tini(r - 1) - Tini(r - 2), yr is the first correction coefficient, yd is the second correction coefficient, and n is the correction coefficient of Q.

8. [Corrected 02.09.2024 according to Rule 26] The method for real-time optimization of flocculant dosage according to claim 4, characterized in that: Determine the values of H and L for the next round of Jar Test flocculation experiment according to whether the state is within the preset temperature and turbidity state, including: Judge whether the monitored water temperature data Temp is less than the low-temperature judgment standard Temp-L, and whether the average influent turbidity Tini(r) of the r-th round of Jar Test flocculation experiment is less than the low-turbidity judgment standard Tini_L; If the monitored water temperature data Temp is less than the low temperature judgment standard Temp-L, and the average turbidity of the inlet water Tini(r) of the rth round of cup flocculation experiment is less than the low turbidity judgment standard Tini_L, then the state is in the preset temperature and turbidity state. When entering the preset temperature and turbidity state for the first time, the value of L is updated to LL, the value of H is updated to HL, and LL and HL are used as the values ​​of H and L in the next round, where LL = L cur -(1-L cur ) / s, HL=1+(H cur -1) / s,L cur is the L value of the current round of cup flocculation experiment, H cur is the H value of the current round of cup flocculation experiment, s is the adjustment coefficient, and f is the Q value adjustment coefficient under the preset temperature and turbidity state. When it is not the first time to enter the preset temperature and turbidity state, L, H, Q L The value of remains unchanged; If the monitored water temperature data Temp is greater than or equal to the low-temperature judgment standard Temp-L, or the average influent turbidity Tini(r) of the r-th round of Jar Test flocculation experiment is greater than or equal to the low-turbidity judgment standard Tini_L, then L, H, and Q are set to the initialization data, and at the same time, the number of times entering the preset temperature and turbidity state is set to 0.

9. The method for real-time optimization of flocculant dosage according to claim 4, characterized in that: Determine the actual dosage of the new flocculant according to the relationship between the predicted supernatant turbidity Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity Td3(r) after the flocculation experiment of the third intelligent Jar Test module, including: Judge the relationship between the predicted supernatant turbidity Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the predicted supernatant turbidity Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the predicted supernatant turbidity Td3(r) after the flocculation experiment of the third intelligent Jar Test module; If Td1(r) > Td2(r) and Td2(r) < Td3(r), then the actual dosage of the flocculant is adjusted to [Pd2(r) + Pd3(r)] / 2; If Td1(r) < Td2(r) and Td2(r) > Td3(r), then the actual dosage of the flocculant is adjusted to [Pd1(r) + Pd2(r)] / 2.

10. A real-time optimization system for flocculant dosage based on multi-cup flocculation experiment, characterized in that: The real-time optimization system for flocculant dosage includes: A basic dosage acquisition module, which is used to acquire the dosage of the flocculant and the coagulant aid during the first run as the basic dosage of the flocculant for the first round of Jar Test flocculation experiment; A data acquisition module, for respectively acquiring the first intelligent Jar Test module, the second intelligent Jar Test module, the first group of flocculent images and water quality index data, the second group of flocculent images and water quality index data, and the third group of flocculent images and water quality index data corresponding to the first intelligent Jar Test module, which simultaneously perform flocculation experiments in the r-round cup-jar flocculation experiment, based on the basic flocculant dosage of the r-round cup-jar flocculation experiment, wherein the flocculant dosage Pd1(r) of the r-round cup-jar flocculation experiment of the first intelligent Jar Test module is less than the current actual flocculant dosage, the flocculant dosage Pd2(r) of the r-round cup-jar flocculation experiment of the second intelligent Jar Test module is equal to the current actual flocculant dosage, and the flocculant dosage Pd3(r) of the r-round cup-jar flocculation experiment of the third intelligent Jar Test module is greater than the current actual flocculant dosage; A prediction value acquisition module, used to input the first group of floc images and water quality index data, the second group of floc images and water quality index data, and the third group of floc images and water quality index data corresponding to the rth round of cup flocculation experiment into a pre-trained deep learning model, and correspondingly obtain the supernatant turbidity prediction value Td1(r) of the first intelligent Jar Test module after the flocculation experiment, the supernatant turbidity prediction value Td2(r) of the second intelligent Jar Test module after the flocculation experiment, and the supernatant turbidity prediction value Td3(r) of the third intelligent Jar Test module after the flocculation experiment of the rth round of cup flocculation experiment; The actual flocculant dosage acquisition module is used to obtain a new actual flocculant dosage according to the supernatant turbidity predicted value Td1(r) after the flocculation experiment of the first intelligent Jar Test module, the supernatant turbidity predicted value Td2(r) after the flocculation experiment of the second intelligent Jar Test module, and the supernatant turbidity predicted value Td3(r) after the flocculation experiment of the third intelligent Jar Test module, and use the new actual flocculant dosage as the basic flocculant dosage for the r+1th round, and repeatedly execute the processing process from the data acquisition module to the flocculant actual dosage acquisition module to determine the new actual flocculant dosage for the r+1th round.

Images