A method for controlling alum addition in a water treatment plant based on feedforward-feedback composite control

Abstract

The present invention provides a method for controlling alum addition in a water treatment plant based on feedforward-feedback composite control, which includes two parts: pre-alum addition and post-alum addition. Pre-alum addition part: First, aiming at the seasonality of the raw water quality change, a feedforward prediction model and a prediction model for the turbidity of the sedimentation tank effluent are established, and thus the feedforward alum addition amount and the predicted value of the turbidity of the sedimentation tank effluent are obtained; then, aiming at the dynamic time-variability of the raw water quality change, the predicted value of the turbidity of the sedimentation tank effluent is corrected, and the feedback alum addition amount is calculated; finally, the feedforward and feedback alum addition amounts are intelligently weighted to obtain the pre-alum addition amount. Post-alum addition part: First, when it is continuously monitored that the pre-alum addition amount and the turbidity of the sedimentation tank effluent in the recent 30 minutes are higher than the threshold, the post-alum addition process is started; then, the relationship curve between the turbidity of the sedimentation tank effluent and the post-alum addition amount is queried to obtain the feedforward alum addition amount; next, the feedback alum addition amount is calculated; finally, the feedforward and feedback alum addition amounts are intelligently weighted to obtain the post-alum addition amount.

Description

Technical Field

[0001] The present invention relates to a method for controlling alum addition in a waterworks, and particularly to a method for controlling alum addition in a waterworks based on feedforward-feedback composite control. Background Art

[0002] The alum addition process is an important link in the water purification of a waterworks. The main function of alum addition is to remove suspended impurities, colloidal particles and harmful substances such as bacteria and viruses attached to the colloidal particles in the turbid water. Strengthening the effective control of the operation of the alum addition process and strictly limiting the turbidity of the effluent from the sedimentation tank and sand filter is beneficial to the effective removal of viruses and can ensure the safety of the water quality of the water leaving the factory.

[0003] Currently, for the manual control based on a fixed ratio of water flow or the traditional feedback control method, the biggest problem lies in the lag time of at least two hours from alum dosage addition, coagulation and sedimentation to sand filtration. Therefore, for such a large-lag process, when the raw water quality suddenly changes (such as caused by heavy rain, strong wind weather and upstream enterprise sewage discharge), it is difficult to make timely and accurate adjustments to the alum dosage, which greatly affects the turbidity removal effect of the alum addition process and even leads to the exceeding of the turbidity of the effluent for a period of time, endangering the drinking water safety of residents. Therefore, it is very crucial to establish a method for controlling alum addition that can adapt to sudden changes in raw water quality and the large-lag characteristics of the alum addition process.

[0004] The research object of the present invention - adopting the manual control method based on a fixed ratio of water flow, the alum dosage cannot be adjusted in time according to sudden changes in raw water quality, it is difficult to ensure the turbidity removal effect of the effluent from the sedimentation tank and sand filter, and it affects the operation of the subsequent process and the safety of the water quality of the water leaving the factory. Summary of the Invention

[0005] Object of the Invention

[0006] According to the process conditions of the waterworks which is the research object of the present invention, the present invention solves the problem of the operation control of the alum addition process with two parts of pre-alum addition and post-alum addition, and proposes a feedforward-feedback composite control method that can quickly adjust the alum dosage according to sudden changes in raw water quality and adapt to the alum addition process with large-lag characteristics, effectively improving the treatment effect of the alum addition process.

[0007] Technical Solution

[0008] The present invention adopts the following technical solution to solve the above technical problems:

[0009] A method for controlling alum addition in a waterworks based on feedforward-feedback composite control includes the following steps:

[0010] Step (I), the specific steps of the pre-alum addition feedforward control are as follows:

[0011] Step 1: In the water plant production operation database, screen out the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and historical data of pre - alum addition amount corresponding to the moments (before the lag time τ of pre - alum addition) within the range that meets the requirements of the sedimentation tank effluent turbidity, and perform data filtering processing:

[0012]

[0013] Where: is the data filtering value at time t, A(t) is the data acquisition value at time t, k i is the weight coefficient, and l is the filtering time length;

[0014] Step 2: Divide the historical data samples obtained after data filtering in Step 1 into two parts: a training set and a test set;

[0015] Step 3: Using the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity) and influent flow rate as input variables, and the pre - alum addition amount as the output variable, based on the historical data samples of the training set and test set obtained in Step 2, adopt the random forest algorithm to train a pre - alum addition feed - forward prediction model;

[0016] Step 4: Use the data filtering method in Step 1 to obtain the real - time data filtering values of the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity) and influent flow rate, input them into the pre - alum addition feed - forward prediction model obtained in Step 3, and calculate the real - time pre - alum addition feed - forward addition amount u FF (t).

[0017] Step (II), the specific steps of pre - alum addition feedback control are as follows:

[0018] Step 1: In the water plant production operation database, screen out the sedimentation tank effluent turbidity within a certain range and its corresponding raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and historical data of pre - alum addition amount at the corresponding moments (before the lag time τ of pre - alum addition), and perform data filtering processing using the data filtering method in Step 1 of Step (I);

[0019] Step 2: Divide the historical data samples obtained after data filtering in Step 1 into two parts: a training set and a test set;

[0020] Step 3: Using the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and pre - alum addition amount as input variables, and the sedimentation tank effluent turbidity as the output variable, based on the historical data samples of the training set and test set obtained in Step 2, adopt the random forest algorithm to train a sedimentation tank effluent turbidity prediction model;

[0021] Step 4: Obtain the real-time data filtering values of the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and pre-alum dosage by using the data filtering method in Step 1 of Step (I), and input them into the sedimentation tank effluent turbidity prediction model obtained in Step 3 to calculate the real-time predicted sedimentation tank effluent turbidity;

[0022] Step 5: Since the sedimentation tank influent flow rate is inversely proportional to the residence time, taking the research object water plant as an example, the measured sedimentation tank residence time corresponding to an influent flow rate of 3100 cubic meters per hour is about 2.3 hours. Therefore, the lag time (i.e., the residence time) can be calculated according to the current influent flow rate Q:

[0023]

[0024] Step 6: Obtain the real-time data filtering value of the sedimentation tank effluent turbidity by using the data filtering method in Step 1 of Step (I), and obtain the sedimentation tank effluent turbidity values at the current time t and its previous l times (y(t), y(t - 1), y(t - 2), …, y(t - l)) in real time. Then, (y(t), y(t - 1), y(t - 2), …, y(t - l)) corresponds to the real-time raw water quality, influent flow rate, and pre-alum dosage conditions at time t - τ and its previous l times. Let the turbidity prediction value corresponding to the real-time raw water quality, influent flow rate, and pre-alum dosage conditions at the same time t - τ and its previous l times be Therefore The deviation of can be used as the offset for correcting the turbidity prediction value corresponding to the real-time raw water quality, influent flow rate, and pre-alum dosage conditions at the current time t :

[0025]

[0026] In the formula: is the corrected turbidity prediction value, and α i is the weight coefficient, and l is the correction time length.

[0027] Step 7: According to the deviation between the corrected turbidity prediction value and the set value, calculate the real-time pre-alum feedback dosage by using PI feedback control:

[0028]

[0029] In the formula: u FB (t - 1) is the pre-alum feedback dosage at the previous time, is the corrected turbidity prediction value, y ref is the turbidity set value, λ is the weight coefficient, and β i is the gain coefficient, and l is the integral time length.

[0030] Step (III), the pre - alum addition feed - forward and feedback compound control step is as follows:

[0031] Step 1: Track and monitor the pre - alum addition feed - forward dosing amount obtained in Step (I), the pre - alum addition feedback dosing amount obtained in Step (II), and the turbidity of the sedimentation tank effluent (obtained by filtering the readings of the turbidity meter of the sedimentation tank effluent according to the data in Step 1 of Step (I)). When it is monitored that the pre - alum addition feed - forward dosing amount or (and) the pre - alum addition feedback dosing amount has been continuously increasing within the most recent 1 hour, while the turbidity of the sedimentation tank effluent is continuously higher than the threshold value, it is determined that the pre - alum addition feed - forward dosing amount tends to be inaccurate at this time (the pre - alum addition feed - forward prediction model is mismatched or an unprecedented water quality mutation phenomenon occurs). Gradually increase the weight of the pre - alum addition feedback dosing amount (adjustment step: 5% per time, adjustment period: 1 hour per time), and decrease the weight of the pre - alum addition feed - forward dosing amount (adjustment step: 5% per time, adjustment period: 1 hour per time). When it is monitored that the pre - alum addition feed - forward dosing amount and the pre - alum addition feedback dosing amount have remained unchanged or changed little within the most recent 2 hours, and the turbidity of the sedimentation tank effluent is also continuously ideal, it is determined that the pre - alum addition feed - forward dosing amount is relatively accurate at this time (the pre - alum addition feed - forward prediction model is matched). Gradually increase the weight of the pre - alum addition feed - forward dosing amount (adjustment step: 2.5% per time, adjustment period: 2 hours per time), and decrease the weight of the pre - alum addition feedback dosing amount (adjustment step: 2.5% per time, adjustment period: 2 hours per time). After performing intelligent weighting on the pre - alum addition feed - forward dosing amount and the pre - alum addition feedback dosing amount, the pre - alum addition amount is obtained:

[0032] u(t) = ω·u FF (t)+(1 - ω)·u FB (t) (5)

[0033] Where: u(t) is the pre - alum addition amount, u FF (t) is the pre - alum addition feed - forward dosing amount, u FB (t) is the pre - alum addition feedback dosing amount, and ω is the weight coefficient.

[0034] Step 2: To ensure the stability of the pre - alum addition process, compare the deviation between the pre - alum addition amount calculated in Step 1 and the actual pre - alum addition amount at the previous moment in the current system. If the absolute value of the deviation between the two is within 0 - 1 mg / L, keep the actual pre - alum addition amount in the current system unchanged. If the deviation between the two is greater than 1 mg / L, add the pre - alum addition amount according to the calculation in Step 1. If the deviation between the two is less than - 1 mg / L, the actual pre - alum addition amount at the current moment is:

[0035]

[0036] Where: is the actual pre - alum addition amount at the current moment, is the actual pre - alum addition amount at the previous moment, and u(t) is the pre - alum addition amount calculated in Step 1.

[0037] Step (IV), the post - addition alum pre - feed control step is as follows:

[0038] Step 1, track and monitor the pre - addition alum dosage and the turbidity of the sedimentation tank effluent (obtained by filtering and processing the readings of the turbidity meter of the sedimentation tank effluent according to the data in Step 1 of Step (I)). When it is monitored that both the pre - addition alum dosage and the turbidity of the sedimentation tank effluent have been continuously higher than the threshold within the most recent 30 minutes, start the post - addition alum process. When starting the post - addition alum process, proceed to Step 2 of Step (IV), as well as Step (V) and Step (VI).

[0039] Step 2, according to the turbidity of the sedimentation tank effluent, query the sedimentation tank effluent turbidity - post - addition alum dosage relationship curve as Figure 8 shown to obtain the post - addition alum pre - feed dosage (i.e., the post - addition alum dosage in Figure 8 ).

[0040] Step (V), the post - addition alum feedback control step is as follows:

[0041] Step 1, since the inlet flow rate of the sand filter is inversely proportional to the residence time, taking the water treatment plant under study as an example, it has been measured that the residence time of the sand filter corresponding to an inlet flow rate of 3100 cubic meters per hour is 0.17 hours. Therefore, the lag time (i.e., the residence time) can be calculated according to the current inlet flow rate Q:

[0042]

[0043] Step 2, according to the deviation between the turbidity of the sand filter effluent (obtained by filtering and processing the readings of the turbidity meter of the sand filter effluent according to the data in Step 1 of Step (I)) and the set value, use the large - lag PI feedback control to calculate the post - addition alum feedback dosage:

[0044]

[0045] In the formula: τ′ is the lag time, is the post - addition alum feedback dosage at the moment of t - τ′, is the turbidity value of the sand filter effluent, is the turbidity set value, λ′ is the weight coefficient, γ i is the gain coefficient, and l is the integral time length.

[0046] Step (VI), the post - addition alum pre - feed - feedback composite control step is as follows:

[0047] Step 1: Track and monitor the pre - dosing alum amount obtained in Step (IV), the feedback - dosing alum amount obtained in Step (V), and the turbidity of the effluent from the sand filter (obtained by filtering and processing the readings of the turbidity meter of the sand - filter effluent according to the data in Step 1 of Step (I)). When it is monitored that the pre - dosing alum amount or (and) the feedback - dosing alum amount of post - dosing alum continues to increase within the recent 5 minutes, while the turbidity of the sand - filter effluent continues to be higher than the threshold, it is determined that the pre - dosing alum amount of post - dosing alum tends to be inaccurate (the relationship curve between the turbidity of the sedimentation - tank effluent and the dosing alum amount is mismatched or an unprecedented water - quality mutation phenomenon occurs). Gradually increase the weight of the feedback - dosing alum amount of post - dosing alum (the adjustment step is 8% per time, and the adjustment period is 5 minutes per time), and decrease the weight of the pre - dosing alum amount of post - dosing alum (the adjustment step is 8% per time, and the adjustment period is 5 minutes per time); when it is monitored that the pre - dosing alum amount and the feedback - dosing alum amount of post - dosing alum remain unchanged or change little within the recent 10 minutes, and the turbidity of the sand - filter effluent is also continuously ideal, it is determined that the pre - dosing alum amount of post - dosing alum is relatively accurate (the relationship curve between the turbidity of the sedimentation - tank effluent and the dosing alum amount is matched). Gradually increase the weight of the pre - dosing alum amount of post - dosing alum (the adjustment step is 4% per time, and the adjustment period is 10 minutes per time), and decrease the weight of the feedback - dosing alum amount of post - dosing alum (the adjustment step is 4% per time, and the adjustment period is 10 minutes per time). After intelligent weighting of the pre - dosing alum amount and the feedback - dosing alum amount of post - dosing alum, the post - dosing alum amount is obtained:

[0048]

[0049] In the formula: is the post - dosing alum amount, is the pre - dosing alum amount of post - dosing alum, is the feedback - dosing alum amount of post - dosing alum, and ω′ is the weight coefficient.

[0050] Step 2: To ensure the stability of the post - dosing alum process, compare the deviation between the post - dosing alum amount calculated in Step 1 and the actual post - dosing alum amount at the previous moment in the current system. If the absolute value of the deviation between the two is within 0 - 1 mg / L, keep the actual post - dosing alum amount in the current system unchanged; if the deviation between the two is greater than 1 mg / L, add the alum according to the post - dosing alum amount calculated in Step 1; if the deviation between the two is less than - 1 mg / L, the actual post - dosing alum amount at the current moment is:

[0051]

[0052] In the formula: is the actual post - dosing alum amount at the current moment, is the actual post - dosing alum amount at the previous moment, is the post - dosing alum amount calculated in Step 1.

[0053] Beneficial effects:

[0054] The present invention can mainly adjust the alum dosage in a timely manner according to the changes in the raw water quality through the feedforward control of the pre - alum addition part, and correct the control effect of the entire alum addition process through the feedback control of the pre - alum addition part and the feedforward - feedback composite control of the post - alum addition part, strictly limiting the turbidity of the effluent from the sedimentation tank and the sand filter. Its main technical features are: the method is practical and reliable, does not require any additional cost, and has wide applicability. Brief Description of the Drawings

[0055] Figure 1 is the process flow diagram of the alum addition process in the water plant;

[0056] Figure 2 is the principle block diagram of the feedforward - feedback composite control of alum addition in the water plant;

[0057] Figure 3 is the simple diagram of the data filtering process;

[0058] Figure 4 is the principle block diagram of the feedforward prediction model of pre - alum addition;

[0059] Figure 5 is the process diagram of the feedforward alum dosage prediction of pre - alum addition;

[0060] Figure 6 is the principle block diagram of the turbidity prediction model of the effluent from the sedimentation tank;

[0061] Figure 7 is the process diagram of the turbidity prediction of the effluent from the sedimentation tank;

[0062] Figure 8 is the relationship curve between the turbidity of the effluent from the sedimentation tank and the post - alum dosage. Detailed Embodiment

[0063] The technical solution of the present invention will be described in detail below in combination with the actual on - line test of the water plant:

[0064] (1) Calculate the pre - alum dosage according to the feedforward - feedback composite control method of pre - alum addition. The specific steps are as follows:

[0065] The first step is to obtain the real - time feedforward alum dosage of pre - alum addition

[0066] Step 1: In the production operation database of the water plant, screen out the raw water quality indexes (pH, water temperature, dissolved oxygen, chemical oxygen demand, turbidity), influent flow rate, and historical data of the alum dosage corresponding to the time within the range of the required turbidity of the effluent from the sedimentation tank (before the pre - alum addition lag time τ), and perform data filtering processing. Divide the obtained historical data samples into two parts: a training set and a test set;

[0067] Step 2: Using the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity) and the influent flow rate as input variables, and the pre - alum dosage as the output variable, a random forest algorithm is used to train a pre - alum feed - forward prediction model.

[0068] Step 3: According to the real - time data filtering values of the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity) and the influent flow rate, input them into the pre - alum feed - forward prediction model, and calculate the real - time pre - alum feed - forward dosage u FF (t).

[0069] Second step: Obtain the real - time pre - alum feedback dosage

[0070] Step 1: In the water plant production operation database, screen out the sedimentation tank effluent turbidity that meets the requirements within a certain range and the corresponding raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and historical data of pre - alum dosage at the corresponding time (before the pre - alum lag time τ), and perform data filtering processing. Divide the obtained historical data samples into two parts: a training set and a test set.

[0071] Step 2: Using the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and pre - alum dosage as input variables, and the sedimentation tank effluent turbidity as the output variable, a random forest algorithm is used to train a sedimentation tank effluent turbidity prediction model.

[0072] Step 3: According to the real - time data filtering values of the raw water quality indicators (pH, water temperature, dissolved oxygen, oxygen consumption, turbidity), influent flow rate, and pre - alum dosage, input them into the sedimentation tank effluent turbidity prediction model, and calculate the real - time predicted sedimentation tank effluent turbidity.

[0073] Step 4: Compare the sedimentation tank effluent turbidity values at the current time t and the previous time 1 after data filtering processing with the corresponding turbidity prediction values under the same raw water quality, influent flow rate, and pre - alum dosage conditions. The deviation value is used as the correction amount for the turbidity prediction value corresponding to the current t - time raw water quality, influent flow rate, and pre - alum dosage conditions.

[0074] Step 5: According to the deviation between the corrected turbidity prediction value and the set value, use PI feedback control to calculate the real - time pre - alum feedback dosage u FB (t).

[0075] Third step: Obtain the pre - alum dosage

[0076] Step 1: Combine the real - time pre - alum feed - forward dosage u FF (t) obtained in the first step and the real - time pre - alum feedback dosage u FB(t) and the turbidity of the sedimentation tank effluent (after data filtering) are monitored. According to the monitoring situation, the weights of the pre - alum addition feed - forward dosing amount and the pre - alum addition feedback dosing amount are adjusted intelligently, and the pre - alum addition amount is calculated;

[0077] Step 2: Compare the calculated pre - alum addition amount with the actual pre - alum addition amount at the previous moment. Make certain constraint processing according to the deviation situation to avoid large fluctuations in the pre - alum addition amount.

[0078] (2) Calculate the post - alum addition amount according to the post - alum addition feed - forward - feedback compound control method. The specific steps are as follows:

[0079] The first step: Obtain the real - time post - alum addition feed - forward dosing amount

[0080] Step 1: Monitor the pre - alum addition amount and the turbidity of the sedimentation tank effluent (after data filtering). When it is monitored that both the pre - alum addition amount and the turbidity of the sedimentation tank effluent are continuously higher than the threshold within the recent 30 minutes, start the post - alum addition process. When starting the post - alum addition process, perform step 2 of the first step, as well as the second step and the third step;

[0081] Step 2: According to the turbidity of the sedimentation tank effluent, query the Figure 8 sedimentation tank effluent turbidity - post - alum addition amount relationship curve as shown, and obtain the real - time post - alum addition feed - forward dosing amount

[0082] The second step: Obtain the real - time post - alum addition feedback dosing amount

[0083] Step 1: According to the deviation between the turbidity of the sand filter effluent (after data filtering) and the set value, use the large - lag PI feedback control to calculate the real - time post - alum addition feedback dosing amount

[0084] The third step: Obtain the post - alum addition amount

[0085] Step 1: Monitor the real - time post - alum addition feed - forward dosing amount obtained in the first step the real - time post - alum addition feedback dosing amount obtained in the second step and the turbidity of the sand filter effluent (after data filtering). According to the monitoring situation, intelligently adjust the weights of the post - alum addition feed - forward dosing amount and the post - alum addition feedback dosing amount, and calculate the post - alum addition amount;

[0086] Step 2: Compare the calculated post - alum addition amount with the actual post - alum addition amount at the previous moment. Make certain constraint processing according to the deviation situation to avoid large fluctuations in the post - alum addition amount.

[0087] The supplementary explanation about calculating the lag time according to the influent flow rate is as follows:

[0088] (1) In the pre - alum addition process, the residence time of the sedimentation tank corresponding to an influent flow rate of 3100 m³ / h has been measured to be 2.3 hours. Since the influent flow rate is inversely proportional to the residence time, the pre - alum addition lag time (i.e., the residence time of the sedimentation tank) when the influent flow rate is Q is

[0089] (2) In the post - alum addition process, the residence time of the sand filter corresponding to an influent flow rate of 3100 m³ / h has been measured to be 0.17 hours. Since the influent flow rate is inversely proportional to the residence time, the post - alum addition lag time (i.e., the residence time of the sand filter) when the influent flow rate is Q′ is

Claims

1. A method for controlling alum addition in a water treatment plant based on feedforward-feedback composite control, characterized in that, It includes two parts: pre - alum addition and post - alum addition. The pre - alum addition part specifically includes the following steps: Step 11: For the seasonal variation of raw water quality, based on the random forest algorithm, establish a pre - alum addition feed - forward prediction model and a sedimentation tank effluent turbidity prediction model respectively. The pre - alum addition feed - forward prediction model can quickly calculate the pre - alum addition feed - forward dosage according to the raw water quality and influent flow rate, and the sedimentation tank effluent turbidity prediction model can quickly calculate the predicted value of the sedimentation tank effluent turbidity according to the raw water quality, influent flow rate and pre - alum addition dosage. Step 12: For the dynamic time - varying nature of the raw water quality change, according to the deviation between the sedimentation tank effluent turbidity meter reading and the turbidity predicted value, use a model corrector to correct the turbidity prediction model. According to the deviation between the corrected turbidity predicted value and the set value, use PI feedback control to calculate the pre - alum addition feedback dosage. Step 13: Intelligently weight the calculated pre - alum addition feed - forward dosage and pre - alum addition feedback dosage to obtain the pre - alum addition dosage. The post - alum addition part specifically includes the following steps: Step 21: Monitor the pre - alum addition dosage and the sedimentation tank effluent turbidity meter reading. When it is detected that the pre - alum addition dosage and the sedimentation tank effluent turbidity have been continuously higher than the threshold within the recent 30 minutes, start the post - alum addition process. Step 22: According to the sedimentation tank effluent turbidity, query the sedimentation tank effluent turbidity - post - alum addition dosage relationship curve to obtain the post - alum addition feed - forward dosage. Step 23: According to the deviation between the sand filter effluent turbidity and the set value, use large - lag PI feedback control to calculate the post - alum addition feedback dosage. where: τ′ is the lag time of the post - added alum, is the post - added alum feedback dosing amount at time t - τ′, is the turbidity value of the effluent from the sand filter, is the turbidity set value, λ′ is the weight coefficient, γ i is the gain coefficient, l is the integral time length; Step 4: Intelligently weight the calculated post - alum addition feed - forward dosage and post - alum addition feedback dosage to obtain the post - alum addition dosage.

2. The alum addition control method for water treatment plants based on feedforward-feedback composite control according to claim 1, wherein The pre - alum addition feed - forward prediction model based on the random forest algorithm described in Step 11 of the pre - alum addition part specifically includes the following steps: Step 111 - 1: In the water plant production operation database, screen out the historical data of raw water quality indicators, influent flow rate, and pre - alum addition dosage corresponding to the moments within the range that meets the sedimentation tank effluent turbidity requirements, and perform data filtering processing to obtain historical data samples. The corresponding moments are before the pre - alum addition lag time τ. The raw water quality indicators include pH, water temperature, dissolved oxygen, chemical oxygen demand, and turbidity. Step 112 - 1: Divide the historical data samples obtained in Step 111 - 1 into a training set and a test set. Step 113 - 1: Using the raw water quality indicators and influent flow rate as input variables and the pre - alum addition dosage as the output variable, based on the training set and test set historical data samples obtained in Step 112 - 1, use the random forest algorithm to train the pre - alum addition feed - forward prediction model.

3. The alum dosing control method for waterworks based on feedforward-feedback composite control according to claim 1, characterized in that, The sedimentation tank effluent turbidity prediction model based on the random forest algorithm described in Step 11 of the pre - alum addition part specifically includes the following steps: Step 111 - 2: In the water plant production operation database, screen out the historical data of sedimentation tank effluent turbidity and its corresponding raw water quality indicators, influent flow rate, and pre - alum addition dosage within a certain range, and perform data filtering processing to obtain historical data samples. The corresponding moments are before the pre - alum addition lag time τ. The raw water quality indicators include pH, water temperature, dissolved oxygen, chemical oxygen demand, and turbidity. Step 112-2: Divide the historical data samples obtained in Step 111-2 into a training set and a test set; Step 113-2: Using the raw water quality index, influent flow rate, and pre-alum dosage as input variables, and the turbidity of the sedimentation tank effluent as the output variable, based on the historical data samples of the training set and test set obtained in Step 112-2, use the random forest algorithm to train a prediction model for the turbidity of the sedimentation tank effluent.

4. The alum dosing control method for waterworks based on feedforward-feedback composite control according to claim 1, characterized in that: The model corrector described in step 12 of the pre - alum addition part calculates the pre - alum addition lag time τ based on the influent flow rate, and obtains the turbidity values of the sedimentation tank effluent at the current time t and its previous l times in real - time (y(t), y(t - 1), y(t - 2), …, y(t - l)). Then, (y(t), y(t - 1), y(t - 2), …, y(t - l)) corresponds to the real - time raw water quality, influent flow rate, and pre - alum addition amount conditions at time t - τ and its previous l times. Let the turbidity prediction value corresponding to the real - time raw water quality, influent flow rate, and pre - alum addition amount conditions at the same time t - τ and its previous l times be Therefore, the deviation between (y(t), y(t - 1), y(t - 2), …, y(t - l)) and can be used as the offset for correcting the turbidity prediction value corresponding to the real - time raw water quality, influent flow rate, and pre - alum addition amount conditions at the current time t : In the formula: is the predicted turbidity value after calibration, and α i is the weight coefficient, and l is the calibration time length.

5. The alum dosing control method for waterworks based on feedforward-feedback composite control according to claim 4, characterized in that: Calculate the lag time based on the influent flow rate. For the research object water plant, in the pre - coagulant addition process, the residence time of the sedimentation tank corresponding to an influent flow rate of 3100 cubic meters per hour has been measured to be approximately 2.3 hours. Since the influent flow rate is inversely proportional to the residence time, the lag time when the influent flow rate is Q is ; in the post - coagulant addition process, the residence time of the sand filter corresponding to an influent flow rate of 3100 cubic meters per hour has been measured to be approximately 0.17 hours. Since the influent flow rate is inversely proportional to the residence time, the lag time when the influent flow rate is Q′ is .

6. The alum dosing control method for waterworks based on feedforward-feedback compound control according to claim 4, characterized in that: In the pre-alum addition part of Step 12, the PI feedback control is used to calculate the pre-alum feedback dosage. According to the deviation between the corrected turbidity prediction value and the set value, the PI feedback control is used to calculate the pre-alum feedback dosage: where: u FB (t - 1) is the alum addition feedback alum addition amount at the previous moment, is the predicted turbidity value after correction, y ref is the turbidity set value, λ is the weight coefficient, β i is the gain coefficient, l is the integral time length.

7. The alum dosing control method for waterworks based on feedforward-feedback compound control according to claim 1, characterized in that: In the post-alum addition part of Step 22, based on the big data analysis in the actual production operation control process, 9 groups of typical turbidity of sedimentation tank effluent - post-alum dosage values are obtained; then, the least squares fitting is performed on the 9 groups of turbidity of sedimentation tank effluent - post-alum dosage values to obtain the relationship curve between the turbidity of sedimentation tank effluent and the post-alum dosage.

Images