Intelligent sewage carbon source adding system and method with load sensing and sharing functions

Abstract

The invention relates to a sewage treatment control technology, and discloses an intelligent sewage carbon source adding system and method with a load sensing and sharing function, and the method comprises the following steps: adding a carbon source: adding the carbon source into an anoxic tank and a denitrification tank through a carbon source dosing pump unit; the carbon source dosing frequency is controlled, and the carbon source dosing frequency of a carbon source dosing pump unit is controlled through a carbon source dosing pump control cabinet; feeding back the carbon source dosage to the carbon source dosing pump control cabinet; the carbon source adding amount is determined, and the carbon source adding amount is controlled by controlling the frequency of the carbon source dosing pump unit through the intelligent carbon source adding system. The carbon source dosing amount is controlled by controlling the carbon source dosing pump unit through the intelligent carbon source dosing system. The carbon source feeding amount can be detected in real time, the safety is good, the cost is low, and the operation is simple and convenient.

Description

Technical Field

[0001] The present invention relates to sewage treatment control technology, and in particular to a sewage intelligent carbon source dosing system and method with load sensing and sharing functions. Background Art

[0002] In the field of sewage treatment, the addition of carbon sources is to provide microorganisms with necessary nutrients, promote microbial absorption and conversion of nitrogen in sewage, and then achieve effective removal of total nitrogen and meet nitrogen emission standards. Therefore, it is very important to add carbon sources reasonably.

[0003] At present, most sewage treatment plants use manual control or single indicator feedback control to control carbon source addition. Manual control methods often involve staff controlling the amount of carbon source addition based on indicators such as total nitrogen in the incoming water or total nitrogen in the effluent, combined with operating experience. Although this method can maintain the denitrification effect, it usually requires excessive addition, resulting in increased operating costs. Single indicator feedback control usually uses the effluent total nitrogen indicator to control the amount of carbon source addition, resulting in a large lag in carbon source addition and the inability to comprehensively consider the effects of multiple water treatment indicators on the effect of microbial nitrogen removal, such as incoming water volume, temperature, ammonia nitrogen content, carbon source addition, water pH, etc.

[0004] Aiming at the drawbacks of manual control and single indicator feedback control, the existing technology has been improved.

[0005] For example, prior art 1: Chinese patent CN202311335942.0 discloses an intelligent method for adding a carbon source. According to the current influent flow rate, the ratio of the carbon source dosage to the denitrification denitrification amount, the nitrate nitrogen concentration difference, and the chemical oxygen demand equivalent of the carbon source, the static correction coefficient, the dynamic correction coefficient, the ratio value, etc. are calculated, and finally the above results are input into the carbon source calculation model to obtain the required carbon source dosage. Although this method takes into account multiple water treatment indicators and reduces the carbon source dosage to a certain extent, the coefficients required by the carbon source calculation model are mostly calculated based on artificial experience, and it is difficult to adaptively handle the changes in the total nitrogen in the incoming water and the calculation is complicated.

[0006] Prior art 2: Chinese patent CN202310744254 discloses a control strategy for obtaining the required carbon source dosage in real time based on the change rules and trends between the input operating indicators and the equipment control values, and then adjusting the carbon source dosage according to the carbon source dosage. It has poor adaptive ability and cannot learn the relationship between total nitrogen and nitrate nitrogen in real time. Summary of the invention

[0007] In view of the problem in the prior art that it is difficult to adaptively and quickly respond to changes in total nitrogen in incoming water and changes in environmental conditions, and it is difficult to obtain optimal treatment results, the present invention provides a sewage intelligent carbon source dosing system and method with load sensing and sharing functions.

[0008] In order to solve the above technical problems, the present invention is solved by the following technical solutions:

[0009] A sewage intelligent carbon source dosing method with load sensing and sharing function includes an anoxic tank, a carbon source dosing pump unit, a carbon source dosing pump intelligent control cabinet and an intelligent carbon source dosing system; the method includes:

[0010] The carbon source is added into the anoxic tank through the carbon source dosing pump unit;

[0011] The frequency of carbon source addition is controlled by controlling the frequency of carbon source addition of the carbon source dosing pump unit through the carbon source dosing pump control cabinet; and the carbon source addition amount is transmitted to the carbon source dosing pump control cabinet;

[0012] The amount of carbon source added is determined by controlling the frequency of the carbon source dosing pump unit through the intelligent carbon source dosing system.

[0013] Preferably, the method for the intelligent carbon source dosing system to control the frequency of the carbon source dosing pump unit to control the carbon source dosage includes:

[0014] Preparation of carbon source addition data set: The carbon source addition data set is obtained by preprocessing the input original time series data set; the input original time series data set includes total nitrogen anoxic pool inlet TN in , concentration time series data, total nitrogen and total effluent TN out Concentration time series data, NO3 concentration time series data of nitrate anoxic pool and influent flow of anoxic pool in Time series data;

[0015] The intelligent carbon source addition algorithm controls the input original time series data set through a machine learning algorithm to obtain the carbon source addition amount Carbon required for the corresponding anoxic pool load.

[0016] Preferably, the preprocessing of the carbon source addition data set is performed by processing the carbon source addition data set through a clustering algorithm, and the clustering algorithm processing includes:

[0017] Obtain the inlet and outlet water indicators and the dosing pump data for the past 24 hours and calculate the average value X ave ,

[0018]

[0019] Among them, x i is the i-th data point in the past;

[0020] Calculate the distance L between the current data and the average value of the past 24h data:

[0021]

[0022] Among them, x is the current data, X ave It is the average value of the data in the past 24 hours.

[0023] Preferably, the carbon source dosage Carbon when obtaining the total nitrogen removal amount by using a machine learning algorithm for the input original time series data set includes:

[0024]

[0025] Among them, R TN is the amount of total nitrogen that needs to be removed, and ω is the model parameter;

[0026] For the amount of total nitrogen that needs to be removed, R TN The amount of total nitrogen in the water TN in (total) and total nitric nitrogen NO3(total) are obtained;

[0027] R TN =TN in (total)-NO3(total);

[0028] Among them, TN in (total) is the total nitrogen content in the incoming water, and NO3(total) is the total nitric nitrogen content.

[0029] As a preference, for TN in (TOTEL) is the amount of total nitrogen in the incoming water through the incoming water TN in and water flow acquisition;

[0030] TN in (total) = TN in *Flow in ;

[0031] The total amount of nitrate nitrogen, NO3 (total), is obtained through nitrate nitrogen and water flow.

[0032] NO3(total)=NO3*Flow in ;

[0033] Among them, NO3 is the amount of nitric nitrogen.

[0034] Preferably, the amount of nitric nitrogen NO3 is obtained:

[0035] The model is built based on the total nitrogen in the incoming water and the nitrate nitrogen in the anoxic pool. The model coefficients are obtained by the least squares method. The real-time calculation method is used to ensure that the model always reflects the relationship between the total nitrogen and nitrate nitrogen:

[0036] TN out =β0+β1TNin +β2NO3+ε

[0037] Among them, ε is the difference between the model prediction value and the actual value; β0, β1 and β2 are the optimal parameters;

[0038] Fitting model coefficients by least squares method Get the optimal parameters β0, β1 and β2:

[0039] By manually setting the water output target TN target , combined with the current water index TN in , reverse the target value of nitrate nitrogen in the anoxic pool NO3:

[0040]

[0041] Among them, TN target Manually set the target value of water output index.

[0042] Preferably, the optimal carbon source dosage is calculated by gradient descent;

[0043] Calculate the loss function;

[0044]

[0045] Among them, λ is an adjustable parameter, is the size of the penalty term.

[0046] Find the corresponding parameter ω when the loss function is minimized, and derive the loss function to obtain the following formula:

[0047]

[0048] Among them, α is the learning rate, find out ω and λ when the loss function converges to the minimum value, confirm the weight and bias function; after obtaining the optimal parameter ω, use ω to convert the latest inflow total nitrogen, nitrate nitrogen and inflow flow and the latest total nitrogen removal amount R TN , get the latest required carbon source dosage Carbon:

[0049]

[0050] Among them, R TN is the latest total nitrogen removal.

[0051] Preferably, the method further includes adjusting the amount of carbon source added, and the adjustment of the amount of carbon source added is performed through a dynamic learning strategy.

[0052] In order to solve the above technical problems, the present invention also provides a sewage intelligent carbon source dosing system with load sensing and sharing function, including an anoxic tank, a denitrification tank, a carbon source dosing pump unit, a carbon source dosing pump intelligent control cabinet and an intelligent carbon source dosing system; it also includes,

[0053] A carbon source dosing module, which adds carbon source to the anoxic tank and the denitrification tank through a carbon source dosing pump unit;

[0054] The carbon source dosing frequency control module controls the carbon source dosing frequency of the carbon source dosing pump unit through the carbon source dosing pump control cabinet; and feeds back the carbon source dosing amount to the carbon source dosing pump control cabinet;

[0055] A module for determining the amount of carbon source added, which controls the amount of carbon source added by controlling the frequency of the carbon source dosing pump unit through an intelligent carbon source addition system;

[0056] The carbon source dosage control module controls the carbon source dosing pump unit through the intelligent carbon source dosage system, thereby controlling the carbon source dosage.

[0057] The present invention has significant technical effects due to the adoption of the above technical solution:

[0058] The present invention can process the change of total nitrogen in sewage in real time and the calculation is simple.

[0059] The present invention adaptively and quickly responds to changes in total nitrogen in the incoming water and changes in environmental conditions. After the total nitrogen is removed in the anoxic tank, there will be no additional total nitrogen in the subsequent process, so the relationship between the total nitrogen in the effluent and the nitrate nitrogen in the anoxic tank is stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a flow chart of the present invention.

[0061] Figure 2 It is a diagram showing the nitric nitrogen control effect of the present invention.

[0062] Figure 3 This is a diagram showing the effect of controlling total nitrogen in effluent water according to the present invention. DETAILED DESCRIPTION

[0063] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0064] Example 1

[0065] A sewage intelligent carbon source dosing method with load sensing and sharing function includes an anoxic tank, a carbon source dosing pump unit, a carbon source dosing pump intelligent control cabinet and an intelligent carbon source dosing system; the method includes:

[0066] The carbon source is added into the anoxic tank through the carbon source dosing pump unit;

[0067] The frequency of carbon source addition is controlled by controlling the frequency of carbon source addition of the carbon source dosing pump unit through the carbon source dosing pump control cabinet; and the carbon source addition amount is transmitted to the carbon source dosing pump control cabinet;

[0068] The amount of carbon source added is determined by controlling the frequency of the carbon source dosing pump unit through the intelligent carbon source dosing system.

[0069] The method for the intelligent carbon source dosing system to control the frequency of the carbon source dosing pump unit to control the carbon source dosage includes:

[0070] Preparation of carbon source addition data set: The carbon source addition data set is obtained by preprocessing the input original time series data set; the input original time series data set includes total nitrogen anoxic pool inlet TN in , concentration time series data, total nitrogen and total effluent TN out Concentration time series data, NO3 concentration time series data of nitrate anoxic pool and influent flow of anoxic pool in Time series data;

[0071] The intelligent carbon source addition algorithm controls the input original time series data set through a machine learning algorithm to obtain the carbon source addition amount Carbon required for the corresponding anoxic pool load.

[0072] The preprocessing of the carbon source addition data set is performed by using a clustering algorithm. The clustering algorithm processing includes:

[0073] Obtain the inlet and outlet water indicators and the dosing pump data for the past 24 hours and calculate the average value X ave ,

[0074]

[0075] Among them, x i is the i-th data point in the past;

[0076] The average value of the past 24 hours of data is needed (the sampling time is once every 1 minute), so the above formula requires 1440 data to calculate. The next step is to calculate the distance L between the current data and the average value of the past 24 hours of data:

[0077]

[0078] Among them, x is the current data, X ave It is the average value of the data in the past 24 hours.

[0079] A total of one day's data is needed, and the data sampling frequency is once per minute. There are a total of 1440 data. The distances D of the 1440 data are calculated separately and arranged in order from small to large. 5% of the data with the largest distance D, that is, 1440*5%=72 data are deleted from the data set. These data are defined as abnormal data because they are significantly different from the average data of the past 24 hours. Incorporating abnormal data into the machine learning algorithm will cause the fitted model to be inconsistent with the actual situation, so the abnormal data needs to be deleted.

[0080] The carbon source dosage Carbon used to obtain the total nitrogen removal amount by machine learning algorithm for the input original time series data set includes:

[0081]

[0082] Among them, R TN is the amount of total nitrogen that needs to be removed, and ω is the model parameter;

[0083] For the amount of total nitrogen that needs to be removed, R TN The amount of total nitrogen in the water TN in (total) and total nitric nitrogen NO3(total) are obtained;

[0084] R TN =TN in (total)-NO3(total);

[0085] Among them, TN in (total) is the total nitrogen content in the incoming water, and NO3(total) is the total nitric nitrogen content.

[0086] Since total nitrogen is the final effluent target, but the total nitrogen value takes a long time to update, if it is regulated according to the total nitrogen value, the system will have lag, so the main invention of this patent is to use nitrate nitrogen as the main control factor. The rational point of this method is that due to process reasons, after the total nitrogen is removed in the anoxic tank, there will be no additional total nitrogen in the subsequent process, so the relationship between the effluent total nitrogen and the nitrate nitrogen in the anoxic tank is stable.

[0087] For TN in (total) is the amount of total nitrogen in the incoming water through the incoming water TN in and water flow acquisition;

[0088] TN in (total) = TN in *Flow in ;

[0089] The total amount of nitrate nitrogen, NO3 (total), is obtained through nitrate nitrogen and water flow.

[0090] NO3(total)=NO3*Flow in ;

[0091] Among them, NO3 is the amount of nitric nitrogen.

[0092] The amount of nitric nitrogen NO3 is obtained,

[0093] Because the factors affecting the total nitrogen in the effluent are the total nitrogen in the incoming water and the amount of carbon source added, but the amount of carbon source added is a variable that needs to be controlled, and according to the process logic, it is known that the nitrate nitrogen in the anoxic pool and the total nitrogen in the effluent are strongly correlated, so we built a model based on the total nitrogen in the incoming water and the nitrate nitrogen in the anoxic pool, and obtained the model coefficients by the least squares method.

[0094] The model is built based on the total nitrogen in the incoming water and the nitrate nitrogen in the anoxic pool. The model coefficients are obtained by the least squares method. The real-time calculation method is used to ensure that the model always reflects the relationship between the total nitrogen and nitrate nitrogen:

[0095] TN out =β0+β1TN in +β2NO3+ε

[0096] Among them, ε is the difference between the model prediction value and the actual value; β0, β1 and β2 are the optimal parameters;

[0097] Fitting model coefficients by least squares method Obtain the optimal parameters β0, β1 and β2: By obtaining the gap between all model predicted values ​​and actual values, find the β0, β1 and β2 corresponding to their minimum moment. This parameter is the optimal parameter of the model because under this parameter, the gap between the model predicted value and the actual value is the smallest.

[0098] By manually setting the water output target TN target , combined with the current water index TN in , reverse the target value of nitrate nitrogen in the anoxic pool NO3:

[0099]

[0100] Among them, TN target The target value of the effluent index is set manually. Because the nitrate nitrogen in the anoxic pool can provide more timely feedback, the subsequent control will be controlled by the nitrate nitrogen target value calculated based on the total effluent total nitrogen target. The following method is how to use machine learning to obtain the relationship between the nitrate nitrogen in the anoxic pool and the carbon source dosage, and control the carbon source dosage through this relationship.

[0101] The optimal carbon source dosage will be calculated by gradient descent;

[0102] Calculate the loss function;

[0103]

[0104] Among them, λ is an adjustable parameter, is the size of the penalty term.

[0105] Find the corresponding parameter ω when the loss function is minimized, and derive the loss function to obtain the following formula:

[0106]

[0107] Among them, α is the learning rate,

[0108] Find out ω and λ when the loss function converges to the minimum value, confirm the weight and bias function; after obtaining the optimal parameter ω, use ω to convert the latest total nitrogen, nitrate nitrogen and water flow and the latest total nitrogen removal R TN , obtain the latest required carbon source dosage Carbon: After obtaining the optimal parameter ω, we use ω to calculate the total nitrogen removal amount R TN It is associated with the carbon source dosage Carbon, and the latest total nitrogen removal R is obtained through the latest inflow total nitrogen, nitrate nitrogen and inflow flow. TN , calculate the latest required carbon source dosage;

[0109]

[0110] Among them, R TN is the latest total nitrogen removal.

[0111] Example 2

[0112] Based on Example 1, this example also includes adjusting the input data set and formulating the input data set screening strategy through dynamic learning.

[0113] Get the common values ​​of the inlet and outlet water indicators and the dosing pump, and remove the abnormal values ​​of the distance L; calculate the distance L between the current data and the average value of the past 24h data:

[0114]

[0115] Among them, x is the current data, X ave It is the average value of the data in the past 24 hours.

[0116] When the distance L data is judged as an outlier, the previous data will be used for filling.

[0117] According to the new total nitrogen removal and historical carbon source dosage, the new carbon source dosage is obtained as shown in Table 1.

[0118] Table 1 Carbon source dosage table

[0119] time Carbon source dosing pump flow 2024 / 08 / 2900:00:00 <![CDATA[x1]]> ... 2024 / 08 / 3023:58:00 <![CDATA[x 2879 <!-- 6 -->]]> 2024 / 08 / 3023:59:00 <![CDATA[x 2880 ]]>

[0120] x is the flow rate of the carbon source dosing pump at that time point. The data is collected once every minute. At the latest data point, it is necessary to determine whether the data is an abnormal value.

[0121] Step 1: Calculate the average carbon source dosage for the past day.

[0122]

[0123] Step 2: Calculate the distance D between each data point in the past day and the average value of the previous 24 hours, and take D 2 ;

[0124]

[0125] Step 3: The calculated L 1440 ,...,L 2880 Sort from small to large, and the top 5% will be removed and treated as abnormal data. Assumption: L 2880 and L 2879 All are within the top 5%, and we will replace them with the previous data, as shown in Table 2.

[0126] Table 2 Data preprocessing table

[0127]

[0128]

[0129] Example 3

[0130] Based on the above embodiments, this embodiment is a sewage intelligent carbon source dosing system with load sensing and sharing function, including an anoxic tank, a denitrification tank, a carbon source dosing pump unit, a carbon source dosing pump intelligent control cabinet and an intelligent carbon source dosing system; it also includes,

[0131] A carbon source dosing module, which adds carbon source to the anoxic tank and the denitrification tank through a carbon source dosing pump unit;

[0132] The carbon source dosing frequency control module controls the carbon source dosing frequency of the carbon source dosing pump unit through the carbon source dosing pump control cabinet; and feeds back the carbon source dosing amount to the carbon source dosing pump control cabinet;

[0133] A module for determining the amount of carbon source added, which controls the amount of carbon source added by controlling the frequency of the carbon source dosing pump unit through an intelligent carbon source addition system;

[0134] The carbon source dosage control module controls the carbon source dosing pump unit through the intelligent carbon source dosage system, thereby controlling the carbon source dosage.

Claims

1. A method for intelligent carbon source dosing in sewage with load sensing and sharing function, comprising an anoxic tank, a carbon source dosing pump unit, a carbon source dosing pump intelligent control cabinet and an intelligent carbon source dosing system; the method comprises: The carbon source is added into the anoxic tank through the carbon source dosing pump unit; The frequency of carbon source addition is controlled by controlling the frequency of carbon source addition of the carbon source dosing pump unit through the carbon source dosing pump control cabinet; and the carbon source addition amount is transmitted to the carbon source dosing pump control cabinet; The amount of carbon source added is determined by controlling the frequency of the carbon source dosing pump unit through the intelligent carbon source dosing system.

2. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 1 is characterized in that: The method for the intelligent carbon source dosing system to control the frequency of the carbon source dosing pump unit to control the carbon source dosage includes: Preparation of carbon source addition data set: The carbon source addition data set is obtained by preprocessing the input original time series data set; the input original time series data set includes total nitrogen anoxic pool inlet TN in , concentration time series data, total nitrogen and total effluent TN out Concentration time series data, NO3 concentration time series data of nitrate anoxic pool and influent flow of anoxic pool in Time series data; The intelligent carbon source addition algorithm controls the input original time series data set through a machine learning algorithm to obtain the carbon source addition amount Carbon required for the corresponding anoxic pool load.

3. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 1 is characterized in that: The preprocessing of the carbon source addition data set is performed by using a clustering algorithm. The clustering algorithm processing includes: Obtain the inlet and outlet water indicators and the dosing pump data for the past 24 hours and calculate the average value X ave , Among them, x i is the i-th data point in the past; Calculate the distance L between the current data and the average value of the past 24h data: Among them, x is the current data, X ave It is the average value of the data in the past 24 hours.

4. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 2 is characterized in that: The carbon source dosage Carbon used to obtain the total nitrogen removal amount by machine learning algorithm for the input original time series data set includes: Among them, R TN is the amount of total nitrogen that needs to be removed, and ω is the model parameter; For the amount of total nitrogen that needs to be removed, R TN The amount of total nitrogen in the water TN in (total) and total nitric nitrogen NO3(total) are obtained; <h2 style=";text-align:left;direction:ltr">R<h2 style=";text-align:left;direction:ltr"> TN <h2 style=";text-align:left;direction:ltr"> =TN<h2 style=";text-align:left;direction:ltr"> in <h2 style=";text-align:left;direction:ltr"> (total)-NO3(total); Among them, TN in (total) is the total nitrogen content in the incoming water, and NO3(total) is the total nitric nitrogen content.

5. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 4 is characterized in that: For TN in (total) is the amount of total nitrogen in the incoming water through the incoming water TN in and water flow acquisition; TN in (total)=TN in *Flow in ; The total amount of nitrate nitrogen, NO3 (total), is obtained through nitrate nitrogen and water flow. NO3(total)=NO3*Flow in ; Among them, NO3 is the amount of nitric nitrogen.

6. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 5 is characterized in that: How to obtain the amount of nitric nitrogen NO3: The model is built based on the total nitrogen in the incoming water and the nitrate nitrogen in the anoxic pool. The model coefficients are obtained by the least squares method. The real-time calculation method is used to ensure that the model always reflects the relationship between the total nitrogen and nitrate nitrogen: TN out =β0+β1TN in +β2NO3+ε Among them, ε is the difference between the model prediction value and the actual value; β0, β1 and β2 are the optimal parameters; Fitting model coefficients by least squares method Get the optimal parameters β0, β1 and β2: By manually setting the water output target TN target , combined with the current water index TN in , reverse the target value of nitrate nitrogen in the anoxic pool NO3: Among them, TN target Manually set the target value of water output index.

7. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 6 is characterized in that: The optimal carbon source dosage will be calculated by gradient descent; Calculate the loss function; Among them, λ is an adjustable parameter, is the size of the penalty term. Find the corresponding parameter ω when the loss function is minimized, and derive the loss function to obtain the following formula: Among them, α is the learning rate, Find out ω and λ when the loss function converges to the minimum value, confirm the weight and bias function; after obtaining the optimal parameter ω, use ω to convert the latest total nitrogen, nitrate nitrogen and water flow and the latest total nitrogen removal R TN , get the latest required carbon source dosage Carbon: Among them, R TN is the latest total nitrogen removal.

8. The intelligent carbon source dosing method for sewage with load sensing and sharing function according to claim 1 is characterized in that: It also includes the adjustment of the amount of carbon source added, which is done through a dynamic learning strategy.

9. A sewage intelligent carbon source dosing system with load sensing and sharing function, comprising an anoxic tank, a denitrification tank, a carbon source dosing pump unit, a carbon source dosing pump intelligent control cabinet and an intelligent carbon source dosing system; characterized in that: Also includes, A carbon source dosing module, which adds carbon source to the anoxic tank and the denitrification tank through a carbon source dosing pump unit; The carbon source dosing frequency control module controls the carbon source dosing frequency of the carbon source dosing pump unit through the carbon source dosing pump control cabinet; and feeds back the carbon source dosing amount to the carbon source dosing pump control cabinet; A module for determining the amount of carbon source added, which controls the amount of carbon source added by controlling the frequency of the carbon source dosing pump unit through an intelligent carbon source addition system; The carbon source dosage control module controls the carbon source dosing pump unit through the intelligent carbon source dosage system, thereby controlling the carbon source dosage.

Images