A method and system for quantifying greenhouse gases in a wastewater system

Abstract

This invention proposes a method and system for quantifying greenhouse gases in wastewater systems. The method includes: establishing a data database; establishing reaction kinetic models for greenhouse gases in each process unit; acquiring greenhouse gas production data in each process unit; constructing and training a neural network model; and solving for the predicted greenhouse gas production. This invention considers both chemical and biological reactions in wastewater by combining wastewater substrate concentration and dissolved oxygen concentration. It also establishes direct correlations between different process characteristics through a database, facilitating targeted optimization measures by wastewater treatment plants in these units. Based on this reliable dataset, a neural network model is built to provide continuous and reliable data support for predicting subsequent greenhouse gas production.

Description

Technical Field

[0001] This invention belongs to the field of wastewater treatment technology, and specifically relates to a method and system for quantifying greenhouse gases in wastewater systems. Background Technology

[0002] Wastewater treatment is a significant source of greenhouse gas emissions. Unlike other sources, many processes in wastewater treatment plants involve anaerobic treatment, which directly generates greenhouse gases, primarily CH4 and N2O. Currently, most wastewater treatment plants use the emission factor method for statistical analysis, which ignores the differences in treatment processes, equipment, management levels, and pollution loads among wastewater treatment plants. Furthermore, it only estimates a few major pollutants (such as COD and ammonia nitrogen), failing to accurately reflect the greenhouse gas emissions from the wastewater system of a wastewater treatment plant.

[0003] Furthermore, traditional sampling methods generally employ flux box methods, which are costly in terms of manpower and resources and cannot achieve real-time continuous detection. At the same time, since the load of wastewater treatment plants often fluctuates, the emission concentration of greenhouse gases fluctuates greatly, depending on various factors such as detection time, monitoring frequency, and sampling conditions, which directly affects the accuracy of reflecting the greenhouse gas emissions in the wastewater system of wastewater treatment plants.

[0004] Therefore, overcoming the shortcomings of the existing technology is an urgent problem to be solved in this technical field. Summary of the Invention

[0005] To address the above problems, this invention proposes a method for quantifying greenhouse gases in wastewater systems, comprising the following steps:

[0006] A database of information was established based on the process characteristics of each process unit in the wastewater treatment system.

[0007] Based on the data information database, the wastewater substrate parameters of each process unit are determined, and a reaction kinetic model of greenhouse gases for each process unit is established to identify the process unit that generates greenhouse gases as the operating process unit.

[0008] Monitor each operational process unit and obtain data on the amount of greenhouse gases generated in each operational process unit;

[0009] A dataset is established based on the data information database and the generated data. A neural network model is then constructed and trained based on the dataset.

[0010] Input the substrate parameters of the wastewater to be tested into the neural network model, and solve to obtain the predicted amount of greenhouse gas production.

[0011] Furthermore, the data information repository is a relational database.

[0012] Furthermore, the data information database includes several related tables, which are linked together by the process number as a key field.

[0013] Furthermore, the reaction kinetic model was established using the Mono model;

[0014] The specific steps for identifying the process unit that generates greenhouse gases as the operating process unit are as follows:

[0015] The wastewater substrate parameters for each process unit are determined based on the data information database;

[0016] Establish Monod equations for greenhouse gases for each process unit and obtain the solution values;

[0017] A process unit whose solution value is greater than 0 is determined as a working process unit.

[0018] Furthermore, the dataset includes a training set, a test set, and a validation set, with a training set:test set:validation set ratio of 80-90%:5-10%:5-10%.

[0019] Furthermore, the neural network model adopts a convolutional neural network model, which includes an input layer, a first convolutional layer, an activation function, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer, and an output layer.

[0020] The kernel size for both the first and second convolutional layers is 3×1.

[0021] This invention also proposes a method for quantifying greenhouse gases in a wastewater system, comprising:

[0022] The data benchmark module is used to establish a data information database based on the process characteristics of each process unit in the wastewater treatment system.

[0023] The unit classification module is used to determine the wastewater substrate parameters of each process unit based on the data information database, and to establish a reaction kinetic model of each process unit with respect to greenhouse gases, and to identify the process unit that generates greenhouse gases as the operating process unit.

[0024] The data monitoring module is used to monitor each operational process unit and acquire data on the amount of greenhouse gases generated in each operational process unit.

[0025] The model building module is used to create a dataset based on the data information database and the generated data, and to build and train a neural network model based on the dataset.

[0026] The model calculation module is used to input the substrate parameters of the wastewater to be tested into the neural network model and solve for the predicted amount of greenhouse gas production.

[0027] Furthermore, the dataset includes a training set, a test set, and a validation set, with a training set:test set:validation set ratio of 80-90%:5-10%:5-10%.

[0028] Furthermore, the neural network model adopts a convolutional neural network model, which includes an input layer, a first convolutional layer, an activation function, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer, and an output layer.

[0029] The kernel size of both the first and second convolutional layers is 3×1.

[0030] Compared with the prior art, the embodiments of the present invention have at least the following advantages:

[0031] The greenhouse gas quantification method and system for wastewater systems of the present invention establishes reaction kinetic models of greenhouse gases for different process units. Based on the concentration of wastewater substrates and dissolved oxygen, it considers the chemical and biological reactions in wastewater, accurately determines the greenhouse gas generation in different process units, and establishes direct correlations between different process characteristics through a database, facilitating targeted optimization measures by wastewater treatment plants in these units. Furthermore, based on this reliable dataset, a neural network model is built to provide continuous and reliable data support for predicting subsequent greenhouse gas generation, enabling real-time acquisition of accurate quantification data on greenhouse gases in wastewater systems. This provides strong support for reducing greenhouse gas emissions, improving wastewater treatment efficiency, and meeting environmental protection requirements.

[0032] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description and the drawings. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 A schematic flowchart of the greenhouse gas quantification method in a wastewater system according to an embodiment of the present invention is shown;

[0035] Figure 2 A block diagram of a greenhouse gas quantification system in a wastewater system according to an embodiment of the present invention is shown;

[0036] Figure 3A block diagram of a greenhouse gas monitoring system in a wastewater system according to an embodiment of the present invention is shown. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] This invention provides a method and system for quantifying greenhouse gases in a wastewater system. Figure 1 A schematic flowchart of a greenhouse gas quantification method in a wastewater system according to an embodiment of the present invention is shown. Figure 1 In China, methods for quantifying greenhouse gases in wastewater systems include:

[0039] S101. Establish a database based on the process characteristics of each process unit in the wastewater treatment system;

[0040] S102. Based on the data information database, determine the wastewater substrate parameters of each process unit, establish a reaction kinetic model of each process unit for greenhouse gases, and determine the process unit that generates greenhouse gases as the operating process unit.

[0041] S103. Monitor each operational process unit and obtain data on the amount of greenhouse gases generated in each operational process unit;

[0042] S104. Establish a dataset based on the data information database and the generated data, and construct and train a neural network model based on the dataset.

[0043] S105. Input the substrate parameters of the wastewater to be tested into the neural network model and solve to obtain the predicted amount of greenhouse gas production.

[0044] The neural network model is constructed and trained to establish a dataset using a data information database and production data. This dataset includes wastewater substrate parameters of the process unit and the operating process unit, as well as the production data of the corresponding operating process unit. The production data of the process unit will be uniformly defined as zero or other fixed values.

[0045] The greenhouse gas quantification method for wastewater systems proposed in this invention establishes reaction kinetic models of greenhouse gases for different process units. Based on detailed parameters of the wastewater substrate, it accurately describes gas generation within different process units and establishes direct correlations between different process characteristics through a database, facilitating targeted optimization measures by wastewater treatment plants in these units. Furthermore, based on this reliable dataset, a neural network model is constructed to provide continuous and reliable data support for predicting subsequent greenhouse gas production. This facilitates the real-time acquisition of accurate quantification data on greenhouse gases in wastewater systems, providing strong support for reducing greenhouse gas emissions, improving wastewater treatment efficiency, and meeting environmental protection requirements.

[0046] The data information database includes several related tables: process table, operation data table, and processing index table.

[0047] The specific classification definitions of the process table are shown in Tables 1 and 2, including: process number (process ID), process name (equipment ID), process type, influent source, effluent destination, pollutant type, and process description; used to describe various treatment units or process flows in the wastewater treatment plant, and record different treatment steps and their relationships.

[0048] Table 1 Process Table

[0049]

[0050]

[0051] Table 2. Examples of Process Representation

[0052]

[0053] For specific classification definitions of the operation data table, please refer to Tables 3 and 4, including: process number (process ID), record date, water flow rate, dissolved oxygen, and temperature; records the actual operation data of the wastewater treatment plant in different processes and time periods.

[0054] Table 3 Operational Data Table

[0055]

[0056] Table 4 Example of running data

[0057]

[0058] The specific classification definitions of the treatment index table are shown in Tables 5 and 6, which include: process number (process ID), recording date, chemical oxygen demand in influent, chemical oxygen demand in effluent, biochemical oxygen demand in influent, biochemical oxygen demand in effluent, ammonia nitrogen concentration in influent, and ammonia nitrogen concentration in effluent; used to record key pollutants and emission indicators of each process section and treatment unit.

[0059] Table 5 Processing Indicators

[0060]

[0061]

[0062] Table 6 Examples of Processing Indicators

[0063]

[0064] Specifically, the data information repository is a relational database, which includes several related tables. These tables are linked by the process ID as a key field. For example, the relational database may be MySQL or PostgreSQL.

[0065] In this embodiment, the data in the data information database is divided into multiple related tables. The process ID is used as the key field to establish relationships between the multiple tables, so that the information of each link in the process can be stored and queried in an orderly and accurate manner, ensuring the standardization of the database structure and the integrity of the data.

[0066] Referring to Table 1-6, the process ID serves as a unique identifier for each process unit or step, used to uniquely identify each process unit, such as "pretreatment", "biological treatment", "sludge treatment", etc.

[0067] By using process ID constraints, the database can automatically maintain data consistency and integrity. For example, if a process is deleted, the database can automatically check for associated parameters or equipment information and delete or prevent deletion based on settings. This structure also facilitates database expansion and maintenance; for instance, new process unit tables can be easily added without altering the overall structure.

[0068] It should be added that, during the establishment of the database, the equipment ID can also be used as a unique identifier for specific equipment in the wastewater treatment plant, such as "aeration tank" or "bioreactor".

[0069] Specifically, the reaction kinetic model was established using the Monod model.

[0070] It should be further clarified that greenhouse gases include carbon dioxide (CO2), ozone (O3), nitrous oxide (N2O), methane (CH4), hydrochlorofluorocarbons (CFCs, HFCs, HCFCs), perfluorocarbons (PFCs), and sulfur hexafluoride (SF6). Of these, carbon dioxide (CO2), nitrous oxide (N2O), and methane (CH4) are generated in wastewater systems, while the remaining greenhouse gases are not emitted during wastewater treatment plant operations. Therefore, this description focuses on wastewater treatment operations. In this embodiment, nitrous oxide (N2O) and methane (CH4) gases, and the treatment units including nitrification tanks, denitrification tanks, and anaerobic digesters, are used as examples for illustration.

[0071] Correspondingly, based on the substrate parameters of each process unit, a reaction kinetic model for greenhouse gases is established, and the specific steps for determining the process unit that generates greenhouse gases as the operating process unit are as follows:

[0072] The wastewater substrate parameters for each process unit are determined based on the data information database;

[0073] Establish Monod equations for greenhouse gases for each process unit;

[0074] For example, the Monod equation is used to establish a reaction kinetic model of N2O in the nitrification tank and the denitrification tank;

[0075] The equation for N2O generation in the nitrification tank is as follows:

[0076]

[0077] in, It is the rate of N2O formation during nitration; It is the maximum specific growth rate of N2O during nitrification; C a It refers to ammonia nitrogen concentration; K a C is the half-saturation constant of ammonia nitrogen; DO It is the dissolved oxygen concentration; K DO It is the half-saturation constant of dissolved oxygen.

[0078] Correspondingly, the equation for N2O generation in the denitrification tank is as follows:

[0079]

[0080] in, It is the rate of N2O formation during denitrification; It is the maximum specific growth rate of N2O during denitrification; C x It refers to ammonia nitrogen concentration; K x C is the half-saturation constant of ammonia nitrogen; COD It is the concentration of chemical oxygen demand (COD); KCOD It is the half-saturation constant of chemical oxygen demand (COD).

[0081] It should be added that in anaerobic digesters, the two main pathways for methane generation include acetic acid decomposition and hydrogen conversion, which are carried out by acetic acid-producing methanogens and hydrogen-producing methanogens, respectively.

[0082] The acetic acid decomposition pathway is the main route for methanogenesis, where acetic acid is broken down into methane and carbon dioxide. Therefore, the first equation regarding CH4 formation in the anaerobic digester is as follows:

[0083]

[0084] in, It is the rate at which acetic acid produces CH4; It is the maximum rate constant for the formation of methane from acetic acid; C y It is the concentration of acetic acid; K y It is the half-saturation constant of acetic acid.

[0085] Correspondingly, hydrogen and carbon dioxide in wastewater will produce methane. Therefore, the first equation regarding CH4 formation in the anaerobic digester is as follows:

[0086]

[0087] in, It is the rate of CH4 formation via the hydrogen pathway; It is the maximum rate constant for the formation of methane from hydrogen; C d2 It is the concentration of hydrogen gas; K s It is the half-saturation constant of hydrogen. It refers to the concentration of carbon dioxide; It is the half-saturation constant of carbon dioxide.

[0088] Numerical solutions were obtained for the Monod equations for CH4 and N2O in the corresponding process units.

[0089] when and / or and / or and / or When the value is greater than 0, it indicates that CH4 and / or N2O are generated, and the process unit is determined to be a working process unit.

[0090] By establishing reaction kinetic models of greenhouse gases for different process units, and taking into account the concentration of wastewater substrates and dissolved oxygen, the chemical and biological reactions in wastewater can be considered, and the gas generation in different process units can be accurately determined.

[0091] After determining the operational process unit, the operational process unit is monitored online, and the amount of greenhouse gas generated is calculated.

[0092] Among them, Figure 3 In the example shown, in this embodiment, an online extraction analysis system is used to monitor each work process unit. The online extraction analysis system includes a sampling probe, a preprocessing system, an online monitoring system, and a parameter monitoring system.

[0093] During the sampling process, the sampling probe includes a sampling probe and a sampling tube. The sampling tube (probe) must be corrosion-resistant and oxidation-reduction resistant.

[0094] The pretreatment system includes a condenser, a filter, and an air pump; the filter is used to remove impurities such as particulate matter from the sampled gas, the condenser is used to condense water vapor in the sampled gas into liquid, and the air pump is used to extract the sampled gas; and a hose connector is used to connect to the sampling probe for sampling.

[0095] The online monitoring system uses a concentration sensor employing tunable semiconductor laser absorption spectroscopy (TDLAS) to continuously acquire the concentration of the injected gas.

[0096] Taking the simultaneous measurement of two gases, CH4 and N2O, as an example, the measurement includes a methane concentration sensor and a nitrous oxide concentration sensor. The methane concentration sensor uses a TDLAS-tuned laser to continuously acquire the methane concentration of the sample gas; the nitrous oxide concentration sensor uses a TDLAS-tuned laser to continuously acquire the nitrous oxide concentration of the sample gas.

[0097] The parameter monitoring system includes a temperature detection unit, a humidity monitoring unit, a pressure monitoring unit, and a gas flow monitoring unit; corresponding to a gas temperature sensor, a gas humidity sensor, a gas pressure sensor, and a gas flow rate sensor, respectively.

[0098] The gas flow rate sensor is used to continuously acquire the flow rate of the injected gas for calculating the gas flow rate; the gas pressure sensor is used to continuously acquire the pressure of the injected gas.

[0099] The gas temperature sensor is used to continuously acquire the temperature of the injected gas;

[0100] A gas humidity sensor is used to continuously acquire the humidity of the injected gas.

[0101] For example, the process units are numbered i, i+1, etc., and the CH4 and N2O gases in the process units are monitored online.

[0102] Specifically, the i-th work process unit, i.e. the i-th work process unit is in online monitoring operation, the running time is determined according to the sampling requirements, and sampling and analysis are performed every Δt minutes;

[0103] Correspondingly, the concentration of CH4 or N2O in the i-th process unit is measured, the generation rate of CH4 or N2O in the i-th process unit is calculated, and the time integral is performed on all process units that generate CH4 or N2O gas in the wastewater system to obtain the total amount of CH4 or N2O generated.

[0104] The specific formula for calculating the generation rate of CH4 or N2O in the i-th process unit is as follows:

[0105]

[0106] Among them, G i,j (t) represents the generation rate of gas j in the i-th process unit at time t, in mg / h; C i,j (t) represents the concentration of gas j in the i-th process unit at time t, in mg / m3; Q i (t) represents the wastewater flow rate of the process unit at time t, in m³ / h; V i Δt is the volume of wastewater in the i-th process unit, in m3; Δt is the time interval for measuring concentration; and j is the symbol for CH4 or N2O.

[0107] The volume of wastewater in the i-th process unit is calculated using the following formula:

[0108] V i =h i ×L i ×W i (6)

[0109] Among them, V i Let h be the volume of wastewater in the i-th process unit, in m³; i L represents the wastewater level in the i-th process unit, in meters (m). i W is the length of the i-th process unit, in m; i Let m be the length of the i-th process unit.

[0110] In addition, it should be noted that before building a neural network model, the input and output data need to be converted in format.

[0111] The input data is preprocessed, which involves extracting data from the data information database to form a total feature set suitable for model input. The total feature set will be converted into a four-dimensional format.

[0112] Specifically, the input data X is defined and format-converted to the four-dimensional format (batchsize, 1, height, width) required by CNN (Convolutional Neural Network). The height and width are defined according to the specific structure of the dataset. For example, the input data X is converted to (batchsize, n, 1, 1) to adapt to the input requirements of the convolutional neural network model. Here, the data shape is (batchsize, n), where batchsize represents the number of samples and n represents the number of features.

[0113] Define the output data Y, which represents the predicted greenhouse gas emissions, with a shape of (batchsize, 2). Taking CH4 and N2O as an example, 2 represents the two target variables of emissions (i.e., CH4 and N2O).

[0114] Correspondingly, the dataset formed by input data X and output data Y is divided into training set, test set, and validation set, with training set:test set:validation set = 80-90%:5-10%:5-10%. Then, the neural network model is built. For example, training set:test set:validation set = 90%:5%:5%.

[0115] In this embodiment, the neural network model is a convolutional neural network model, which includes an input layer, a convolutional layer, an activation function, a pooling layer, a fully connected layer, and an output layer.

[0116] The convolutional layer includes a first convolutional layer and a second convolutional layer, and the corresponding pooling layer includes a first pooling layer and a second pooling layer.

[0117] The input layer is used to accept data from the dataset;

[0118] The first convolutional layer is used to extract local features from the data;

[0119] Activation functions are used for nonlinear transformations of feature data;

[0120] The first pooling layer is used to pool the features output by the first convolutional layer, reducing the number of features.

[0121] The second convolutional layer is used to extract higher-level features;

[0122] The second pooling layer is used to pool the features output by the second convolutional layer, thereby reducing the number of features parameters for the second time.

[0123] The flattening layer is used to flatten the multidimensional features pooled by the second pooling layer into one-dimensional data.

[0124] Fully connected layers are used to integrate flattened one-dimensional data and make predictions through nonlinear combinations;

[0125] The output layer is used to output the predicted emission levels of the gas.

[0126] The number of neurons in the output layer should be kept consistent with the number of output variables. For example, in the task of predicting the generation of CH4 and N2O in wastewater treatment, the corresponding number of neurons in the output layer is 2, used to predict the emissions of greenhouse gases (CH4 and N2O).

[0127] The number of neurons in the input layer is consistent with the number of input process features. Taking Table 6 as an example, the available process features are: process ID, record date, influent COD, effluent COD, influent BOD, effluent BOD, influent ammonia nitrogen, and effluent ammonia nitrogen. Therefore, the number of neurons in the input layer is 8.

[0128] By setting two convolutional layers, the model is allowed to gradually transition from low-level features to high-level features, enhancing its ability to capture complex patterns in the data. Furthermore, by using two pooling layers, the size and complexity of the feature maps are reduced, removing some redundant information and thus mitigating the risk of overfitting. The combination of two convolutional layers and two pooling layers enables progressive feature extraction and compression, allowing the model to retain key features while optimizing computational efficiency, providing rich and highly abstract feature representations for subsequent fully connected layer predictions.

[0129] Furthermore, building upon the first and second convolutional layers, the kernel size is set to 3×1. Here, "3" indicates that the kernel continuously covers three features in the feature dimension of the input data, while "1" indicates that the kernel performs a single-step operation in the other dimension. That is, either the first or second convolutional layer covers three features at a time, and each change only modifies one feature. This size design can capture local patterns or relationships between adjacent features, thereby helping the model identify the microscopic correlation between wastewater features and operating conditions.

[0130] For ease of explanation, the model is constructed as follows:

[0131] Select 80% of the dataset as the training set, convert the input data X to (batchsize, 1, height, width) format, and define height and width as 1 according to the specific structure of the dataset; convert the output data Y to (batchsize, 2) format.

[0132] Input data X is passed into the input layer of the model to predict output data Y;

[0133] A first convolutional layer is established and applied to the input data X to extract local feature patterns of the process characteristics;

[0134] The ReLU (Rectified LinearUnit) activation function is applied to perform a non-linear transformation on the features after convolution in the first convolutional layer to improve the model's expressive power.

[0135] Add a first pooling layer to pool the features output by the first convolutional layer, thereby reducing the size of the features and the number of parameters;

[0136] A second convolutional layer is added, which accepts the input features from the first pooling layer and further extracts higher-level features. The output of the second convolutional layer further enhances the feature representation capability of the data.

[0137] A second pooling layer is added, which accepts the input features from the second convolutional layer and performs pooling operations to reduce the size of the features, thereby further reducing the number of parameters.

[0138] Add a flattening layer, which accepts the input features from the second pooling layer, flattens the multidimensional features output by the second pooling layer into one-dimensional data, and passes it to the fully connected layer;

[0139] A fully connected layer is added to receive input features from the flattened layer. The fully connected layer integrates the flattened one-dimensional data and completes the prediction through nonlinear combination.

[0140] Add an output layer to provide continuous value predictions, outputting predicted CH4 and N2O emissions to support greenhouse gas emission predictions.

[0141] It should be noted that the model is trained using supervised learning methods based on the data information database and the generated data.

[0142] After training, the model's hyperparameters are tuned using a validation set, and the model's prediction accuracy on unknown data is evaluated using a test set to ensure the model's generalization performance. This completes the model construction.

[0143] The convolutional neural network model of this invention can accurately predict the generation of CH4 and N2O based on online monitoring data and historical data. Furthermore, it can achieve continuous value prediction through its own output layer, making the model applicable to continuous value regression tasks. This provides quantitative prediction of greenhouse gas emissions in wastewater treatment, enabling managers to monitor and control emissions in real time, optimize process conditions, reduce greenhouse gas emissions, and provide data support and technical reference for greenhouse gas emission control in wastewater treatment processes.

[0144] In addition, refer to Figure 2 The present invention also proposes a greenhouse gas quantification system for a wastewater system, comprising:

[0145] The data benchmark module is used to establish a data information database based on the process characteristics of each process unit in the wastewater treatment system.

[0146] The unit classification module is used to determine the wastewater substrate parameters of each process unit based on the data information database, and to establish a reaction kinetic model of each process unit with respect to greenhouse gases, and to identify the process unit that generates greenhouse gases as the operating process unit.

[0147] The data monitoring module is used to monitor each operational process unit and acquire data on the amount of greenhouse gases generated in each operational process unit.

[0148] The model building module is used to create a dataset based on the data information database and the generated data, and to build and train a neural network model based on the dataset.

[0149] The model calculation module is used to input the substrate parameters of the wastewater to be tested into the neural network model and solve for the predicted amount of greenhouse gas production.

[0150] The dataset includes a training set, a test set, and a validation set, with a ratio of 80-90% : 5-10% : 5-10%.

[0151] The neural network model adopts a convolutional neural network model, which includes an input layer, a first convolutional layer, an activation function, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer, and an output layer.

[0152] The kernel size of both the first and second convolutional layers is 3×1.

[0153] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for quantifying greenhouse gases in a wastewater system, characterized in that, Includes the following steps: A database of information was established based on the process characteristics of each process unit in the wastewater treatment system. Based on the data information database, the wastewater substrate parameters of each process unit are determined, and a reaction kinetic model of greenhouse gases for each process unit is established to identify the process unit that generates greenhouse gases as the operating process unit. Monitor each operational process unit and obtain data on the amount of greenhouse gases generated in each operational process unit; A dataset is established based on the data information database and the generated data. A neural network model is then constructed and trained based on the dataset. Input the substrate parameters of the wastewater to be tested into the neural network model, and solve to obtain the predicted amount of greenhouse gas production.

2. The method for quantifying greenhouse gases in a wastewater system according to claim 1, characterized in that, The data information repository is a relational database.

3. The method for quantifying greenhouse gases in a wastewater system according to claim 2, characterized in that, The data information database includes several related tables, which are linked together by the process number as a key field.

4. The method for quantifying greenhouse gases in a wastewater system according to claim 1, characterized in that, The reaction kinetics model was established using the Monod model. The specific steps for identifying the process unit that generates greenhouse gases as the operating process unit are as follows: The wastewater substrate parameters for each process unit are determined based on the data information database; Establish Monod equations for greenhouse gases for each process unit and obtain the solution values; A process unit whose solution value is greater than 0 is determined as a working process unit.

5. The method for quantifying greenhouse gases in a wastewater system according to claim 1, characterized in that, The dataset includes a training set, a test set, and a validation set, with a training set:test set:validation set ratio of 80-90%:5-10%:5-10%.

6. The method for quantifying greenhouse gases in a wastewater system according to claim 5, characterized in that, The neural network model is a convolutional neural network model, which includes an input layer, a first convolutional layer, an activation function, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer, and an output layer. The kernel size of both the first and second convolutional layers is 3×1.

7. A greenhouse gas quantification system for a wastewater system, characterized in that, include: The data benchmark module is used to establish a data information database based on the process characteristics of each process unit in the wastewater treatment system. The unit classification module is used to determine the wastewater substrate parameters of each process unit based on the data information database, and to establish a reaction kinetic model of each process unit with respect to greenhouse gases, and to identify the process unit that generates greenhouse gases as the operating process unit. The data monitoring module is used to monitor each operational process unit and acquire data on the amount of greenhouse gases generated in each operational process unit. The model building module is used to create a dataset based on the data information database and the generated data, and to build and train a neural network model based on the dataset. The model calculation module is used to input the substrate parameters of the wastewater to be tested into the neural network model and solve for the predicted amount of greenhouse gas production.

8. The greenhouse gas quantification system in the wastewater system according to claim 7, characterized in that, The data information repository is a relational database.

9. The greenhouse gas quantification system in the wastewater system according to claim 7, characterized in that, The dataset includes a training set, a test set, and a validation set, with a training set:test set:validation set ratio of 80-90%:5-10%:5-10%.

10. The greenhouse gas quantification system in the wastewater system according to claim 9, characterized in that, The neural network model is a convolutional neural network model, which includes an input layer, a first convolutional layer, an activation function, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer, and an output layer. The kernel size of both the first and second convolutional layers is 3×1.

Images