Automatic monitoring system and distribution prediction method for greenhouse gas concentration in inland water body

Abstract

The embodiment of the application discloses an automatic monitoring system and a distribution prediction method for the concentration of greenhouse gases in inland water bodies. The system comprises: an acidity sensor and a temperature sensor, which are used for collecting the pH value and temperature of water at a monitoring point, respectively; a carbon dioxide sensor, which is used for collecting the concentration of bicarbonate in water at the monitoring point; a methane sensor and a nitrous oxide sensor, which are used for collecting the content of dissolved greenhouse gases CH4 and N2O in water at the monitoring point, respectively; and an operator device, which is used for calculating the concentration of greenhouse gases in water according to the sensor data. The embodiment can accurately monitor the concentration of various greenhouse gases in water under actual conditions.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of gas concentration detection in water, and in particular to an automatic monitoring system, monitoring method, and distribution prediction method for greenhouse gas concentration in inland water bodies. Background Art

[0002] Under the dual influence of global climate change and human activities, the challenges facing the aquatic environment are becoming increasingly severe. Greenhouse gases dissolved in water, such as carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), not only enter the atmosphere through exchange at the air-water interface, enhancing the greenhouse effect and further exacerbating global warming, but also changes in their concentrations can affect the metabolic processes of aquatic plants and microorganisms, thereby altering the structure and function of entire ecosystems. These parameters are crucial for aquatic environments and microbial activity in wetlands, rivers, lakes, reservoirs, and other areas.

[0003] Currently, the quantitative technology for greenhouse gas concentrations in water bodies mostly uses on-site gas collection in closed boxes or headspace extraction of gases. The basic principle is to establish a two-phase equilibrium between the substance being measured in the gas-liquid or gas-solid phase, and then use experimental instruments (such as gas chromatographs) to test and analyze the gas. First, these two technologies not only require specific scientific experimental instruments to detect and quantitatively analyze the gas, but the instruments and equipment are complex and the testing is expensive, making them inconvenient for field operation and unable to achieve in-situ continuous automatic detection. Second, both technologies ignore the impact of wind fields on the gas exchange rate at the water-air interface. Especially for waters with high water flow rates, this will cause a significant difference between the equilibrium gas in the enclosed space and the actual water environment conditions, leading to huge errors in the greenhouse gas concentration detection data in water bodies.

[0004] Patent application CN112255385A discloses a system, method and application for in-situ continuous measurement of greenhouse gas concentrations in highly turbid water bodies. A diaphragm pump is used to pump water through an atomizing nozzle at the top of a balancing chamber to atomize the water and exchange gas with the gas in the balancing chamber. A greenhouse gas meter measures the greenhouse gas concentration in the balancing chamber. Although this overcomes the greenhouse gas measurement technologies in water bodies represented by float tank technology and headspace technology, the extraction of water samples destroys the original water environment state, resulting in huge errors in the measured parameters of water pressure, saturated water vapor pressure and balancing gas in the balancer. In addition, since the greenhouse gas concentration in the water environment changes in real time and over a long period of time, it takes a specific long time to maintain the Henry equilibrium mode of water-gas in the balancing chamber, which makes this technology have certain limitations in the time-frequency domain.

[0005] Patent application CN117347456A discloses an online automatic detection device for the partial pressure of carbon dioxide in inland water bodies and a detection method thereof. The device uses three sensors to perform online detection of the partial pressure of carbon dioxide in water bodies, overcoming the greenhouse gas measurement technologies in water bodies represented by floating tank technology and headspace technology. However, the device still needs to further utilize water environment parameters to quantify the concentration of carbon dioxide in water bodies, and lacks monitoring of the concentrations of two other important greenhouse gases in water bodies - methane and nitrous oxide. These gases have greater global warming potential, which are 25 times and 298 times that of carbon dioxide, respectively. Summary of the Invention

[0006] The embodiments of the present invention provide an automatic monitoring system for greenhouse gas concentration in inland water bodies and a distribution prediction method to solve the above technical problems.

[0007] In a first aspect, an embodiment of the present invention provides an automatic monitoring system for greenhouse gas concentration in inland water bodies, which is applied to inland water bodies, and the system includes:

[0008] The acidity sensor and temperature sensor are used to collect the pH value S1 and temperature S2 of the water at the monitoring point respectively;

[0009] Carbon dioxide sensor, used to collect bicarbonate concentration S in water at the monitoring point CO2 ;

[0010] Methane sensor and nitrous oxide sensor are used to collect the content of CH4 and N2O dissolved in water at the monitoring point respectively. CH4 and S N2O ;

[0011] Operator device for calculating the concentration of greenhouse gases in water by:

[0012] S110. Calculate the solubility K of CO2, CH4 and N2O in surface water according to the following formula: CO2 , K CH4 and K N2O , and Henry's equilibrium constant k for bicarbonate concentration in water h :

[0013]

[0014] Among them, A, B, C, D, E, F, G, H, I, J, K, L, M, N, P, Q, and R are fixed constants that are calibrated or derived;

[0015] S120. Calculate the dissolved CO2, CH4 and N2O concentrations C in the water at the monitoring point according to the following formula: CO2 、C CH4 and C N2O :

[0016] C CO2 = S CO2 × K CO2 × k h

[0017] C CH4 = S CH4 × K CH4

[0018] C N2O = S N2O × K N2O

[0019] In a second aspect, the embodiments of the present application provide a method for predicting the concentration distribution of greenhouse gases in inland water bodies, comprising:

[0020] S210, dividing the water area to be monitored into a plurality of grids;

[0021] S220, installing the above-mentioned automatic monitoring system for the concentration of greenhouse gases in inland water bodies in each grid as a monitoring point, respectively, to monitor the concentration of any one of the greenhouse gases in the water body in each grid in real time; and taking the concentration of each grid at the same time as a sample to construct a limited sample set;

[0022] S230, averaging the concentration distribution monitored in a period of time, taking a plurality of extreme grids and large gradient grids in the average concentration distribution as a plurality of initial grids; and constructing a neural network model with the concentration of the plurality of initial grids at the same time as input and the concentration distribution of all grids at the same time as output;

[0023] S240, training and testing the neural network model using the limited sample set, if the model accuracy does not meet the requirements, clustering the grids with concentration prediction values and monitoring values that differ by more than a set threshold in the test results according to geographical location, and taking each cluster center into the model input;

[0024] S250, returning to S240 according to the new neural network model, and repeating the cycle until the model accuracy meets the requirements;

[0025] S260, using a genetic algorithm to select an optimal input grid combination from the input grid set of the final model that can make the model accuracy meet the requirements and has the least number of grids, and using the optimal model corresponding to the optimal input grid combination to predict the concentration distribution in the water body.

[0026] In summary, the automatic monitoring system for the concentration of greenhouse gases in inland water bodies and the distribution prediction method can achieve the following beneficial effects:

[0027] 1. The existing technology uses observation and sampling, which changes the original environmental conditions. This is also affected by the basic principle of the technology, the gas phase equilibrium, and the greenhouse gas concentration will have a large error compared to the actual value. However, this embodiment does not change the original environmental conditions of the water body, and the automatic online monitoring data is more accurate.

[0028] 2. The existing technology requires the use of specific gas analysis instruments, such as gas chromatographs, for testing and analysis. The testing is expensive, the equipment is complex, and it is not convenient for field testing operations. However, the monitoring system of this embodiment is easy to use, and the monitoring data can be transmitted wirelessly and received by the PC.

[0029] 3. Existing greenhouse gas monitoring technologies have temporal and spatial limitations, and are unable to simultaneously monitor the three greenhouse gases carbon dioxide, methane, and nitrous oxide simultaneously in real time and over a long period of time. However, this embodiment utilizes online water temperature calibration to measure dissolved CO2, CH4, and N2O in water, enabling automatic real-time, fixed-point, and long-term concentration monitoring of CO2, CH4, and N2O in water.

[0030] 4. This embodiment is based on a concentration distribution prediction method for an automatic greenhouse gas concentration monitoring system in inland water bodies. By combining partial point monitoring with overall water area prediction, the greenhouse gas concentration distribution of the entire water area is obtained using as few automatic greenhouse gas concentration monitoring systems in inland water bodies as possible, thereby reducing the number of devices used for long-term monitoring, improving equipment utilization, and reducing the cost of greenhouse gas concentration monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 This is a schematic structural diagram of an automatic monitoring system for greenhouse gas concentrations in inland water bodies provided by an embodiment of the present invention;

[0033] Figure 2 This is a schematic diagram of the installation of an automatic monitoring system for greenhouse gas concentrations provided by an embodiment of the present invention;

[0034] Figure 3 This is a comparison chart of box statistics and linear correlation of data results of two methods provided by an embodiment of the present invention;

[0035] Figure 4 This is a flow chart of a method for predicting greenhouse gas concentration distribution in inland water bodies provided by an embodiment of the present invention;

[0036] Figure 5 This is a schematic diagram of grid division of a water area to be monitored provided by an embodiment of the present invention;

[0037] Reference numerals:

[0038] 1. Power supply device;

[0039] 2. Warning lights;

[0040] 3. Signal transmitter;

[0041] 4. Industrial computer;

[0042] 5. Acidity sensor;

[0043] 6. Temperature sensor;

[0044] 7. Carbon dioxide sensor;

[0045] 8. Methane sensor;

[0046] 9. Nitrous oxide sensor. DETAILED DESCRIPTION

[0047] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention are described clearly and completely below. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are also within the scope of protection of the present invention.

[0048] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0049] In the description of the present invention, it should also be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood broadly. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0050] Figure 1 and Figure 2 They are respectively a structural diagram and an installation diagram of an automatic monitoring system for greenhouse gas concentration in inland water bodies provided by an embodiment of the present invention. Figure 1 and Figure 2 The system mainly includes a floating box, a power supply device 1, a series of sensors and an industrial computer 4. The power supply device 1 mainly provides power to the system through the photovoltaic panels above and the batteries under the panels. The warning light 2 is used to prevent collisions with water vehicles.

[0051] When in use, the floating box can be fixed at the location of the water body to be monitored with reinforced concrete anchor points in the field. The float-type automatic regulator of the floating box is used to make the probes of the five sensors located 5-10 cm below the water surface. Photovoltaic solar panels are used to power the entire system. The monitoring principle of greenhouse gas concentration is based on the water chemical ion balance relationship and Henry's law in the water body. The sensor signals are collected and calculated by the data acquisition module of the industrial computer 4, and then the result data is output using the signal transmitter 3 of the remote communication module (wired RS485 communication interface and wireless GPRS transmission).

[0052] Before installing the system in the field, the acidity sensor 5 is first calibrated with pH buffer solutions (2.00, 4.01, 7.00, 9.21, and 10.00) to accurately measure the pH value (unit, dimensionless) of the surface water environment parameter. Then, the carbon dioxide sensor 7 is calibrated with ion electrode standards using NaHCO3 standard solutions (0.1mM, 0.5mM, 1.0mM, 5.0mM, and 10.0mM, where mM represents millimoles per liter) to accurately measure the bicarbonate content (unit, mM) in the surface water.

[0053] The system uses the precise signal (±0.01°) transmitted by the temperature sensor 6 to automatically perform temperature compensation correction on the signals of the acidity sensor 5 and the carbon dioxide sensor 7; the system collects the signals of the five sensors, wherein the signal S1 of the acidity sensor 5 is the pH value of the water at the monitoring point (unit, dimensionless), the signal S2 of the temperature sensor 6 is the Kelvin temperature of the water at the monitoring point (unit, °K), and the signal S of the carbon dioxide sensor 7 is the Kelvin temperature of the water at the monitoring point (unit, °K). CO2 is the bicarbonate concentration in the water at the monitoring point (unit, mM), the signal S of the methane sensor 8 CH4 is the content of CH4 dissolved in water at the monitoring point (unit, ppm), the signal S of the nitrous oxide sensor 9 N2OThe content of dissolved greenhouse gas N2O in the water at the monitoring point is monitored (unit, ppm); and based on the water chemical equilibrium equation and Henry's law, the sensor signals collected are calculated through the embedded operator device in the collection module to obtain the concentrations of dissolved CO2, CH4 and N2O in the water.

[0054] In a specific embodiment, the calculation method can include the following steps:

[0055] S110, the solubility K of CO2, CH4 and N2O in the surface water is calculated according to the following formula: CO2 CH4 and K N2O (unit, mole·L -1 ·atm -1 ), and the concentration of bicarbonate in the water body Henry's equilibrium constant k h (unit, atm·L·mol -1 ):

[0056]

[0057]

[0058] S120, the concentrations C of dissolved CO2, CH4 and N2O in the water body at the monitoring point are calculated according to the following formula: CO2 CH4 and C N2O (unit, μM, micromole):

[0059] C CO2 = S CO2 × K CO2 × k h (5)

[0060] C CH4 = S CH4 × K CH4 (6)

[0061] C N2O = S N2O × K N2O (7)

[0062] Optionally, the data collection module can store or wirelessly transmit the signals C CO2 , C CH4 and C N2O to the office PC for storage.

[0063] ​​Furthermore, in order to verify the reliability of the automatic monitoring system for greenhouse gas concentration in inland water bodies of this embodiment, this embodiment also compares and analyzes the difference between the gas concentrations measured by the system of this application and the traditional headspace method. In a specific embodiment, the bicarbonate signal S can be obtained by titration at 78 sampling points along the river surface water body in the field. CO2 The value of the pH sensor pH electrode signal S1 and the value of the temperature sensor signal S2 are used to obtain the carbon dioxide concentration signal C in the water body through the above formula (5): CO2 At the same time, water samples were collected at the sensor detection position, and the carbon dioxide partial pressure P in the sample was obtained using the headspace method and indoor instrument gas chromatography analysis test. CO2 The concentration of dissolved carbon dioxide in the water is calculated by combining the value of signal S2 obtained by the field temperature sensor (μatm). CO2 (unit, mM), the calculation formula is:

[0064] C CO2-sample =P CO2 ×K CO2 (8)

[0065] The data obtained by the two technical methods were subjected to one-way ANOVA, F>Fcrit, and the p value was 0.0356, which was obviously less than 0.05, indicating that there was a significant difference between the two groups of data. Figure 3 As shown, the method of this patent has a very strong correlation with the data obtained by the sample headspace-gas chromatography analysis test method, and the correlation coefficient R 2 It reaches 0.83, indicating that the overall data change trend is consistent; however, since the sample headspace method changes the original water environment, the ambient temperature caused by bringing the water sample container back to the indoor operation tends to be consistent, and the headspace method cannot fully separate the dissolved gas in the water, and it is difficult to completely remove all the gas in the headspace after gas equilibrium to measure the gas volume, therefore, compared with the method of this patent, the data change amplitude and detection value obtained by the sample headspace-gas chromatography analysis test method are smaller; indicating that the method adopted by this patent obtains data in the original water environment with better representativeness and accuracy.

[0066] In summary, this embodiment provides an automatic monitoring system for greenhouse gas concentrations in inland water bodies, which can achieve the following beneficial effects:

[0067] 1. The existing technology uses observation and sampling, which changes the original environmental conditions. This is also affected by the basic principle of the technology, the gas phase equilibrium, and the greenhouse gas concentration will have a large error compared to the actual value. However, this embodiment does not change the original environmental conditions of the water body, and the automatic online monitoring data is more accurate.

[0068] 2. The existing technology requires the use of specific gas analysis instruments, such as gas chromatographs, for testing and analysis. The testing is expensive, the equipment is complex, and it is not convenient for field testing operations. However, the monitoring system of this embodiment is easy to use, and the monitoring data can be transmitted wirelessly and received by the PC.

[0069] 3. Existing greenhouse gas monitoring technologies have temporal and spatial limitations, and are unable to simultaneously monitor the three greenhouse gases carbon dioxide, methane, and nitrous oxide simultaneously in real time and over a long period of time. However, this embodiment utilizes online water temperature calibration to measure dissolved CO2, CH4, and N2O in water, enabling automatic real-time, fixed-point, and long-term concentration monitoring of CO2, CH4, and N2O in water.

[0070] Based on the above-mentioned automatic monitoring system for greenhouse gas concentrations in inland water bodies, Figure 4 This is a flow chart of a method for predicting greenhouse gas concentration distribution in inland waters. This method can predict the greenhouse gas concentration distribution in the entire water area using as few monitoring points as possible, thus reducing the number of in-service equipment in long-term monitoring. Figure 4 As shown, the method specifically includes:

[0071] S210: Divide the water area to be monitored into multiple grids.

[0072] Figure 5 Taking a regular rectangle as an example, a top view of a water area to be monitored is shown as follows: Figure 5 As shown in Figure 2, the plane can be divided into multiple grids for zone monitoring. For irregularly shaped water planes, the division method is similar.

[0073] S220. Each grid is used as a monitoring point, and the above-mentioned automatic monitoring system for greenhouse gas concentration in inland water bodies is installed respectively to monitor the concentration of any greenhouse gas in the water body of each grid in real time; and the concentration of each grid at the same time is used as a sample to construct a finite sample set.

[0074] Combine Figure 2 , in each grid according to Figure 2 In the manner shown, automatic greenhouse gas concentration monitoring systems are installed to collect sensor data in real time and calculate the greenhouse gas concentration of each grid at each moment.

[0075] After monitoring for a period of time, for any greenhouse gas, the greenhouse gas concentration of each grid at the same time in the monitoring data is taken as a sample. Multiple samples can be obtained from the data of multiple historical moments to form a finite sample set. The finiteness of the sample set is emphasized here because, for waters with a large area, the cost of installing a set of automatic greenhouse gas concentration monitoring systems for each grid is very high. Therefore, this embodiment can only achieve sample collection for a limited time while ensuring that all grids have equipment. In subsequent steps, a prediction model for the greenhouse gas concentration distribution will be trained using limited samples, and the concentration distribution of the entire water area will be predicted by the monitoring concentration of some grids, thereby reducing the number of equipment in service for a long time and reducing the concentration monitoring cost of the entire water area.

[0076] S230. Average the concentration distribution monitored over a period of time, and take multiple extreme value grids and large gradient grids in the averaged concentration distribution as multiple initial grids; and construct a neural network model that uses the concentrations of the multiple initial grids at the same time as input and the concentration distribution of all grids at the same time as output.

[0077] The time period can be selected as a period of time when the water flow, temperature, etc. are relatively stable. The concentration distribution at each moment in the period is taken as a two-dimensional matrix arranged according to the geographical location of the grid (in irregular waters, the grid elements corresponding to the land position are permanently set to 0). The two-dimensional matrices of multiple moments are added together and then divided by the number of moments in the period to obtain the average concentration distribution in the period. From the average concentration distribution, the grids where the extreme points are located and the grids where the gradient of change exceeds the set threshold are selected to form the initial set of input grids of the neural network model. For ease of description, the grids in these initial sets are referred to as initial grids in this embodiment. These grids are selected because they often represent key positions where turning points or mutations occur in the concentration distribution of the entire water area, and can provide key information for the later prediction of the concentration distribution of the entire water area.

[0078] A neural network model is constructed based on this initial set. This embodiment trains the model so that, after inputting the greenhouse gas concentrations of these initial grids at the same moment, it can output the distribution of greenhouse gas concentrations in water for all grids at that moment. Optionally, the model can utilize a multi-layer convolutional neural network, followed by a fully connected layer to generate results of a specific dimension. The dimensions of the input and output layers can be adjusted based on the amount of input and output data. The number of convolutional layers can be greater than five, and the convolution kernel can be 3×3, expanding or extracting data features layer by layer. The fully connected layer is used to connect the dimensions of the last convolutional layer and the output layer.

[0079] S240. The neural network model is trained and tested using the finite sample set. If the model accuracy does not meet the requirements, the grids whose concentration prediction values ​​and monitoring values ​​in the test results differ by more than a set threshold are clustered according to their geographical locations, and the centers of the clusters are included in the model input.

[0080] As mentioned above, the model needs to learn the data relationship between the concentrations of some grids and the concentrations of all grids. Due to the large amount of missing grid data, the model is very difficult to learn. When the number of samples is limited, the accuracy of the trained model is likely to fall short of the ideal requirements. Therefore, this embodiment increases the number of input grids when the model accuracy is poor, providing more data information to the model and gradually reducing the learning difficulty.

[0081] In a specific implementation, in order to make full use of limited sample data, model training can be completed through the following steps:

[0082] Step 1: Divide the finite sample set into a first training set, a second training set, and a test set. The first training set and the second training set are used to correspond to and distinguish the two subsequent training stages. The samples in the two training sets can be the same, different, or partially the same, and this embodiment does not impose any specific restrictions.

[0083] Step 2: Use the first training set to train the neural network model once, and update the model parameters by minimizing the concentration changes of adjacent grids during the training. This step corresponds to the first training stage of the model, which uses the rules followed by the input data itself to update the model parameters so that the model output has a certain regularity. Due to the continuity of the water body, the change in greenhouse gas concentration in the water should also be relatively uniform. Therefore, this embodiment updates the model parameters by constraining the concentration changes of adjacent grids to be minimal, so as to minimize random changes without any rules. Optionally, the following loss function can be constructed:

[0084]

[0085] Where L1 represents the loss function value, i and j represent the grid indexes along the horizontal and vertical directions respectively, and F(i,j) represents the concentration prediction value of the grid in the i-th row and j-th column. represents the sum of all grids in the water column (excluding land grids where the concentration is permanently 0), and abs represents the absolute value. The model parameters are updated by minimizing this loss function.

[0086] Step three, the model after the first training is secondarily trained by using the second training set, and the model parameters are updated by minimizing the difference between the concentration prediction value of each grid and the monitoring value in the secondary training. This step corresponds to the second training stage of the model, which uses the actually monitored concentration data as the true value, and continuously approaches the true value by constraining the model output to perform supervised learning on the model after the first training. Optionally, the following loss function can be constructed:

[0087]

[0088] wherein L2 represents the loss function value, represents the concentration true value of the grid in the i-th row and the j-th column.

[0089] Further, the present embodiment also considers the case that part of the grids have no true value data due to equipment failure or communication failure, etc. In this case, although the sample data is missing part of the grid data, it can still be used for secondary training of the model, and only the minimization of the concentration change of the part of the grid and the adjacent grid and the minimization of the difference between the concentration prediction value and the monitoring value of the other grid in the loss function is required. Optionally, the following loss function can be constructed for this kind of sample:

[0090]

[0091] wherein L3 represents the loss function value, i1 and j1 represent the two-dimensional index of the data complete grid, i2 and j2 represent the two-dimensional index of the data missing grid, F(,) represents the concentration prediction value of the corresponding grid, represents the concentration monitoring value of the corresponding grid, and a and β represent the weight coefficients respectively. In this way, for the sample with complete data, the network parameters can be updated by using L2, and for the sample with missing data, the network parameters can be updated by using L3, further avoiding the waste of sample data.

[0092] The above two training stages first let the model learn the law of the input data itself, and then perform one-to-one supervised training through the true value of each grid. Compared with directly using the true value data for supervised training, the training method of the present embodiment can accelerate the convergence speed of the model and achieve the best training effect with limited sample set.

[0093] Step four, after the training is completed, the model after the secondary training is tested by using the test set. If the model accuracy still does not meet the requirements, the grids with the difference between the concentration prediction value and the monitoring value exceeding the set threshold in the test results are clustered according to the geographical position, and the cluster centers are included in the model input. This step selects the grids with large prediction errors from the prediction results of each test sample as the input grids of the model for the case that the model accuracy does not reach a good level after two training stages.

[0094] Optionally, if some grids exhibit large errors across multiple test samples, these grids are clustered by geographic location to form multiple spatial clusters. Each spatial cluster represents a relatively close area where grids within this area repeatedly exhibit large measurement errors, indicating that the model cannot effectively learn the data patterns of this area from the existing input grids. Therefore, it is necessary to select one or more grids within this area as model input to provide the model with effective information about this area. Optionally, the DBSCAN clustering method can be used, ultimately selecting the cluster center of each cluster as the model input.

[0095] Furthermore, when adding new model inputs, you can add convolution kernels connected to the newly added network and subsequent calculation layers in the first convolution layer based on the original model; or add a fully connected layer between the input layer and the first convolution layer to transform the new input layer dimension to the dimension of the original hidden layer; the rest of the structure in the model remains unchanged and inherits the original parameters as the initial parameters.

[0096] S250. Return to S240 based on the new neural network model, and repeat this process until the model accuracy meets the requirements.

[0097] Specifically, the finite sample set is used again to train and test the new neural network model; if the accuracy of the new model still does not meet the requirements, the grids in which the difference between the concentration prediction value and the monitoring value in the test results exceeds the set threshold are clustered according to the geographical location, and the centers of each cluster are included in the model input to obtain an updated neural network, and then return to S240. This cycle is repeated until the final model accuracy meets the required indicators. At this time, the neural network model has sufficient valid grid data. According to these data and the change law of the concentration itself, the concentration distribution of the entire water plane can be predicted more accurately. For the sake of distinction and description, the neural network model obtained at this time is called the full model.

[0098] S260. Using a genetic algorithm, select an optimal input grid combination from the input grid set of the full model that can ensure the model accuracy meets the requirements and has the least number of grids, and use the optimal model corresponding to the optimal input grid combination to predict the concentration distribution in the water body.

[0099] As mentioned above, during the multiple iterations of S240-S250, the model accuracy was improved by continuously increasing the number of input grids. To prevent redundancy in the final input grids, this step uses a genetic algorithm to filter the input grids of the full model to remove redundant grids. This allows the prediction of concentration distribution across the entire water surface to be achieved with minimal model input and the greenhouse gas concentration automatic monitoring system.

[0100] In a specific implementation, the above screening process can be completed in the following manner:

[0101] Step 1: Construct a chromosome sequence structure of equal length based on the number of grids in the full model input grid set. For example, if the full model input grids have 15, a vector of length 15 can be constructed as the chromosome sequence structure in the genetic algorithm, where each bit in the vector corresponds to a grid in the input grid set; a value of 1 indicates that the corresponding grid is retained as input to the neural network model, and a value of 0 indicates that the corresponding grid is deleted from the neural network model input grid.

[0102] Step 2: Generate the current population of the genetic algorithm based on the chromosome sequence structure. For example, a random initialization method can be used to generate multiple chromosomes to form the current population.

[0103] Step 3: Prune the full model based on each chromosome in the current population, and train the pruned model three times using the finite sample set. Specifically, for a particular chromosome, remove the grid corresponding to the bit with a value of 0 in that chromosome from the input grid of the full model, and remove all branches in the first hidden layer of the model that connect to that input grid, to obtain a new model. This new model is then trained using the finite sample set. In this embodiment, this training is referred to as three training cycles.

[0104] Step 4: Determine whether the model after three trainings meets the optimal goal. If not, select, crossover, and mutate the current population using a genetic algorithm. Select excellent chromosomes from the mutated current population to form a new current population. Return to step 3, prune the full model based on the chromosomes in the new current population, and train the pruned model three more times using the limited sample set. This cycle repeats until the model after three trainings meets the optimal goal. The optimal goal is: the model accuracy after three trainings meets the requirements, and the model with the minimum number of input grids has been obtained in the three trainings so far.

[0105] Optionally, each time a chromosome that meets the model accuracy requirements is obtained, the optimal chromosome with the smallest number of input grids can be selected from all the chromosomes that meet the model accuracy requirements. If the optimal chromosome remains unchanged after three consecutive training cycles, the model with the smallest number of input grids is considered to have been obtained. Of course, it is also possible to iterate directly to the maximum number of iterations, and after the iterations terminate, select the model that meets the model accuracy requirements and has the smallest number of input grids as the optimal model.

[0106] Furthermore, when selecting excellent chromosomes, a fitness function can be constructed based on the variation of the predicted gas concentration values ​​between the input grids in the chromosomes; based on the fitness function, excellent chromosomes are selected from the mutated current population to construct a new current population. Optionally, the following fitness function can be constructed:

[0107]

[0108] Where fit is the fitness value, num(G) indicates the number of extreme value grids in the concentration distribution predicted by the model after three training runs for the input grids with a value of 1 for the current chromosome, i3 and j3 are the two-dimensional indices of the input grids with a value of 1 for the current chromosome, num(0) indicates the number of grids with a value of 0 in the current chromosome, and γ, δ, and ε are weight coefficients. The larger the fitness value, the better the chromosome.

[0109] Through the above fitness function, chromosomes with a small number of input grids and input grids covering more extreme grids and large gradient grids (these grids represent key morphological changes in concentration distribution) can be preferentially selected to evolve the new current population, which helps to obtain the optimal chromosome and optimal model as soon as possible.

[0110] Finally, after obtaining the optimal input grid combination, only the automatic monitoring systems for greenhouse gas concentrations in inland water bodies in the grids within that combination are retained, while the automatic monitoring systems for greenhouse gas concentrations in inland water bodies in the remaining grids are removed. The remaining systems are used to monitor the gas concentrations in each target grid in real time and input into the optimal model to obtain the real-time concentration distribution for all grids. The removed equipment systems can be used for ensemble prediction of concentrations in other water bodies, improving equipment utilization and reducing greenhouse gas concentration monitoring costs.

[0111] Furthermore, the above method uses a certain greenhouse gas as an example to illustrate a specific method for predicting the distribution of greenhouse gas concentrations in the entire water area using as few monitoring points as possible. In practical applications, operations S220-S260 can be performed for each greenhouse gas; wherein, in S260, when the optimal model corresponding to the optimal input grid combination is used to predict the concentration distribution in the water body, the optimal input grid combinations corresponding to various greenhouse gases are taken as a union, and the automatic monitoring system for greenhouse gas concentrations in inland water bodies of each target grid in the union is retained, and the automatic monitoring system for greenhouse gas concentrations in inland water bodies of the remaining grids is removed; the greenhouse gas concentrations of each target grid are monitored in real time using the remaining systems, and the optimal models corresponding to various greenhouse gases are input respectively to obtain the real-time greenhouse gas concentration distribution of all grids. In this way, the concentration distribution of the three greenhouse gases in the entire water area can be predicted using as few monitoring points as possible.

[0112] It should be noted that, in the above-mentioned method for predicting greenhouse gas concentration distribution at the water-air interface, except for the steps of installing and dismantling the automatic monitoring system for greenhouse gas concentration at the water-air interface, the remaining steps can be automatically performed by electronic equipment.

[0113] In summary, this embodiment also provides a method for predicting greenhouse gas concentration distribution in inland water bodies, which can achieve the following beneficial effects:

[0114] 1. By combining partial point monitoring with overall water body prediction, the greenhouse gas concentration distribution of the entire water body can be obtained using as few automatic greenhouse gas concentration monitoring systems as possible in inland water bodies, thereby reducing the number of equipment used for long-term monitoring, improving equipment utilization, and reducing greenhouse gas concentration monitoring costs;

[0115] 2. Because the neural network model needs to learn complex concentration distribution patterns based on a subset of grids, the learning difficulty is high and the number of samples is limited, making it difficult for the trained model to achieve ideal accuracy all at once. To address this difficulty, this embodiment selectively selects representative grids from areas with large prediction errors as model input when the model accuracy is poor. By increasing the number of input grids, the model is provided with more effective information, gradually reducing the learning difficulty.

[0116] 3. To address the limited number of grid-wide data samples, this embodiment uses multi-stage training to enable the model to first learn the patterns of the input data itself, and then conducts one-on-one supervised training using the monitoring values ​​of each grid. Compared with directly using monitoring data for supervised training, this method can accelerate model convergence and achieve the best possible training results using a limited sample set.

[0117] 4. For samples with missing data due to equipment or communication failures in some grids, this embodiment integrates uniform concentration changes with monitoring values ​​and adopts different loss value calculation methods for different grids, making full use of the remaining data information in the sample and further avoiding sample waste;

[0118] 5. After obtaining a full model with sufficient valid input grids, this embodiment uses a genetic algorithm to screen the input grids and remove redundant grids to prevent redundancy. This allows the automatic greenhouse gas concentration monitoring system to predict concentration distributions across the entire water surface with minimal model input. This can significantly reduce the number of devices needed and lower monitoring costs when the water surface is large and the concentration distribution of each greenhouse gas needs to be predicted.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. An automatic monitoring system for greenhouse gas concentration in inland water bodies, characterized in that: include: The acidity sensor and temperature sensor are used to collect the pH value S1 and temperature S2 of the water at the monitoring point respectively; Carbon dioxide sensor, used to collect bicarbonate concentration S in water at the monitoring point CO2 ; Methane sensor and nitrous oxide sensor are used to collect the content of CH4 and N2O dissolved in water at the monitoring point respectively. CH4 and S N2O ; Operator device for calculating the concentration of greenhouse gases in water by: S110. Calculate the solubility K of CO2, CH4 and N2O in surface water according to the following formula: CO2 , K CH4 and K N2O , and Henry's equilibrium constant k for bicarbonate concentration in water h : Among them, A, B, C, D, E, F, G, H, I, J, K, L, M, N, P, Q, and R are fixed constants that are calibrated or derived; S120. Calculate the dissolved CO2, CH4 and N2O concentrations C in the water at the monitoring point according to the following formula: CO2 、C CH4 and C N2O : C CO2 =S CO2 ×K CO2 ×k h C CH4 =S CH4 ×K CH4 C N2O =S N2O ×K N2O。 2. A method for predicting greenhouse gas concentration distribution in inland water bodies, characterized in that: include: S210, dividing the water area to be monitored into a plurality of grids; S220: Using each grid as a monitoring point, installing the automatic monitoring system for greenhouse gas concentration in inland water bodies as claimed in claim 1, and monitoring the concentration of any greenhouse gas in the water body of each grid in real time; and using the concentration of each grid at the same time as a sample to construct a finite sample set; S230, averaging the concentration distribution monitored over a period of time, and taking multiple extreme value grids and large gradient grids in the averaged concentration distribution as multiple initial grids; and constructing a neural network model that takes the concentrations of the multiple initial grids at the same moment as input and takes the concentration distribution of all grids at the same moment as output; S240, training and testing the neural network model using the finite sample set; if the model accuracy does not meet the requirements, clustering the grids whose concentration prediction values ​​and monitored values ​​differ by more than a set threshold in the test results by geographical location, and incorporating the centers of the clusters into the model input; S250, returning to S240 according to the new neural network model, and repeating this cycle until the model accuracy meets the requirements; S260. Using a genetic algorithm, select an optimal input grid combination from the input grid set of the final model that can meet the model accuracy requirements and has the least number of grids, and use the optimal model corresponding to the optimal input grid combination to predict the concentration distribution in the water body.

3. The method according to claim 2, characterized in that The training and testing of the neural network model using the finite sample set includes: Dividing the finite sample set into a first training set, a second training set and a test set; Training the neural network model once using the first training set, and updating model parameters by minimizing the concentration change of adjacent grids during the training; The second training set is used to perform a second training on the model after the first training, and in the second training, the model parameters are updated by minimizing the difference between the concentration prediction value and the monitoring value of each grid; The test set is used to test the model after secondary training.

4. The method according to claim 3, characterized in that The second training set is used to perform a second training on the model after the first training, wherein the model parameters are updated by minimizing the difference between the concentration prediction value and the monitoring value of each grid in the second training, including: If some grids in the samples of the second training set are missing monitoring data, the model parameters are updated in the secondary training by minimizing the concentration changes between the grids and adjacent grids, and minimizing the differences between the concentration prediction values ​​and the monitoring values ​​of other grids.

5. The method according to claim 4, characterized in that The updating of model parameters by minimizing the concentration change between the partial grid and the adjacent grids and minimizing the difference between the concentration prediction value and the monitoring value of other grids includes: Construct the following loss function: Among them, L3 represents the loss function value, i1 and j1 represent the two-dimensional index of the data complete grid, i2 and j2 represent the two-dimensional index of the data missing grid, and F(,) represents the concentration prediction value of the corresponding grid. Indicates the concentration monitoring value of the corresponding grid, abs represents the absolute value, and α and β represent the weight coefficients respectively.

6. The method according to claim 2, characterized in that The method of using a genetic algorithm to select an optimal input grid combination from the input grid set of the final model that can meet the model accuracy requirements and has the least number of grids includes: According to the size of the input grid set of the final model, a chromosome sequence structure of equal length is constructed, wherein each numerical bit in the sequence structure corresponds one-to-one to each grid in the set, and the 1 / 0 value of the numerical bit represents whether the corresponding grid is used as the input grid of the neural network model; generating a current population of a genetic algorithm according to the chromosome sequence structure; Pruning the final model according to each chromosome in the current population, and training the pruned model three times using the limited sample set; Determine whether the model after three trainings meets the optimal goal. If not, generate a new current population and return to the model pruning operation until the model after three trainings meets the optimal goal. The optimal goal is that the model accuracy after three trainings meets the requirements, and a model with the least number of input grids has been obtained in the three trainings conducted so far.

7. The method according to claim 6, characterized in that The generating of a new current population includes: Construct a fitness function based on the variation of the concentration prediction value between each input grid in the chromosome; According to the fitness function, excellent chromosomes are selected from the mutated current population to construct a new current population.

8. The method according to claim 7, characterized in that The fitness function is constructed according to the variation of the concentration prediction value between each input grid in the chromosome, including: Construct the following fitness function: Among them, fit is the fitness value, num(G) represents the number of extreme grids in the concentration distribution predicted by the model after three trainings with the input grids of 1 in the chromosome, i3 and j3 are the two-dimensional indexes of the input grids of 1 in the chromosome, F(,) represents the concentration prediction value of the corresponding grid, abs represents the absolute value, num(0) represents the number of grids of 0 in the chromosome, γ, δ and ε represent the weight coefficients respectively; the larger the fitness value, the better the chromosome.

9. The method according to claim 2, characterized in that The method of using the optimal model corresponding to the optimal input grid combination to predict the concentration distribution in the water body includes: Only the automatic monitoring system for greenhouse gas concentration in inland water bodies of each target grid in the optimal input grid combination is retained, and the automatic monitoring system for greenhouse gas concentration in inland water bodies of the remaining grids is removed; The remaining systems are used to monitor the gas concentration of each target grid in real time, and the gas concentration is input into the optimal model to obtain the real-time concentration distribution of all grids.

10. The method according to claim 2, characterized in that include: For each greenhouse gas, operations S220-S260 are performed; The method of using the optimal model corresponding to the optimal input grid combination to predict the concentration distribution in the water body includes: Take the union of the optimal input grid combinations corresponding to various greenhouse gases; retaining and concentrating the automatic monitoring systems for greenhouse gas concentrations in inland water bodies in the target grids, and dismantling the automatic monitoring systems for greenhouse gas concentrations in inland water bodies in the remaining grids; The remaining systems are used to monitor the greenhouse gas concentration of each target grid in real time, and the optimal models corresponding to various greenhouse gases are input respectively to obtain the real-time greenhouse gas concentration distribution of all grids.

Images