Industrial gas composition monitoring system and method based on big data analysis

Abstract

The application discloses an industrial gas component monitoring system and method based on big data analysis, and relates to the technical field of big data, comprising: establishing a three-dimensional model of an industrial site, obtaining historical leakage records, and judging and extracting target records; analyzing the component concentration of the gas to obtain the target concentration and target distance of the monitoring items corresponding to the storage tank; dividing the industrial site into a plurality of clustering areas to obtain the clustering points of each clustering area and calculate the early warning degree of the storage tank; dividing the industrial site into a plurality of sub-areas to obtain the characteristic ratio of each sub-area, performing regional iterative judgment, extracting the characteristic points corresponding to each sub-area, and then sampling at the characteristic points. The application analyzes the historical leakage records of the storage tank in the industrial site, sets a reasonable and reliable sampling point position, can effectively make the sampled data more representative and more in line with the actual monitoring requirements, avoids data distortion, and improves the sample reliability.

Description

Technical Field

[0001] This invention relates to the field of big data technology, specifically to an industrial gas composition monitoring system and method based on big data analysis. Background Technology

[0002] To ensure the stability and safety of industrial gases, they are generally stored in storage tanks. Monitoring the gases is crucial to determine if leaks or other anomalies exist. Gas monitoring instruments are directly installed in industrial production processes to continuously or intermittently monitor gas composition or physical properties in real time. Their core function is to achieve real-time monitoring and optimization of the production process through online acquisition, processing, and analysis of sampled data. However, current methods for sampling industrial gases using gas monitoring instruments often involve randomly selected sampling points without analysis based on storage tank distribution and historical leak data. This leads to sampled data that fails to reflect the true state and lacks sample reliability. Summary of the Invention

[0003] The purpose of this invention is to provide an industrial gas composition monitoring system and method based on big data analysis to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] The method for monitoring industrial gas composition based on big data analysis includes the following steps:

[0006] Obtain the layout drawings of the industrial site and the arrangement of storage tanks, create a 3D model of the industrial site, and mark the location of the storage tanks in the 3D model;

[0007] Acquire historical leakage records of storage tanks, extract and analyze gas physical parameters recorded during gas monitoring, and identify and extract target records from the leakage records;

[0008] Extract the storage tank corresponding to the target record, and analyze the composition concentration of gas at various locations in the industrial site during the monitoring process to obtain the target concentration and target distance of the monitoring item in the target record corresponding to the storage tank; acquire historical monitoring videos of the industrial site, divide the industrial site into several cluster areas according to the location of the staff in the monitoring videos, and obtain the cluster point of each cluster area.

[0009] The warning level of the storage tank is calculated based on the target concentration and target distance of the monitoring items and the cluster points of each cluster area;

[0010] Based on the current number of industrial gas samples, the industrial site is divided into several sub-regions. According to the warning level and location of each storage tank, the feature ratio of each sub-region is obtained. Based on the feature ratio, it is determined whether to perform regional iteration on each sub-region to obtain the final sub-regions. Then, the feature points corresponding to each sub-region are extracted, and sampling is performed on the feature points.

[0011] Preferably, the steps for judging and extracting the target record are as follows:

[0012] Obtain historical leakage records of the storage tank. Leakage records are generated when the gas monitoring instrument detects that the gas concentration around the target storage tank exceeds the preset concentration threshold, and after investigation, it is confirmed that only the target storage tank is leaking.

[0013] Gas monitoring instruments perform gas monitoring at fixed time intervals. During the monitoring process, the gas physical parameters corresponding to each monitoring item of a certain leak record are extracted. The gas physical parameters include pressure and flow rate. The variance between all pressures is calculated as the first target value, and the variance between all flow rates is calculated as the second target value. According to the preset weights corresponding to pressure and flow rate, the total target value of a certain leak record is obtained. If the total target value is less than the preset value threshold, then the leak record is taken as the target record.

[0014] Pressure and flow rate provide crucial environmental and dynamic parameter support for gas monitoring instruments during gas monitoring. They are important auxiliary indicators to ensure the accuracy, completeness, and adaptability of monitoring data. Gas concentration is related to pressure, and gas velocity is related to flow rate. During gas monitoring, sudden changes in pressure and flow rate may be caused by external factors such as momentary equipment failure, poor sensor contact, and environmental interference. These factors can lead to deviations between the monitored gas composition data and the actual situation. Therefore, records with such pressure and flow rate changes cannot provide reliable data support for the following calculations. Thus, it is necessary to identify and extract the target records in this step to make the calculation results more reliable.

[0015] Preferably, the steps for obtaining the target concentration and target distance of the monitoring entries in the target record corresponding to the storage tank are as follows:

[0016] The industrial site includes several storage tanks storing the same gas. The stored gas is designated as Q, and the concentration C of gas Q in the air is obtained in advance. Q ; Obtain the position P of the storage tank T corresponding to a target record in the 3D model. T Extract the location P of the gas monitoring instrument at time t for a certain monitoring item. t And the component concentration C of gas Q recorded by the gas monitoring instrument in monitoring item t. tThe concentration of component C t Subtract component concentration C Q The gas Q at position P is obtained. t Leakage concentration C T t ;

[0017] Based on gas Q at position P T Leakage concentration C T The target concentration of monitoring item t was obtained as C. T -C T t Position P T and position P t The distance between them is used as the target distance for monitoring item t.

[0018] Preferably, the steps for obtaining the cluster points of each cluster region are as follows: establish a planar coordinate system for the industrial site, extract several historical monitoring images of the industrial site, obtain the location of the staff in the monitoring images, and mark them in the planar coordinate system; use a clustering algorithm to aggregate the staff location points in the planar coordinate system to obtain several cluster regions, and randomly extract several staff location points from the cluster regions, calculate the average coordinate value, obtain the cluster points, and obtain the cluster points of each cluster region.

[0019] In this scheme, a clustering algorithm is required that does not require presetting the number of categories and does not change the original coordinates of sample points during the clustering process;

[0020] Preferably, the steps for calculating the warning level of a storage tank are as follows: Based on the target concentration and target distance of several monitoring items corresponding to a certain storage tank T, a function is established to show the change of target concentration with target distance, and a linear function is fitted to obtain the slope K corresponding to the fitted function. The coordinates of storage tank T in the plane coordinate system are then used as F. T Based on the area of ​​each cluster region and the number of personnel location points within it, as well as the coordinates of each cluster point and coordinate F... T The distance between them determines the warning level of storage tank T. Where W1 and W2 are the first and second weights, respectively, e is the natural index, D is the number of cluster regions, and S is the number of cluster regions. d Let N be the area of ​​the d-th cluster region. d Let L be the number of personnel location points within the d-th cluster region. d Let F be the coordinates of the cluster point corresponding to the d-th cluster region. T The distance between them.

[0021] Preferably, the steps for extracting feature points corresponding to each sub-region are as follows:

[0022] Obtain the current number of industrial gas samples M, and divide the industrial site into M equal sub-regions; extract several location points within a sub-region r, and calculate the average coordinates as the target point; obtain the distance H between a target point G and a storage tank X, and determine the warning level Y of storage tank X. X The degree of influence of storage tank X on target point G is obtained as follows: Sum the influence of all storage tanks on target point G to obtain the total influence of target point G.

[0023] The total influence of each target point is obtained and summed to obtain the target influence. The total influence is then divided by the target influence to obtain the feature ratio of each sub-region. The variance among all feature ratios is calculated. If the variance is greater than a preset variance threshold, the area A of a sub-region r is used as the determining factor. r eigenratio R r The iterative area of ​​subregion r is obtained as: A r (1-R r +1 / M), and then obtain the iteration area of ​​all sub-regions, and perform region iteration on each sub-region according to the iteration area;

[0024] A larger eigenvalue ratio indicates a greater impact of the storage tank on the sub-region, thus requiring greater monitoring efforts and denser sampling points. Therefore, a larger eigenvalue ratio should correspond to a smaller iteration area for the sub-region to ensure the reliability of the sampling points. The variance obtained from the eigenvalue ratio is an important factor in measuring whether the monitoring efforts of each sub-region are balanced.

[0025] The feature ratio is obtained again until the number of iterations is greater than the preset threshold or the variance is not greater than the variance threshold. The final sub-regions are obtained, and the target points of the last iteration are used as the feature points of the corresponding sub-regions.

[0026] The industrial gas composition monitoring system based on big data analysis includes a target record extraction module, an early warning level calculation module, and a feature point extraction module.

[0027] Target record extraction module: used to acquire layout drawings and storage tank layout diagrams of industrial sites, build a 3D model of the industrial site, and mark the location of storage tanks in the 3D model; acquire historical leakage records of storage tanks, extract and analyze the gas physical parameters recorded during gas monitoring, and realize the judgment and extraction of target records in leakage records;

[0028] Warning Level Calculation Module: This module is used to extract the storage tanks corresponding to the target records, analyze the gas composition and concentration at various locations within the industrial site during the monitoring process, and obtain the target concentration and target distance of the monitoring items within the target records corresponding to the storage tanks; acquire historical monitoring videos of the industrial site, divide the industrial site into several clustered areas based on the location of personnel in the monitoring videos, and obtain the cluster points of each clustered area; calculate the warning level of the storage tanks based on the target concentration and target distance of the monitoring items and the cluster points of each clustered area.

[0029] Feature point extraction module: Based on the current number of industrial gas samples, the industrial site is divided into several sub-regions. According to the warning level and location of each storage tank, the feature ratio of each sub-region is obtained. Based on the feature ratio, it is determined whether to perform region iteration on each sub-region to obtain the final sub-regions. The feature points corresponding to each sub-region are extracted and then sampled.

[0030] Preferably, the early warning level calculation module includes a monitoring item analysis unit, a clustering region division unit, and an early warning level calculation unit;

[0031] Monitoring item analysis unit: used to extract the storage tank corresponding to the target record, and analyze the composition concentration of gas at various locations in the industrial site during the monitoring process, to obtain the target concentration and target distance of the monitoring item in the target record corresponding to the storage tank;

[0032] Clustering region division unit: Used to acquire historical surveillance videos of industrial sites, divide the industrial sites into several clustering regions based on the location of workers in the surveillance videos, and obtain the cluster points of each clustering region;

[0033] Warning level calculation unit: used to calculate the warning level of the storage tank based on the target concentration and target distance of the monitoring item and the cluster points of each cluster area.

[0034] Preferably, the feature point extraction module includes a total influence calculation unit and a feature point extraction unit;

[0035] Total Impact Calculation Unit: Used to obtain the number of samples of industrial gas currently being sampled, to divide the industrial site into several sub-regions of equal area, to obtain the target points of the sub-regions; to obtain the distance of the target points from the storage tanks, to obtain the impact of the storage tanks on the target points based on the warning level of the storage tanks, and then to obtain the total impact of the target points.

[0036] Feature point extraction unit: used to obtain the total influence of each target point, obtain the feature ratio of each sub-region, and determine whether to perform region iteration on each sub-region based on the feature ratio, thereby obtaining the final sub-regions and extracting the feature points corresponding to each sub-region.

[0037] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides an industrial gas composition monitoring system and method based on big data analysis, including: establishing a three-dimensional model of the industrial site, acquiring historical leakage records, and judging and extracting target records; analyzing the composition concentration of the gas to obtain the target concentration and target distance of the monitoring item corresponding to the storage tank; dividing the industrial site into several clustered regions, obtaining the cluster points of each clustered region, and calculating the warning level of the storage tank; dividing the industrial site into several sub-regions, obtaining the feature ratio of each sub-region, performing iterative judgment of the region, extracting the feature points corresponding to each sub-region, and then sampling at the feature points. This invention, by analyzing historical leakage records of storage tanks in industrial sites and setting reasonable and reliable sampling point locations, can effectively make the sampled data more representative, more in line with actual monitoring needs, avoid data distortion, and improve sample reliability. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the process for monitoring industrial gas composition based on big data analysis according to the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Example: Figure 1 As shown, this invention provides a technical solution for industrial gas composition monitoring based on big data analysis, including the following steps:

[0041] 1. Obtain the layout drawings of the industrial site and the arrangement of storage tanks, create a 3D model of the industrial site, and mark the location of the storage tanks in the 3D model.

[0042] 2. Obtain historical leakage records of the storage tank, extract and analyze the gas physical parameters recorded during gas monitoring, and realize the judgment and extraction of target records in the leakage records.

[0043] Obtain historical leakage records of the storage tank. Leakage records are generated when the gas monitoring instrument detects that the gas concentration around the target storage tank exceeds the preset concentration threshold, and after investigation, it is confirmed that only the target storage tank is leaking.

[0044] Gas monitoring instruments perform gas monitoring at fixed time intervals. During the monitoring process, the gas physical parameters corresponding to each monitoring item of a certain leak record are extracted. The gas physical parameters include pressure and flow rate. The variance between all pressures is calculated as the first target value, and the variance between all flow rates is calculated as the second target value. According to the preset weights corresponding to pressure and flow rate, the total target value of a certain leak record is obtained. If the total target value is less than the preset value threshold, then the leak record is taken as the target record.

[0045] Pressure and flow rate provide crucial environmental and dynamic parameter support for gas monitoring instruments during gas monitoring. They are important auxiliary indicators to ensure the accuracy, completeness, and adaptability of monitoring data. Gas concentration is related to pressure, and gas velocity is related to flow rate. During gas monitoring, sudden changes in pressure and flow rate may be caused by external factors such as momentary equipment failure, poor sensor contact, and environmental interference. These factors can lead to deviations between the monitored gas composition data and the actual situation. Therefore, records with such pressure and flow rate changes cannot provide reliable data support for the following calculations. Thus, it is necessary to identify and extract the target records in this step to make the calculation results more reliable.

[0046] 3. Extract the storage tank corresponding to the target record, and analyze the composition concentration of the gas at various locations in the industrial site during the monitoring process to obtain the target concentration and target distance of the monitoring item in the target record corresponding to the storage tank.

[0047] The industrial site includes several storage tanks storing the same gas. The stored gas is designated as Q, and the concentration C of gas Q in the air is obtained in advance. Q ; Obtain the position P of the storage tank T corresponding to a target record in the 3D model. T Extract the location P of the gas monitoring instrument at time t for a certain monitoring item. t And the component concentration C of gas Q recorded by the gas monitoring instrument in monitoring item t. t The concentration of component C t Subtract component concentration C Q The gas Q at position P is obtained. t Leakage concentration C T t ;

[0048] Based on gas Q at position P T Leakage concentration C T The target concentration of monitoring item t was obtained as C. T -C T t Position P T and position P tThe distance between them is used as the target distance for monitoring item t.

[0049] 4. Obtain historical surveillance videos of the industrial site, divide the industrial site into several cluster areas based on the location of the staff in the surveillance videos, and obtain the cluster point of each cluster area.

[0050] Establish a planar coordinate system for the industrial site, extract several historical surveillance images of the industrial site, obtain the location of the staff in the surveillance images, and mark them in the planar coordinate system; use a clustering algorithm to aggregate the staff location points in the planar coordinate system to obtain several cluster regions, and randomly extract several staff location points from the cluster regions, calculate the average coordinates to obtain the cluster points, and obtain the cluster points for each cluster region.

[0051] In this solution, a clustering algorithm is required that does not require presetting the number of categories and does not change the original coordinates of the sample points during the clustering process. In this embodiment, the DBSCAN algorithm is used. After the clustering is divided by DBSCAN, there are several personnel location points. Using the DBSCAN algorithm to cluster personnel location points in the planar coordinate system is an existing technology, which will not be described in detail here.

[0052] 5. Calculate the warning level of the storage tank based on the target concentration and target distance of the monitoring items and the cluster points of each cluster area.

[0053] Based on the target concentration and target distance of several monitoring items corresponding to a certain storage tank T, a function of target concentration versus target distance is established, and a linear function is fitted to obtain the slope K corresponding to the fitted function. The coordinates of storage tank T in the plane coordinate system are then used as F. T Based on the area of ​​each cluster region and the number of personnel location points within it, as well as the coordinates of each cluster point and coordinate F... T The distance between them determines the warning level of storage tank T. Where W1 and W2 are the first and second weights, respectively, e is the natural index, D is the number of cluster regions, and S is the number of cluster regions. d Let N be the area of ​​the d-th cluster region. d Let L be the number of personnel location points within the d-th cluster region. d Let F be the coordinates of the cluster point corresponding to the d-th cluster region. T The distance between them.

[0054] A smaller slope K indicates that a greater amount of leaking gas can be detected at locations farther from the leaking tank, suggesting a more severe leak. Conversely, a larger slope K indicates that less leaking gas can be detected at locations farther from the leaking tank, suggesting a less severe leak. Therefore, a smaller slope K indicates a higher level of warning for the storage tank T. (Formula: y=e) -x If y is a function that takes integer values ​​from 0 to 1 when x > 0, and y decreases as x increases, then the formula y = e -x Used to analyze slope K;

[0055] In this scheme, the clustering region is the area where people are densely concentrated; the clustering region can be understood as each work area. Therefore, S... d / N d This represents the population density corresponding to the d-th cluster region, when S d / N d The larger the value, the greater the number of people or the duration of their stay in that area, while L... d The smaller the value of S, the closer the d-th cluster region is to storage tank T, and the greater its influence on storage tank T. Therefore, when S... d / N d The larger L is d The smaller the value, the greater the warning level of storage tank T; formula y=1-e -x If y is a function that takes integer values ​​from 0 to 1 when x > 0, and y increases as x increases, then the formula y = 1 - e^x can be applied. -x Used to analyze S d / N d and L d For details, please refer to the calculation of the early warning level Y of storage tank T. T .

[0056] 6. Based on the current number of industrial gas samples, the industrial site is divided into several sub-regions. According to the warning level and location of each storage tank, the feature ratio of each sub-region is obtained. Based on the feature ratio, it is determined whether to perform regional iteration on each sub-region to obtain the final sub-regions. The feature points corresponding to each sub-region are extracted, and then sampling is performed on the feature points.

[0057] 6-1 Obtain the current number of industrial gas samples M, and divide the industrial site into M equal sub-regions; extract several location points within a sub-region r, and calculate the average coordinates as the target point; obtain the distance H between a target point G and a storage tank X, and determine the warning level Y of storage tank X. X The degree of influence of storage tank X on target point G is obtained as follows: The total influence of all storage tanks on target point G is obtained by summing the influence of all storage tanks on target point G.

[0058] The meaning of the formula for the influence of target point G can be found in the above analysis of the early warning level of storage tanks, and will not be repeated here.

[0059] 6-2 Obtain the total influence degree of each target point and sum them to obtain the target degree. Divide each total influence degree by the target degree to obtain the feature ratio of each sub-region. Calculate the variance among all feature ratios. If the variance is greater than a preset variance threshold, determine the area A of a sub-region r. r eigenratio R r The iterative area of ​​subregion r is obtained as: A r (1-R r +1 / M), and then obtain the iteration area of ​​all sub-regions, and perform region iteration on each sub-region according to the iteration area;

[0060] A larger eigenvalue ratio indicates a greater impact of the storage tank on the sub-region, thus requiring stronger monitoring and denser sampling. Therefore, a larger eigenvalue ratio necessitates a smaller iteration area for that sub-region to ensure sampling reliability. The variance of the eigenvalue ratio is a crucial factor in assessing the balance of monitoring efforts across sub-regions. For example, setting M=10 (10 sampling points) divides the industrial site into 10 sub-regions. Since the eigenvalue ratio is based on the total influence of each target point, if the total influence of each target point is balanced (i.e., all target points have the same total influence), the eigenvalue ratio for each sub-region should be 0.1. However, if a sub-region has a eigenvalue ratio of 0.12, it indicates a greater monitoring effort and a smaller iteration area. Therefore, based on the area A before iteration... r The area obtained after iteration is: A r (1-R r +1 / M)=0.98A r Similarly, if the feature ratio of this sub-region is 0.08, then the area after iteration is: A r (1-R r +1 / M)=1.02A r .

[0061] The feature ratio is obtained again until the number of iterations is greater than the preset threshold or the variance is not greater than the variance threshold. The final sub-regions are obtained, and the target points of the last iteration are used as the feature points of the corresponding sub-regions.

[0062] This invention also provides an industrial gas composition monitoring system based on big data analysis, including a target record extraction module, an early warning level calculation module, and a feature point extraction module. The early warning level calculation module includes a monitoring item analysis unit, a clustering region division unit, and an early warning level calculation unit. The feature point extraction module includes a total impact level calculation unit and a feature point extraction unit. When the system executes the computer program, it implements the aforementioned industrial gas composition monitoring method based on big data analysis. Since this method has been described in detail above, it will not be repeated here.

[0063] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for monitoring industrial gas composition based on big data analysis, characterized in that, Includes the following steps: Obtain the layout drawings of the industrial site and the arrangement of storage tanks, create a 3D model of the industrial site, and mark the location of the storage tanks in the 3D model; Acquire historical leakage records of storage tanks, extract and analyze gas physical parameters recorded during gas monitoring, and identify and extract target records from the leakage records; Extract the storage tank corresponding to the target record, and analyze the composition concentration of gas at various locations in the industrial site during the monitoring process to obtain the target concentration and target distance of the monitoring item in the target record corresponding to the storage tank; acquire historical monitoring videos of the industrial site, divide the industrial site into several cluster areas according to the location of the staff in the monitoring videos, and obtain the cluster point of each cluster area. The warning level of the storage tank is calculated based on the target concentration and target distance of the monitoring items and the cluster points of each cluster area; Based on the current number of industrial gas samples, the industrial site is divided into several sub-regions. According to the warning level and location of each storage tank, the feature ratio of each sub-region is obtained. Based on the feature ratio, it is determined whether to perform regional iteration on each sub-region to obtain the final sub-regions. Then, the feature points corresponding to each sub-region are extracted, and sampling is performed on the feature points.

2. The industrial gas composition monitoring method based on big data analysis according to claim 1, characterized in that, The steps for judging and extracting the target record are as follows: Obtain historical leakage records of the storage tank. The leakage records are generated when the gas monitoring instrument detects that the gas concentration around the target storage tank exceeds a preset concentration threshold, and after investigation, it is confirmed that only the target storage tank is leaking. The gas monitoring instrument performs gas monitoring at fixed time intervals. During the monitoring process, the gas physical parameters corresponding to each monitoring item of a certain leak record are extracted. The gas physical parameters include pressure and flow rate. The variance between all pressures is calculated as the first target value, and the variance between all flow rates is calculated as the second target value. According to the preset weights corresponding to pressure and flow rate, the total target value of the certain leak record is obtained. If the total target value is less than the preset value threshold, the certain leak record is taken as the target record.

3. The industrial gas composition monitoring method based on big data analysis according to claim 1, characterized in that, The steps to obtain the target concentration and target distance of the monitoring entries in the target record corresponding to the storage tank are as follows: The industrial site includes several storage tanks storing the same gas. The stored gas is designated as Q, and the concentration C of gas Q in the air is obtained in advance. Q ; Obtain the position P of the storage tank T corresponding to a target record in the 3D model. T Extract the location P of the gas monitoring instrument at time t for a certain monitoring item. t And the component concentration C of gas Q recorded by the gas monitoring instrument in monitoring item t. t The concentration of component C t Subtract component concentration C Q The gas Q at position P is obtained. t Leakage concentration C T t ; Based on gas Q at position P T Leakage concentration C T The target concentration of monitoring item t was obtained as C. T -C T t Position P T and position P t The distance between them is used as the target distance for monitoring item t.

4. The industrial gas composition monitoring method based on big data analysis according to claim 3, characterized in that, The steps to obtain the cluster points of each cluster region are as follows: establish a planar coordinate system for the industrial site, extract several historical surveillance images of the industrial site, obtain the location of the staff in the surveillance images, and mark them in the planar coordinate system; use a clustering algorithm to aggregate the staff location points in the planar coordinate system to obtain several cluster regions, and randomly extract several staff location points from the cluster regions, calculate the average coordinates, obtain the cluster points, and obtain the cluster points of each cluster region.

5. The industrial gas composition monitoring method based on big data analysis according to claim 4, characterized in that, The steps for calculating the warning level of a storage tank are as follows: Based on the target concentration and target distance of several monitoring items corresponding to a certain storage tank T, establish a function of target concentration changing with target distance, and perform a linear function fitting to obtain the slope K corresponding to the fitting. Then, use the coordinates of storage tank T in the plane coordinate system as F. T Based on the area of ​​each cluster region and the number of personnel location points within it, as well as the coordinates of each cluster point and coordinate F... T The distance between them determines the warning level of storage tank T. Where W1 and W2 are the first and second weights, respectively, e is the natural index, D is the number of cluster regions, and S is the number of cluster regions. d Let N be the area of ​​the d-th cluster region. d Let L be the number of personnel location points within the d-th cluster region. d Let F be the coordinates of the cluster point corresponding to the d-th cluster region. T The distance between them.

6. The industrial gas composition monitoring method based on big data analysis according to claim 1, characterized in that, The steps for extracting feature points corresponding to each sub-region are as follows: Obtain the current number of industrial gas samples M, and divide the industrial site into M equal sub-regions; extract several location points within a certain sub-region r, and calculate the average coordinates as the target point; Obtain the distance H from a target point G to a storage tank X, and then determine the warning level Y of storage tank X. X The degree of influence of storage tank X on the target point G is obtained as follows: Sum the influence of all storage tanks on target point G to obtain the total influence of target point G. The total influence of each target point is obtained and summed to obtain the target influence. The total influence is then divided by the target influence to obtain the feature ratio of each sub-region. The variance among all feature ratios is calculated. If the variance is greater than a preset variance threshold, the area A of a sub-region r is used as the determining factor. r eigenratio R r The iterative area of ​​subregion r is obtained as: A r (1-R r +1 / M), and then obtain the iteration area of ​​all sub-regions, and perform region iteration on each sub-region according to the iteration area; The feature ratio is obtained again until the number of iterations is greater than the preset threshold or the variance is not greater than the variance threshold. The final sub-regions are obtained, and the target points of the last iteration are used as the feature points of the corresponding sub-regions.

7. An industrial gas composition monitoring system, used to execute the industrial gas composition monitoring method based on big data analysis as described in any one of claims 1-6, characterized in that, The system includes a target record extraction module, an early warning level calculation module, and a feature point extraction module; Target record extraction module: used to acquire layout drawings and storage tank layout diagrams of industrial sites, build a 3D model of the industrial site, and mark the location of storage tanks in the 3D model; acquire historical leakage records of storage tanks, extract and analyze the gas physical parameters recorded during gas monitoring, and realize the judgment and extraction of target records in leakage records; Warning Level Calculation Module: This module is used to extract the storage tanks corresponding to the target records, analyze the gas composition and concentration at various locations within the industrial site during the monitoring process, and obtain the target concentration and target distance of the monitoring items within the target records corresponding to the storage tanks; acquire historical monitoring videos of the industrial site, divide the industrial site into several clustered areas based on the location of personnel in the monitoring videos, and obtain the cluster points of each clustered area; calculate the warning level of the storage tanks based on the target concentration and target distance of the monitoring items and the cluster points of each clustered area. Feature point extraction module: Based on the current number of industrial gas samples, the industrial site is divided into several sub-regions. According to the warning level and location of each storage tank, the feature ratio of each sub-region is obtained. Based on the feature ratio, it is determined whether to perform region iteration on each sub-region to obtain the final sub-regions. The feature points corresponding to each sub-region are extracted and then sampled.

8. The industrial gas composition monitoring system according to claim 7, characterized in that, The early warning level calculation module includes a monitoring item analysis unit, a clustering region division unit, and an early warning level calculation unit. Monitoring item analysis unit: used to extract the storage tank corresponding to the target record, and analyze the composition concentration of gas at various locations in the industrial site during the monitoring process, to obtain the target concentration and target distance of the monitoring item in the target record corresponding to the storage tank; Clustering region division unit: Used to acquire historical surveillance videos of industrial sites, divide the industrial sites into several clustering regions based on the location of workers in the surveillance videos, and obtain the cluster points of each clustering region; Warning level calculation unit: used to calculate the warning level of the storage tank based on the target concentration and target distance of the monitoring item and the cluster points of each cluster area.

9. The industrial gas composition monitoring system according to claim 7, characterized in that, The feature point extraction module includes a total influence calculation unit and a feature point extraction unit; Total Impact Calculation Unit: Used to obtain the number of samples of industrial gas currently being sampled, to divide the industrial site into several sub-regions of equal area, to obtain the target points of the sub-regions; to obtain the distance of the target points from the storage tanks, to obtain the impact of the storage tanks on the target points based on the warning level of the storage tanks, and then to obtain the total impact of the target points. Feature point extraction unit: used to obtain the total influence of each target point, obtain the feature ratio of each sub-region, and determine whether to perform region iteration on each sub-region based on the feature ratio, thereby obtaining the final sub-regions and extracting the feature points corresponding to each sub-region.

Images