Characteristic prediction method and system for reducing multiple antibiotic resistance genes of biochar

Abstract

The invention provides a biochar feature prediction method and system for reducing various antibiotic resistance genes, and the method comprises the steps: obtaining data required by a prediction model, preprocessing the required data to obtain processed data, and dividing the processed data into a training set and a test set; wherein the prediction model comprises a biochar feature prediction model and a biochar pyrolysis condition prediction model; respectively training and verifying the prediction model by utilizing the training set and the test set to obtain a final biochar feature prediction model and a final biochar pyrolysis condition prediction model; obtaining a biochar characteristic range for reducing the various antibiotic resistance genes in the sewage through the final biochar characteristic prediction model based on actual demands; the method further comprises the step of inputting the biochar characteristic range into the final biochar pyrolysis condition prediction model to obtain an optimal pyrolysis condition range corresponding to the biochar related characteristics.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biochar research, and particularly relates to a biochar feature prediction method and system for reducing multiple antibiotic resistance genes. BACKGROUND

[0002] Antibiotic resistance genes (ARGs) are a specific DNA sequence in the genome of microorganisms. Such genes can encode specific proteins, enabling microorganisms to resist antibiotics and survive and reproduce even in the presence of antibiotics. Human and animal excrement discharge, industrial agricultural production pollution, and other reasons cause antibiotic resistance genes to exist widely in wastewater. In particular, during the process of livestock and poultry breeding, the overuse of antibiotics leads to a large amount of antibiotic residues, further accelerating the spread of resistance genes. Resistance genes are common pollutants in wastewater. Whether it is used for biogas irrigation or discharged wastewater, the resistance pollutants contained therein can spread and cause bacterial drug resistance problems, posing a potential threat to the ecological environment and human health.

[0003] Biochar is a carbon-rich product produced by heating biomass (such as wood, manure, or crop residues) in a closed container (under anaerobic conditions). As an environmentally friendly carbon material, biochar with different characteristics can improve agriculture and the environment in various ways. It has been used to reduce antibiotic resistance genes in wastewater. However, there are many types of antibiotic resistance genes, and most biochar research is limited to eliminating certain resistance genes, such as tetracycline resistance genes and quinolone resistance genes. Since the characteristics (types, specific surface area, carbon, hydrogen, oxygen, and nitrogen) of biochar for reducing antibiotic resistance genes in wastewater are uncertain, the preparation or modification of biochar often lacks directionality. Therefore, to obtain the best biochar for reducing resistance genes, multiple modifications or production comparisons of biochar are required. Traditional biochar preparation and modification methods rely on experimental tests, which are time-consuming and inefficient.

[0004] Compared with the traditional trial and error experiment to compare the reduction effect of different characteristic biochar on antibiotic resistance genes, the method of machine learning assisted modeling is more efficient; at the same time, through machine learning model to predict the pyrolysis condition of biomass characteristics, can better save the cost of manpower. Machine learning technology can simplify the experimental process and improve the efficiency through data-driven method, using this advantage, researchers can screen out materials that can reduce a certain pollutant according to known experimental data, such as the rapid material screening database established by researchers of Nanjing University, which is used to screen the best adsorbent for removing complex organic matter in wastewater, and provides a strong guarantee for accurate screening of adsorption materials; in addition, some scholars use machine learning to study the adsorption of biochar on phosphorus, and get the importance ranking of biochar characteristics.

[0005] Although the above technology provides research for the rapid screening of materials for adsorbing pollutants in wastewater and the adsorption capacity of biochar on personal care products (PPCPs), on the one hand, the above adsorption technology is limited to the screening of one kind of organic pollutant corresponding to one kind of adsorption material, and cannot determine the removal effect of multiple pollutants such as antibiotic resistance genes, and there is a blank in the synergistic reduction of multiple pollutants by the same material; on the other hand, the above biochar technology focuses on the adsorption research of one kind of pollutant (such as phosphorus), without considering the elimination research of multiple pollutants such as antibiotic resistance genes, in addition, the research is only limited to the removal effect of biochar on phosphorus, and the result output of the model stops at the importance ranking of biochar characteristics, without forming a complete technical chain from biochar characteristic model construction to biochar pyrolysis process optimization. In recent years, the types of biochar are increasing, and the data of different characteristics of biochar on the reduction effect of pollutants in wastewater are increasing, therefore, it is urgent to develop a technology that uses machine learning to predict the target biochar characteristics for reducing antibiotic resistance genes in wastewater, and uses machine learning to predict the corresponding pyrolysis conditions of the target characteristic biochar, optimizes the preparation process of biochar, and improves the removal efficiency of antibiotic resistance genes in wastewater. SUMMARY

[0006] The purpose of the present application is to provide a biochar characteristic prediction method and system for reducing multiple antibiotic resistance genes, aiming to solve the above problems in the prior art.

[0007] The biochar characteristic prediction method for reducing multiple antibiotic resistance genes provided by the embodiments of the present application comprises: Obtain the data required by the prediction model, preprocess the required data to obtain the processed data, and divide the processed data into a training set and a test set; wherein the prediction model comprises a biochar characteristic prediction model and a biochar pyrolysis condition prediction model; training and verifying the prediction model respectively by using the training set and the test set, to obtain a final biochar feature prediction model and a final biochar pyrolysis condition prediction model; obtaining a biochar feature range for reducing multiple antibiotic resistance genes in sewage based on actual requirements through the final biochar feature prediction model.

[0008] The embodiment of the present application provides a biochar feature prediction system for reducing multiple antibiotic resistance genes, comprising: A data module is configured to acquire data required by a prediction model, pre-process the required data to obtain processed data, and divide the processed data into a training set and a test set; wherein the prediction model comprises a biochar feature prediction model and a biochar pyrolysis condition prediction model. A construction module is configured to train and verify the prediction model respectively by using the training set and the test set, to obtain a final biochar feature prediction model and a final biochar pyrolysis condition prediction model. A biochar feature prediction module is configured to obtain a biochar feature range for reducing multiple antibiotic resistance genes in sewage based on actual requirements through the final biochar feature prediction model.

[0009] The embodiment of the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is executed by the processor to implement the steps of the biochar feature prediction method for reducing multiple antibiotic resistance genes.

[0010] The embodiment of the present application also provides a computer readable storage medium, wherein the computer readable storage medium stores an information transmission implementation program, and the program is executed by a processor to implement the steps of the biochar feature prediction method for reducing multiple antibiotic resistance genes.

[0011] The embodiment of the present application can have the following beneficial effects: the embodiment of the present application can predict biochar related features with significant effect on reducing antibiotic resistance genes in sewage through machine learning, and can predict corresponding pyrolysis conditions of biochar related features with significant effect on reducing antibiotic resistance genes through machine learning, which is not limited to biochar for reducing multiple types of antibiotic resistance genes, but can also obtain biochar for removing target types of antibiotic resistance genes flexibly according to different sewage conditions and different types of antibiotic resistance genes, and can realize dynamic reduction of antibiotic resistance genes in sewage by biochar. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to make one or more embodiments of the present specification or the prior art clearer, the drawings needed in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the specification, and other drawings can be obtained by those skilled in the art without creative labor.

[0013] Figure 1 is a biochar feature prediction method for reducing multiple antibiotic resistance genes of an embodiment of the present application; Figure 2 is a prediction method flowchart for designing biochar to dynamically reduce multiple antibiotic resistance genes in sewage based on machine learning assistance of an embodiment of the present application; Figure 3 is a data selection example diagram for establishing a random forest model of an embodiment of the present application; Figure 4 is a data selection example diagram for establishing a deep neural network model of an embodiment of the present application; Figure 5 is a biochar feature prediction system diagram for reducing multiple antibiotic resistance genes of an embodiment of the present application. DETAILED DESCRIPTION

[0014] In order to make one or more embodiments of the present specification or the prior art clearer, the drawings needed in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the specification, and other drawings can be obtained by those skilled in the art without creative labor.

[0015] Method embodiment According to an embodiment of the present application, a biochar feature prediction method for reducing multiple antibiotic resistance genes is provided, Figure 1 is a biochar feature prediction method flowchart of an embodiment of the present application, as Figure 1 shown, the biochar feature prediction method for reducing multiple antibiotic resistance genes according to an embodiment of the present application specifically includes: Step S101, obtaining data required by a prediction model, pre-processing the required data to obtain processed data, and dividing the processed data into a training set and a test set; wherein the prediction model includes a biochar feature prediction model and a biochar pyrolysis condition prediction model, specifically including: The required data includes sewage-related indicators, biochar-related characteristics, antibiotic resistance gene-related indicators, and biochar pyrolysis conditions. The sewage-related indicators include sewage matrix, chemical oxygen demand, pH, nitrate nitrogen, and ammonia nitrogen. The biochar-related characteristics include species, specific surface area, and carbon, hydrogen, oxygen, and nitrogen content. The antibiotic resistance gene-related indicators include antibiotic resistance gene species, antibiotic resistance gene sub-species, and antibiotic resistance gene removal rate. The biochar pyrolysis conditions include pyrolysis temperature, pyrolysis time, and heating rate. The missing values of the antibiotic resistance gene species are supplemented by the antibiotic resistance gene database, and the missing values of the remaining data are supplemented by the R language using the non-linear relationship between variables. The processed sewage-related indicators, biochar-related characteristics, antibiotic resistance gene species, and antibiotic resistance gene sub-species are used as independent variables, and the processed antibiotic resistance gene removal rate is used as the dependent variable to construct a biochar characteristic prediction model dataset, and the biochar characteristic prediction model dataset is divided into a first training set and a first test set. The processed biochar-related characteristics are used as independent variables, and the processed biochar pyrolysis conditions are used as dependent variables to construct a biochar pyrolysis condition prediction model dataset, and the biochar pyrolysis condition prediction model dataset is divided into a second training set and a second test set.

[0016] Step S102, the training set and the test set are used to train and verify the prediction model respectively, and the final biochar characteristic prediction model and the final biochar pyrolysis condition prediction model are obtained, specifically including: The first training set is input into the random forest model for training, and the trained biochar characteristic prediction model is obtained, and the first test set is used to cross-validate the trained biochar characteristic prediction model, and the final biochar characteristic prediction model is obtained. The second training set is dynamically preprocessed, and the normalized data is input into the deep neural network model for training, and the trained biochar pyrolysis condition prediction model is obtained, and the second test set is used to verify the trained biochar pyrolysis condition prediction model, and the final biochar characteristic prediction model is obtained. The dynamic preprocessing includes multi-source heterogeneous data fusion, feature extraction, and deletion of missing rows.

[0017] Step S103, based on actual demand, the biochar characteristic range for reducing multiple antibiotic resistance genes in sewage is obtained by the final biochar characteristic prediction model, specifically including: According to actual needs, a biochar feature prediction data table is constructed with sewage-related indicators, biochar-related characteristics, large categories of resistance genes, and small categories of resistance genes, the biochar feature prediction data table is input into a final biochar feature prediction model, a corresponding resistance gene removal rate is obtained, the corresponding resistance gene removal rate is sorted from high to low according to the numerical value, and the biochar-related characteristics corresponding to the first k rows of data are selected as the biochar feature range for reducing multiple antibiotic resistance genes in sewage.

[0018] The method further comprises: The biochar feature range is input into the final biochar pyrolysis condition prediction model to obtain a best pyrolysis condition range corresponding to the biochar-related characteristics, and specifically comprises: The biochar feature range is selected from the minimum value according to a preset span ladder until the maximum value of the biochar feature range, a biochar pyrolysis condition prediction data table is generated according to the selected data, the biochar pyrolysis condition prediction data table is input into the final biochar pyrolysis condition prediction model, and a best pyrolysis condition range corresponding to the biochar-related characteristics is obtained.

[0019] The following describes the specific conditions of the biochar feature prediction method for reducing multiple antibiotic resistance genes according to the embodiments of the application, as shown in Figure 2 The above technical solutions of the embodiments of the application are described in detail.

[0020] The embodiments of the application provide a prediction method for dynamically reducing multiple antibiotic resistance genes in sewage based on machine learning assisted design of biochar, which predicts the best biochar feature through a machine learning model, optimizes the biochar preparation process, and improves the removal efficiency of antibiotic resistance genes.

[0021] To achieve the above technical purposes, the embodiments of the application achieve the following steps: 1. Model data selection To obtain the biochar feature range with significant effect on eliminating antibiotic resistance genes in sewage, a random forest model (RF model) needs to be established. The data required by the model is selected from the database already sorted by the research group (the specific index content is as shown in Figure 3 The database data comes from Chinese and English literature in China Knowledge Network and web of science, and if the data is contained in the literature pictures, the picture is converted using the image digitization tool of Origin2022, and the search keywords are respectively: “water / Sewage, biochar, resistance gene / antibiotic resistance gene” (“water / Sewage, biochar, resistance gene / antibiotic resistance gene”).

[0022] The specific indexes selected are: ① Sewage-related indicators: Sewage substrate, chemical oxygen demand (COD), pH, nitrate nitrogen ( -N), ammonia nitrogen ( -N); ② Biochar-related characteristics: type (Sample 1), specific surface area (SSA), carbon (C), hydrogen (H), oxygen (O), and nitrogen (N) content; ③Resistance gene related indicators: Resistance gene large species (Sample2), resistance gene small species (Sample3), resistance gene removal rate ( -ARGs).

[0023] Figure 3 The resistance gene removal rate indicator can be replaced by the initial / final relative abundance of the resistance gene or the initial / final absolute abundance of the resistance gene through the formula conversion: (1); Among them, the indicator data of sewage conditions and biochar-related characteristics may contain missing values, while the data related to resistance gene removal rate do not contain missing values.

[0024] In order to obtain the pyrolysis conditions corresponding to the target biochar characteristics, a deep neural network model (DNN model) needs to be established. The model data is selected from the database organized by the research team (the specific indicators selected are as follows Figure 4 The database data is sourced from Chinese and English literature on CNKI and Web of Science. Images included in the literature were converted using Origin2022's image digitization tool. The search keywords were "biochar, produce / pyrolysis / preparation." The specific indicators selected were: ① Biochar-related characteristics: type (Sample 1), specific surface area (SSA), carbon (C), hydrogen (H), oxygen (O), and nitrogen (N) content; ② Biochar pyrolysis conditions: pyrolysis temperature (T), pyrolysis time (RT), and heating rate (HR).

[0025] in, Figure 4 The relevant indicator data in the dataset can contain less than 20% missing values.

[0026] 2. Model data processing Missing data for major types of resistance genes were supplemented through the resistance gene database, and missing data values ​​for other data were supplemented through R code (random forest missing value interpolation). This code uses the nonlinear relationship between variables to predict missing values, which is more scientific than mean / median interpolation.

[0027] 3. Constructing a random forest model (predicting target biochar related characteristics) The model is used to predict the removal rate of biochar characteristics on resistant genes in wastewater, and to clarify the relationship between wastewater, biochar and resistant gene removal rate. The output y value of this model is the removal rate of resistant genes, and the rest are independent variables x. 523 sets of data are divided into training set and test set according to 8:2, and the max_depth parameter is optimized by 5-fold cross validation. Random forest regression + cross validation is adopted, so that the prediction accuracy of the model reaches R²=0.88 (nonlinear ensemble learning), which is significantly better than traditional multiple linear regression, and the prediction error (MAE) of tetA gene is reduced by 37% compared with traditional method.

[0028] Unlike the application scene of traditional methods applied to a single wastewater type, the embodiment of the application innovatively integrates 23 kinds of wastewater substrates for cross-scene prediction, solves the problem of ignoring sample source differences in traditional environmental research, and adds biochar species preparation process and resistant gene types and subtypes in the scheme. The multi-heterogeneous data collection constructs a cross-scene prediction model that has not been seen in the prior art. Creatively, physical and chemical parameters (COD, pH, SSA, etc.) and molecular identifiers (such as Sample3 column tetX, tnpA_1, etc. ARGs subtypes) are jointly modeled, which is the first time to realize prediction from macro conditions to gene level, forming a model framework of the relationship between wastewater-related indicators, biochar characteristics, and resistant gene-related indicators.

[0029] Fill in the data table of the random forest model, and fill in the wastewater-related indicators (wastewater substrate (Substrate), chemical oxygen demand (COD), pH, nitrate nitrogen (NO3-N), ammonia nitrogen (NH4-N)), biochar-related characteristics (species (Sample1)), and resistant gene-related indicators (large species, small species) according to the test needs or actual environmental conditions. Biochar-related characteristics (specific surface area, carbon, hydrogen, oxygen, nitrogen content), and resistant gene removal rate are filled in with 5, 0.5, 0.1, 0.2, and 0.1 as the span ladder (the span ladder can be adjusted as needed), and the table is substituted into the random forest model to obtain the corresponding resistant gene removal rate. According to the numerical size of the resistant gene removal rate, select the top 20% data rows, find the corresponding biochar-related characteristic (specific surface area, carbon, hydrogen, oxygen, nitrogen content) data value, and inversely select the biochar-related characteristic data to finally obtain the biochar-related characteristic (specific surface area, carbon, hydrogen, oxygen, nitrogen content) data value range. The characteristic range is the target biochar characteristic range for reducing resistant genes in wastewater.

[0030] According to the numerical size of the resistant gene removal rate, select the top 20% data rows, find the corresponding biochar-related characteristic (specific surface area, carbon, hydrogen, oxygen, nitrogen content) data value, and inversely select the biochar-related characteristic data to finally obtain the biochar-related characteristic (specific surface area, carbon, hydrogen, oxygen, nitrogen content) data value range. The characteristic range is the target biochar characteristic range for reducing resistant genes in wastewater. ​

[0031] 4. Constructing a deep neural network model (predicting pyrolysis conditions corresponding to target biochar-related features) The model is used to predict the pyrolysis conditions corresponding to biochar features, and to clarify the relationship between biochar-related features and pyrolysis conditions for producing biochar. The output y value of this model is the pyrolysis condition of biochar (pyrolysis temperature (T), pyrolysis time (RT), heating rate (HR)), and the biochar-related features (specific surface area, carbon, hydrogen, oxygen, nitrogen content) are independent variables x.

[0032] 501 data are used to establish the model: ① Dynamic data preprocessing (multi-source heterogeneous data fusion, accurate extraction of 9-dimensional features: use `usecols='C:J'` to read 9 columns of data from Excel, the first 6 columns are input features, and the last 3 columns are output targets. When predicting new data, read columns A to F to form a mapping with the training input features. Robust processing: delete missing rows by `dropna()` to ensure data purity); ② Dual-channel normalization engine (to avoid the distortion of input features (such as COD concentration) and output targets (such as η-ARGs removal rate) caused by traditional methods, this scheme can maintain the numerical distribution characteristics of input data and output targets, allowing the model to handle COD mg / L type pollutant data and pH dimensionless parameters simultaneously. Independent MinMaxScaler instances are used for input and output to avoid distortion of the output target by the dimension of the input feature); ③ Lightweight DNN architecture (compared to traditional MLP, the training speed is 3 times faster); ④ Intelligent training system (500 rounds of training, the loss reduction curve is smoother than the learning rate of traditional methods, and the automatic decay strategy avoids falling into local optimum); ⑤ Predicting biochar pyrolysis conditions (using torch.no_grad() for context management).

[0033] Fill in the data table for the input deep neural network model, and replace the target biochar feature (specific surface area, carbon, hydrogen, oxygen, nitrogen content) range data in step 3 with the deep neural network model data table. The feature range is filled in from the minimum value with a span of 5, 0.5, 0.1, 0.2, 0.1 as a step ladder to the maximum value (the span ladder can be adjusted as needed). Through the deep neural network prediction model, the pyrolysis conditions corresponding to the biochar-related features can be obtained, including pyrolysis temperature, heating rate, and residence time. This condition range value is the best pyrolysis condition range corresponding to the target biochar-related features for reducing resistant genes in wastewater.

[0034] In summary, the prediction method for assisting in designing biochar for dynamically reducing multiple antibiotic resistance genes in sewage based on machine learning provided by the embodiment of the present application can obtain target biochar for reducing different antibiotic resistance genes under different sewage conditions and preparation pyrolysis conditions thereof, improve the removal efficiency of antibiotics and resistance genes in farm sewage, and achieve the protection of ecological environment and human health.

[0035] System embodiment According to the embodiment of the present application, a biochar feature prediction system for reducing multiple antibiotic resistance genes is provided, Figure 5 is a biochar feature prediction system for reducing multiple antibiotic resistance genes according to the embodiment of the present application, as Figure 5 shown, the biochar feature prediction system for reducing multiple antibiotic resistance genes according to the embodiment of the present application specifically comprises: The data module 50 is used to obtain the data required by the prediction model, pre-process the required data to obtain processed data, and divide the processed data into a training set and a test set; wherein the prediction model comprises a biochar feature prediction model and a biochar pyrolysis condition prediction model; The construction module 52 is used to train and verify the prediction model using the training set and the test set, respectively, to obtain a final biochar feature prediction model and a final biochar pyrolysis condition prediction model; The biochar feature prediction module 54 is used to obtain the biochar feature range for reducing multiple antibiotic resistance genes in sewage based on actual demand through the final biochar feature prediction model; The system further comprises: The biochar pyrolysis condition prediction module is used to input the biochar feature range into the final biochar pyrolysis condition prediction model to obtain the optimal pyrolysis condition range corresponding to the biochar related features.

[0036] The embodiment of the present application is a system embodiment corresponding to the above-mentioned method embodiment, and the specific operation of each module can be understood with reference to the description of the method embodiment, which will not be repeated here.

[0037] In summary, the embodiment of the present application specifically includes the following beneficial effects: 1、The embodiment of the present application is no longer limited to the screening of one kind of organic pollutant corresponding to one kind of adsorption removal material, but precisely targets the reduction effect of pollutants under different feature conditions of the same material (biochar); 2、The related biochar technology in the embodiment of the present application is not a simple adsorption removal research for one kind of pollutant (for example: phosphorus), but a complex research on the reduction effect of biochar on multiple types of pollutants such as antibiotic resistance genes; 3. Unlike the application scenarios of traditional methods applied to a single wastewater type, the embodiment of the application innovatively integrates 23 kinds of wastewater substrate matrices for cross-scenario prediction, solves the problem of ignoring sample source differences in traditional environmental research, and adds biochar species preparation process and resistance gene types and subtypes in the scheme. The multi-element heterogeneous data collection constructs a cross-scenario prediction model that has not been seen in previous schemes. 4. A complete technical chain from model data collection, biochar model architecture to biochar pyrolysis process optimization is constructed: the feature range value (specific surface area, element composition, etc.) and contribution rate sorting output by the random forest model can directly guide the design of biochar materials; the pyrolysis condition prediction model converts the theoretical characteristics into operable process parameters (temperature / heating rate / residence time), shortening the time of traditional experimental method; 5. In addition, the embodiment of the application is not limited to the synergistic reduction of multiple types of resistance genes by biochar, but can also flexibly obtain biochar for removing target types of resistance genes according to different wastewater conditions and different types of resistance genes existing in the field, and realizes the dynamic reduction of antibiotic resistance genes in wastewater by biochar.

[0038] Device embodiment one The embodiment of the application provides an electronic device, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the computer program is executed by the processor to realize the steps as described in the method embodiment.

[0039] Device embodiment two The embodiment of the application provides a computer readable storage medium, wherein the computer readable storage medium stores an implementation program of information transmission, and the program is executed by a processor to realize the steps as described in the method embodiment.

[0040] The computer readable storage medium described in the embodiment includes but is not limited to ROM, RAM, magnetic disk or optical disk, etc.

[0041] Finally, it should be pointed out that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solution deviate from the scope of the technical solutions of the embodiments of the application.

Claims

1. A method for predicting biochar characteristics for eliminating multiple antibiotic resistance genes, characterized in that include: Obtaining data required for a prediction model, preprocessing the required data to obtain processed data, and dividing the processed data into a training set and a test set; wherein the prediction model includes a biochar characteristic prediction model and a biochar pyrolysis condition prediction model; The prediction model is trained and verified using the training set and the test set, respectively, to obtain a final biochar characteristic prediction model and a final biochar pyrolysis condition prediction model; Based on actual needs, the final biochar characteristic prediction model is used to obtain the biochar characteristic range for eliminating multiple antibiotic resistance genes in sewage.

2. The method according to claim 1, characterized in that The method further comprises: The biochar characteristic range is input into the final biochar pyrolysis condition prediction model to obtain the optimal pyrolysis condition range corresponding to the biochar related characteristics.

3. The method according to claim 1, characterized in that Preprocessing the required data to obtain processed data, and dividing the processed data into a training set and a test set specifically includes: The required data include sewage-related indicators, biochar-related characteristics, resistance gene-related indicators, and biochar pyrolysis conditions; The sewage-related indicators include sewage matrix, chemical oxygen demand, pH, nitrate nitrogen, and ammonia nitrogen; The biochar-related characteristics include type, specific surface area, and carbon, hydrogen, oxygen, and nitrogen content; The resistance gene related indicators include major resistance gene types, minor resistance gene types and resistance gene removal rate; The biochar pyrolysis conditions include pyrolysis temperature, pyrolysis time, and heating rate; The missing values ​​of the major types of resistance genes were supplemented by using the resistance gene database, and the missing values ​​of the remaining data were supplemented by using the nonlinear relationship between variables using R language; A biochar characteristic prediction model dataset is constructed using treated sewage-related indicators, biochar-related characteristics, large resistance gene species, and small resistance gene species as independent variables, and the resistance gene removal rate after treatment as a dependent variable, and the biochar characteristic prediction model dataset is divided into a first training set and a first test set; A biochar pyrolysis condition prediction model dataset is constructed by taking the processed biochar-related characteristics as independent variables and the processed biochar pyrolysis conditions as dependent variables, and the biochar pyrolysis condition prediction model dataset is divided into a second training set and a second test set.

4. The method according to claim 3, characterized in that The prediction model is trained and verified using the training set and the test set, respectively, to obtain a final biochar characteristic prediction model and a final biochar pyrolysis condition prediction model, specifically including: Inputting the first training set into a random forest model for training to obtain a trained biochar characteristic prediction model, and cross-validating the trained biochar characteristic prediction model using the first test set to obtain a final biochar characteristic prediction model; Dynamically preprocessing the second training set and normalizing the dynamically preprocessed data using a dual-channel normalization engine, inputting the normalized data into a deep neural network model for training to obtain a trained biochar pyrolysis condition prediction model, and validating the trained biochar pyrolysis condition prediction model using the second test set to obtain a final biochar characteristic prediction model; The dynamic preprocessing includes multi-source heterogeneous data fusion, feature extraction and deletion of missing rows.

5. The method according to claim 1, wherein Based on actual needs, the final biochar characteristic prediction model is used to obtain the biochar characteristic range for reducing multiple antibiotic resistance genes in sewage, which specifically includes: According to actual needs, a biochar characteristic prediction data table is constructed using sewage-related indicators, biochar-related characteristics, major types of resistance genes, and minor types of resistance genes. The biochar characteristic prediction data table is input into the final biochar characteristic prediction model to obtain the corresponding resistance gene removal rate. The corresponding resistance gene removal rates are sorted from high to low according to the numerical value, and the biochar-related characteristics corresponding to the first k rows of data are selected as the biochar characteristic range for eliminating multiple antibiotic resistance genes in sewage.

6. The method according to claim 2, characterized in that Inputting the biochar characteristic range into the final biochar pyrolysis condition prediction model to obtain the optimal pyrolysis condition range corresponding to the biochar related characteristics specifically includes: The biochar characteristic range is selected from the minimum value according to the preset span ladder to the maximum value of the biochar characteristic range, and a biochar pyrolysis condition prediction data table is generated according to the selected data. The biochar pyrolysis condition prediction data table is input into the final biochar pyrolysis condition prediction model to obtain the optimal pyrolysis condition range corresponding to the biochar related characteristics.

7. A biochar feature prediction system for eliminating multiple antibiotic resistance genes, characterized by include: A data module is used to obtain data required by the prediction model, preprocess the required data to obtain processed data, and divide the processed data into a training set and a test set; wherein the prediction model includes a biochar characteristic prediction model and a biochar pyrolysis condition prediction model; A construction module is used to train and verify the prediction model using the training set and the test set, respectively, to obtain a final biochar characteristic prediction model and a final biochar pyrolysis condition prediction model; The biochar characteristic prediction module is used to obtain the biochar characteristic range for reducing multiple antibiotic resistance genes in sewage through the final biochar characteristic prediction model based on actual needs.

8. The system according to claim 7, characterized in that The system further comprises: The biochar pyrolysis condition prediction module is used to input the biochar characteristic range into the final biochar pyrolysis condition prediction model to obtain the optimal pyrolysis condition range corresponding to the biochar related characteristics.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the method for predicting biochar characteristics for eliminating multiple antibiotic resistance genes as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an information transmission implementation program, and when the program is executed by a processor, the steps of the method for predicting biochar characteristics for eliminating multiple antibiotic resistance genes according to any one of claims 1 to 6 are implemented.

Images