A method and device for screening a denitrification high-efficiency liquid composite carbon source formula

Abstract

The present invention provides a method and device for screening a liquid composite carbon source formula with high denitrification efficiency. The method divides a sample space of multiple candidate liquid composite carbon source formulas containing liquid composite carbon source formulas into an experimental sample space and a sample space to be predicted. The experimental sample space is tested for denitrification performance. The denitrification performance of the sample space to be predicted is predicted using a denitrification performance prediction model. A genetic algorithm is used to iterate based on the denitrification performance test results and the prediction results to screen out the liquid composite carbon source formula corresponding to the candidate liquid composite carbon source formula sample space with the best denitrification performance. In this solution, prediction is performed using the denitrification performance prediction model, and high-throughput screening is performed using the genetic algorithm, the prediction results of the denitrification performance, and the test results to obtain the liquid composite carbon source formula with the best performance. This achieves the purpose of quickly and high-throughput screening of the optimal and highly universal liquid composite carbon source formula from a variety of liquid composite carbon source formulas.

Description

Technical Field

[0001] The invention relates to the technical field of wastewater treatment, and in particular to a method and device for screening a denitrification high-efficiency liquid composite carbon source formula. Background Art

[0002] Denitrification, as an important denitrification technology in the sewage treatment process, is often limited by the insufficient C / N ratio of the wastewater itself. In order to solve the above problem, in recent years, people have considered increasing the C / N ratio by adding additional carbon sources, thereby achieving better denitrification and denitrification effects and improving sewage treatment capacity.

[0003] At present, the addition of organic acid salts (sodium formate, sodium acetate, sodium propionate, sodium citrate), organic alcohols (methanol, ethanol, glycerol, n-butanol) and sugars (sucrose, glucose, lactose) as external carbon sources has certain application prospects. According to the liquid composite carbon source, according to the type of single carbon source involved and the difference in the carbon source ratio, a large number of liquid composite carbon source formula results can be obtained. How to screen a carbon source formula with both excellent performance, low sludge yield and cost advantages is crucial for the economical and efficient operation of sewage treatment. In addition, the effectiveness of the carbon source ratio is closely related to the type of inoculated sludge microorganisms. Based on the above considerations, it is of great significance to prepare a broad-spectrum liquid composite carbon source formula.

[0004] Existing technologies often use a single chemical as a liquid carbon source to enhance denitrification, as single carbon sources offer lower performance and cost than composite carbon sources. Traditional experimental methods, which test the denitrification performance of all possible composite liquid carbon sources before selecting the optimal formula, are time-consuming and difficult to achieve high-throughput screening. Therefore, rapidly identifying the optimal formula from a diverse range of composite liquid carbon source formulations through high-throughput screening is an urgent challenge. Summary of the Invention

[0005] In view of this, the embodiments of the present invention provide a method and apparatus for screening a denitrification high-efficiency liquid composite carbon source formula, so as to achieve the purpose of quickly screening out the optimal liquid composite carbon source formula from a variety of liquid composite carbon source formulas.

[0006] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0007] The first aspect of the embodiment of the present invention discloses a method for screening a denitrification high-efficiency liquid composite carbon source formula, comprising:

[0008] Based on the combination of the selection space of the liquid composite carbon source and the selection space of the inoculated sludge, a plurality of sample spaces of liquid composite carbon source formulas to be selected are constructed, wherein the selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources according to different combinations and proportions, the selection space of the inoculated sludge is greater than 2, and each liquid composite carbon source formula corresponds to a sample space of liquid composite carbon source formula to be selected;

[0009] Randomly extracting a preset proportion of the candidate liquid composite carbon source formula sample space from a plurality of candidate liquid composite carbon source formula sample spaces as the experimental sample space;

[0010] The denitrification performance test is performed on the experimental sample space to obtain the denitrification performance test results corresponding to each experimental sample space. The denitrification performance test results include: NO3 - -N removal rate and ΔOD 600 ;

[0011] The remaining unselected sample spaces of the liquid composite carbon source formula to be selected are used as sample spaces to be predicted, and are input into the denitrification performance prediction model to predict the denitrification performance, and obtain the denitrification performance prediction results. The denitrification performance prediction model is pre-built based on the regression algorithm and the experimental sample space. The denitrification performance prediction results include: NO3 - -N removal rate and ΔOD 600 ;

[0012] Iterative calculation is performed using a genetic algorithm and the denitrification performance test results and the denitrification performance prediction results to obtain a liquid composite carbon source formula corresponding to a sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different inoculated sludge selection spaces.

[0013] Preferably, the single liquid carbon source comprises one of sodium formate, sodium acetate, sodium propionate, sodium citrate, methanol, ethanol, glycerol, n-butanol, sucrose, glucose and lactose.

[0014] Preferably, the selection space of the liquid composite carbon source includes Possibly, the sample space of liquid composite carbon source formula to be selected includes possibilities;

[0015] in, It represents the number of selection spaces of liquid composite carbon sources formed by combining two single liquid carbon sources from 11 single liquid carbon sources C according to five different combination ratios. Indicates the number of selection spaces of liquid composite carbon sources composed of three single liquid carbon sources among 11 single liquid carbon sources C according to 10 different combination ratios, It represents the number of sample spaces of liquid composite carbon source formulas to be selected by combining the selection space of each liquid composite carbon source with the selection space of the three inoculated sludges.

[0016] Preferably, the denitrification performance test is performed on the experimental sample space to obtain the denitrification performance test result corresponding to each experimental sample space, including:

[0017] For each experimental sample space, prepare the initial solution with consistent initial conditions;

[0018] After the initial solution is deoxygenated for a preset time under a nitrogen atmosphere, a preset volume of the initial solution is placed in a 96-well plate and sealed with paraffin oil;

[0019] The 96-well plate was cultured in a constant temperature box based on a preset temperature and a preset culture time. During the culture process, the initial solution was monitored using a microplate reader to obtain the initial OD value. 600 and the final OD 600 Calculate ΔOD 600 ;

[0020] Detect and based on the current initial solution NO3 - -N concentration and initial solution NO3 before incubation - -N concentration, calculated as NO3 - -N removal rate.

[0021] Preferably, the initial conditions of the initial solution include: chemical oxygen demand concentration of 300 mg / L, NO3 - -N concentration was 50 mg / L and sludge OD 600 =0.04;

[0022] The preset time for deoxygenating the initial solution under nitrogen atmosphere includes: any time from 15 minutes to 30 minutes;

[0023] After the initial solution was deoxygenated for a predetermined time under a nitrogen atmosphere, the volumes of the initial solutions selected and placed in the 96-well plate included: 200 μL;

[0024] After the initial solution was placed in the 96-well plate, the volume of paraffin oil used to seal the 96-well plate included: 80 μL;

[0025] The preset temperatures for culturing the 96-well plate in the incubator include: 30°C;

[0026] The preset culture time for the 96-well plate to be cultured in a constant temperature box includes: 48 hours.

[0027] Preferably, the process of constructing the denitrification performance prediction model includes:

[0028] Divide the experimental sample space into training set and test set according to the preset ratio;

[0029] Encoding the training set and the test set using a topological structure descriptor and a functional group descriptor;

[0030] Constructing a denitrification performance prediction model to be trained;

[0031] Inputting the encoded training set into the denitrification performance prediction model, and training using a regression algorithm to obtain a trained denitrification performance prediction model; the regression algorithm includes: decision tree algorithm, linear regression algorithm, support vector regression algorithm, K nearest neighbor algorithm, random forest algorithm, adaptive boosting algorithm, gradient boosting algorithm, bagging regression algorithm, extreme tree algorithm and multilayer perceptron algorithm;

[0032] Input the encoded test set into the trained denitrification performance prediction model, and judge whether the accuracy of the trained denitrification performance prediction model meets the requirements based on the output denitrification performance prediction results;

[0033] If so, determine to obtain a denitrification performance prediction model with accuracy that meets the requirements;

[0034] If not, continue inputting the training set into the denitrification performance prediction model for training until it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements.

[0035] Preferably, inputting the test set into the trained denitrification performance prediction model, and judging whether the accuracy of the trained denitrification performance prediction model meets the requirements based on the output denitrification performance prediction result, includes:

[0036] Inputting the test set into the trained denitrification performance prediction model to perform denitrification performance prediction to obtain a denitrification performance prediction result;

[0037] The denitrification performance prediction result and the denitrification performance test result are input into the determination coefficient formula to calculate the determination coefficient value; the determination coefficient formula is: Among them, R 2 is the determination coefficient, N is the total number of liquid composite carbon source formulas in the test set, and y pi is the denitrification performance prediction result, y ai is the denitrification performance test result, is the average value of the test results of denitrification performance; the determination coefficient values ​​include: NO3 - -N removal coefficient and ΔOD 600 The coefficient of determination value of

[0038] When the NO3 - -N removal efficiency determination coefficient value is greater than or equal to 0.8, and the ΔOD 600 When the determination coefficient value is greater than or equal to 0.85, it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements;

[0039] When the NO3 - -N removal rate determination coefficient value is less than 0.8, or ΔOD 600 When the determination coefficient value is less than 0.85, it is determined that the accuracy of the trained denitrification performance prediction model does not meet the requirements.

[0040] Preferably, the iterative calculation using the genetic algorithm and the denitrification performance test results and the denitrification performance prediction results to obtain the liquid composite carbon source formula corresponding to the sample space of the candidate liquid composite carbon source formula with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges includes:

[0041] Query and obtain the carbon source price of each single liquid carbon source;

[0042] The Z-score normalization formula, the denitrification performance test results, the denitrification performance prediction results, and the carbon source price of each single liquid carbon source are used to calculate and obtain the standardized denitrification performance test results, the denitrification performance prediction results, and the carbon source price; the Z-score normalization formula is: Where x is the denitrification performance, μ is the mean value of the denitrification performance, and σ represents the standard deviation;

[0043] Using a genetic algorithm with pre-set parameter conditions, as well as standardized denitrification performance test results, denitrification performance prediction results, and carbon source prices, an iterative calculation is performed to obtain the liquid composite carbon source formula corresponding to the sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different inoculated sludge selection spaces;

[0044] The pre-set parameter conditions in the genetic algorithm include: the initial number of species is 100, the number of iterations ranges from 40 to 120, and the mutation probability is 0.08. The carbon source score formula is: Carbon source score = K1*NO3 - -N removal rate + K2*ΔOD 600 +K3*carbon source price, where K1, K2 and K3 are parameters, and the sum of the absolute values ​​of K1, K2 and K3 is 1.

[0045] Preferably, the liquid composite carbon source formula corresponding to the sample space of the candidate liquid composite carbon source formula with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges includes: glycerol: sodium acetate: ethanol = 4:1:1.

[0046] A second aspect of an embodiment of the present invention discloses a screening device for a denitrification high-efficiency liquid composite carbon source formula, comprising:

[0047] A construction unit is used to construct a plurality of sample spaces of liquid composite carbon source formulas to be selected, comprising liquid composite carbon source formulas, based on a combination of a selection space of a liquid composite carbon source and a selection space of an inoculum sludge, wherein the selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources in different combinations and proportions, the selection space of the inoculum sludge is greater than two, and each liquid composite carbon source formula corresponds to a sample space of liquid composite carbon source formula to be selected;

[0048] An extraction unit is used to randomly extract a preset proportion of the candidate liquid composite carbon source formula sample space from a plurality of candidate liquid composite carbon source formula sample spaces as an experimental sample space;

[0049] The testing unit is used to perform a denitrification performance test on the experimental sample space to obtain a denitrification performance test result corresponding to each experimental sample space. The denitrification performance test result includes: NO3 - -N removal rate and ΔOD 600 ;

[0050] The prediction unit is used to input the remaining unextracted sample space of the selected liquid composite carbon source formula as the sample space to be predicted into the denitrification performance prediction model to predict the denitrification performance and obtain the denitrification performance prediction result. The denitrification performance prediction model is pre-built based on the regression algorithm and the experimental sample space. The denitrification performance prediction result includes: NO3 - -N removal rate and ΔOD 600 ;

[0051] The screening unit is used to perform iterative calculations using a genetic algorithm and the denitrification performance test results and the denitrification performance prediction results to obtain a liquid composite carbon source formula corresponding to a sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different selection spaces of inoculated sludge.

[0052] Based on the above-mentioned embodiment of the present invention, a screening method and device for a denitrification efficient liquid composite carbon source formula is provided. Based on the combination of the selection space of the liquid composite carbon source and the selection space of the inoculated sludge, a plurality of sample spaces of liquid composite carbon source formulas to be selected containing liquid composite carbon source formulas are constructed. The selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources according to different combinations and proportions. The selection space of the inoculated sludge is greater than 2. Each of the liquid composite carbon source formulas corresponds to a sample space of liquid composite carbon source formula to be selected. From the plurality of sample spaces of liquid composite carbon source formulas to be selected, a preset proportion of sample spaces of liquid composite carbon source formula to be selected is randomly selected as an experimental sample space. A denitrification performance test is performed on the experimental sample space to obtain a denitrification performance test result corresponding to each experimental sample space. The denitrification performance test result includes: NO3 - -N removal rate and ΔOD 600 The remaining unextracted sample space of the selected liquid composite carbon source formula is used as the sample space to be predicted, and is input into the denitrification performance prediction model to predict the denitrification performance, and obtain the denitrification performance prediction result. The denitrification performance prediction model is pre-constructed based on the regression algorithm and the experimental sample space. The denitrification performance prediction result includes: NO3 - -N removal rate and ΔOD 600 ; Using a genetic algorithm, as well as the denitrification performance test results and the denitrification performance prediction results, iterative calculations are performed to obtain the liquid composite carbon source formula corresponding to the sample space of the candidate liquid composite carbon source formula with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges. In this solution, the experimental sample space is tested by experimental means, and the sample space to be predicted is predicted using a denitrification performance prediction model. Finally, high-throughput screening is performed based on the genetic algorithm, the denitrification performance prediction results, and the test results to obtain the liquid composite carbon source formula with the best performance, thereby achieving the purpose of quickly and high-throughput screening of the optimal and highly universal liquid composite carbon source formula from a variety of liquid composite carbon source formulas. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0054] Figure 1 This is a flow chart of a method for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in an embodiment of the present invention;

[0055] Figure 2This is a flow chart of a method for constructing a denitrification performance prediction model disclosed in an embodiment of the present invention;

[0056] Figure 3 A flow chart of an application method for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in an embodiment of the present invention;

[0057] Figure 4 This is a structural diagram of a screening device for a denitrification high-efficiency liquid composite carbon source formula disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] In this application, the terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0060] As can be seen from the background art, the traditional experimental method is to test the denitrification performance of all possible liquid composite carbon sources and then screen them, which is too time-consuming and difficult to implement.

[0061] Therefore, the embodiment of the present invention discloses a method and device for screening a liquid composite carbon source formula with high denitrification efficiency. In this scheme, a denitrification performance test is performed on the liquid composite carbon source formula in the experimental sample space to obtain a denitrification performance test result. A denitrification performance prediction model is used to predict the liquid composite carbon source formula in the sample space to be predicted, and a corresponding denitrification performance prediction result is obtained. A genetic algorithm is used to perform high-throughput screening based on the denitrification performance prediction result and the test result to obtain the liquid composite carbon source formula with the best performance, thereby achieving the purpose of quickly and high-throughput screening of the optimal and highly universal liquid composite carbon source formula from a variety of liquid composite carbon source formulas.

[0062] like Figure 1 FIG. 1 is a flow chart of a method for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in an embodiment of the present invention, and the method mainly comprises the following steps:

[0063] Step S101: constructing a plurality of sample spaces of liquid composite carbon source formulas to be selected, including liquid composite carbon source formulas, based on a combination of the selection space of the liquid composite carbon source and the selection space of the inoculated sludge.

[0064] In step S101 , the combination formula of the selection space of the liquid composite carbon source and the selection space of the inoculated sludge is: sample space of liquid composite carbon source formula to be selected = selection space of liquid composite carbon source * selection space of inoculated sludge.

[0065] The selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources in different combinations and proportions, and each liquid composite carbon source formula corresponds to a sample space of liquid composite carbon source formulas to be selected.

[0066] In order to ensure that the screened liquid composite carbon source formula has good universality for the field operation of various sewage treatment plants, the selection space of inoculated sludge is greater than 2 types, and comes from different sewage treatment plants, covering the reagent application conditions of sewage treatment plants of all denitrification processes, so as to screen out the liquid composite carbon source formula with the best performance in the selection space of various inoculated sludges in the subsequent process.

[0067] The single liquid carbon source includes, but is not limited to, one of 11 types: sodium formate, sodium acetate, sodium propionate, sodium citrate, methanol, ethanol, glycerol, n-butanol, sucrose, glucose and lactose. Any existing single liquid carbon source that can be used in the denitrification process can be used in the embodiments of the present invention.

[0068] Considering that the evaluation of the denitrification performance of a carbon source often needs to be considered from three aspects: denitrification response rate, sludge production, and price, the traditional single liquid carbon source cannot take all three aspects into account at the same time. Therefore, this application combines the traditional single liquid carbon source into a liquid composite carbon source. While ensuring excellent sewage treatment effect, it reduces the carbon source addition cost of the sewage treatment plant, further reduces sludge production and thus reduces sludge treatment costs, and realizes efficient and comprehensive utilization of resources. The specific combination method is as follows:

[0069] When the selection space of the liquid composite carbon source is composed of two single liquid carbon sources, the combination form is (A, B, 0); when the selection space of the liquid composite carbon source is composed of three single liquid carbon sources, the combination form is (A, B, C), where A, B and C represent different types of single liquid carbon sources.

[0070] When the selection space of the liquid composite carbon source is composed of a combination of two single liquid carbon sources, the combination ratios A:B include: 5:1, 1:5, 2:4, 4:2 and 3:3; when the selection space of the liquid composite carbon source is composed of a combination of three single liquid carbon sources, the combination ratios A:B:C include: 4:1:1, 1:4:1, 1:1:4, 3:2:1, 3:1:2, 2:3:1, 2:1:3, 1:2:3, 1:3:2 and 2:2:2.

[0071] It should be noted that the ratio of the combination of single liquid carbon sources in the selection space of liquid composite carbon sources is determined by calculation based on the chemical oxygen demand of each single liquid carbon source.

[0072] It is understood that under the aforementioned conditions of single liquid carbon source, combination form and combination ratio, the selection space of liquid composite carbon source includes Possible, the sample space of liquid composite carbon source formula to be selected includes possibilities.

[0073] in, It represents the number of selection spaces of liquid composite carbon sources composed of two single liquid carbon sources among 11 single liquid carbon sources according to five different combination ratios. Indicates the number of selection spaces of liquid composite carbon sources composed of three single liquid carbon sources among 11 single liquid carbon sources according to 10 different combination ratios, It represents the number of sample spaces of liquid composite carbon source formulas to be selected by combining the selection space of each liquid composite carbon source with the selection space of the three inoculated sludges.

[0074] Step S102: randomly extracting a preset proportion of the candidate liquid composite carbon source formula sample spaces from a plurality of candidate liquid composite carbon source formula sample spaces as experimental sample spaces.

[0075] In the specific implementation process of step S102, a random sampling method is adopted to randomly select 30% of the candidate liquid composite carbon source formula sample spaces from a plurality of candidate liquid composite carbon source formula sample spaces as the experimental sample space.

[0076] Step S103: performing a denitrification performance test on the experimental sample space to obtain a denitrification performance test result corresponding to each experimental sample space.

[0077] In step S103, the denitrification performance test results include: NO3 - -N removal rate and ΔOD 600 .

[0078] In the specific implementation process of step S103, the denitrification performance test of the experimental sample space is performed using a 96-well plate high-throughput method, which mainly includes the following steps:

[0079] Step S11: For each experimental sample space, prepare an initial solution with consistent initial conditions.

[0080] In step S11, the initial conditions of the initial solution include: chemical oxygen demand concentration of 300 mg / L, NO3 - -N concentration was 50 mg / L and sludge OD 600 =0.04.

[0081] Step S12: After the initial solution is deoxygenated for a preset time under a nitrogen atmosphere, a preset volume of the initial solution is placed in a 96-well plate and sealed with paraffin oil.

[0082] In step S12 , the initial solution is deoxygenated under a nitrogen atmosphere for a preset time period ranging from 15 minutes to 30 minutes, and the volume of the selected initial solution is 200 μL.

[0083] Preferably, the initial solution is deoxygenated under nitrogen atmosphere for a preset time of 20 minutes.

[0084] Step S13: The 96-well plate is cultured in a constant temperature box based on a preset temperature and a preset culture time. During the culture process, the initial solution is monitored using a microplate reader to obtain the initial OD value. 600 and the final OD 600 Calculate ΔOD 600 .

[0085] In the specific implementation process of step S13, the 96-well plate was cultured in a constant temperature box at 30°C for 48 hours, and during the culture process, the initial solution was monitored in real time using an enzyme marker to obtain the initial OD 600 and the final OD 600 , calculate the initial OD 600 and the final OD 600 The difference between the two values ​​is ΔOD 600 .

[0086] Step S14: Detect and based on the current initial solution NO3 - -N concentration and initial solution NO3 before incubation - -N concentration, calculated as NO3 - -N removal rate.

[0087] In step S14, the initial solution NO3 before incubation - -N concentration, which is the NO3 in the initial conditions of the initial solution- -N concentration.

[0088] In the specific implementation process of step S14, the current initial solution NO3 is detected. - -N concentration, using the formula: And the initial solution NO3 before culture - -N concentration, calculated as NO3 - -N removal rate.

[0089] Step S104: input the remaining unextracted sample spaces of the candidate liquid composite carbon source formula as sample spaces to be predicted into the denitrification performance prediction model to perform denitrification performance prediction and obtain denitrification performance prediction results.

[0090] It should be noted that since manual denitrification performance testing of all candidate liquid composite carbon source formula sample spaces takes too long and high-throughput screening cannot be achieved, and considering that machine learning has the ability to efficiently predict and summarize big data sample results, this application adopts a machine learning method (i.e., the denitrification performance prediction model in this application) to predict the denitrification performance of the candidate liquid composite carbon source formula sample space, thereby providing conditions for subsequent genetic algorithm high-throughput screening.

[0091] In the specific implementation process of step S104, the remaining unextracted sample space of the selected liquid composite carbon source formula is used as the sample space to be predicted, and the sample space to be predicted is encoded using the topological structure descriptor and the functional group descriptor. The encoded sample space to be predicted is input into the denitrification performance prediction model to predict the denitrification performance, and the denitrification performance prediction result is obtained.

[0092] Among them, the topological descriptors include RDKit, Atom Pairs, Morgan and Pattern, the functional group descriptor is MACCSkeys, and the classification descriptor is one-hot.

[0093] Preferably, Atom Pairs are used to encode the sample space to be predicted.

[0094] It should be noted that the purpose of encoding the sample space to be predicted is to convert chemical language information such as the chemical molecular formula of the single liquid carbon source itself into mathematical matrix information that can be recognized by a computer.

[0095] Preferably, in the sample space of liquid composite carbon source formulas to be selected, the sample space of liquid composite carbon source formulas to be selected that is extracted accounts for 30%, and the remaining sample space of liquid composite carbon source formulas to be selected that is not extracted accounts for 70%.

[0096] In step S104, the denitrification performance prediction model is pre-built based on the regression algorithm and the experimental sample space. The prediction results of the denitrification performance include: NO3 - -N removal rate and ΔOD 600 .

[0097] Among them, the regression algorithms include: decision tree algorithm, linear regression algorithm, support vector regression (SVR) algorithm, K-nearest neighbor (KNN) algorithm, random forest algorithm, adaptive boosting (AdaBoost) algorithm, gradient boosting algorithm, bagging regression algorithm, extra tree algorithm and multilayer perceptron (MLP) algorithm, a total of 10 types.

[0098] It should be noted that when constructing the denitrification performance prediction model, it is preferred to use a multi-layer perceptron algorithm and a training set encoded with AtomPairs to construct the denitrification performance prediction model, so that the denitrification performance prediction model has excellent prediction performance.

[0099] Among them, the denitrification performance prediction model with the best prediction performance predicted NO3 - -N removal rate and NO3 obtained in actual experiment - -N removal rate between the coefficient of determination R 2 Up to 0.8, the predicted ΔOD 600 Compared with the ΔOD obtained in the actual experiment 600 The coefficient of determination R 2 As high as 0.85.

[0100] The formula for the coefficient of determination is: Among them, R 2 is the determination coefficient, N is the total number of liquid composite carbon source formulas in the test set, y pi is the denitrification performance prediction result, y ai is the denitrification performance test result, It is the average value of the test results of denitrification performance.

[0101] It should be noted that since the test results and prediction results of denitrification performance include NO3 - -N removal rate and ΔOD 600 , so the coefficient of determination includes: NO3 - -N removal coefficient and ΔOD 600The coefficient of determination of .

[0102] Step S105: using a genetic algorithm, denitrification performance test results, and denitrification performance prediction results to perform iterative calculations to obtain a liquid composite carbon source formula corresponding to a sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different inoculated sludge selection spaces.

[0103] In step S105 , the genetic algorithm is a genetic algorithm with pre-set parameter conditions.

[0104] Preferably, the parameter conditions are set specifically including: the initial number of species is 100, the number of iterations ranges from 40 to 120, and the mutation probability is 0.08.

[0105] The specific implementation process of step S105 includes the following steps:

[0106] Step S21: query and obtain the carbon source price of each single liquid carbon source.

[0107] In step S21 , the carbon source price of each single liquid carbon source refers to the current market price of each single liquid carbon source.

[0108] Step S22: Calculate using the Z-score standardization formula, the denitrification performance test results, the denitrification performance prediction results, and the carbon source price of each single liquid carbon source to obtain standardized denitrification performance test results, denitrification performance prediction results, and carbon source prices.

[0109] In step S22, the Z-score normalization formula is: Where x is the denitrification performance, μ is the mean value of the denitrification performance, and σ is the standard deviation.

[0110] To illustrate the purpose of Z-score standardization, the present invention provides the following examples:

[0111] Taking glycerol as an example, assuming that its NO3 - -N removal rate was 85% (0.85), ΔOD 600 is 0.4, and the price of glycerol is 3.99 yuan / kg COD. If Z-score standardization is not performed, it is not difficult to see that the actual value of glycerol price is NO3 - -N removal rate is 4.7 times the actual value, which is ΔOD 600 Even if the carbon source score formula (carbon source score = K1*NO3 - -N removal rate + K2*ΔOD 600+K3*carbon source price) to correct the parameters K1, K2, and K3. The value of the glycerol price parameter itself directly affects the carbon source score calculation result, while NO3 is ignored. - -N removal rate and ΔOD 600 The influence weight of the data itself is far greater than the influence of different parameter data.

[0112] NO3 - -N removal rate, ΔOD 600 After Z-score normalization with carbon source price, the deviation degree of data from the mean is used to describe the data itself, eliminating the influence of the size of the data itself, making NO3 - -N removal rate, ΔOD 600 The data scale range, namely the Z value, is the same as that of the carbon source price, and the Z value has the same value range. The carbon source score calculation result is no longer directly dominated by the size of the data itself, making the model screening result reliable.

[0113] Step S23: Using a genetic algorithm with pre-set parameter conditions, as well as the standardized denitrification performance test results, denitrification performance prediction results and carbon source prices, iterative calculations are performed to obtain the liquid composite carbon source formula corresponding to the sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges.

[0114] In step S23, the carbon source score formula is: Carbon source score = K1*NO3 - -N removal rate + K2*ΔOD 600 +K3*carbon source price, where K1, K2 and K3 are parameters, and the sum of the absolute values ​​of K1, K2 and K3 is 1.

[0115] It should be noted that the purpose of using carbon source price as a condition affecting carbon source score is to ensure excellent NO3 - -N removal rate, while trying to find a liquid composite carbon source formula with a lower cost.

[0116] Preferably, the values ​​of K1, K2 and K3 include: |k1|=0.4, |k2|=0.4, |k3|=0.2.

[0117] Preferably, the values ​​of K1, K2 and K3 include: k1 = 0.4, k2 = -0.4, k3 = -0.2.

[0118] The pre-set parameter conditions in the genetic algorithm include: the initial number of species, the number of iterations and the mutation probability.

[0119] In one embodiment, the initial number of species is 100, the number of iterations is 40 to 120, and the mutation probability is 0.08.

[0120] Preferably, the score of the genetic algorithm begins to stabilize after 40 iterations, and it is considered that the optimal liquid composite carbon source formula result corresponding to the highest score is found.

[0121] In one embodiment, the genetic algorithm performs iterative calculations under the scikit-opt data package.

[0122] In one embodiment, the highest carbon source score obtained by iterative calculation of the genetic algorithm is 1.160, and the optimal liquid composite carbon source formula corresponding to the highest carbon source score is glycerol: sodium acetate: ethanol = 4:1:1.

[0123] Based on the above-mentioned method for screening a denitrification-efficient liquid composite carbon source formula disclosed in the embodiment of the present invention, a plurality of candidate liquid composite carbon source formula sample spaces containing liquid composite carbon source formulas are constructed, and divided into an experimental sample space and a sample space to be predicted, the experimental sample space is tested for denitrification performance, the denitrification performance of the sample space to be predicted is predicted using a denitrification performance prediction model, and finally a genetic algorithm is used to perform iterative calculations based on the test results and prediction results of the denitrification performance to screen out the liquid composite carbon source formula corresponding to the candidate liquid composite carbon source formula sample space with the best denitrification performance and the highest carbon source score. In this scheme, the denitrification performance of the experimental sample space is obtained through actual testing, the denitrification performance of the sample space to be predicted is predicted using a denitrification performance prediction model, and finally a genetic algorithm is used to perform high-throughput screening based on the test results and prediction results of the denitrification performance to obtain the liquid composite carbon source formula with the best performance, thereby achieving the purpose of quickly and high-throughput screening of the optimal and highly universal liquid composite carbon source formula from a variety of liquid composite carbon source formulas.

[0124] In order to illustrate the construction process of the denitrification test model in the screening method of a denitrification high-efficiency liquid composite carbon source formula disclosed in the embodiment of the present invention, as shown in FIG. Figure 2 FIG. 1 is a flow chart of a method for constructing a denitrification performance prediction model disclosed in an embodiment of the present invention, the method comprising the following steps:

[0125] Step S201: Divide the experimental sample space into a training set and a test set according to a preset ratio.

[0126] In step S201, the experimental sample space is created by executing Figure 1 The corresponding steps S101 to S102 in the embodiment are obtained.

[0127] Step S202: Encode the training set and the test set using the topology descriptor and the functional group descriptor.

[0128] Among them, the topological descriptors include RDKit, Atom Pairs, Morgan and Pattern, the functional group descriptor is MACCSkeys, and the classification descriptor is one-hot.

[0129] Preferably, Atom Pairs are used to encode the training set and the test set.

[0130] It should be noted that the purpose of encoding the training set and the test set is to convert chemical language information such as the chemical molecular formula of the single liquid carbon source itself into mathematical matrix information that can be recognized by the computer.

[0131] Step S203: constructing a denitrification performance prediction model to be trained.

[0132] Step S204: inputting the encoded training set into the denitrification performance prediction model, and performing training using a regression algorithm to obtain a trained denitrification performance prediction model.

[0133] In step S203, the regression algorithm includes but is not limited to: decision tree algorithm, linear regression algorithm, support vector regression algorithm, K nearest neighbor algorithm, random forest algorithm, adaptive boosting algorithm, gradient boosting algorithm, bagging regression algorithm, extreme tree algorithm and multilayer perceptron algorithm.

[0134] Preferably, step S203 adopts a multi-layer perceptron algorithm.

[0135] Step S205: Input the encoded test set into the trained denitrification performance prediction model. Based on the output denitrification performance prediction results, determine whether the accuracy of the trained denitrification performance prediction model meets the requirements. If so, proceed to step S206; otherwise, proceed to step S207.

[0136] In step S205, the process of determining whether the accuracy of the trained denitrification performance prediction model meets the requirements is as follows:

[0137] The denitrification performance prediction results and the denitrification performance test results are input into the determination coefficient formula to calculate the determination coefficient value.

[0138] Among them, the denitrification performance prediction result is the denitrification performance prediction result output by the denitrification performance prediction model for the test set, and the denitrification performance test result is the execution Figure 1 The denitrification performance test result obtained in step S103 in the illustrated embodiment.

[0139] The coefficient of determination formula is: Among them, R 2 is the determination coefficient, N is the total number of liquid composite carbon source formulas in the test set, ypi is the denitrification performance prediction result, y ai is the denitrification performance test result, It is the average value of the test results of denitrification performance.

[0140] It should be noted that since the test results and prediction results of denitrification performance include NO3 - -N removal rate and ΔOD 600 , so the coefficient of determination includes: NO3 - Determination coefficient of -N removal rate and ΔOD 600 The coefficient of determination value.

[0141] When NO3 - -N removal rate determination coefficient value is greater than or equal to the first preset value, and ΔOD 600 When the determination coefficient value is greater than or equal to the second preset value, it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements.

[0142] When NO3 - -N removal rate determination coefficient value is less than the first preset value, or ΔOD 600 When the determination coefficient value is less than the second preset value, it is determined that the accuracy of the trained denitrification performance prediction model does not meet the requirement.

[0143] Preferably, the first preset value is 0.8 and the second preset value is 0.85.

[0144] Step S206: Determine a denitrification performance prediction model with accuracy that meets the requirements.

[0145] Step S207: Continue inputting the training set into the denitrification performance prediction model for training until it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements.

[0146] Based on the method for constructing a denitrification performance prediction model disclosed in the above-mentioned embodiment of the present invention, the experimental sample space is proportionally divided into a training set and a test set, a denitrification performance prediction model to be trained is constructed, and then the denitrification performance prediction model to be trained is trained using the training set and a regression algorithm to obtain a trained denitrification performance prediction model. In this solution, the denitrification performance prediction model is constructed using the experimental samples and the regression algorithm, thereby achieving the purpose of using the denitrification performance prediction model to predict the denitrification performance of the experimental sample space to be predicted.

[0147] Based on the above-mentioned method for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in the embodiment of the present invention, Figure 3 FIG. 1 is a flow chart of an application method for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in an embodiment of the present invention. The application method includes the following steps:

[0148] Step S301: combining two or three of the 11 single liquid carbon sources in a combination ratio of 5 or 10 to obtain a selection space of 1925 liquid composite carbon sources.

[0149] In step S301 , the 11 single liquid carbon sources include: sodium formate, sodium acetate, sodium propionate, sodium citrate, methanol, ethanol, glycerol, n-butanol, sucrose, glucose and lactose.

[0150] When the selection space of the liquid composite carbon source is composed of two single liquid carbon sources, the combination form is (A, B, 0); when the selection space of the liquid composite carbon source is composed of three single liquid carbon sources, the combination form is (A, B, C), where A, B and C represent different types of single liquid carbon sources.

[0151] When the selection space of liquid composite carbon sources is composed of two single liquid carbon sources, the combination ratio A:B includes 5 kinds: 5:1, 1:5, 2:4, 4:2 and 3:3. When the selection space of liquid composite carbon sources is composed of three single liquid carbon sources, the combination ratio A:B:C includes 10 kinds: 4:1:1, 1:4:1, 1:1:4, 3:2:1, 3:1:2, 2:3:1, 2:1:3, 1:2:3, 1:3:2 and 2:2:2.

[0152] Therefore, the selection space of liquid composite carbon sources includes possibilities.

[0153] in, It represents the number of selection spaces of liquid composite carbon sources composed of two single liquid carbon sources among 11 single liquid carbon sources according to five different combination ratios. It represents the number of selection spaces of liquid composite carbon sources formed by combining three single liquid carbon sources among 11 single liquid carbon sources according to 10 different combination ratios.

[0154] Step S302: Based on the combination of the selection space of 1925 liquid composite carbon sources and the selection space of 3 inoculated sludges, a sample space of liquid composite carbon source formulas to be selected is constructed, which includes 5775 liquid composite carbon source formulas.

[0155] In step S302, the sample space of liquid composite carbon source formula to be selected is constructed based on the formula: sample space of liquid composite carbon source formula to be selected = selection space of liquid composite carbon source * selection space of inoculated sludge.

[0156] In order to ensure that the screened liquid composite carbon source formula has good universality for the field operation of each sewage treatment plant, the selection space of the three types of inoculated sludge comes from different sewage treatment plants. In the embodiment of the present invention, Class A sludge, Class B sludge and Class C sludge are used to represent the selection space of the three types of inoculated sludge.

[0157] Therefore, the sample space of liquid composite carbon source formulas to be selected includes possibilities.

[0158] in, It represents the number of sample spaces of liquid composite carbon source formulas to be selected by combining the selection space of each liquid composite carbon source with the selection space of the three inoculated sludges.

[0159] Step S303: randomly selecting 1773 candidate liquid composite carbon source formula sample spaces from the 5775 candidate liquid composite carbon source formula sample spaces as experimental sample spaces.

[0160] Step S304: performing a denitrification performance test on each experimental sample space to obtain a denitrification performance test result corresponding to each experimental sample space.

[0161] In step S304, the denitrification performance test results include: NO3 - -N removal rate and ΔOD 600 .

[0162] The specific process of denitrification performance test results is as follows:

[0163] For each experimental sample space, COD 300mg / L, NO3 - -N is 50mg / L, sludge OD 600 = 0.04 initial turbid solution.

[0164] Subsequently, the obtained turbid liquid was deoxygenated under N2 atmosphere for 20 min for anaerobically treatment.

[0165] 200 μL of the above turbid solution was placed in a 96-well plate and anaerobically sealed with 80 μL of paraffin oil.

[0166] Then, the 96-well plate was placed in a 30°C constant temperature incubator for 48 h, and the biomass (OD 600) Continuous monitoring and calculation of NO3 at the end of the culture - -N removal rate; finally, the denitrification performance test results corresponding to each experimental sample space were obtained.

[0167] Step S305: The remaining 4002 unselected liquid composite carbon source formula sample spaces are used as sample spaces to be predicted, and are input into the denitrification performance prediction model to perform denitrification performance prediction and obtain denitrification performance prediction results.

[0168] In step S305 , a denitrification performance prediction model is pre-constructed based on a regression algorithm and 1773 experimental sample spaces. The specific construction process is described in the above embodiment of the present invention and will not be described again here.

[0169] Denitrification performance prediction results include: NO3 - -N removal rate and ΔOD 600 .

[0170] Step S306: Using a genetic algorithm with preset parameter conditions of 100 initial species, 120 iterations, and a mutation probability of 0.08, and a carbon source scoring formula, 5775 liquid composite carbon source formula sample spaces were screened to obtain the optimal liquid composite carbon source formula corresponding to the highest carbon source score of 1.160: glycerol: sodium acetate: ethanol = 4:1:1.

[0171] In step S306, the carbon source score formula is: Carbon source score = K1*NO3 - -N removal rate + K2*ΔOD 600 +K3*carbon source price, where K1, K2, and K3 are parameters, and the values ​​of K1, K2, and K3 are 0.4, -0.4, and -0.2, respectively. The carbon source price is the current market price, which can be obtained through query.

[0172] When using the carbon source formula to calculate, the NO3 - -N removal rate, OD 600 and carbon source prices, both are Z-score standardized values.

[0173] In one embodiment, the process for obtaining the second best liquid composite carbon source formula with the second best carbon source score is as follows:

[0174] Change Figure 3 The sample space of the liquid composite carbon source formula to be selected in the corresponding embodiment is to exclude the optimal liquid composite carbon source formula: glycerol: sodium acetate: ethanol = 4:1:1, and retain the other 1772 sample spaces of the liquid composite carbon source formula to be selected as the experimental sample space. The remaining steps are carried out according to Figure 3 The second highest carbon source score obtained in the corresponding example was 0.773, which corresponds to the suboptimal liquid composite carbon source formula of glycerol: ethanol: lactose = 4:1:1.

[0175] In one embodiment, the process for obtaining the composite carbon source formula with the third highest carbon source score is as follows:

[0176] Change Figure 3 The sample space of the liquid composite carbon source formula to be selected in the corresponding embodiment, that is, the optimal liquid composite carbon source formula: glycerol: sodium acetate: ethanol = 4:1:1 and the suboptimal liquid composite carbon source formula: glycerol: ethanol: lactose = 4:1:1 are eliminated, and the other 1771 sample spaces of the liquid composite carbon source formula to be selected are retained as the experimental sample space, and the remaining steps are carried out according to Figure 3 According to the corresponding example, the third highest carbon source score was 0.764, that is, the third best liquid composite carbon source formula was ethanol: glycerol: glucose = 1:2:3.

[0177] In order to verify the liquid composite carbon source formula of the first carbon source score, the second carbon source score, and the third carbon source score obtained above, the embodiment of the present invention further discloses the following verification process:

[0178] Verification Example 1:

[0179] The liquid composite carbon source formula corresponding to the first carbon source score, i.e., glycerol: sodium acetate: ethanol = 4:1:1, was applied to a continuous flow reactor for heterotrophic denitrification.

[0180] The COD concentration of the reactor inlet water is 300 mg / L (including 163.93 mg / L glycerol, 64.10 mg / L sodium acetate, and 24.04 mg / L ethanol, i.e., glycerol COD equivalent: sodium acetate COD equivalent: ethanol COD equivalent = 4:1:1), NO3 - -N concentration is 50 mg / L, the hydraulic retention time of the reactor is 8 h, the MLVSS of Class A sludge is 3000 mg / L, and the final nitrate NO3 in the reactor is measured. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0181] Verification Example 2:

[0182] The Class A sludge in the heterotrophic denitrification continuous flow reactor of Verification Example 1 was changed to Class B sludge. The remaining steps were carried out according to Verification Example 1. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0183] Verification Example 3:

[0184] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Verification Example 1 was changed to Class C sludge. The remaining steps were carried out according to Verification Example 1. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0185] Verification Example 4:

[0186] The liquid composite carbon source formula corresponding to the second carbon source score, i.e., glycerol:ethanol:lactose=4:1:1, was applied to a continuous flow reactor for heterotrophic denitrification.

[0187] The COD concentration of the reactor inlet water is 300 mg / L (including 163.93 mg / L glycerol, 24.04 mg / L ethanol, and 44.64 mg / L lactose, i.e., glycerol COD equivalent: ethanol COD equivalent: lactose COD equivalent = 4:1:1), NO3 - -N concentration is 50 mg / L, the hydraulic retention time of the reactor is 8 h, the MLVSS of Class A sludge is 3000 mg / L, and the final nitrate NO3 in the reactor is measured. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0188] Verification Example 5:

[0189] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Verification Example 2 was changed to Class B sludge. The remaining steps were carried out according to Verification Example 2. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0190] Verification Example 6:

[0191] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Verification Example 2 was changed to Class C sludge. The remaining steps were carried out according to Verification Example 2. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0192] Verification Example 7:

[0193] The liquid composite carbon source formula corresponding to the third carbon source score, i.e., ethanol: glycerol: glucose = 1:2:3, was applied to a continuous flow reactor for heterotrophic denitrification. The reactor inlet COD concentration was 300 mg / L (including 24.04 mg / L ethanol, 81.97 mg / L glycerol, and 140.19 mg / L glucose, i.e., ethanol COD equivalent: glycerol COD equivalent: glucose COD equivalent = 4:1:1), NO3 - -N concentration is 50 mg / L, the hydraulic retention time of the reactor is 8 h, the MLVSS of Class A sludge is 3000 mg / L, and the final nitrate NO3 in the reactor is measured. --N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0194] Verification Example 8:

[0195] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Verification 3 was changed to Class B sludge. The rest of the steps were carried out according to Verification 3. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0196] Verification Example 9:

[0197] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Verification Example 3 was changed to Class C sludge. The rest of the steps were carried out according to Verification 3. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0198] Verification of denitrification performance of single sodium acetate liquid carbon source:

[0199] Comparative Example 1:

[0200] The influent COD concentration of the heterotrophic denitrification continuous flow reactor is 300 mg / L (i.e., 384.62 mg / L of sodium acetate is added), NO3 - -N concentration is 50 mg / L, the hydraulic retention time of the reactor is 8 h, the MLVSS of Class A sludge is 3000 mg / L, and the final nitrate NO3 in the reactor is measured. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0201] Comparative Example 2:

[0202] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Comparative Example 1 was changed to Class B sludge, and the remaining steps were carried out according to Comparative Example 1. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0203] Comparative Example 3:

[0204] The Class A sludge in the continuous flow reactor of heterotrophic denitrification in Comparative Example 1 was changed to Class C sludge, and the remaining steps were carried out according to Comparative Example 1. The final nitrate NO3 in the reactor was tested. - -N removal rate and sludge volume change ΔOD at the beginning and end of the reaction 600 , the results are shown in Table 1.

[0205] Table 1:

[0206] <![CDATA[NO3 - -N removal rate (%)]]> <![CDATA[ΔOD 600 ]]> Verification Example 1 70.88 0.028 Verification Example 2 75.96 0.026 Verification Example 3 66.44 0.020 Verification Example 4 64.59 0.014 Verification Example 5 62.42 0.062 Verification Example 6 69.55 0.019 Verification Example 7 60.77 0.052 Verification Example 8 63.18 0.010 Verification Example 9 61.03 0.055 Comparative Example 1 58.10 0.073 Comparative Example 2 60.19 0.080 Comparative Example 3 63.12 0.065

[0207] As can be seen from Table 1, the liquid composite carbon sources of Verification Examples 1, 2 and 3 compared with Comparative Example 1, the liquid composite carbon sources of Verification Examples 4, 5 and 6 compared with Comparative Example 2, and the liquid composite carbon sources of Verification Examples 7, 8 and 9 compared with Comparative Example 3 improved the nitrate removal rate in the heterotrophic denitrification continuous flow reactor while having a lower biomass increment ΔOD 600 It is believed that the liquid composite carbon source obtained by screening has better denitrification performance than the single liquid carbon source.

[0208] In addition, compared with comparative examples 1 to 3, verification examples 1 to 9 all showed better denitrification performance. Therefore, it is believed that the screened liquid composite carbon source has the effect of efficiently and stably improving the denitrification performance of different sludges, and the results are considered to be universal.

[0209] Based on the above-mentioned screening application method of a denitrification efficient liquid composite carbon source formula disclosed in the embodiment of the present invention, by constructing a plurality of candidate liquid composite carbon source formula sample spaces containing liquid composite carbon source formulas, and dividing them into experimental sample spaces and to-be-predicted sample spaces, the experimental sample spaces are tested for denitrification performance, the to-be-predicted sample spaces are predicted for denitrification performance using a denitrification performance prediction model, and finally a genetic algorithm is used to iteratively calculate based on the test results and prediction results of the denitrification performance to screen out the liquid composite carbon source formula corresponding to the candidate liquid composite carbon source formula sample space with the best denitrification performance and the highest carbon source score. In this scheme, the denitrification performance of the experimental sample space is obtained through actual testing, the denitrification performance of the to-be-predicted sample space is predicted using a denitrification performance prediction model, and finally a genetic algorithm is used to perform high-throughput screening based on the test results and prediction results of the denitrification performance to obtain the liquid composite carbon source formula with the best performance, thereby achieving the purpose of quickly and high-throughput screening of the optimal and highly universal liquid composite carbon source formula from a variety of liquid composite carbon source formulas.

[0210] Based on the above-mentioned method for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in the embodiment of the present invention, Figure 4 , which is a structural diagram of a device for screening a denitrification high-efficiency liquid composite carbon source formula disclosed in an embodiment of the present invention, the device includes: a construction unit 401, an extraction unit 402, a testing unit 403, a prediction unit 404 and a screening unit 405.

[0211] Among them, construction unit 401 is used to construct a plurality of sample spaces of liquid composite carbon source formulas to be selected based on the combination of the selection space of the liquid composite carbon source and the selection space of the inoculated sludge, and the selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources according to different combinations and proportions.

[0212] Specifically, the selection space of the inoculated sludge is greater than 2, and each liquid composite carbon source formula corresponds to a sample space of liquid composite carbon source formulas to be selected.

[0213] In one embodiment, the single liquid carbon source includes one of sodium formate, sodium acetate, sodium propionate, sodium citrate, methanol, ethanol, glycerol, n-butanol, sucrose, glucose and lactose.

[0214] In one embodiment, the selection space of the liquid composite carbon source is a combination of two single liquid carbon sources, and the combination ratio of the two single liquid carbon sources includes: 5:1, 1:5, 2:4, 4:2 and 3:3.

[0215] In one embodiment, the selection space of the liquid composite carbon source is a combination of three single liquid carbon sources, and the combination ratios of the three single liquid carbon sources include: 4:1:1, 1:4:1, 1:1:4, 3:2:1, 3:1:2, 2:3:1, 2:1:3, 1:2:3, 1:3:2 and 2:2:2.

[0216] In one embodiment, the combination ratio of the single liquid carbon sources in the selection space of the liquid composite carbon source is determined by calculation based on the chemical oxygen demand of each single liquid carbon source.

[0217] The extraction unit 402 is configured to randomly extract a preset proportion of the candidate liquid composite carbon source formula sample spaces from a plurality of candidate liquid composite carbon source formula sample spaces as the experimental sample space.

[0218] The testing unit 403 is used to perform a denitrification performance test on the experimental sample space and obtain a denitrification performance test result corresponding to each experimental sample space.

[0219] Denitrification performance test results include: NO3 - -N removal rate and ΔOD 600 ;

[0220] In one embodiment, the testing unit 403 is specifically configured to:

[0221] For each experimental sample space, an initial solution with consistent initial conditions is prepared.

[0222] After the initial solution was deoxygenated for a preset time under a nitrogen atmosphere, a preset volume of the initial solution was placed in a 96-well plate and sealed with paraffin oil.

[0223] The 96-well plate was cultured in a constant temperature box based on the preset temperature and the preset culture time. During the culture process, the initial solution was monitored using a microplate reader to obtain the initial OD value. 600 and the final OD 600 Calculate ΔOD 600 .

[0224] Detect and based on the current initial solution NO3 - -N concentration and initial solution NO3 before incubation - -N concentration, calculated as NO3 - -N removal rate.

[0225] The prediction unit 404 is used to input the remaining unextracted sample space of the selected liquid composite carbon source formula as the sample space to be predicted into the denitrification performance prediction model to predict the denitrification performance and obtain the denitrification performance prediction result. The denitrification performance prediction model is pre-constructed based on the regression algorithm and the experimental sample space.

[0226] Denitrification performance prediction results include: NO3 - -N removal rate and ΔOD 600 ;

[0227] The screening unit 405 is used to perform iterative calculations using a genetic algorithm, denitrification performance test results, and denitrification performance prediction results to obtain a liquid composite carbon source formula corresponding to a sample space of candidate liquid composite carbon source formulas having the best denitrification performance and the highest carbon source score under different selection spaces of inoculated sludge.

[0228] In one embodiment, the screening unit 405 is specifically used to: query and obtain the carbon source price of each single liquid carbon source, use the Z-score standardization formula, and the denitrification performance test results, the denitrification performance prediction results and the carbon source price of each single liquid carbon source to calculate, to obtain the standardized denitrification performance test results, the denitrification performance prediction results and the carbon source price, use a genetic algorithm with pre-set parameter conditions, and the standardized denitrification performance test results, the denitrification performance prediction results and the carbon source price to perform iterative calculations, to obtain the liquid composite carbon source formula corresponding to the candidate liquid composite carbon source formula sample space with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges.

[0229] Among them, the Z-score standardization formula is: Among them, x is the denitrification performance, μ is the average value of the denitrification performance, and σ represents the standard deviation. The pre-set parameter conditions in the genetic algorithm include: the initial number of species is 100, the number of iterations ranges from 40 to 120, and the mutation probability is 0.08. The carbon source score formula is: Carbon source score = K1*NO3 --N removal rate + K2*ΔOD 600 +K3*carbon source price, where K1, K2 and K3 are parameters, and the sum of the absolute values ​​of K1, K2 and K3 is 1.

[0230] In one embodiment, the screening device for a denitrification high-efficiency liquid composite carbon source formula further includes a model construction unit, which is used to divide the experimental sample space into a training set and a test set according to a preset ratio, encode the training set and the test set using a topological structure descriptor and a functional group descriptor, construct a denitrification performance prediction model to be trained, input the encoded training set into the denitrification performance prediction model, and train it using a regression algorithm to obtain a trained denitrification performance prediction model. The regression algorithm includes: a decision tree algorithm, a linear regression algorithm, a support vector regression algorithm, a K-nearest neighbor algorithm, a random forest algorithm, an adaptive boosting algorithm, a gradient boosting algorithm, a bagging regression algorithm, an extreme tree algorithm and a multi-layer perceptron algorithm. The encoded test set is input into the trained denitrification performance prediction model, and based on the output denitrification performance prediction result, it is determined whether the accuracy of the trained denitrification performance prediction model meets the requirements. If so, it is determined that a denitrification performance prediction model with the accuracy meeting the requirements is obtained. If not, the training set is continued to be input into the denitrification performance prediction model for training until it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements.

[0231] Based on the above-mentioned invention embodiment disclosed a screening device for a denitrification efficient liquid composite carbon source formula, by constructing a plurality of candidate liquid composite carbon source formula sample spaces containing liquid composite carbon source formulas, and dividing them into an experimental sample space and a sample space to be predicted, the experimental sample space is subjected to a denitrification performance test, the denitrification performance of the sample space to be predicted is predicted using a denitrification performance prediction model, and finally a genetic algorithm is used to perform iterative calculations based on the denitrification performance test results and the prediction results to screen out the liquid composite carbon source formula corresponding to the candidate liquid composite carbon source formula sample space with the best denitrification performance and the highest carbon source score. In this scheme, the denitrification performance of the experimental sample space is obtained through actual testing, the denitrification performance of the sample space to be predicted is predicted using a denitrification performance prediction model, and finally a genetic algorithm is used to perform high-throughput screening based on the denitrification performance test results and the prediction results to obtain the liquid composite carbon source formula with the best performance, thereby achieving the purpose of quickly and high-throughput screening of the optimal and highly universal liquid composite carbon source formula from a variety of liquid composite carbon source formulas.

[0232] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple. For relevant parts, refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0233] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0234] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for screening a denitrification high-efficiency liquid composite carbon source formula, characterized in that: include: Based on the combination of the selection space of the liquid composite carbon source and the selection space of the inoculated sludge, a plurality of sample spaces of liquid composite carbon source formulas to be selected are constructed, wherein the selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources according to different combinations and proportions, the selection space of the inoculated sludge is greater than 2, and each liquid composite carbon source formula corresponds to a sample space of liquid composite carbon source formula to be selected; Randomly extracting a preset proportion of the candidate liquid composite carbon source formula sample space from a plurality of candidate liquid composite carbon source formula sample spaces as the experimental sample space; The denitrification performance test is performed on the experimental sample space to obtain the denitrification performance test results corresponding to each experimental sample space. The denitrification performance test results include: NO3 - -N removal rate and ΔOD 600 ; The remaining unselected sample spaces of the liquid composite carbon source formula to be selected are used as sample spaces to be predicted, and are input into the denitrification performance prediction model to predict the denitrification performance, and obtain the denitrification performance prediction results. The denitrification performance prediction model is pre-built based on the regression algorithm and the experimental sample space. The denitrification performance prediction results include: NO3 - -N removal rate and ΔOD 600 ; Iterative calculation is performed using a genetic algorithm and the denitrification performance test results and the denitrification performance prediction results to obtain a liquid composite carbon source formula corresponding to a sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different inoculated sludge selection spaces.

2. The method according to claim 1, characterized in that The single liquid carbon source comprises one of sodium formate, sodium acetate, sodium propionate, sodium citrate, methanol, ethanol, glycerol, n-butanol, sucrose, glucose and lactose.

3. The method according to claim 1 or 2, characterized in that The selection space of the liquid composite carbon source includes Possibly, the sample space of liquid composite carbon source formula to be selected includes possibilities; in, It represents the number of selection spaces of liquid composite carbon sources formed by combining two single liquid carbon sources from 11 single liquid carbon sources C according to five different combination ratios. Indicates the number of selection spaces of liquid composite carbon sources composed of three single liquid carbon sources among 11 single liquid carbon sources C according to 10 different combination ratios, It represents the number of sample spaces of liquid composite carbon source formulas to be selected by combining the selection space of each liquid composite carbon source with the selection space of the three inoculated sludges.

4. The method according to claim 1, wherein The denitrification performance test is performed on the experimental sample space to obtain the denitrification performance test results corresponding to each experimental sample space, including: For each experimental sample space, prepare the initial solution with consistent initial conditions; After the initial solution is deoxygenated for a preset time under a nitrogen atmosphere, a preset volume of the initial solution is placed in a 96-well plate and sealed with paraffin oil; The 96-well plate was cultured in a constant temperature box based on a preset temperature and a preset culture time. During the culture process, the initial solution was monitored using a microplate reader to obtain the initial OD value. 600 and the final OD 600 Calculate ΔOD 600 ; Detect and based on the current initial solution NO3 - -N concentration and initial solution NO3 before incubation - -N concentration, calculated as NO3 - -N removal rate.

5. The method according to claim 4, characterized in that The initial conditions of the initial solution include: chemical oxygen demand concentration of 300 mg / L, NO3 - -N concentration was 50 mg / L and sludge OD 600 =0.04; The preset time for deoxygenating the initial solution under nitrogen atmosphere includes: any time from 15 minutes to 30 minutes; After the initial solution was deoxygenated for a predetermined time under a nitrogen atmosphere, the volumes of the initial solutions selected and placed in the 96-well plate included: 200 μL; After the initial solution was placed in the 96-well plate, the volume of paraffin oil used to seal the 96-well plate included: 80 μL; The preset temperatures for culturing the 96-well plate in the incubator include: 30°C; The preset culture time for the 96-well plate to be cultured in a constant temperature box includes: 48 hours.

6. The method according to claim 1, characterized in that The construction process of the denitrification performance prediction model includes: Divide the experimental sample space into training set and test set according to the preset ratio; Encoding the training set and the test set using a topological structure descriptor and a functional group descriptor; Constructing a denitrification performance prediction model to be trained; Inputting the encoded training set into the denitrification performance prediction model, and training using a regression algorithm to obtain a trained denitrification performance prediction model; the regression algorithm includes: decision tree algorithm, linear regression algorithm, support vector regression algorithm, K nearest neighbor algorithm, random forest algorithm, adaptive boosting algorithm, gradient boosting algorithm, bagging regression algorithm, extreme tree algorithm and multilayer perceptron algorithm; Input the encoded test set into the trained denitrification performance prediction model, and judge whether the accuracy of the trained denitrification performance prediction model meets the requirements based on the output denitrification performance prediction results; If so, determine to obtain a denitrification performance prediction model with accuracy that meets the requirements; If not, continue inputting the training set into the denitrification performance prediction model for training until it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements.

7. The method according to claim 6, characterized in that Inputting the test set into the trained denitrification performance prediction model, and judging whether the accuracy of the trained denitrification performance prediction model meets the requirements based on the output denitrification performance prediction result, includes: Inputting the test set into the trained denitrification performance prediction model to perform denitrification performance prediction to obtain a denitrification performance prediction result; The denitrification performance prediction result and the denitrification performance test result are input into the determination coefficient formula to calculate the determination coefficient value; the determination coefficient formula is: Among them, R 2 is the determination coefficient, N is the total number of liquid composite carbon source formulas in the test set, and y pi is the denitrification performance prediction result, y ai is the denitrification performance test result, is the average value of the test results of denitrification performance; the determination coefficient values ​​include: NO3 - -N removal rate determination coefficient and ΔOD 600 The coefficient of determination value of When the NO3 - -N removal efficiency determination coefficient value is greater than or equal to 0.8, and the ΔOD 600 When the determination coefficient value is greater than or equal to 0.85, it is determined that the accuracy of the trained denitrification performance prediction model meets the requirements; When the NO3 - -N removal rate determination coefficient value is less than 0.8, or ΔOD 600 When the determination coefficient value is less than 0.85, it is determined that the accuracy of the trained denitrification performance prediction model does not meet the requirements.

8. The method according to claim 1, characterized in that The genetic algorithm, the denitrification performance test results and the denitrification performance prediction results are used to perform iterative calculations to obtain the liquid composite carbon source formula corresponding to the sample space of the candidate liquid composite carbon source formula with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges, including: Query and obtain the carbon source price of each single liquid carbon source; The Z-score normalization formula, the denitrification performance test results, the denitrification performance prediction results, and the carbon source price of each single liquid carbon source are used to calculate and obtain the standardized denitrification performance test results, the denitrification performance prediction results, and the carbon source price; the Z-score normalization formula is: Where x is the denitrification performance, μ is the mean value of the denitrification performance, and σ represents the standard deviation; Using a genetic algorithm with pre-set parameter conditions, as well as standardized denitrification performance test results, denitrification performance prediction results, and carbon source prices, an iterative calculation is performed to obtain the liquid composite carbon source formula corresponding to the sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different inoculated sludge selection spaces; The pre-set parameter conditions in the genetic algorithm include: the initial number of species is 100, the number of iterations ranges from 40 to 120, and the mutation probability is 0.

08. The carbon source score formula is: Carbon source score = K1*NO3 - -N removal rate + K2*ΔOD 600 +K3*carbon source price, where K1, K2 and K3 are parameters, and the sum of the absolute values ​​of K1, K2 and K3 is 1.

9. The method according to claim 1 or 2, characterized in that The liquid composite carbon source formula corresponding to the sample space of the candidate liquid composite carbon source formula with the best denitrification performance and the highest carbon source score under the selection space of different inoculated sludges includes: glycerol: sodium acetate: ethanol = 4:1:

1.

10. A screening device for a denitrification high-efficiency liquid composite carbon source formula, characterized in that: include: A construction unit is used to construct a plurality of sample spaces of liquid composite carbon source formulas to be selected, comprising liquid composite carbon source formulas, based on a combination of a selection space of a liquid composite carbon source and a selection space of an inoculum sludge, wherein the selection space of the liquid composite carbon source is determined by a plurality of single liquid carbon sources in different combinations and proportions, the selection space of the inoculum sludge is greater than two, and each liquid composite carbon source formula corresponds to a sample space of liquid composite carbon source formula to be selected; An extraction unit is used to randomly extract a preset proportion of the candidate liquid composite carbon source formula sample space from a plurality of candidate liquid composite carbon source formula sample spaces as an experimental sample space; The testing unit is used to perform a denitrification performance test on the experimental sample space to obtain a denitrification performance test result corresponding to each experimental sample space. The denitrification performance test result includes: NO3 - -N removal rate and ΔOD 600 ; The prediction unit is used to input the remaining unextracted sample space of the selected liquid composite carbon source formula as the sample space to be predicted into the denitrification performance prediction model to predict the denitrification performance and obtain the denitrification performance prediction result. The denitrification performance prediction model is pre-built based on the regression algorithm and the experimental sample space. The denitrification performance prediction result includes: NO3 - -N removal rate and ΔOD 600 ; The screening unit is used to perform iterative calculations using a genetic algorithm and the denitrification performance test results and the denitrification performance prediction results to obtain a liquid composite carbon source formula corresponding to a sample space of candidate liquid composite carbon source formulas with the best denitrification performance and the highest carbon source score under different selection spaces of inoculated sludge.

Images