Prediction method of degradation rate of petroleum hydrocarbon in groundwater based on complex water chemical condition and efficient degradation bacterial community of BP neural network

Abstract

This invention belongs to the field of groundwater pollution remediation technology and discloses a method for predicting the degradation rate of petroleum hydrocarbons in groundwater by highly efficient degrading microbial communities based on complex hydrochemical conditions and a backpropagation (BP) neural network. Addressing the technical challenge of predicting the remediation effect of highly efficient degrading microbial communities due to differences in hydrochemical conditions and their combined influence in groundwater at different sites, this invention constructs a dataset of petroleum hydrocarbon degradation rates in relation to dissolved oxygen, water temperature, salinity, NH3-N concentration, and remediation time through laboratory simulations. Furthermore, a BP neural network model is established to predict the degradation rate of petroleum hydrocarbons in groundwater using dissolved oxygen, water temperature, salinity, NH3-N concentration, and remediation time. The model demonstrates excellent predictive ability and can effectively predict the degradation rate of petroleum hydrocarbons by microbial communities under different groundwater hydrochemical conditions, providing a reference for predicting the effectiveness of microbial remediation of petroleum hydrocarbons in groundwater.

Description

Technical Field

[0001] This invention belongs to the field of groundwater pollution and remediation, and relates to a method for predicting the degradation rate of petroleum hydrocarbons in groundwater in a site based on complex hydrochemical conditions and a BP neural network of highly efficient degrading bacteria. Background Technology

[0002] During the extraction, transportation, storage, and processing of oil, leaks and spills can occur. When oil seeps into the soil and groundwater of petrochemical sites, it can cause serious damage to the groundwater environment. Bioremediation, especially microbial remediation, is currently the most promising strategy for remediating petroleum hydrocarbon-contaminated groundwater due to its environmental friendliness, low cost, and lack of secondary pollution.

[0003] Coastal petrochemical sites exhibit complex and highly variable hydrological conditions. Groundwater temperatures are typically low, between 10-20°C, and dissolved oxygen levels are low, usually between 0.1-2 mg / L. Salinity and ammonia nitrogen levels are high, far exceeding the Class IV groundwater standard. These factors significantly impact the degradation rate of petroleum hydrocarbons during microbial remediation, making it difficult to assess remediation feasibility and predict remediation effectiveness in practical applications. Furthermore, conducting experimental assessments of microbial remediation is costly, time-consuming, and susceptible to environmental factors, leading to data bias and delays. Therefore, establishing a highly efficient predictive model for the remediation of petroleum hydrocarbon pollution in groundwater using biodegrading microorganisms is crucial for assessing remediation feasibility and predicting petroleum hydrocarbon degradation rates under complex hydrochemical conditions.

[0004] Currently, the main models for predicting the degradation rate of petroleum hydrocarbons in groundwater include kinetic models, multiple quadratic regression models, and response surface methodology. The literature (Khan AA, et al. "Establishing correlations and scale-up factor for estimating the petroleum biodegradation rate in soil." Bioremediation Journal, 2015, 19(1):32-46.) constructed a regression model that estimates the degradation rate of petroleum hydrocarbons in soil by incorporating soil physicochemical properties (soil texture, soil organic matter, soil pH, and soil moisture content) and the initial inoculation amount of petroleum-degrading microorganisms. R0 2The value was 0.93, but the model only screened a few parameters related to soil physicochemical properties and did not consider the influence of groundwater hydrochemical conditions on the degradation of petroleum hydrocarbons by microorganisms. The literature (Liang Qingxia, et al. "Comparison of detection methods for petroleum hydrocarbons in soil and analysis of degradation kinetics." China Environmental Science, 2023, 43(08):4194-4201.) evaluated the fitting effect of zero-order reaction kinetic equation, first-order reaction kinetic equation, pseudo-first-order reaction kinetic equation and second-order reaction kinetic equation model on the biodegradation of petroleum hydrocarbons in polluted soil. It was found that the second-order reaction kinetics can better simulate the degradation effect of microorganisms on petroleum hydrocarbons. However, the second-order reaction kinetics is only a simplified model and is usually only applicable to specific conditions and ranges. It has great limitations and the prediction results deviate greatly from the actual situation in complex environments. The literature (He Jiani, et al. "Research on High-Throughput Evaluation Method of Functional Characteristics of Petroleum Degrading Bacteria." Modern Chemical Industry, 2024, 53(08):1772-1779.) established a multiple linear regression model for predicting the degradation rate of petroleum hydrocarbons, including four parameters: C23O enzyme activity, surfactant emulsification performance, dehydrogenase activity, and lipase activity of the bacterial community. This regression model was optimized by removing two parameters with insignificant correlation, lipase and dehydrogenase, and establishing an optimized multiple linear regression equation with C23O enzyme activity and surfactant emulsification index as parameters. The coefficient of determination R of this model was... 2 The value was 0.731, indicating poor fitting accuracy and prediction performance. Furthermore, it only considered the influence of the microbial community's own characteristics on the petroleum hydrocarbon degradation rate. The literature (Yang Y, et al. "Efficient Bioremediation of Petroleum-Contaminated Soil by Immobilized Bacterial Agent of Gordonia alkanivorans W33." Bioengineering, 10(5). (2023)) used a response surface methodology model to optimize the fermentation conditions (including the concentrations of MgCl2 and CaCl2 in the culture medium) of the petroleum hydrocarbon degrading bacterium Gordonia alkanivorans W33, predicting the biomass after fermentation and thus its degradation effect on petroleum in the soil. However, this model did not consider other potential environmental factors, thus affecting the model's completeness and accuracy.

[0005] In actual contaminated sites, the hydrochemical conditions of groundwater are complex and vary significantly across different areas and sites. Currently, there are no parameter prediction models that comprehensively consider the hydrochemical conditions of groundwater, especially those focusing on important influencing parameters such as groundwater temperature, dissolved oxygen content, salinity, and NH3-N concentration.

[0006] Therefore, this invention addresses the technical challenge of predicting the remediation effect of highly efficient degrading microbial communities on groundwater remediation due to differences in hydrochemical conditions and their combined effects across various sites. It establishes a method for predicting the degradation rate of petroleum hydrocarbons (PHH) in groundwater using highly efficient PHH based on complex hydrochemical conditions and a backpropagation (BP) neural network. First, a dataset of PHH degradation rate and parameters such as dissolved oxygen, water temperature, salinity, NH3-N concentration, and remediation time is constructed through laboratory simulations. Second, a BP neural network model is established to predict the PHH degradation rate in groundwater using five parameters: dissolved oxygen, water temperature, salinity, NH3-N concentration, and remediation time, providing a reference for predicting the effectiveness of microbial remediation of PHH in groundwater. Summary of the Invention

[0007] To address the technical challenge of predicting the effectiveness of existing microbial remediation methods due to the complex influence of hydrological conditions, this invention constructs a BP neural network model that predicts the degradation rate of petroleum hydrocarbons in groundwater based on dissolved oxygen, water temperature, salinity, NH3-N concentration, and remediation time.

[0008] The technical solution of the present invention is as follows:

[0009] A method for predicting the degradation rate of petroleum hydrocarbons in groundwater in a site using highly efficient degrading microbial communities based on complex hydrochemical conditions and a backpropagation neural network includes the following steps:

[0010] (1) The effect of microbial community on petroleum hydrocarbon degradation under different hydrochemical conditions

[0011] Petroleum-degrading bacteria were cultured in a simulated groundwater culture medium with diesel as the sole carbon source. The amount of diesel added was 1% of the volume of the simulated groundwater culture medium, and the culture was carried out in a constant temperature shaking incubator.

[0012] The petroleum-degrading bacteria are the highly efficient petroleum-degrading bacteria disclosed in patent CN 116410864 A.

[0013] Based on the hydrochemical conditions of the groundwater at the petrochemical site, the environmental parameters are set as follows: dissolved oxygen content is 1.8 mg / L-7.2 mg / L, culture temperature is 10-25℃, salinity in the system (calculated as NaCl) is 0-80 g / L, and NH3-N concentration in the system is set to 0-5 g / L.

[0014] The cultivation period was 28 days, and samples were taken every 7 days. The residual petroleum hydrocarbon content was determined by gravimetric method to obtain the petroleum hydrocarbon degradation rate.

[0015] (2) Establishment of a prediction model for the degradation rate of petroleum hydrocarbons in groundwater

[0016] (2.1) Take each environmental parameter as input variable and the petroleum hydrocarbon degradation rate as output indicator;

[0017] Input variables: x1 = dissolved oxygen content, x2 = culture temperature, x3 = salinity, x4 = NH3-N concentration, x5 = repair time;

[0018] Output variable: y1 = petroleum hydrocarbon degradation rate;

[0019] The obtained input and output variables are used as the sample dataset X for the petroleum hydrocarbon degradation rate prediction model. The data in the sample dataset are then normalized using the mapminmax normalization function. The process is as follows:

[0020]

[0021] In the formula, X i The environmental data for the sample dataset X; x imin and x imax These are the minimum and maximum values ​​before normalization, respectively. The normalization operation is completed on the sample dataset. The normalized data sample set X′={x′1,x′2,x′3,x′4,x′5};

[0022] (2.2) Determine the network structure. Based on the input and output indicators selected in step (2.1), determine the number of input, output, and hidden layers of the petroleum hydrocarbon degradation rate prediction model based on the BP neural network. The number of nodes in the input layer of the petroleum hydrocarbon degradation rate prediction model is m=5, and the number of nodes in the output layer is n=1. The number of hidden layer neurons is determined by the following formula:

[0023]

[0024] In the formula, S is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and a is a constant between 1 and 10;

[0025] (2.3) Selection of activation function: Tansig activation function is used for input layer, Purelin activation function is used for output layer, and trainlm function is selected as training function for petroleum hydrocarbon degradation rate prediction model.

[0026] (2.4) The obtained sample dataset is used to train a model for predicting the degradation rate of petroleum hydrocarbons in the groundwater of the site based on complex hydrochemical conditions and efficient degradation bacteria.

[0027] The relevant parameter settings during the training process of the petroleum hydrocarbon degradation rate prediction model are as follows: learning rate is 0.01, training times are 2000, and minimum training error is 0.00001.

[0028] (2.5) The coefficient of determination R is used as a statistical indicator. 2 To evaluate the prediction results of the petroleum hydrocarbon degradation rate prediction model; coefficient of determination R2 The mathematical form is as follows:

[0029]

[0030] In the formula, y i It is the actual value, f i It is a predicted value. It is the average of the actual values, and i is the data index in the sample dataset;

[0031] The closer the predicted value is to the actual value, the higher the goodness of fit. Therefore, R... 2 It can effectively reflect the degree of fit between predicted and actual values ​​and evaluate the quality of model prediction results.

[0032] The beneficial effects of this invention are as follows: This invention considers the influence of dissolved oxygen content, water temperature, salinity, and NH3-N concentration in actual petrochemical contaminated sites on the degradation rate of petroleum hydrocarbon pollution by microbial community remediation. A BP neural network model is used to construct a model for predicting the petroleum hydrocarbon degradation rate based on dissolved oxygen content, water temperature, salinity, NH3-N concentration, and remediation time. This invention demonstrates good predictive performance and solves the problem of difficulty in predicting the degradation effect of microbial communities due to the combined influence of hydrological conditions. Attached Figure Description

[0033] Figure 1 This is a comparison chart of the actual and predicted values ​​in the model training set.

[0034] Figure 2 This is a model validation graph. Detailed Implementation

[0035] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0036] Example 1: The effect of bacterial community on the degradation of petroleum hydrocarbons under different dissolved oxygen and temperature levels

[0037] (1) Source of microbial community

[0038] The highly efficient petroleum-degrading bacterial community disclosed in patent CN 116410864 A consists of various bacteria and fungi. The dominant bacterial genera are *Pseudomonas*, *Achromobacter*, and *Stenotrophomonas*, all of which are petroleum hydrocarbon-degrading bacteria. The dominant fungal genera are *Mortierella* and *Fusarium*, both of which are petroleum hydrocarbon-degrading bacteria. This community contains as many as 25 petroleum hydrocarbon-degrading bacterial genera with a relative abundance greater than 0.1%.

[0039] (2) The effect of microbial community on the degradation of petroleum hydrocarbons under different dissolved oxygen and temperature levels

[0040] The activated microbial community was inoculated into 50 mL of sterilized simulated groundwater culture medium at a concentration of 5% (v / v), with 1% (v / v) diesel oil added as the sole carbon source. Dissolved oxygen concentrations in the system were set at 2 (±0.2) mg / L, 5 (±0.2) mg / L, and 7 (±0.2) mg / L, and incubation temperatures were set at 10℃, 15℃, and 20℃, respectively. Dissolved oxygen concentration and temperature were combined in pairs, and the system was incubated at 180 rpm for 28 days. Samples were taken at 7, 14, 21, and 28 days to determine the degradation rate of petroleum hydrocarbons.

[0041] The remaining petroleum hydrocarbons in the sample were extracted with dichloromethane, and the content of residual petroleum hydrocarbons was determined by gravimetric method to obtain the petroleum hydrocarbon degradation rate.

[0042] The simulated groundwater culture medium contains: MgSO4·H2O 0.3g / L, NH4Cl 1g / L, CaCl2·H2O 0.3g / L, KH2PO4 0.4g / L, Na2SO4 0.4g / L, MnSO4·H2O 0.0015g / L, and FeSO4·7H2O 0.002g / L.

[0043] (3) The effect of microbial community on the degradation of petroleum hydrocarbons at different salinity levels

[0044] The activated microbial community was inoculated into 50 mL of sterilized simulated groundwater culture medium at a dosage of 5% (v / v), with 1% (v / v) diesel oil as the sole carbon source. A certain amount of NaCl was added to achieve NaCl concentrations of 0, 20 g / L, 40 g / L, and 60 g / L in the culture medium. The activated microbial community was inoculated into 50 mL of sterilized simulated groundwater culture medium at a dosage of 5% (v / v), maintaining a dissolved oxygen content of 2 (±0.2) mg / L. The culture was carried out at 20℃ and 180 rpm for 28 days. Samples were taken at 7, 14, 21, and 28 days of culture to determine the petroleum hydrocarbon degradation rate.

[0045] (4) The effect of microbial community on the degradation of petroleum hydrocarbons at different NH3-N concentration levels

[0046] The activated microbial community was inoculated into 50 mL of sterilized simulated groundwater culture medium at a dosage of 5% (v / v), with 1% (v / v) diesel oil added as the sole carbon source. A certain amount of NH3-N (NH4Cl was used in this example) was added to achieve NH3-N concentrations of 0, 1 g / L, and 3 g / L in the culture medium. The activated microbial community was inoculated into 50 mL of sterilized simulated groundwater culture medium at a dosage of 5% (v / v), maintaining a dissolved oxygen content of 2 (±0.2) mg / L. The culture was carried out at 20℃ and 180 rpm for 28 days. Samples were taken at 7, 14, 21, and 28 days of culture to detect the degradation rate of petroleum hydrocarbons.

[0047] Example 2: Establishment of a predictive model for the degradation rate of petroleum hydrocarbon pollution by microbial community remediation

[0048] (1) Construct a model dataset from the obtained input and output data, using dissolved oxygen content, water temperature, salinity, NH3-N concentration, and remediation time as input variables, and petroleum hydrocarbon degradation rate as output variable. Train the model using the obtained dataset. The specific symbols representing the model construction are shown in Table 1.

[0049] Table 1. Symbols representing the model parameters

[0050]

[0051] The input and output variables obtained in the above steps are used to build a sample dataset for the model, and the data in the dataset is normalized using the mapminmax normalization function. The process is as follows:

[0052]

[0053] In the formula, X i The environmental data for the sample dataset X; x imin and x imax These are the minimum and maximum values ​​before normalization, respectively. This completes the normalization operation on the sample dataset. The normalized data sample set X′={x′1,x′2,x′3,x′4,x′5} can be directly used as the input and output variables required for subsequent model building.

[0054] (2) Determine the strategy for network structure. Based on the input and output indicators selected in step (1), determine the input, output, and number of hidden layers of the network structure. For the BP neural network used to predict the degradation rate of petroleum hydrocarbons, the number of nodes in the input layer is m=5, and the number of nodes in the output layer is n=1. The number of hidden layer neurons is determined by the following formula.

[0055]

[0056] In the formula, S is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and a is an adjustment constant between 1 and 10;

[0057] (3) Selection of activation function: Tansig activation function is used for input layer, Purelin activation function is used for output layer, and trainlm function is selected as training function for petroleum hydrocarbon degradation rate prediction model.

[0058] (4) The sample dataset of the obtained input and output variables is used to build a model. Through training, a BP neural network model is obtained to predict the degradation rate of petroleum hydrocarbons in groundwater in the site based on complex hydrochemical conditions.

[0059] The relevant parameter settings during the training of the BP neural network model are as follows: learning rate is 0.01, training iterations are 2000, and minimum training error is 0.00001.

[0060] (5) Using the data statistical indicator, the coefficient of determination R 2 To evaluate the prediction results of the petroleum hydrocarbon degradation rate prediction model; coefficient of determination R 2 The mathematical form is as follows:

[0061]

[0062] In the formula, y i It is the actual value, f i It is a predicted value. It is the average of the actual values, and i is the data index in the sample dataset;

[0063] The closer the predicted value is to the actual value, the higher the goodness of fit, the higher the R-value. 2 The larger it is, the greater R is. 2 It can effectively reflect the degree of fit between predicted and actual values ​​and evaluate the quality of model prediction results.

[0064] Model fitting results: the coefficient of determination R on the model training set 2 It is 0.90.

[0065] Example 3: Validation of a predictive model for the degradation rate of petroleum hydrocarbon pollution by microbial community remediation

[0066] Petroleum hydrocarbon degradation experiments were conducted under hydrochemical conditions different from those in Example 1 to obtain the actual petroleum hydrocarbon degradation rate in groundwater. Simultaneously, the experimental parameters were substituted into a pre-established prediction model to obtain the predicted values. Table 2 summarizes the actual and predicted values ​​of petroleum hydrocarbon degradation rates by highly efficient degrading microbial communities under different remediation conditions.

[0067] Table 2. Actual and predicted values ​​of degradation rate of petroleum hydrocarbons by highly efficient degrading bacteria under different conditions.

[0068]

[0069] Model fitting results: Coefficient of determination R 2 =0.97, with a coefficient of determination of over 0.9, indicating that this method is effective in predicting the degradation rate of petroleum hydrocarbons by highly efficient degrading bacteria in groundwater at the site.

Claims

1. A method for predicting the degradation rate of petroleum hydrocarbons in groundwater in a site based on complex hydrochemical conditions and a BP neural network, characterized in that, Includes the following steps: (1) The effect of microbial community on petroleum hydrocarbon degradation under different hydrochemical conditions Petroleum-degrading bacteria were cultured in a simulated groundwater culture medium with diesel as the sole carbon source. The amount of diesel added was 1% of the volume of the simulated groundwater culture medium, and the culture was carried out in a constant temperature shaking incubator. The cultivation period was 28 days, and samples were taken every 7 days. The residual petroleum hydrocarbon content was determined by gravimetric method to obtain the petroleum hydrocarbon degradation rate. (2) Establishment of a prediction model for the degradation rate of petroleum hydrocarbons in groundwater (2.1) Take each environmental parameter as input variable and the petroleum hydrocarbon degradation rate as output indicator; Input variables: x1 = dissolved oxygen content, x2 = culture temperature, x3 = salinity, x4 = NH3-N concentration, x5 = repair time; Output variable: y1 = petroleum hydrocarbon degradation rate; (2.2) Determine the network structure. Based on the input and output indicators selected in step (2.1), determine the number of input, output, and hidden layers of the petroleum hydrocarbon degradation rate prediction model based on the BP neural network. The number of nodes in the input layer of the petroleum hydrocarbon degradation rate prediction model is m=5, and the number of nodes in the output layer is n=1. The number of hidden layer neurons is determined by the following formula: In the formula, S is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and a is a constant between 1 and 10; (2.3) Selection of activation function: Tansig activation function is used for input layer, Purelin activation function is used for output layer, and trainlm function is selected as training function for petroleum hydrocarbon degradation rate prediction model. (2.4) The obtained sample dataset is used to train a model for predicting the degradation rate of petroleum hydrocarbons in the groundwater of the site based on complex hydrochemical conditions and efficient degradation bacteria. (2.5) The coefficient of determination R is used as a statistical indicator. 2 To evaluate the prediction results of the petroleum hydrocarbon degradation rate prediction model.

2. The method for predicting the degradation rate of petroleum hydrocarbons in groundwater by highly efficient degrading microbial communities according to claim 1, characterized in that, Based on the hydrochemical conditions of the groundwater at the petrochemical site, the environmental parameters are set as follows: dissolved oxygen content is 1.8 mg / L-7.2 mg / L, culture temperature is 10-25℃, salinity in the system (calculated as NaCl) is 0-80 g / L, and NH3-N concentration in the system is set to 0-5 g / L.

3. The method for predicting the degradation rate of petroleum hydrocarbons in groundwater by highly efficient degrading microbial communities according to claim 1, characterized in that, The obtained input and output variables are used as the sample dataset X for the petroleum hydrocarbon degradation rate prediction model. The data in the sample dataset are then normalized using the mapminmax normalization function. The process is as follows: In the formula, X i The environmental data for the sample dataset X; x imin and x imax These are the minimum and maximum values ​​before normalization, respectively. The normalization operation is then performed on the sample dataset, resulting in the normalized data sample set X′={x′1,x′2,x′3,x′4,x′5}.

4. The method for predicting the degradation rate of petroleum hydrocarbons in groundwater by highly efficient degrading microbial communities according to claim 1, characterized in that, The relevant parameter settings during the training process of the petroleum hydrocarbon degradation rate prediction model are as follows: learning rate is 0.01, training times are 2000, and minimum training error is 0.00001.

5. The method for predicting the degradation rate of petroleum hydrocarbons in groundwater by highly efficient degrading microbial communities according to claim 1, characterized in that, The simulated groundwater culture medium contains: MgSO4·H2O 0.3g / L, NH4Cl 1g / L, CaCl2·H2O 0.3g / L, KH2PO4 0.4g / L, Na2SO4 0.4g / L, MnSO4·H2O 0.0015g / L, and FeSO4·7H2O 0.002g / L.

Images