A system and method for determining the optimal H2S monitoring point in a drainage pipe

Abstract

The present invention relates to the technical field of drainage pipeline monitoring and discloses a method for determining H in a drainage pipeline. 2 The system and method for optimal monitoring point S. The system includes a numerical simulation module, a data processing module, a machine learning module, a point selection module, and a H 2 S mobile monitoring device; the numerical simulation module is used to simulate the H at the sewage level of the drainage pipe in the application scenario 2 S concentration value; the data processing module is used to organize and construct the liquid level H 2 The data set of S concentration simulation value and boundary conditions; the machine learning module is used to use artificial neural network to perform ANN machine learning; the point selection module is used to calculate multiple H 2 S concentration detection value and liquid level H 2 The root mean square error (RMSE) between the predicted values ​​of S concentration. The method provided by the present invention can accurately determine the H concentration in the drainage pipe by combining CFD numerical simulation and artificial neural network machine learning technology. 2 The best monitoring point for S concentration is the selected point that can better reflect the liquid surface H 2 S concentration.

Description

Technical Field

[0001] The present invention relates to the technical field of drainage pipe monitoring, and in particular to a system and method for determining an optimal H2S monitoring point in a drainage pipe. Background Art

[0002] In municipal drainage systems, the concentration of hydrogen sulfide (H2S) has always been a focus of attention. In urban drainage pipes, the corrosiveness of H2S will accelerate the corrosion process of the inner wall of the pipe, shorten the service life of the pipe, increase maintenance costs and the risk of system failure. In addition, the foul smell of H2S will seriously affect the quality of life of residents, especially in high temperature and low wind speed environments, when H2S in the pipeline environment escapes into the ground air, which can easily cause odor pollution problems in the surrounding areas. In view of the serious hazards of H2S gas, many countries and regions have formulated strict emission and monitoring standards. For example, the "Emission Standards for Odor Pollutants" clearly stipulates the maximum allowable emission concentration of hydrogen sulfide and requires regular monitoring of areas prone to escape. Accurate monitoring of H2S concentration changes is an important means of timely warning, avoiding escape and even accidents.

[0003] In order to meet the above policy requirements, H2S monitors are usually installed in the drainage system to monitor the concentration changes of gas in the pipeline in real time. However, how to determine the best location for the monitor to ensure that it can accurately reflect the actual amount of H2S escape is a technical problem.

[0004] Although the H2S monitoring equipment currently available on the market can provide real-time data to a certain extent, its monitoring results often deviate greatly from the actual situation. The main reason is that the monitoring points are improperly selected and fail to accurately capture the diffusion law of H2S gas. Since the issuance of the "Detection Methods for Gas in Urban Drainage Facilities", gas monitoring in drainage pipes has followed its standards, such as the hydrogen sulfide sampling point is located at one-half of the sampling space from bottom to top, and the methane sampling point is located at four-fifths from bottom to top. However, in a complex pipeline flow field, gas diffusion is affected by multiple factors such as liquid level fluctuations and flow rate changes. Fixed points are difficult to adapt to the dynamic changes in H2S concentration distribution in a complex flow field, resulting in insufficient representativeness of monitoring data. This has led to doubts about the reliability of monitoring data, which is not conducive to the maintenance and management of pipeline systems, and it is difficult to meet the requirements of policies and regulations for emission monitoring. Therefore, how to determine the optimal layout location of H2S monitors through scientific methods to ensure the accuracy and representativeness of monitoring data has become a technical problem that needs to be solved urgently.

[0005] In order to meet this challenge, researchers used CFD (computational fluid dynamics) numerical simulation to model the drainage pipe fluid and study the diffusion process of H2S in the drainage pipe. CFD simulation can reproduce complex flow fields and gas diffusion processes on the computer, thus providing a theoretical basis for the selection of H2S monitoring points. However, relying solely on numerical simulation is relatively complicated and still has certain limitations. Summary of the invention

[0006] The purpose of the present invention is to provide a system and method for determining the optimal monitoring point of H2S in a drainage pipe based on the above technical problem.

[0007] In order to achieve the above-mentioned object, a first aspect of the present invention provides a system for determining the optimal monitoring point of H2S in a drainage pipe, the system comprising a numerical simulation module, a data processing module, a machine learning module, a point selection module, and an H2S mobile monitoring device;

[0008] The numerical simulation module is used to simulate the H2S concentration value at the sewage surface of the drainage pipe under the application scenario; the data processing module is used to organize and construct a data set including the simulated H2S concentration value and boundary conditions of the liquid surface; the machine learning module is used to use an artificial neural network to perform ANN machine learning, and train the network model by inputting the data set and boundary conditions of the data processing module, and establish a mathematical relationship between the simulated H2S concentration value and the boundary conditions of the liquid surface; the point selection module is used to calculate the root mean square error RMSE between the multiple H2S concentration detection values ​​of each point under the application scenario and the predicted H2S concentration value of the liquid surface, and select the point with the minimum RMSE value as the optimal monitoring point;

[0009] The H2S mobile monitoring device comprises an H2S monitoring and control component 1, an intelligent monitoring and mobile component 2, and a grille protection net 3;

[0010] The numerical simulation module includes a boundary condition setting unit, a flow field simulation unit, a concentration distribution calculation unit, and a simulation result output unit;

[0011] The boundary condition setting unit is used to set the boundary condition parameters in the drainage pipe; the flow field simulation unit is used to simulate the flow field in the drainage pipe using CFD simulation software and generate H2S concentration distribution data in the area; the concentration distribution calculation unit is used to report the simulated value of H2S concentration on the liquid surface in the pipe; the simulation result output unit is used to output the simulated value of H2S concentration on the liquid surface to the data processing module for machine learning analysis and point selection.

[0012] Further preferably, the boundary condition parameters include actual H2S production rate, flow rate, pipeline inner wall friction coefficient, gas-liquid interaction, ambient temperature, and air pressure.

[0013] Preferably, the data processing module includes a data set construction unit, a data set division unit, a data preprocessing unit, and a data storage unit;

[0014] The data set construction unit is used to obtain the simulated value of the liquid surface H2S concentration from the numerical simulation module, and in combination with the boundary conditions in the actual scene, construct a data set containing the simulated value of the liquid surface H2S concentration and the boundary conditions in different scenes; the data set division unit is used to divide the data set into a training set, a validation set, and a test set, so as to perform model training and performance evaluation in the machine learning module; the data preprocessing unit is used to preprocess the data set, including data cleaning, normalization and other steps, so as to improve the training effect and prediction accuracy of the machine learning model; the data storage unit is used to store the data set preprocessed by the data preprocessing unit, so as to be used in the machine learning module and the point selection module later.

[0015] Preferably, the machine learning module includes a model building unit, a model training unit, a model verification unit, a model testing unit, and a model storage unit; the model building unit is used to build an ANN model based on a training set generated by the data processing module; the model training unit is used to train the artificial neural network using the training set, using forward propagation and back propagation algorithms, and iteratively optimizing the weights of the model, selecting the mean square error MSE as the minimization loss function, and optimizing the performance of the model by minimizing the MSE value to improve the prediction accuracy of the model;

[0016] The calculation formula of the mean square error MSE is as follows: Among them, y k is the actual value of the kth sample, x k is the predicted value of the kth sample, and m is the number of samples.

[0017] Preferably, the model verification unit is used to apply the neural network model to the verification set in each training iteration process, calculate the MSE value of the neural network model on the verification set, and judge whether the neural network model is overfitting or underfitting according to the change of the MSE value on the verification set. If it is found that the error of the neural network model on the verification set begins to increase, that is, overfitting, the training is stopped or the hyperparameters of the training process are adjusted to prevent the model from performing well on the training data but not performing well in actual applications. According to the verification results, the number of neural network layers and the number of neurons are increased or decreased to optimize the model performance and ensure that the model has good generalization ability. The model testing unit is used to evaluate the finally trained model on the test set to ensure that it has good generalization ability, that is, the model can handle various unknown scenarios in actual applications and accurately predict the H2S concentration on the liquid surface.

[0018] More preferably, the model storage unit is used to store the trained neural network model. After the training and testing are completed, the best model finally selected is stored in the form of a file or database to ensure the persistence and reusability of the model, and manage different versions of the model, including recording the training date of each model, the usage of the training set and the validation set, the hyperparameter setting, and the test set evaluation results, so as to have a basis for future review or update. When actually selecting the monitoring points, the stored model can be quickly loaded and applied, and its prediction ability can be directly used to select the best H2S monitoring point to ensure the efficiency and accuracy of the system operation.

[0019] Preferably, in the model building unit of the machine learning module, the ANN model includes several hidden layers, and the number of neurons and the activation function of each hidden layer are set according to actual needs so as to capture the mathematical relationship between the simulated value of the liquid surface H2S concentration and the boundary conditions; the mathematical relationship is to train the ANN model through a large number of data sets, and finally to achieve the input boundary conditions, so that the ANN model can output the effect of the simulated value of the liquid surface concentration.

[0020] Preferably, the point selection module includes a data input unit, an RMSE calculation unit, a point determination unit, and a result output unit;

[0021] The data input unit is used to receive the model trained by the machine learning module and receive the boundary conditions obtained by actual monitoring to calculate the RMSE value;

[0022] The RMSE calculation unit is used to calculate the root mean square error RMSE between the H2S concentration monitoring values ​​at different points and the liquid surface H2S concentration simulation values ​​under specific boundary conditions;

[0023] The point determination unit is used to select the point with the smallest RMSE value as the best monitoring point according to the calculation result of the root mean square error RMSE, so as to ensure that the H2S concentration monitoring value of the point can accurately reflect the actual liquid surface H2S concentration;

[0024] The result output unit is used to output the analysis result of the selected optimal monitoring point, so as to determine the layout position of the H2S monitoring probe in practical application.

[0025] According to a particularly preferred embodiment, in the H2S mobile monitoring device, the H2S monitoring and control component 1 is fixed to the pipe wall of the inspection well by a fixing column 11, and the H2S monitoring and control component 1 includes a power supply 12, a spring reel 13, and an H2S monitoring control device 14;

[0026] The intelligent monitoring and moving component 2 is fixed to the wall of the inspection well by screws 21, and the intelligent monitoring and moving component 2 comprises a vertical pipe 22 fixed to the wall of the inspection well, a movable circular pipe 23, an H2S concentration monitoring probe 25 and a water level monitoring meter 26 fixed to the circular pipe 23; and a moving controller 24 for vertically moving and rotating the circular pipe is provided at the junction of the vertical pipe 22 and the circular pipe 23;

[0027] The grille protection net 3 is arranged below the manhole cover of the inspection manhole.

[0028] A second aspect of the present invention provides a method for determining an optimal H2S monitoring point in a drainage pipe, the method comprising the following steps:

[0029] (1) Modeling the gas phase flow in the drainage system pipe to be studied using CFD software, setting parameters according to the boundary condition I obtained from actual monitoring, and setting the parameter distribution form to a normal distribution, performing random sampling using the Monte Carlo method, and generating a new set of boundary condition combinations each time the sampling is performed. After multiple samplings, the obtained boundary condition combinations are input into the Fluent software, and the fluid movement in the drainage system pipe is simulated using Fluent, and the numerical simulation module is used for calculation to obtain the liquid surface H2S concentration simulation value, and a data processing module is used to construct a data set including the liquid surface H2S concentration simulation value and the boundary condition I parameters obtained from actual monitoring;

[0030] (2) using a data processing module to organize the above data set into a training set, a validation set, and a test set, and perform normalization processing; with the support of the data processing module, in a machine learning module, using an artificial neural network to perform machine learning, construct a neural network model, and train it through forward propagation and back propagation algorithms to obtain a trained neural network model;

[0031] (3) Inputting the boundary condition II obtained from actual monitoring into the trained neural network model, calculating the predicted value of H2S concentration on the liquid surface, and stratifying the inspection wells according to the accuracy limit of the H2S probe, and using the point selection module to select different points, and using the H2S mobile monitoring device to monitor the H2S concentration value on each layer in the inspection well to obtain the H2S concentration monitoring values ​​at different points; calculating the root mean square error RMSE between the multiple H2S concentration monitoring values ​​at the different points and the predicted value of H2S concentration on the liquid surface, and performing RMSE analysis, and selecting the point with the smallest RMSE value as the best monitoring point;

[0032] Among them, the system described in the first aspect of the present invention includes the numerical simulation module, the data processing module, the machine learning module, the point selection module, and the H2S mobile monitoring device.

[0033] Preferably, in step (1), the parameters include actual H2S production rate, flow rate, pipeline inner wall friction coefficient, ambient temperature, and flow rate.

[0034] Preferably, in step (2), the training is to train the neural network model using a training set, and evaluate the performance of the neural network model through a validation set, ultimately determining the optimal structure and parameters of the neural network model, and finally testing the neural network model using a test set to ensure its prediction accuracy in practical applications.

[0035] Further preferably, in step (3), the calculation formula of the root mean square error RMSE is as follows: Where C0 is the predicted value of H2S concentration at the liquid surface, C i is the i-th H2S concentration monitoring value of the point in a day, and n is the number of monitoring times of the point in a day.

[0036] The method provided by the present invention has at least the following beneficial effects:

[0037] (1) The method provided by the present invention uses CFD numerical simulation to input boundary conditions under different conditions in the pipe section, such as H2S production, sewage flow, wind speed, temperature and air pressure, to obtain the simulated value of H2S concentration on the liquid surface in the pipe section. Based on machine learning, mathematical modeling of the H2S concentration value on the liquid surface and the boundary conditions is performed to obtain the point with the least error between the simulated H2S concentration on the liquid surface and the H2S monitoring concentration at the monitoring point under different application scenarios, so as to realize accurate monitoring of the H2S generation inside the pipeline.

[0038] (2) The method provided by the present invention can accurately determine the optimal monitoring point for H2S gas concentration in the drainage pipe by combining CFD numerical simulation and artificial neural network machine learning technology. The selected point can better reflect the concentration of H2S on the liquid surface, thereby improving the accuracy and reliability of the monitoring data. Traditional methods usually use fixed standards to select monitoring points, lacking comprehensive consideration of the distribution of H2S concentration in complex flow environments. The present invention provides a scientific point selection method through systematic simulation and analysis, making the layout of monitoring probes more reasonable and avoiding blind deployment and waste of resources. Since the present invention can accurately determine the optimal monitoring point, the cost of frequent maintenance and replacement of probes due to incorrect point selection is reduced.

[0039] (3) The optimized monitoring scheme in the method provided by the present invention can warn of corrosion and odor problems more promptly, thereby reducing the overall maintenance cost of the system; through the application of artificial neural networks, the present invention establishes a mathematical relationship between the predicted value of H2S concentration on the liquid surface and the boundary conditions, so that the system can effectively measure the H2S concentration and its changing trend in different scenarios; this capability has significant advantages in the management of complex drainage systems; the method of the present invention is suitable for different types of drainage systems and a variety of complex scenarios, and has broad application prospects; whether in an intercepting combined system, a diversion system or other types of drainage systems, the method can be flexibly applied and provide accurate monitoring solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of the present invention for determining the optimal monitoring point of H2S in a drainage pipe;

[0041] Figure 2 It is a schematic diagram of the module composition of the system for determining the optimal monitoring point of H2S in the drainage pipe of the present invention;

[0042] Figure 3 is a flow chart of using an artificial neural network to perform ANN machine learning in step (2) of an embodiment of the present invention;

[0043] Figure 4 is a schematic diagram of a H2S mobile monitoring device in step (3) of an embodiment of the present invention;

[0044] Figure 5 is a schematic diagram of the H2S monitoring and control component 1 in the H2S mobile monitoring device in step (3) of an embodiment of the present invention;

[0045] Figure 6 It is a schematic diagram of the intelligent monitoring and mobile component 2 in the H2S mobile monitoring device in step (3) of an embodiment of the present invention.

[0046] Description of Reference Numerals

[0047] 1. H2S monitoring and control components 2. Intelligent monitoring and mobile components 3. Grille protection net

[0048] 11. Fixed column 12. Power supply 13. Spring reel

[0049] 14. H2S monitoring and control device 21. Screw 22. Vertical pipe

[0050] 23. Circular pipe 24. Mobile controller 25. H2S concentration monitoring probe

[0051] 26. Water level monitoring meter DETAILED DESCRIPTION

[0052] The endpoints and any values ​​of the ranges disclosed in this article are not limited to the precise ranges or values, and these ranges or values ​​should be understood to include values ​​close to these ranges or values. For numerical ranges, the endpoint values ​​of each range, the endpoint values ​​of each range and the individual point values, and the individual point values ​​can be combined with each other to obtain one or more new numerical ranges, which should be considered as specifically disclosed in this article.

[0053] In the present invention, the boundary conditions for predicting the H2S concentration value in the point selection module are selected according to the field conditions, and are recorded as the boundary conditions II obtained by actual monitoring. If the boundary conditions II obtained by actual monitoring fluctuate slightly, the predicted value of the liquid surface H2S concentration from a single boundary condition monitoring is selected as the overall predicted value; if the boundary conditions II obtained by actual monitoring fluctuate greatly, the average value of the predicted values ​​of the liquid surface H2S concentration from multiple boundary condition monitorings is taken as the overall predicted value.

[0054] As mentioned above, the first aspect of the present invention provides a system for determining the optimal monitoring point of H2S in a drainage pipe, the system comprising a numerical simulation module, a data processing module, a machine learning module, a point selection module, and an H2S mobile monitoring device;

[0055] The numerical simulation module is used to simulate the H2S concentration value at the sewage surface of the drainage pipe under the application scenario; the data processing module is used to organize and construct a data set including the simulated H2S concentration value and boundary conditions of the liquid surface; the machine learning module is used to use an artificial neural network to perform ANN machine learning, and train the network model by inputting the data set and boundary conditions of the data processing module, and establish a mathematical relationship between the simulated H2S concentration value and the boundary conditions of the liquid surface; the point selection module is used to calculate the root mean square error RMSE between the multiple H2S concentration detection values ​​of each point under the application scenario and the predicted H2S concentration value of the liquid surface, and select the point with the minimum RMSE value as the optimal monitoring point;

[0056] The H2S mobile monitoring device comprises an H2S monitoring and control component 1, an intelligent monitoring and mobile component 2, and a grille protection net 3;

[0057] The numerical simulation module includes a boundary condition setting unit, a flow field simulation unit, a concentration distribution calculation unit, and a simulation result output unit;

[0058] The boundary condition setting unit is used to set the boundary condition parameters in the drainage pipe; the flow field simulation unit is used to simulate the flow field in the drainage pipe using CFD simulation software and generate H2S concentration distribution data in the area; the concentration distribution calculation unit is used to report the simulated value of H2S concentration on the liquid surface in the pipe; the simulation result output unit is used to output the simulated value of H2S concentration on the liquid surface to the data processing module for machine learning analysis and point selection.

[0059] Further preferably, the boundary condition parameters include actual H2S production rate, flow rate, pipeline inner wall friction coefficient, gas-liquid interaction, ambient temperature, and air pressure.

[0060] Further preferably, the data processing module includes a data set construction unit, a data set division unit, a data preprocessing unit, and a data storage unit;

[0061] The data set construction unit is used to obtain the simulated value of the liquid surface H2S concentration from the numerical simulation module, and in combination with the boundary conditions in the actual scene, construct a data set containing the simulated value of the liquid surface H2S concentration and the boundary conditions in different scenes; the data set division unit is used to divide the data set into a training set, a validation set, and a test set, so as to perform model training and performance evaluation in the machine learning module; the data preprocessing unit is used to preprocess the data set, including data cleaning, normalization and other steps, so as to improve the training effect and prediction accuracy of the machine learning model; the data storage unit is used to store the data set preprocessed by the data preprocessing unit, so as to be used in the machine learning module and the point selection module later.

[0062] Preferably, the machine learning module includes a model building unit, a model training unit, a model verification unit, a model testing unit, and a model storage unit; the model building unit is used to build an ANN model based on a training set generated by the data processing module; the model training unit is used to train the artificial neural network using the training set, using forward propagation and back propagation algorithms, and iteratively optimizing the weights of the model, selecting the mean square error MSE as the minimization loss function, and optimizing the performance of the model by minimizing the MSE value to improve the prediction accuracy of the model;

[0063] The calculation formula of the mean square error MSE is as follows: Among them, y k is the actual value of the kth sample, x k is the predicted value of the kth sample, and m is the number of samples.

[0064] Preferably, the model verification unit is used to apply the neural network model to the verification set in each training iteration process, calculate the MSE value of the neural network model on the verification set, and judge whether the neural network model is overfitting or underfitting according to the change of the MSE value on the verification set. If it is found that the error of the neural network model on the verification set begins to increase, that is, overfitting, the training is stopped or the hyperparameters of the training process are adjusted to prevent the model from performing well on the training data but not performing well in actual applications. According to the verification results, the number of neural network layers and the number of neurons are increased or decreased to optimize the model performance and ensure that the model has good generalization ability. The model testing unit is used to evaluate the finally trained model on the test set to ensure that it has good generalization ability, that is, the model can handle various unknown scenarios in actual applications and accurately predict the H2S concentration on the liquid surface.

[0065] Preferably, the model storage unit is used to store the trained neural network model. After the training and testing are completed, the best model finally selected is stored in the form of a file or database to ensure the durability and reusability of the model, and manage different versions of the model, including recording the training date of each model, the usage of the training set and the validation set, the hyperparameter setting, and the test set evaluation results, so that there is a basis for future review or update. When actually selecting monitoring points, the stored model can be quickly loaded and applied, and its prediction ability can be directly used to select the best H2S monitoring point to ensure the efficiency and accuracy of the system operation.

[0066] Preferably, in the model building unit of the machine learning module, the ANN model includes several hidden layers, and the number of neurons and the activation function of each hidden layer are set according to actual needs so as to capture the mathematical relationship between the simulated value of the liquid surface H2S concentration and the boundary conditions; the mathematical relationship is to train the ANN model through a large number of data sets, and finally to achieve the input boundary conditions, so that the ANN model can output the effect of the simulated value of the liquid surface concentration.

[0067] Further preferably, the point selection module includes a data input unit, an RMSE calculation unit, a point determination unit, and a result output unit;

[0068] The data input unit is used to receive the model trained by the machine learning module and receive the boundary conditions obtained by actual monitoring to calculate the RMSE value;

[0069] The RMSE calculation unit is used to calculate the root mean square error RMSE between multiple H2S concentration monitoring values ​​at different points and the liquid surface H2S concentration simulation value under specific boundary conditions;

[0070] The point determination unit is used to select the point with the smallest RMSE value as the best monitoring point according to the calculation result of the root mean square error RMSE, so as to ensure that the H2S concentration monitoring value of the point can accurately reflect the actual liquid surface H2S concentration;

[0071] The result output unit is used to output the analysis result of the selected optimal monitoring point, so as to determine the layout position of the H2S monitoring probe in practical application.

[0072] Preferably, in the H2S mobile monitoring device, the H2S monitoring and control component 1 is fixed to the pipe wall of the inspection well through a fixing column 11, and the H2S monitoring and control component 1 includes a power supply 12, a spring reel 13, and an H2S monitoring control device 14;

[0073] The intelligent monitoring and moving component 2 is fixed to the wall of the inspection well by screws 21, and the intelligent monitoring and moving component 2 comprises a vertical pipe 22 fixed to the wall of the inspection well, a movable circular pipe 23, an H2S concentration monitoring probe 25 and a water level monitoring meter 26 fixed to the circular pipe 23; and a moving controller 24 for vertically moving and rotating the circular pipe is provided at the junction of the vertical pipe 22 and the circular pipe 23;

[0074] The grille protection net 3 is arranged below the manhole cover of the inspection manhole.

[0075] In the present invention, in the H2S mobile monitoring device, the circular pipe 23 in the intelligent monitoring and mobile component 2 can realize self-spinning through the mobile controller 24, and can move up and down along the vertical pipe 22, and the H2S concentration at different heights and different positions in the inspection well is monitored by the H2S concentration monitoring probe 25 fixed on the circular pipe 23, and the height of the inspection well is layered according to the accuracy limit of the H2S concentration monitoring probe 25. When the circular pipe 23 moves on the vertical pipe 22, it stops accurately according to the preset layering height, and spins at each height point to measure the H2S concentration at different positions at different heights. The water level monitoring meter 26 is used to accurately record the position of the probe and correspond to the concentration information at the position. At the same time, when the water level monitoring meter 26 in the intelligent monitoring and mobile component 2 detects that the circular pipe 23 is too close to the liquid surface, the water level monitoring meter 26 will trigger the mobile controller 24 to automatically stop detection, and let the circular pipe 23 rise, so as to avoid the H2S concentration monitoring probe 25 from contacting the liquid surface, thereby ensuring the monitoring accuracy and equipment safety.

[0076] In the present invention, in the H2S mobile monitoring device, the H2S monitoring and control component 1 is used for the power supply of the device and the arrangement of the power line and the signal transmission line, and the spring reel 13 is used to maintain the orderly management of the power line and the signal line, ensuring that the cable is stable when the probe moves on the vertical pipeline without being entangled or pulled, thereby ensuring the smooth operation of the equipment and the stability of data transmission.

[0077] In the present invention, in the H2S mobile monitoring device, the grille protection net 3 is used to prevent road garbage from entering the inspection well, ensure the cleanliness of the drainage system operating environment, ensure the long-term stable operation of the equipment, and avoid the influence of external factors on the monitoring equipment.

[0078] As mentioned above, the second aspect of the present invention provides a method for determining the optimal monitoring point of H2S in a drainage pipe, the method comprising the following steps:

[0079] Use CFD software to model the gas phase flow in the drainage system pipe to be studied, set parameters according to the boundary condition I obtained from actual monitoring, and set the parameter distribution form to normal distribution. Perform random sampling through the Monte Carlo method. Each sampling will generate a new set of boundary condition combinations. After multiple samplings, input the obtained boundary condition combinations into the Fluent software, and simulate the fluid movement in the drainage system pipe through Fluent. Use the numerical simulation module to calculate and obtain the liquid surface H2S concentration simulation value, and use the data processing module to construct a data set including the liquid surface H2S concentration simulation value and the boundary condition I parameters obtained from actual monitoring;

[0080] (2) using a data processing module to organize the above data set into a training set, a validation set, and a test set, and perform normalization processing; with the support of the data processing module, in a machine learning module, using an artificial neural network to perform machine learning, construct a neural network model, and train it through forward propagation and back propagation algorithms to obtain a trained neural network model;

[0081] (3) Inputting the boundary condition II obtained from actual monitoring into the trained neural network model, calculating the predicted value of H2S concentration on the liquid surface, and stratifying the inspection wells according to the accuracy limit of the H2S probe, and selecting different points using the point selection module, and using the H2S mobile monitoring device to monitor the H2S concentration value on each layer in the inspection well, to obtain the H2S concentration monitoring values ​​at different points; calculating the root mean square error RMSE between the H2S concentration monitoring values ​​of the points for multiple times on the same day and the predicted value of H2S concentration on the liquid surface, and performing RMSE analysis, and selecting the point with the smallest RMSE value as the best monitoring point;

[0082] Among them, the system described in the first aspect of the present invention includes the numerical simulation module, the data processing module, the machine learning module, the point selection module, and the H2S mobile monitoring device.

[0083] Preferably, in step (1), the parameters include actual H2S production rate, flow rate, pipeline inner wall friction coefficient, ambient temperature, and flow rate.

[0084] Preferably, in step (2), the training is to train the neural network model using a training set, and evaluate the performance of the neural network model through a validation set, ultimately determining the optimal structure and parameters of the neural network model, and finally testing the neural network model using a test set to ensure its prediction accuracy in practical applications.

[0085] Further preferably, in step (3), the calculation formula of the root mean square error RMSE is as follows: Where C0 is the predicted value of H2S concentration at the liquid surface, C i is the i-th H2S concentration monitoring value of the point in a day, and n is the number of monitoring times of the point in a day.

[0086] The present invention will be described in detail below by way of examples. In the following examples, unless otherwise specified, the raw materials are all commercially available.

[0087] In the following examples, unless otherwise specified, in a certain intercepting combined sewer system A in the central urban area of ​​Shanghai, domestic sewage and rainwater are discharged through the same pipe, and are composed of a main sewage pipe along the X-axis direction, and two rainwater pipes A and rainwater pipe B branches distributed on both sides and along the Z-axis direction, wherein the main sewage pipe is about 69m long, about 1m in diameter, and has a slope of about 0.001; the rainwater pipe A is about 8.5m long, about 0.3m in diameter, and has a slope of about 0.05; the rainwater pipe B is about 14m long, about 0.3m in diameter, and has a slope of about 0.03; the inspection well is 5m deep and 1.5m in diameter, and the nodes are distributed at the inlet and outlet of the main sewage pipe, and the inlets of the two rainwater pipes.

[0088] In the following examples, unless otherwise specified, the CFD software model is ANSYS Fluent 2024R1.

[0089] In the following examples, unless otherwise specified, the smaller the RMSE calculation result, the smaller the difference between the actual H2S concentration monitoring value at the point and the H2S concentration value on the liquid surface. If the RMSE is close to or equal to 0, it means that the H2S concentration at the monitoring point can best reflect the actual H2S concentration on the liquid surface.

[0090] Example:

[0091] This example is used to provide a combination Figure 1 The method for determining the optimal monitoring point of H2S in a drainage pipe according to the present invention is described, and the method comprises the following steps:

[0092] (1) An intercepting combined sewer system A in the central urban area of ​​Shanghai was selected as the research object. The gas phase flow in the pipeline of the drainage system A was modeled using CFD software. According to the boundary conditions I obtained from the actual monitoring of the drainage system A, the flow rate was set to 408.96-575.99 l / s, the flow velocity was set to 0.98-1.14 m / s, the friction coefficient of the inner wall of the pipeline was set to 0.011, the ambient temperature was set to 285-291 K, and the actual H2S yield was set to 1×10 -7 ~2×10 -7 kg / m 2 / s, the distribution form of the above parameters is set to normal distribution, and random sampling is performed through the Monte Carlo method. Each sampling will generate a new set of boundary condition combinations. After multiple samplings, the obtained boundary condition combinations are input into the Fluent software, and the fluid movement in the drainage system pipe is simulated by Fluent. The numerical simulation module is used to perform calculations and sort the results to obtain the simulated value of the H2S concentration on the liquid surface. The data processing module is used to construct a data set including the simulated value of the H2S concentration on the liquid surface and the boundary condition I obtained by actual monitoring, and the data combination, the boundary condition I obtained by actual monitoring, and the simulated H2S concentration on the liquid surface are sorted.

[0093] (2) The data processing module is used to organize the above data sets into training sets, validation sets and test sets, and the above training sets, validation sets and test sets are normalized. With the support of the data processing module, artificial neural networks are used in the machine learning module to perform ANN machine learning (the flowchart is shown in FIG. Figure 3 As shown), a neural network model including several hidden layers is constructed (the number of neurons in each hidden layer is determined according to actual needs), and a trained neural network model is obtained by training through forward propagation and back propagation algorithms;

[0094] The specific content of the training is: using the training set to train the neural network model, and evaluating the performance of the neural network model through the validation set, finally determining the optimal structure and parameters of the neural network model, and using the test set to perform a final test on the neural network model to ensure its prediction accuracy in practical applications.

[0095] (3) The boundary condition II obtained by actual monitoring is input into the trained neural network model to calculate the predicted value of H2S concentration on the liquid surface (due to the small change range of boundary conditions in the pipeline within a day and considering the convenience and feasibility, the predicted value under the monitoring time is used as the overall predicted value of the day). -3 ppm), the inspection well is divided into 15 layers, each layer is 20 cm high, and the point selection module is used to select different points, and the H2S concentration value on each layer in the inspection well is monitored using the H2S mobile monitoring device, which is the H2S concentration monitoring value at different points;

[0096] The root mean square error (RMSE) of the H2S concentration monitoring values ​​at the different points and the predicted value of the H2S concentration at the liquid surface is calculated, and an RMSE analysis is performed to evaluate whether the point can accurately reflect the H2S concentration value at the liquid surface; wherein the calculation formula of the root mean square error (RMSE) is as follows: Where: C0 is the predicted value of H2S concentration at the liquid level, C i is the i-th H2S concentration monitoring value in a day, n is the number of monitoring times in a day, in this embodiment, monitoring is performed every 15 minutes, n=96;

[0097] Based on the above analysis results, the monitoring point 50 cm below the inspection well mouth is determined to be the best point, and the RMSE between the H2S concentration and the liquid surface concentration at this point is the lowest 0.

[0098] In order to more clearly express the method of the present invention, the present invention adopts Figure 2 The schematic diagram of the module composition of the system for determining the optimal monitoring point of H2S in the drainage pipe is shown. Figure 4 A schematic diagram of the H2S mobile monitoring device in step (3) of the embodiment is shown. Figure 5 The schematic diagram of the H2S monitoring and control component 1 in the H2S mobile monitoring device in step (3) of the embodiment of the present invention is shown. Figure 6 A schematic diagram of the intelligent monitoring and moving component 2 in the H2S mobile monitoring device in step (3) of an embodiment of the present invention is shown.

[0099] Depend on Figure 2It can be seen that the system for determining the optimal monitoring point of H2S in the drainage pipe of the present invention includes a numerical simulation module, a data processing module, a machine learning module, and a point selection module, wherein the numerical simulation module includes a boundary condition setting unit, a flow field simulation unit, a concentration distribution calculation unit, and a simulation result output unit; the data processing module includes a data set construction unit, a data set division unit, a data preprocessing unit, and a data storage unit; the machine learning module includes a model construction unit, a model training unit, a model verification unit, a model testing unit, and a model storage unit; the point selection module includes a data input unit, an RMSE calculation unit, a point determination unit, and a result output unit.

[0100] Depend on Figure 4 It can be seen that the H2S mobile monitoring device includes an H2S monitoring and control component 1, an intelligent monitoring and mobile component 2, and a grille protection net 3.

[0101] Depend on Figure 5 It can be seen that the H2S monitoring and control component 1 is fixed on the pipe wall of the inspection well through a fixing column 11, and the H2S monitoring and control component 1 includes a power supply 12, a spring reel 13, and an H2S monitoring control device 14.

[0102] Depend on Figure 6 It can be seen that the intelligent monitoring and moving component 2 is fixed to the wall of the inspection well by screws 21, and the intelligent monitoring and moving component 2 includes a vertical pipe 22 fixed on the wall of the inspection well, a movable circular pipe 23, an H2S concentration monitoring probe 25 and a water level monitoring meter 26 fixed on the circular pipe 23; and a mobile controller 24 for vertically moving and rotating the circular pipe is provided at the junction of the vertical pipe 22 and the circular pipe 23.

[0103] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited thereto. Within the technical concept of the present invention, the technical solution of the present invention can be subjected to a variety of simple modifications, including the combination of various technical features in any other suitable manner, and these simple modifications and combinations should also be regarded as the contents disclosed by the present invention and belong to the protection scope of the present invention.

Claims

1. A system for determining the optimal H2S monitoring point in a drainage pipe, characterized in that: The system includes numerical simulation module, data processing module, machine learning module, point selection module, and H2S mobile monitoring device; The numerical simulation module is used to simulate the H2S concentration value at the sewage surface of the drainage pipe under the application scenario; the data processing module is used to organize and construct a data set including the simulated H2S concentration value and boundary conditions of the liquid surface; the machine learning module is used to use an artificial neural network to perform ANN machine learning, and train the network model by inputting the data set and boundary conditions of the data processing module, and establish a mathematical relationship between the simulated H2S concentration value and the boundary conditions of the liquid surface; the point selection module is used to calculate the root mean square error RMSE between multiple H2S concentration detection values ​​of each point under the application scenario and the predicted H2S concentration value of the liquid surface, and select the point with the minimum RMSE value as the optimal monitoring point; The H2S mobile monitoring device comprises an H2S monitoring and control component (1), an intelligent monitoring and mobile component (2), and a grille protection net (3); The numerical simulation module includes a boundary condition setting unit, a flow field simulation unit, a concentration distribution calculation unit, and a simulation result output unit; The boundary condition setting unit is used to set the boundary condition parameters in the drainage pipe; the flow field simulation unit is used to simulate the flow field in the drainage pipe using CFD simulation software and generate H2S concentration distribution data in the area; the concentration distribution calculation unit is used to report the simulated value of H2S concentration on the liquid surface in the pipe; the simulation result output unit is used to output the simulated value of H2S concentration on the liquid surface to the data processing module for machine learning analysis and point selection.

2. The system according to claim 1, wherein: The boundary condition parameters include actual H2S production rate, flow rate, pipeline inner wall friction coefficient, gas-liquid interaction, ambient temperature, and air pressure.

3. The system according to claim 1 or 2, wherein: The data processing module includes a data set construction unit, a data set division unit, a data preprocessing unit, and a data storage unit; The data set construction unit is used to obtain the simulated value of the liquid surface H2S concentration from the numerical simulation module, and in combination with the boundary conditions in the actual scene, construct a data set containing the simulated value of the liquid surface H2S concentration and the boundary conditions in different scenes; the data set division unit is used to divide the data set into a training set, a validation set, and a test set, so as to perform model training and performance evaluation in the machine learning module; the data preprocessing unit is used to preprocess the data set, including data cleaning, normalization and other steps, so as to improve the training effect and prediction accuracy of the machine learning model; the data storage unit is used to store the data set preprocessed by the data preprocessing unit, so as to be used in the machine learning module and the point selection module later.

4. The system according to claim 1 or 2, wherein: The machine learning module includes a model building unit, a model training unit, a model verification unit, a model testing unit, and a model storage unit; the model building unit is used to build an ANN model based on a training set generated by a data processing module; the model training unit is used to train an artificial neural network using the training set, adopting forward propagation and back propagation algorithms, and optimizing the weights of the model through iteration, selecting mean square error MSE as a minimization loss function, and optimizing the performance of the model by minimizing the MSE value to improve the prediction accuracy of the model; The calculation formula of the mean square error MSE is as follows: Among them, y k is the actual value of the kth sample, x k is the predicted value of the kth sample, and m is the number of samples.

5. The system according to claim 4, wherein: The model verification unit is used to apply the neural network model to the verification set in each training iteration process, calculate the MSE value of the neural network model on the verification set, and judge whether the neural network model is overfitting or underfitting according to the change of the MSE value on the verification set. If it is found that the error of the neural network model on the verification set begins to increase, that is, overfitting, the training is stopped or the hyperparameters of the training process are adjusted to prevent the model from performing well on the training data but not performing well in actual applications. According to the verification results, the number of neural network layers and the number of neurons are increased or decreased to optimize the model performance and ensure that the model has good generalization ability. The model testing unit is used to evaluate the finally trained model on the test set to ensure that it has good generalization ability, that is, the model can handle various unknown scenarios in actual applications and accurately predict the H2S concentration on the liquid surface.

6. The system according to claim 4, wherein: The model storage unit is used to store the trained neural network model. After the training and testing are completed, the best model finally selected is stored in the form of a file or database to ensure the persistence and reusability of the model, and manage different versions of the model, including recording the training date of each model, the usage of the training set and the validation set, the hyperparameter setting, and the test set evaluation results, so that there is a basis for future review or update. When actually selecting the monitoring point, the stored model can be quickly loaded and applied, and its prediction ability can be directly used to select the best H2S monitoring point, ensuring the efficiency and accuracy of the system operation; And / or, in the model building unit of the machine learning module, the ANN model includes several hidden layers, and the number of neurons and activation function of each hidden layer are set according to actual needs so as to capture the mathematical relationship between the simulated value of the liquid surface H2S concentration and the boundary conditions; the mathematical relationship is to train the ANN model through a large number of data sets, and finally meet the input boundary conditions, so that the ANN model can output the effect of the liquid surface concentration simulation value.

7. The system according to claim 1 or 2, wherein: The point selection module includes a data input unit, an RMSE calculation unit, a point determination unit, and a result output unit; The data input unit is used to receive the model trained by the machine learning module and the boundary conditions obtained by actual monitoring to calculate the RMSE value; The RMSE calculation unit is used to calculate the root mean square error RMSE between the H2S concentration monitoring values ​​monitored multiple times at different points and the liquid surface H2S concentration simulation value under specific boundary conditions; The point determination unit is used to select the point with the smallest RMSE value as the best monitoring point according to the calculation result of the root mean square error RMSE, so as to ensure that the H2S concentration monitoring value of the point can accurately reflect the actual liquid surface H2S concentration; The result output unit is used to output the analysis result of the selected optimal monitoring point, so as to determine the layout position of the H2S monitoring probe in practical application.

8. The system according to claim 1 or 2, wherein: In the H2S mobile monitoring device, the H2S monitoring and control component (1) is fixed to the pipe wall of the inspection well through a fixing column (11), and the H2S monitoring and control component (1) comprises a power supply (12), a spring reel (13), and an H2S monitoring control device (14); The intelligent monitoring and moving component (2) is fixed to the wall of the inspection well by screws (21), and the intelligent monitoring and moving component (2) comprises a vertical pipe (22) fixed to the wall of the inspection well, a movable circular pipe (23), an H2S concentration monitoring probe (25) and a water level monitoring meter (26) fixed to the circular pipe (23); and a moving controller (24) for vertically moving and rotating the circular pipe is provided at the junction of the vertical pipe (22) and the circular pipe (23); The grille protection net (3) is arranged below the manhole cover of the inspection manhole.

9. A method for determining the optimal H2S monitoring point in a drainage pipe, characterized in that: The method comprises the following steps: (1) Modeling the gas phase flow in the drainage system pipe to be studied using CFD software, setting parameters according to the boundary condition I obtained from actual monitoring, and setting the parameter distribution form to a normal distribution, performing random sampling using the Monte Carlo method, and generating a new set of boundary condition combinations each time the sampling is performed. After multiple samplings, the obtained boundary condition combinations are input into the Fluent software, and the fluid movement in the drainage system pipe is simulated using Fluent, and the numerical simulation module is used for calculation to obtain the liquid surface H2S concentration simulation value, and a data processing module is used to construct a data set including the liquid surface H2S concentration simulation value and the boundary condition I parameters obtained from actual monitoring; (2) using a data processing module to organize the above data set into a training set, a validation set, and a test set, and perform normalization processing; with the support of the data processing module, in a machine learning module, using an artificial neural network to perform machine learning, construct a neural network model, and train it through forward propagation and back propagation algorithms to obtain a trained neural network model; (3) Inputting the boundary condition II obtained from actual monitoring into the trained neural network model, calculating the predicted value of H2S concentration on the liquid surface, and stratifying the inspection wells according to the accuracy limit of the H2S probe, and selecting different points using the point selection module, and using the H2S mobile monitoring device to monitor the H2S concentration value on each layer in the inspection well to obtain the H2S concentration monitoring values ​​at different points; calculating the root mean square error RMSE of the H2S concentration monitoring values ​​at the different points and the predicted value of H2S concentration on the liquid surface, and performing RMSE analysis, and selecting the point with the smallest RMSE value as the best monitoring point; Among them, the system described in any one of claims 1-8 includes the numerical simulation module, the data processing module, the machine learning module, the point selection module, and the H2S mobile monitoring device.

10. The method according to claim 9, wherein: In step (1), the parameters include actual H2S production rate, flow rate, pipeline inner wall friction coefficient, ambient temperature, and flow rate; And / or, in step (2), the training is to train the neural network model using the training set, and evaluate the performance of the neural network model through the validation set, ultimately determine the optimal structure and parameters of the neural network model, and finally test the neural network model using the test set to ensure its prediction accuracy in practical applications; And / or, in step (3), the root mean square error RMSE is calculated as follows: Where C0 is the predicted value of H2S concentration at the liquid surface, C i is the i-th H2S concentration monitoring value of the point in a day, and n is the number of monitoring times of the point in a day.

Images