Underway detection method for rapidly monitoring fluorine-containing greenhouse gas
Through the combination of gas sampler and deep learning algorithm, the rapid monitoring of fluorine-containing greenhouse gases is achieved, solving the problems of long and high cost in the detection in the prior art, and improving the analysis accuracy and efficiency.
Patent Information
- Application Number
- CN202411109879.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-22
AI Technical Summary
The existing fluorine-containing greenhouse gas navigation detection methods require highly professional equipment and time, resulting in high testing time and cost.
Gas samplers are used to collect greenhouse gas samples, perform pre-processing and quantitative analysis, and analyze data in combination with deep learning algorithms to shorten detection time and reduce costs.
Through automated processes, significantly improve analysis accuracy and sensitivity, shorten detection time and reduce costs, while ensuring the accuracy and reliability of results.
Smart Images

Figure CN119000929A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of greenhouse gas cruise detection, and in particular relates to a cruise detection method for rapidly monitoring fluorine-containing greenhouse gases. Background Art
[0002] Common fluorinated greenhouse gases include hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), sulfur hexafluoride (SF6) and nitrogen trifluoride (NF3). Since fluorinated gases have extremely high global warming potential, such as the GWP100 of NF3 is as high as 15750, and the GWP100 of SF6 is as high as 23500, therefore, even though the concentration of fluorinated greenhouse gases in the atmosphere is extremely low, it is necessary to detect the movement of fluorinated greenhouse gases.
[0003] The current cruise detection method for fluorinated greenhouse gases is mainly through high-precision monitoring systems, such as the GHG5-Pro high-precision fluorinated greenhouse gas monitoring system. This type of system can achieve real-time and accurate measurement of fluorinated greenhouse gases in the atmosphere by automatically and accurately monitoring fluorinated greenhouse gases in the ambient air.
[0004] However, since fluorinated greenhouse gases have high global warming potential, their detection requires highly professional equipment and technology, and the deployment and operation of these equipment requires a certain amount of time to ensure the accuracy and reliability of the data, thereby extending the detection time of fluorinated greenhouse gases and increasing the detection cost of fluorinated greenhouse gases. Therefore, we need to propose a rapid monitoring cruise detection method for fluorinated greenhouse gases to solve the above-mentioned problems, so that it can shorten the cruise detection time of fluorinated greenhouse gases and reduce a certain gas cruise detection cost. Summary of the invention
[0005] The object of the present invention is to provide a cruise detection method for rapidly monitoring fluorine-containing greenhouse gases, which can shorten the cruise detection time of fluorine-containing greenhouse gases and reduce the cruise detection cost of certain gases, so as to solve the problems raised in the above-mentioned background technology.
[0006] To achieve the above object, the present invention adopts the following technical solutions:
[0007] A cruise detection method for rapidly monitoring fluorinated greenhouse gases comprises the following steps:
[0008] S1. Install a gas sampler in the greenhouse so that the air inlet of the gas sampler is placed in the air flow to be measured in the greenhouse, and collect gas samples through the gas sampler;
[0009] S2. Pre-process the gas sample to remove impurities in the gas sample to obtain a test gas;
[0010] S3, performing quantitative analysis on the detected gas to obtain gas sample data;
[0011] S4. Analyze and process the gas sample data through a deep learning algorithm to obtain analysis results;
[0012] S5. Organize and interpret the analysis results and form a test result report.
[0013] Preferably, when the gas sampler is in use, the gas sampler is first connected to a gas chromatograph and a computer, and then the sampling parameters of the gas sampler are set according to the sampling requirements, the sampling parameters including the sampling time, sampling rate and sampling capacity, and the gas sampler is started for gas sampling. The gas sampler transmits the collected gas samples to the gas chromatograph after preprocessing.
[0014] Preferably, in step S2, the process of pre-treatment of the gas sample includes particulate matter filtration, moisture removal and adsorption treatment, wherein particulate matter filtration is performed by installing a filter membrane at the air inlet of the gas sampler to filter out particulate matter in the gas; moisture removal is performed by installing a cooling mechanism in the collection chamber of the gas sampler, and utilizing the cooling mechanism to cool the gas sample so that the moisture in the gas sample condenses and drips into the collection chamber of the gas collector; adsorption treatment is performed by installing an activated carbon filter at the air outlet of the collection chamber of the gas sampler, and the gas sample collected by the gas sampler is adsorbed by the activated carbon in the activated carbon filter before being transmitted, so as to remove interfering components in the gas sample.
[0015] Preferably, the filter membrane is configured as one of a polytetrafluoroethylene filter membrane, a polypropylene filter membrane, a polyethersulfone filter membrane and a polyvinylidene fluoride filter membrane.
[0016] Preferably, in step S3, when the detection gas is quantitatively analyzed, the following steps are included:
[0017] A1. First calibrate the gas chromatograph, which is equipped with a gas sensor and a temperature and humidity sensor;
[0018] A2. Inject the pretreated sample into the gas chromatograph;
[0019] A3. Obtain the gas concentration data of the current environment through the gas sensor;
[0020] A4. Obtain the temperature and humidity data of the current environment through the temperature and humidity sensor;
[0021] A5. Correct the gas concentration data according to the environmental data and the preset concentration correction coefficient;
[0022] A6. Use SoilFluxPro to process and optimize data and obtain accurate gas concentration data.
[0023] Preferably, in step S4, the deep learning algorithm includes deep model selection, model training and tuning, model integration and multi-task learning when analyzing and processing gas sample data, wherein the deep model includes a convolutional neural network model, a recursive neural network model, a long short-term memory network model and a multi-task learning model.
[0024] Preferably, the process of model training and tuning is as follows:
[0025] B1. Divide the data set into training set, validation set and test set;
[0026] B2. Find the optimal hyperparameter combination through grid search method;
[0027] B3. Use the training set data to train the selected deep learning model, and continuously adjust the model parameters through the back propagation algorithm to minimize the loss function;
[0028] B4. Use the validation set data to verify the trained model and select the model with the best performance on the validation set as the final model.
[0029] Preferably, the model integration is to select an appropriate integration strategy according to the problem requirements, integrate the trained single models using the selected integration strategy to form an integrated model, and tune the weights and combination methods in the integrated model to improve the overall performance;
[0030] The multi-task learning is to first clarify the multiple tasks to be solved, analyze the relationship between these tasks, and then design the network structure so that multiple tasks can share certain representation layers to capture the shared features between tasks. Then, a specific output layer or head is designed for each task to ensure that the network can be optimized for each task. Finally, the data of multiple tasks are simultaneously input into the network for joint training, and the parameters of the shared layer and the task-specific layer are optimized through back propagation and gradient descent.
[0031] Preferably, the analysis results obtained after the gas sample data is analyzed and processed by the deep learning algorithm include gas types and concentrations, spatial distribution and changing trends, environmental impact and safety assessment, data verification and quality control, wherein the types of gases include hydrofluorocarbons, perfluorocarbons, sulfur hexafluoride and nitrogen trifluoride, the spatial distribution and changing trends are to show the spatial distribution and changing trends of fluorine-containing greenhouse gases in different regions and time ranges, the environmental impact and safety assessment is to evaluate the environmental impact of fluorine-containing greenhouse gases and evaluate the potential risks of these gases to human health and ecosystems, and data verification and quality control are to calibrate the monitoring equipment using standard gases and cross-validate through laboratory quality control samples.
[0032] Preferably, in step S5, when the detection result report is formed, the data of gas concentration, time and location are first derived based on the analysis results, and then the types and concentrations of various fluorinated greenhouse gases detected are clearly indicated. The gas concentration data are then visualized on a map, and the spatial distribution and change trends within the time range of different regions are analyzed. Finally, the destructive effects of fluorinated greenhouse gases on ozone and climate systems are evaluated, and the potential risks to human health and ecosystems are analyzed. Then, the structure of the report is designed, and the structure of the report includes introduction, methods, results, discussion, conclusions and suggestions. The report is generated according to the structure of the report.
[0033] The cruise detection method for rapidly monitoring fluorinated greenhouse gases proposed in the present invention has the following advantages compared with the prior art:
[0034] 1. The present invention collects greenhouse gas samples through a gas sampler, and then pre-processes the gas samples to remove impurities in the gas samples to obtain test gas, quantitatively analyzes the obtained test gas, obtains gas sample data, analyzes and processes the gas sample data through a deep learning algorithm to obtain analysis results, and finally organizes and interprets the analysis results to form a test result report. The deep learning algorithm can automatically learn features from complex gas sample data, which can significantly improve the accuracy and sensitivity of the analysis compared to traditional methods. Once the deep learning model is trained, its prediction speed is usually much faster than traditional methods, thereby shortening the time from sampling to obtaining analysis results. The entire process from gas collection, pre-processing, quantitative analysis to data processing can be automated, reducing human operating errors and improving efficiency. It can not only shorten the time for cruise detection of fluorine-containing greenhouse gases, but also reduce the detection cost to a certain extent, while ensuring the accuracy and reliability of the detection results. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 It is a flowchart of the present invention;
[0036] Figure 2 It is a flowchart of the quantitative analysis process of the present invention;
[0037] Figure 3 This is a flowchart of the model training and tuning process of the present invention. DETAILED DESCRIPTION
[0038] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0039] The present invention provides Figure 1-3 A cruise detection method for rapidly monitoring fluorinated greenhouse gases is shown, comprising the following steps:
[0040] S1. Install a gas sampler in the greenhouse so that the air inlet of the gas sampler is placed in the air flow to be measured in the greenhouse, and collect gas samples through the gas sampler;
[0041] When the gas sampler is in use, the gas sampler is first connected to the gas chromatograph and the computer, and then the sampling parameters of the gas sampler are set according to the sampling requirements. The sampling parameters include sampling time, sampling rate and sampling capacity. The gas sampler is started for gas sampling. The gas sampler transmits the collected gas samples to the gas chromatograph after preprocessing.
[0042] S2. Pre-process the gas sample to remove impurities in the gas sample to obtain a test gas;
[0043] The process of pre-treatment of the gas sample includes particulate matter filtration, moisture removal and adsorption treatment, wherein particulate matter filtration is performed by installing a filter membrane at the air inlet of the gas sampler to filter out particulate matter in the gas; moisture removal is performed by installing a cooling mechanism in the collection chamber of the gas sampler, and using the cooling mechanism to cool the gas sample so that the moisture in the gas sample condenses and drips into the collection chamber of the gas collector; adsorption treatment is performed by installing an activated carbon filter at the air outlet of the collection chamber of the gas sampler, and the gas sample collected by the gas sampler is adsorbed by the activated carbon in the activated carbon filter before being transmitted to remove interfering components in the gas sample.
[0044] The filter membrane is set to be one of a polytetrafluoroethylene filter membrane, a polypropylene filter membrane, a polyethersulfone filter membrane and a polyvinylidene fluoride filter membrane;
[0045] The polytetrafluoroethylene filter membrane has extremely high chemical stability and can withstand corrosion from various chemical solvents, including strong acids, strong alkalis and organic solvents. The filter membrane can be used for a long time in the temperature range of 90°C to 260°C and is suitable for gas filtration in high temperature environments. The pore size of the filter membrane ranges from 0.03 microns to 5.0 microns, which can efficiently filter out tiny particles in the gas. The surface of the polytetrafluoroethylene filter membrane has extremely low protein adsorption and dissolution, ensuring the purity of the sample.
[0046] Polypropylene filter membrane has good chemical stability, can resist the erosion of acids, alkalis and certain organic solvents, and is also hydrophobic, suitable for filtering impurities in hydrophobic gases. The pore size of the filter membrane ranges from 0.1 microns to 10 microns, which can meet the filtering needs of different precisions.
[0047] Polyethersulfone filter membrane has high mechanical strength, can work stably under high pressure environment, and has good biocompatibility, making it suitable for gas filtration in medical and biological fields.
[0048] PVDF membranes can resist corrosion from most chemical solvents, including DMSO, THF, DMF, etc. PVDF membranes are highly hydrophobic and can effectively prevent water molecules and other liquid impurities from penetrating.
[0049] S3, performing quantitative analysis on the detected gas to obtain gas sample data;
[0050] The quantitative analysis of the detected gas includes the following steps:
[0051] A1. First calibrate the gas chromatograph, which is equipped with gas sensors and temperature and humidity sensors to ensure the accuracy of the measurement results. Usually, standard gases with known concentrations are used for calibration to establish a reliable standard curve.
[0052] A2. Inject the pretreated sample into the gas chromatograph. This step requires precise control of the sample injection volume and injection speed to avoid errors.
[0053] A3. Obtain the gas concentration data of the current environment through gas sensors. These sensors can sense the concentration of the target gas and convert the data into electrical signals;
[0054] A4. Obtain the temperature and humidity data of the current environment through the temperature and humidity sensor;
[0055] A5. Correct the gas concentration data according to the environmental data and the preset concentration correction coefficient. This step is to eliminate the interference of environmental data on the gas concentration data, thereby improving its accuracy. For example, the environmental data can be compensated by the preset temperature and humidity compensation formula, and then the gas concentration data can be corrected according to these compensated data.
[0056] A6. Use SoilFluxPro to process and optimize data and obtain accurate gas concentration data; the software can calculate flux and perform quality control on long-term measurement data sets and survey measurement data sets. It can also revise the set variables of the data set (such as start time, end time, dead zone length, etc.) and immediately view their impact on the flux calculation. In addition, multiple gas fluxes can be calculated through post-data processing, and complex data sets can be imported into the software for analysis.
[0057] S4. Analyze and process the gas sample data through a deep learning algorithm to obtain analysis results;
[0058] The deep learning algorithm includes deep model selection, model training and tuning, model integration and multi-task learning when analyzing and processing gas sample data, wherein the deep model includes a convolutional neural network model, a recursive neural network model, a long short-term memory network model and a multi-task learning model.
[0059] The model training and tuning process is as follows:
[0060] B1. Divide the dataset into training set, validation set, and test set to ensure that the performance of the model on different data subsets can be comprehensively evaluated.
[0061] B2. Find the optimal hyperparameter combination through grid search method to improve model performance.
[0062] B3. Use the training set data to train the selected deep learning model, and continuously adjust the model parameters through the back propagation algorithm to minimize the loss function.
[0063] B4. Use the validation set data to verify the trained model and select the model with the best performance on the validation set as the final model.
[0064] The model integration is to select an appropriate integration strategy, such as Bagging, Boosting or Stacking, according to the problem requirements, integrate the trained single models using the selected integration strategy to form an integrated model, and optimize the weights and combination methods in the integrated model to improve the overall performance;
[0065] The multi-task learning is to first clarify the multiple tasks to be solved, analyze the relationship between these tasks, and then design the network structure so that multiple tasks can share certain representation layers to capture the shared features between tasks. Then, a specific output layer or head is designed for each task to ensure that the network can be optimized for each task. Finally, the data of multiple tasks are simultaneously input into the network for joint training, and the parameters of the shared layer and the task-specific layer are optimized through back propagation and gradient descent.
[0066] The analysis results obtained after the gas sample data is analyzed and processed by the deep learning algorithm include gas types and concentrations, spatial distribution and changing trends, environmental impact and safety assessment, data verification and quality control, where the types of gases include hydrofluorocarbons, perfluorocarbons, sulfur hexafluoride and nitrogen trifluoride. The spatial distribution and changing trends show the spatial distribution and changing trends of fluorinated greenhouse gases in different regions and time ranges. The environmental impact and safety assessment is to evaluate the environmental impact of fluorinated greenhouse gases and the potential risks of these gases to human health and ecosystems. Data verification and quality control is to calibrate the monitoring equipment using standard gases and cross-validate through laboratory quality control samples.
[0067] S5. Organize and interpret the analysis results and form a test result report;
[0068] When the test result report is formed, the data on gas concentration, time and location are first derived based on the analysis results. Then the types and concentrations of various fluorinated greenhouse gases detected are clearly indicated. The gas concentration data are then visualized on a map, and the spatial distribution and changing trends within the time range of different regions are analyzed. Finally, the destructive effects of fluorinated greenhouse gases on ozone and climate systems are evaluated, and the potential risks to human health and ecosystems are analyzed. The structure of the report is then designed. The structure of the report includes introduction, methods, results, discussion, conclusions and recommendations, and the report is generated based on its structure.
[0069] In summary, the use of deep learning algorithms can automatically learn features from complex gas sample data. Compared with traditional methods, this can significantly improve the accuracy and sensitivity of analysis. Once the deep learning model is trained, its prediction speed is usually much faster than traditional methods, thereby shortening the time from sampling to obtaining analysis results. The entire process from gas collection, pretreatment, quantitative analysis to data processing can be automated, reducing human operational errors and improving efficiency. It can not only shorten the time for cruise detection of fluorinated greenhouse gases, but also reduce the detection cost to a certain extent, while ensuring the accuracy and reliability of the detection results.
[0070] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A cruise detection method for rapid monitoring of fluorinated greenhouse gases, characterized in that: The steps include: S1. Install a gas sampler in the greenhouse so that the air inlet of the gas sampler is placed in the air flow to be measured in the greenhouse, and collect gas samples through the gas sampler; S2. Pre-process the gas sample to remove impurities in the gas sample to obtain a test gas; S3, performing quantitative analysis on the detected gas to obtain gas sample data; S4. Analyze and process the gas sample data through a deep learning algorithm to obtain analysis results; S5. Organize and interpret the analysis results and form a test result report.
2. A cruise detection method for rapid monitoring of fluorinated greenhouse gases according to claim 1, characterized in that: In step S1, when the gas sampler is in use, the gas sampler is first connected to the gas chromatograph and the computer, and then the sampling parameters of the gas sampler are set according to the sampling requirements. The sampling parameters include sampling time, sampling rate and sampling capacity. The gas sampler is started for gas sampling. The gas sampler transmits the collected gas samples to the gas chromatograph after preprocessing.
3. A cruise detection method for rapid monitoring of fluorine-containing greenhouse gases according to claim 2, characterized in that: In step S2, the process of pre-treatment of the gas sample includes particulate matter filtration, moisture removal and adsorption treatment, wherein particulate matter filtration is performed by installing a filter membrane at the air inlet of the gas sampler to filter out particulate matter in the gas; moisture removal is performed by installing a cooling mechanism in the collection chamber of the gas sampler, and using the cooling mechanism to cool the gas sample so that the moisture in the gas sample condenses and drips into the collection chamber of the gas collector; adsorption treatment is performed by installing an activated carbon filter at the air outlet of the collection chamber of the gas sampler, and the gas sample collected by the gas sampler is adsorbed by the activated carbon in the activated carbon filter before being transmitted to remove interfering components in the gas sample.
4. A cruise detection method for rapid monitoring of fluorinated greenhouse gases according to claim 3, characterized in that: The filter membrane is configured as one of a polytetrafluoroethylene filter membrane, a polypropylene filter membrane, a polyethersulfone filter membrane and a polyvinylidene fluoride filter membrane.
5. A cruise detection method for rapid monitoring of fluorine-containing greenhouse gases according to claim 4, characterized in that: In step S3, when the detection gas is quantitatively analyzed, the following steps are included: A1. First calibrate the gas chromatograph, which is equipped with a gas sensor and a temperature and humidity sensor; A2. Inject the pretreated sample into the gas chromatograph A3. Obtain the gas concentration data of the current environment through the gas sensor; A4. Obtain the temperature and humidity data of the current environment through the temperature and humidity sensor; A5. Correct the gas concentration data according to the environmental data and the preset concentration correction coefficient; A6. Use SoilFluxPro to process and optimize data and obtain accurate gas concentration data.
6. A cruise detection method for rapid monitoring of fluorinated greenhouse gases according to claim 5, characterized in that: In step S4, the deep learning algorithm includes deep model selection, model training and tuning, model integration and multi-task learning when analyzing and processing gas sample data, wherein the deep model includes a convolutional neural network model, a recursive neural network model, a long short-term memory network model and a multi-task learning model.
7. A cruise detection method for rapid monitoring of fluorine-containing greenhouse gases according to claim 6, characterized in that: The model training and tuning process is as follows: B1. Divide the data set into training set, validation set and test set; B2. Find the optimal hyperparameter combination through grid search method; B3. Use the training set data to train the selected deep learning model, and continuously adjust the model parameters through the back propagation algorithm to minimize the loss function; B4. Use the validation set data to verify the trained model and select the model with the best performance on the validation set as the final model.
8. A cruise detection method for rapid monitoring of fluorine-containing greenhouse gases according to claim 7, characterized in that: The model integration is to select an appropriate integration strategy according to the problem requirements, integrate the trained single models using the selected integration strategy to form an integrated model, and optimize the weights and combination methods in the integrated model to improve the overall performance; The multi-task learning is to first clarify the multiple tasks to be solved, analyze the relationship between these tasks, and then design the network structure so that multiple tasks can share certain representation layers to capture the shared features between tasks. Then, a specific output layer or head is designed for each task to ensure that the network can be optimized for each task. Finally, the data of multiple tasks are simultaneously input into the network for joint training, and the parameters of the shared layer and the task-specific layer are optimized through back propagation and gradient descent.
9. A cruise detection method for rapid monitoring of fluorinated greenhouse gases according to claim 8, characterized in that: The analysis results obtained after the gas sample data is analyzed and processed by the deep learning algorithm include gas types and concentrations, spatial distribution and changing trends, environmental impact and safety assessment, data verification and quality control, where the types of gases include hydrofluorocarbons, perfluorocarbons, sulfur hexafluoride and nitrogen trifluoride. The spatial distribution and changing trends show the spatial distribution and changing trends of fluorinated greenhouse gases in different regions and time ranges. The environmental impact and safety assessment is to evaluate the environmental impact of fluorinated greenhouse gases and the potential risks of these gases to human health and ecosystems. Data verification and quality control is to calibrate the monitoring equipment using standard gases and cross-validate through laboratory quality control samples.
10. A cruise detection method for rapid monitoring of fluorine-containing greenhouse gases according to claim 9, characterized in that: In step S5, when the detection result report is formed, the data of gas concentration, time and location are first derived based on the analysis results, and then the types and concentrations of various fluorinated greenhouse gases detected are clearly indicated. The gas concentration data are then visualized on a map, and the spatial distribution and change trends within the time range of different regions are analyzed. Finally, the destructive effects of fluorinated greenhouse gases on ozone and climate systems are evaluated, and the potential risks to human health and ecosystems are analyzed. The structure of the report is then designed, and the structure of the report includes introduction, methods, results, discussion, conclusions and recommendations. The report is generated according to the structure of the report.
Citation Information
Cited By
Device and method for detecting suffocating gas in deep pit operation
CN120121794A
