Pesticide advanced oxidative degradation efficiency prediction method and device based on molecular fingerprints and machine learning, and medium

Abstract

The invention provides a pesticide advanced oxidative degradation efficiency prediction method and device based on molecular fingerprints and machine learning and a medium, and the method comprises the following steps: S1, collecting and preprocessing pesticide degradation data; s2, extracting an SMILES format of pesticide molecules, converting the SMILES format into Morgan and counting molecular fingerprints, and combining the Morgan and counting molecular fingerprints with degradation data to form a data set; s3, constructing an extreme gradient lifting model to predict degradation efficiency, and evaluating to obtain an optimal model; s4, identifying key variables through a Shapley value method and a partial dependency graph; and S5, optimizing reaction conditions by using a genetic algorithm-optimal model. According to the model, the advanced oxidative degradation rate is predicted through counting molecular fingerprints and an extreme gradient lifting algorithm, the precision of the model is evaluated by using R2 and RMSE, and feature analysis is performed through a Shapley value and a part of dependency graph. The method is low in cost, simple, convenient and rapid, saves manpower and material resources, is high in accuracy, and has both statistical significance and chemical significance.

Description

Technical Field

[0001] The present invention relates to a method for predicting the advanced oxidative degradation rate, and in particular to an intelligent method for predicting the advanced oxidative degradation efficiency of pesticides and optimizing working conditions by combining molecular fingerprints and machine learning. Background Art

[0002] Advanced oxidation technologies, primarily free radical oxidation, Fenton oxidation, and ozone oxidation, are widely used in practical projects as an important means of removing recalcitrant organic matter (such as pesticides). In practical wastewater treatment, degradation rate is often used as a parameter to assess the effectiveness of research methods, and its level is directly related to the design of degradation conditions. Therefore, accurately predicting degradation rates and optimizing operating conditions are of great significance to the operation and management of actual wastewater treatment.

[0003] Currently, experimental studies mainly use gas chromatography or spectrophotometry to evaluate the degradation of pesticides. Although these methods provide high accuracy, they also have disadvantages such as long experimental cycles and high costs. In addition, factors affecting the degradation rate include solution pH, oxidant dosage, UV light intensity, and organic properties. These factors are not completely independent and their influencing mechanisms are complex, which makes it challenging to accurately evaluate the degradation efficiency. Machine learning, as a highly intelligent method that can autonomously identify the relationship between input parameters and output values and accurately predict new data, has been widely used. Machine learning makes it possible to quickly and accurately predict degradation rates and adjust operating conditions.

[0004] Furthermore, the accuracy of machine learning is significantly dependent on the input data, making quantifying the properties of organic compounds another challenge. Traditional physicochemical parameters cannot intuitively and fully capture the characteristics of organic compounds, which can limit the accuracy and interpretability of models. Molecular fingerprints can intuitively, concisely, and comprehensively express the structural properties of organic compounds using numbers, and can be used to quantify their properties.

[0005] Therefore, in order to solve the above problems, the present invention proposes a method for predicting the advanced oxidative degradation efficiency of pesticides based on molecular fingerprints and machine learning. Summary of the Invention

[0006] In order to solve the above problems, a method for predicting the advanced oxidation degradation efficiency of pesticides based on molecular fingerprints and machine learning was proposed to predict the degradation rates of different oxidation modes, optimize the reaction conditions, and improve the automation and management levels of the advanced oxidation process.

[0007] To this end, the specific technical solutions adopted in the present invention are as follows:

[0008] A method for predicting the efficacy of pesticide advanced oxidative degradation based on molecular fingerprints and machine learning, comprising the following steps:

[0009] Step S1, collecting original detection data in the pesticide degradation reaction and grouping and preprocessing it to obtain degradation reaction data;

[0010] Step S2: extract pesticide molecular samples from the chemical database, obtain the SMILES format of each pesticide, convert it into a molecular fingerprint, combine it with the degradation reaction data to form a complete data set, and divide the data set into a training set and a test set;

[0011] Step S3: constructing random forest, gradient boosting decision tree, extreme gradient boosting, and lightweight gradient boosting machine models, performing target output based on the models and key input variables, predicting the degradation efficiency of pesticides by different advanced oxidation processes, and using evaluation indicators to evaluate and compare the model prediction effects to obtain the optimal model;

[0012] Step S4, based on the optimal model, using Shapley value method and partial dependence diagram to assist in verification and identification of key variables to verify the correctness of the model;

[0013] Step S5, inputting the pollutant type and initial concentration into the obtained optimal model to obtain the operating conditions when the removal efficiency is the highest;

[0014] Step S6: Using the Tanimoto coefficient to define the applicable range of the obtained optimal model.

[0015] In a possible design, in step S1, preprocessing includes one or more of data cleaning, data noise reduction, data misalignment or null value filling; and degradation reaction data are grouped according to advanced oxidation methods.

[0016] In one possible design, step S2 specifically includes:

[0017] Step S21: Determine the pesticide type based on the degradation reaction data in S1, extract pesticide molecular samples from the Drugbank database, obtain the SMILES format of the corresponding drug samples from the pubchempy package in the pandas library, and then call the RDkit package in the pandas library to convert the SMILES format of each pesticide sample into different molecular fingerprints;

[0018] Step S22, combining the pre-processed degradation reaction data with the corresponding two molecular fingerprints to form a complete data set;

[0019] In step S23, each data set is randomly divided into a training set and a test set according to a certain proportion.

[0020] In one possible design, step S3 specifically includes:

[0021] Step S31: input the training set into the random forest, gradient boosting decision tree, extreme gradient boosting and lightweight gradient boosting machine models, and use the grid search-cross validation algorithm to optimize the hyperparameters of the model to obtain the best model;

[0022] Step S32: Use the obtained better model to predict the test set and use the coefficient of determination R 2 The accuracy of the best model obtained by evaluating the root mean square error (RMSE) is compared with the accuracy of the best models established based on the same data set to determine the optimal model. The calculation formula is as follows:

[0023]

[0024] Where: n is the number of samples, is the predicted value of the degradation rate of the i-th sample, y i is the actual degradation rate of the i-th sample, is the average value of the actual samples.

[0025] In a possible design, in step S5, the initial concentration value and type of pollutants are first determined, and the value ranges of other parameters are set and input into the optimal model - genetic algorithm. The data are recombined through the genetic algorithm's inheritance, crossover and mutation methods, and the predicted value of the pesticide advanced oxidation degradation efficiency obtained by the optimal model is used as the fitness. The parameter combination corresponding to the maximum fitness is obtained, that is, the optimal operating condition is obtained.

[0026] In one possible design, step S6 specifically includes the following steps:

[0027] S61, performing iterative calculation on the data in the training set to calculate the Tanimoto coefficient between each compound and the compounds in other training sets;

[0028] S62, by setting different Tanimoto coefficient thresholds, deleting the data below these thresholds, using the optimal model to predict the remaining data, and calculating the root mean square error (RMSE) between the predicted value and the experimental value under each setting;

[0029] S63, determining the Tanimoto coefficient corresponding to the obtained lowest RMSE value as a threshold standard;

[0030] When verifying whether the model can be used for a new data set, the Tanimoto coefficient is calculated through steps S61 to S63. If the calculated Tanimoto coefficient is higher than the preset threshold, the tested compound is within the applicability domain and the prediction is reliable; otherwise, the prediction is unreliable, thereby defining the scope of applicability of the obtained optimal model.

[0031] In a possible design, the pesticide degradation reaction conditions in step S1 include but are not limited to initial concentration, pH, reaction time, temperature, ultraviolet light intensity, and ultraviolet light wavelength.

[0032] The present invention also provides a device for predicting the effectiveness of pesticide advanced oxidative degradation based on molecular fingerprints and machine learning, comprising a memory, a control processor, and a computer program stored in the memory and executable on the control processor. The control processor executes the program to implement the method for predicting the effectiveness of pesticide advanced oxidative degradation based on molecular fingerprints and machine learning as described above.

[0033] The present invention also provides a system comprising the aforementioned device for predicting the pesticide advanced oxidative degradation efficiency based on molecular fingerprints and machine learning.

[0034] The present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to implement the pesticide advanced oxidative degradation efficiency prediction method based on molecular fingerprint and machine learning as described above.

[0035] The principles and effects of the present invention are as follows:

[0036] 1. This paper proposes a method for predicting the efficiency of pesticide advanced oxidation degradation based on molecular fingerprints and machine learning. The model predicts the advanced oxidation degradation rate by counting molecular fingerprints and extreme gradient boosting algorithm. 2 The model accuracy was evaluated using the RMSE and Shapley values, and feature analysis was performed using partial dependence plots. This method is low-cost, simple, fast, and labor-saving, with high accuracy and both statistical and chemical significance.

[0037] 2. The present invention combines the optimal model of advanced oxidation degradation with the genetic algorithm, which can accurately and quickly predict the degradation rate based on process detection indicators and obtain the optimal operating conditions, thereby optimizing the operating conditions and improving the automation and management level of sewage treatment plant control.

[0038] 3. The method proposed in this invention for predicting the advanced oxidation degradation efficiency of pesticides based on molecular fingerprints and machine learning is not only applicable to the degradation rate prediction of pesticides, but also has the potential to be extended to the prediction of the advanced oxidation degradation rate of other pollutants, providing a broader application prospect for environmental pollution control. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1Schematic diagram of a method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning in an embodiment of the present invention;

[0041] Figure 2 1 is a ranking diagram of the importance of degradation rate correlation variables of the three oxidation methods in the embodiment of the present invention, where (a) is free radical oxidation, (b) is Fenton oxidation, and (c) is ozone oxidation;

[0042] Figure 3 Graph showing the prediction results of the degradation rate using the optimal counting molecular fingerprint-extreme gradient boosting model for the three oxidation methods in the embodiments of the present invention, where (a) is free radical oxidation, (b) is Fenton oxidation, and (c) is ozone oxidation. DETAILED DESCRIPTION

[0043] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. By referring to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0044] The following drawings and embodiments provide a detailed and complete description of the technical solutions of the present invention. Obviously, the embodiments described are only a portion of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are also within the scope of protection of the present invention.

[0045] like Figure 1 As shown, the present invention provides a method for predicting the efficacy of pesticide advanced oxidation degradation based on molecular fingerprint and machine learning, comprising the following steps:

[0046] Step S1: collecting original detection data of pesticide degradation reactions from published literature, pre-processing them, and grouping them according to the processing methods.

[0047] This example is based on published papers and collected a total of 4749 data sets, including 2084 sets of persulfate radical oxidation, 1345 sets of Fenton oxidation and 1320 sets of ozone oxidation.

[0048] The reaction condition parameters collected in this example are: initial concentration, reaction time, initial pH, reaction temperature, oxidation dosage, UV intensity, UV wavelength, Fe 2+ Fe 3+ 、Cl - and other ion concentrations.

[0049] Step S2: Extract pesticide molecule samples from the chemical database, obtain the SMILES format of each pesticide, convert it into a count molecular fingerprint, combine it with the raw degradation reaction data, and form a data set. The data set is divided into a training set and a test set in a ratio of 8:2. Specifically, it includes:

[0050] Step S21, extracting 39 pesticide molecular samples in S1 from the Drugbank database;

[0051] Step S22: Based on the Python programming language, obtain the SMILES format of the corresponding drug sample from the pubchempy package in the pandas library; then call the RDkit package in the pandas library, use "AllChem.GetMorganFingerprintAsBitVect" to convert the SMILES symbol into a Morgan molecular fingerprint, and use "AllChem.GetHashedMorganFingerprint" to convert it into a count molecular fingerprint. The length of both molecular fingerprints is selected to be 256 bits. The main difference between the Morgan molecular fingerprint and the technical molecular fingerprint is that the former can only use "0" and "1" to indicate the presence of a certain structure in the organic compound, while the count molecular fingerprint can count the structure;

[0052] Step S23, combining the three pre-processed oxidation reaction condition data sets with the corresponding two molecular fingerprints to form a total of six data sets;

[0053] Step S24: Each data set is randomly divided into a training set and a test set in a ratio of 8:2.

[0054] Step S3: construct random forest, gradient boosting decision tree, extreme gradient boosting and lightweight gradient boosting machine models, perform target output based on the models and key input variables, predict the degradation efficiency of pesticides in advanced oxidation, and use evaluation indicators to evaluate and compare the model prediction effects to obtain the optimal model, specifically including:

[0055] S31. Build random forest, gradient boosting decision tree, extreme gradient boosting and lightweight gradient boosting machine models through RandomForestRegressor, xgboost, GradientBoostingRegressorlightgbm, numpy and sklearn modules.

[0056] The training set was fed into four models, and the model hyperparameters were optimized using a grid search-cross-validation algorithm. A 9-fold cross-validation was used for model evaluation. The hyperparameter ranges were as follows: n_estimators ranged from 10 to 10,000, max_depth ranged from 3 to 20, learning_rate ranged from 0.05 to 0.5, min_samples_leaf ranged from 1 to 10, and min_samples_split ranged from 2 to 20; colsample_bytree ranged from 0.1 to 1, min_child_weight ranged from 1 to 10, reg_lambda ranged from 1 to 5, num_leaves ranged from 4 to 10, and colsample_bytree ranged from 0.4 to 1.

[0057] Step S32: Target output is performed based on the optimal model of each model and the input key variables, degradation rate is predicted, and the model prediction effect is evaluated.

[0058] Using the correlation coefficient R 2 And the root mean square error is used to evaluate the prediction accuracy of the model, and the calculation formula is as follows:

[0059]

[0060] Where: n is the number of samples, is the predicted value of degradation rate of sample i, y i is the actual degradation rate of the i-th sample, is the average value of the actual sample. 2 If the improvement is less than 0.001 and the decrease of RMSE is less than 0.001, the model can be considered to be optimal and no further search is performed.

[0061] After tuning the model hyperparameters through grid search and cross validation, the extreme gradient boosting model based on the three datasets using count molecular fingerprints performed the best. The optimal hyperparameter values for the extreme gradient boosting model for free radical oxidation are as follows: learning_rate is 0.15, n_estimators is 494, max_depth=6, min_child_weight is 1, subsample is 0.9, colsample_bytree is 1, and reg_lambda is 1. The model prediction effect is as follows Figure 2 As shown, the training set R 2 =0.991, RMSE of training set =3.101, R 2=0.914, test set RMSE = 7.748. The hyperparameter values of the Fenton oxidation optimal extreme gradient boosting model are as follows: learning_rate = 0.05, n_estimators = 157, max_depth = 11, min_child_weight = 10, subsample = 0.7, colsample_bytree = 1, and reg_lambda = 4. The model prediction effect is as follows Figure 2 As shown, the training set R 2 =0.979, RMSE of training set =4.752, R 2 =0.934, test set RMSE =8.718. The hyperparameter values of the optimal extreme gradient boosting model for ozone oxidation are as follows: learning_rate = 0.4, n_estimators = 1000, max_depth = 8, min_child_weight = 2, subsample = 1, colsample_bytree = 1, and reg_lambda = 0.6. The model prediction effect is as follows Figure 2 As shown, the training set R 2 =0.996, training set RMSE =4.697, test set R 2 =0.969, and the test set RMSE =3.842, indicating that the above model has excellent prediction effect on degradation rate.

[0062] Step S4: Based on the optimal model, Shapley value method and partial dependence diagram are used to assist in verifying and identifying key variables, and output the importance ranking of key variables, such as Figure 3 As shown in the figure, the importance of free radical oxidation characteristics from high to low is reaction time, organic structure parameters, UV light intensity, oxidant dosage, temperature, Fe 2+ Concentration, pH, initial concentration of pollutants and amount of H2O2. The importance of Fenton oxidation characteristics is from high to low: amount of H2O2, reaction time, organic matter structure parameters, pH, Fe 2+ concentration, UV light intensity, temperature, initial concentration of pollutants and Fe 3+ The importance of ozone oxidation characteristics from high to low is reaction time, structural parameters, initial concentration, pH, O3 usage, H2O2 amount, UV light intensity, temperature and Fe 2+ concentration.

[0063] The feature importance ranking and the direction of its impact on degradation rate are consistent with the results of the Pearson correlation coefficient. The verification results show that the identification of key variables is accurate and reasonable.

[0064] Building a model requires not only high accuracy but also variable interpretability. Step S3 evaluates the model's accuracy, and step S4 evaluates its interpretability. If the results do not meet expectations, return to step S1 and re-clean the data. This involves tracing back to the article where the data was collected, assessing the feasibility of the experimental data in the article, confirming that no errors occurred during the data collection process, and performing data cleaning, removing duplicate values, etc.

[0065] Step S5: Fix the initial concentration, set a certain value range for the input parameter, input it into the genetic algorithm-optimal model, predict the degradation rate under different input parameters, and obtain the working conditions with the highest removal efficiency. Specifically including:

[0066] Step S51: the initial concentration is set to 100 μmol / L, the pollutant is selected as atrazine, and the ultraviolet wavelength is 254 nm;

[0067] Step S52: Establish a genetic algorithm based on Python, use the predicted value of the optimal model as the fitness, and use the maximum fitness as the model target;

[0068] Step S53: Input the data set into the model to obtain the optimal parameter combination. For free radical oxidation, when the reaction time is set to 92 minutes, the pH value is 5.1, the PMS concentration is 6.0 mmol, the PDS concentration is 1.6 mmol, and the Fe 2+ The concentration is 0.35 mg / L and the UV intensity is 373 mJ / cm 2 The degradation rate reached the highest 95.3% when the reaction time was about 28.3 minutes, the pH value was about 2.8, the H2O2 concentration was 13.7 mmol, and the Fe 2+ The concentration is 10.1mg / L and the UV intensity reaches 462mJ / cm 2 , the degradation rate was significantly increased to 99.7%. In the O3 oxidation, the reaction time was about 109.7 minutes, the pH value was about 11.6, the H2O2 concentration was about 0.04 mmol, and Fe 2+ The concentration is about 0.4 mg / L, the ozone concentration is about 6.7 mg / L, and the ultraviolet light intensity is 281 mJ / cm 2 The expected degradation rate can reach 95.8%.

[0069] Step S6: Use the Tanimoto coefficient to define the applicable range of the model. In this example, the Tanimoto coefficient thresholds for the free radical oxidation, Fenton oxidation, and ozone oxidation models are 0.185, 0.373, and 0.422, respectively.

[0070] The present invention proposes a method for predicting the efficiency of pesticide advanced oxidation degradation based on molecular fingerprint and machine learning. The model predicts the advanced oxidation degradation rate by counting molecular fingerprints and extreme gradient boosting algorithm.2 The model accuracy was evaluated using the RMSE and Shapley values, and feature analysis was performed using partial dependence plots. This method is low-cost, simple, fast, and labor-saving, with high accuracy and both statistical and chemical significance.

[0071] The present invention combines the optimal model of advanced oxidation degradation with a genetic algorithm, which can accurately and quickly predict the degradation rate based on process detection indicators and obtain the optimal operating conditions, thereby optimizing the operating conditions and improving the automation and management levels of sewage plant control.

[0072] The method proposed in this invention for predicting the advanced oxidation degradation efficiency of pesticides based on molecular fingerprints and machine learning is not only applicable to the prediction of the degradation rate of pesticides, but also has the potential to be extended to the prediction of the advanced oxidation degradation rate of other pollutants, providing a broader application prospect for environmental pollution control.

[0073] The present invention also provides a device for predicting the effectiveness of pesticide advanced oxidative degradation based on molecular fingerprints and machine learning, comprising a memory, a control processor, and a computer program stored in the memory and executable on the control processor. The control processor executes the program to implement the method for predicting the effectiveness of pesticide advanced oxidative degradation based on molecular fingerprints and machine learning as described above.

[0074] The present invention also provides a control system, comprising the pesticide advanced oxidation degradation efficiency prediction device based on molecular fingerprint and machine learning as described above.

[0075] The present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to implement the pesticide advanced oxidative degradation efficiency prediction method based on molecular fingerprint and machine learning as described above.

[0076] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in different orders and / or concurrently with other actions from the diagrams and descriptions herein or not illustrated and described herein but understood by those skilled in the art. Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of the two. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Technicians can implement the described functionality in different ways for each specific application, but such implementation decisions should not be interpreted as resulting in a departure from the scope of the present invention. The various illustrative logic blocks, modules, and circuits described in conjunction with the embodiments disclosed herein may be implemented or executed using a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, a battery compartment control board, a micro-battery compartment control board, or a state machine. A processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read and write information from / to the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and storage medium may reside in the user terminal as discrete components. In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functions may be stored or transmitted as one or more instructions or codes on a computer-readable medium.Computer-readable media include both computer storage media and communication media, including any media that facilitates the transfer of a computer program from one place to another. Storage media can be any available media that can be accessed by a computer. As an example and not limitation, such computer-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer. Any connection is also properly referred to as a computer-readable medium. For example, if the software is transmitted from a website, a central control computer, or other remote source using a coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of medium. As used herein, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, where disks typically reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0077] Although the above methods are illustrated and described as a series of acts for simplicity of explanation, it is to be understood and appreciated that these methods are not limited by the order of the acts, as some acts may occur in a different order and / or concurrently with other acts from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art according to one or more embodiments.

[0078] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for predicting the efficacy of pesticide advanced oxidation degradation based on molecular fingerprints and machine learning, characterized in that: The following steps are involved: Step S1, collecting original detection data in the pesticide degradation reaction and grouping and preprocessing it to obtain degradation reaction data; Step S2: extract pesticide molecular samples from the chemical database, obtain the SMILES format of each pesticide, convert it into a molecular fingerprint, combine it with the degradation reaction data to form a complete data set, and divide the data set into a training set and a test set; Step S3: constructing random forest, gradient boosting decision tree, extreme gradient boosting, and lightweight gradient boosting machine models, performing target output based on the models and key input variables, predicting the degradation efficiency of pesticides by different advanced oxidation processes, and using evaluation indicators to evaluate and compare the model prediction effects to obtain the optimal model; Step S4, based on the optimal model, using Shapley value method and partial dependence diagram to assist in verification and identification of key variables to verify the correctness of the model; In step S5, the pollutant type and initial concentration are input into the obtained optimal model to obtain the operating conditions when the removal efficiency is the highest.

2. The method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning according to claim 1, characterized in that: In step S1, preprocessing includes one or more of data cleaning, data noise reduction, data dislocation or null value filling; and degradation reaction data are grouped according to advanced oxidation methods.

3. The method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning according to claim 1 or 2, characterized in that: Step S2 specifically includes: Step S21: Determine the pesticide type based on the degradation reaction data in S1, extract pesticide molecular samples from the Drugbank database, obtain the SMILES format of the corresponding drug samples from the pubchempy package in the pandas library, and then call the RDkit package in the pandas library to convert the SMILES format of each pesticide sample into different molecular fingerprints; Step S22, combining the pre-processed degradation reaction data with the corresponding two molecular fingerprints to form a complete data set; In step S23, each data set is randomly divided into a training set and a test set according to a certain proportion.

4. The method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning according to claim 3, characterized in that: Step S3 specifically includes: Step S31: input the training set into the random forest, gradient boosting decision tree, extreme gradient boosting and lightweight gradient boosting machine models, and use the grid search-cross validation algorithm to optimize the hyperparameters of the model to obtain the best model; Step S32: Use the obtained better model to predict the test set and use the coefficient of determination R 2 The accuracy of the best model obtained by evaluating the root mean square error (RMSE) is compared with the accuracy of the best models established based on the same data set to determine the optimal model. The calculation formula is as follows: Where: n is the number of samples, is the predicted value of the degradation rate of the i-th sample, y i is the actual degradation rate of the i-th sample, is the average value of the actual samples.

5. The method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning according to claim 4, characterized in that: In step S5, the initial concentration value and type of pollutants are first determined, and the value ranges of other parameters are set and input into the optimal model - genetic algorithm. The data are recombined through the genetic algorithm's inheritance, crossover and mutation methods. The predicted value of the pesticide advanced oxidation degradation efficiency obtained by the optimal model is used as the fitness, and the parameter combination corresponding to the maximum fitness is obtained, that is, the optimal operating condition is obtained.

6. The method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning according to claim 5, characterized in that: The specific steps of step S6 are as follows: Step S61, performing iterative calculations on the data in the training set to calculate the Tanimoto coefficient between each compound and the compounds in other training sets; Step S62, by setting different Tanimoto coefficient thresholds, deleting the data below these thresholds, using the optimal model to predict the remaining data, and calculating the root mean square error (RMSE) between the predicted value and the experimental value under each setting; Step S63, determining the Tanimoto coefficient corresponding to the obtained lowest RMSE value as a threshold standard; When verifying whether a new data set can be used with the model, the Tanimoto coefficient is calculated through steps S61 to S63. If the calculated Tanimoto coefficient is higher than a preset threshold, the tested compound is within the applicability domain and the prediction is reliable; Otherwise, the predictions are unreliable; thus limiting the scope of applicability of the resulting optimal model.

7. The method for predicting pesticide advanced oxidation degradation efficiency based on molecular fingerprint and machine learning according to any one of claims 1, 2, or 4-6, characterized in that: The pesticide degradation reaction conditions in step S1 include but are not limited to initial concentration, pH, reaction time, temperature, ultraviolet light intensity, and ultraviolet light wavelength.

8. A device for predicting the effectiveness of pesticide advanced oxidation degradation based on molecular fingerprints and machine learning, characterized in that: The method comprises a memory, a control processor, and a computer program stored in the memory and executable on the control processor, wherein the control processor executes the program to implement the method for predicting the pesticide advanced oxidative degradation efficiency based on molecular fingerprint and machine learning as described in any one of claims 1 to 7.

9. A system, characterized in that: Including the pesticide advanced oxidation degradation efficiency prediction device based on molecular fingerprint and machine learning as described in claim 8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to implement the method for predicting the pesticide advanced oxidative degradation efficiency based on molecular fingerprints and machine learning as described in any one of claims 1 to 7.

Images