Method and system for nh3, no2, ch4 concentration inversion and leakage detection based on polarized light field analysis

Abstract

The present application relates to the technical field of optical detection, and discloses a method and system for NH3, NO2 and CH4 concentration inversion and leakage detection based on polarized light field analysis, which combines the reflection, scattering and absorption differences of gases on polarized light to analyze the polarization state, classifies the gases by using Poincare sphere mapping Stokes vector; quantifies the polarization state disorder degree by using polarization entropy, builds a differentiated model to invert the concentration; and detects the leakage by combining the polarization degree change amount. Meanwhile, the wind speed, humidity and temperature are synchronously collected, the polarization state drift is corrected by dynamic turbulence and aerosol extinction compensation, and the environmental compensation matrix double-stage correction interference is integrated, the matrix condition number is controlled to ensure numerical stability, and the calculation error caused by environmental fluctuations is avoided. The present application realizes the synchronous identification, concentration inversion and leakage detection of the three gases, solves the problems of difficult synchronous differentiation, poor anti-interference and low precision in industrial scene gas detection, and is suitable for chemical pipelines and natural gas stations.

Description

Technical Field

[0001] This invention relates to the field of optical gas detection, specifically to a method and system for inverting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis. Background Technology

[0002] In the field of optical gas detection, traditional solutions for gases such as NH3, NO2, and CH4 mainly include sensor electrochemical solutions, point infrared technology, and Fourier transform infrared technology. Multi-sensor collaborative solutions have also been developed to alleviate the limitations of single sensors. Polarized light imaging is used due to its advantages of low cost and large field of view, and tunable diode laser absorption spectroscopy (TDLAS) technology has also matured.

[0003] However, traditional solutions have significant drawbacks: electrochemical solutions have short lifespans and are prone to cross-sensitivity; point-based infrared spectroscopy cannot image; Fourier transform equipment is expensive; and under high humidity, gas characteristic signals are easily overlapped with water vapor, leading to poor accuracy and misjudgments. Multi-sensor collaborative systems are complex, costly, and unstable under high temperature and humidity. In industrial environments, aerosols, temperature, humidity, and wind speed interference can cause polarization state drift and light intensity attenuation, increasing detection and concentration inversion errors. Polarized light imaging is difficult to detect weak signals in the early stages of a leak; most technologies can only determine the presence of gas and lack multi-gas concentration inversion solutions. Traditional inversion errors are large, and algorithms are difficult to adapt to environmental changes. To address these issues, a solution based on polarized light field analysis is proposed: gas polarization characteristics are classified using a Poincaré sphere, and a differentiated inversion model is built using polarization entropy. Environmental parameters are collected simultaneously to construct a compensation matrix M. env Achieving dual-level correction and ensuring stability through the control matrix condition number, combined with polarization degree leakage detection, can overcome the shortcomings of traditional methods, unleash the potential of polarized light imaging, and meet the needs of precise industrial testing. Summary of the Invention

[0004] Therefore, the purpose of this invention is to provide a method and system for inverting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis, aiming to solve the problems of difficulty in synchronously distinguishing multiple gases, poor stability due to environmental interference, large concentration inversion errors, and low leak detection accuracy in existing optical gas detection technologies.

[0005] To achieve the aforementioned objectives, the technical solution adopted is as follows: A method for inverting NH3, NO2, and CH4 concentrations and detecting leaks based on polarization field analysis includes the following steps: S1, Multi-source data collaborative acquisition Synchronously acquire the intensity of four-directional polarized light at 0°, 45°, 90°, and 135°. At the same time, wind speed (v) and humidity are obtained through environmental parameter sensors. The ambient temperature T is used as the input data for the four-directional polarized light intensity and the ambient data. Unlike the traditional single light intensity acquisition, the four-directional polarized light intensity and the ambient parameters are acquired in conjunction, providing multi-dimensional input for subsequent environmental adaptive correction and solving the problem of insufficient stability caused by ignoring environmental interference. S2, Polarization State Conversion and Two-Stage Environmental Correction Through Stokes transformation matrix Convert the four-axis polarized light intensity into a Stokes vector. The Stokes transformation matrix The expression is: ; Construct an environmental compensation matrix that integrates turbulence phase correction and aerosol extinction compensation. The expression is: ; pass Complete polarization state correction; among which, turbulent phase delay amount And wind speed hour Increased by 30%, aerosol extinction coefficient and humidity hour Increase by 12%, number of control conditions To ensure computational stability; a "dual-level dynamic correction mechanism" is used to specifically suppress interference from aerosol scattering and turbulent drift, avoiding the polarization distortion that is common in traditional technologies in complex industrial environments; S3, Polarization Entropy Feature Quantization Analysis Constructing polarization state vectors ,in , Illumination fluctuation interference is eliminated through normalization, and the covariance matrix is ​​used. eigenvalues Calculate polarization entropy The polarization entropy, an innovative physical parameter introduced in this invention, can quantify the degree of gas perturbation on polarized light, providing a basis for differential characteristics for subsequent concentration inversion and avoiding the problem of difficulty in distinguishing multiple gases due to reliance on a single optical feature. S4, Poincaré Sphere Spatial Clustering and Intelligent Classification normalized parameters Mapped to a two-dimensional Poincaré sphere, a pre-trained SVM classifier is used to identify gas types. The SVM classifier employs a radial basis function kernel. The classification function is The three types of gases form specific clusters on the Poincaré sphere, with NO2 concentrated in [specific region]. NH3 is distributed in CH4 is located The confidence-weighted algorithm is used to optimize the judgment in areas with ambiguous boundaries; the innovation lies in the "fusion of spatial features and machine learning" to achieve accurate differentiation of multiple gases and overcome the defects of misjudgment in cross-identification of traditional technologies. S5, Differential Concentration Inversion Based on the gas characteristics, the inversion model is adapted. NO2 (strong absorption) adopts the exponential growth curve in the high entropy region, NH3 (weak scattering) adopts the linear relationship, and methane (hydrocarbon) is normalized after baseline compensation. The concentration inversion is achieved through the specific correlation between polarization entropy and gas characteristics, which solves the problem of large inversion error. S6, Environmentally Adaptive Leak Detection Calculate the degree of polarization and spatial variation : ; ; Constructing threshold equations that are dynamically correlated with environmental parameters Among them, the threshold increases by 4% for every 1 m / s increase in wind speed, and decreases by 15% when humidity is >80%. The system marks the leakage area in time, and outputs the leakage coordinates after filtering out false targets by combining connected component analysis, with a coordinate error ≤0.1m; the innovation lies in the "dynamic threshold mechanism", which makes the detection accuracy adapt to environmental changes; S7, Multi-parameter fusion and hierarchical alarm By integrating data on gas type, concentration, and leak location, the system converts the data into real geographic coordinates and generates a 3D distribution map. Based on changes in concentration and leak area, it triggers graded alarms. For example, a Level I alarm corresponds to a concentration >100 ppm and leak expansion, while a Level II alarm corresponds to methane >5% LEL (Lower Explosive Limit). Through multi-dimensional information fusion, it provides intuitive and decision-making detection results for industrial scenarios, meeting the needs of high-precision monitoring.

[0006] As a further improvement of the present invention, in step S1, a polarization imaging acquisition module is used to synchronously acquire the intensity of four-directional polarized light. The polarization imaging acquisition module includes a polarization light source and a detector. The detector is used to capture optical signals in the gas overflow path. The environmental parameter sensor is linked with the polarization imaging acquisition module to realize the synchronous acquisition of polarization light intensity and environmental parameters.

[0007] As a further improvement of the present invention, in step S2, the turbulence phase correction uses a Kalman filter to track the phase delay in real time. The aerosol extinction compensation is based on an aerosol transmittance correction model. ,in,d Optical path length θ The angle of incidence is denoted as .

[0008] As a further improvement of the present invention, in step S3, the covariance matrix The eigenvalues ​​were obtained through statistical analysis of the polarization state vector P. The eigenvalue decomposition algorithm is used to quantify the degree of gas disturbance to polarized light.

[0009] As a further improvement of the present invention, in step S4, the pre-trained SVM classifier is trained using a large amount of labeled gas polarization feature data, and the parameters of the radial basis kernel function are optimized using cross-validation during the training process. and the weights in the classification function With bias The pre-trained SVM classifier is trained on a large amount of sample data and can accurately capture the clustering characteristics of different gases on the Poincaré sphere, achieving accurate differentiation of multiple gases and overcoming the shortcomings of traditional techniques in cross-identification and misjudgment.

[0010] As a further improvement of the present invention, in step S5, the adaptation and inversion model based on gas characteristics specifically states that NO2 concentration is exponentially inversely proportional to polarization entropy, as expressed by: The concentration of NH3 is linearly proportional to the polarization entropy, as expressed by: The CH4 concentration is linearly related to the normalized entropy value, as expressed by: Where k1, k2, k3, and k4 are model coefficients, and H is the polarization entropy. For the minimum polarization entropy, This represents the maximum polarization entropy.

[0011] As a further improvement of the present invention, in step S6, the connected component analysis adopts the eight-neighbor connected region labeling algorithm to analyze the area and shape characteristics of the labeled leakage area, and filter out false targets whose area is less than a preset threshold and whose shape does not conform to the gas leakage diffusion characteristics.

[0012] As a further improvement of the present invention, in step S7, the three-dimensional distribution map is generated using three-dimensional visualization technology to intuitively display the spatial distribution of gas concentration and the relationship between the leakage location. The graded alarm is realized through an audible and visual alarm device, and the alarm information and detection data are uploaded to the industrial monitoring platform.

[0013] A system for retrieving and detecting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis, comprising: Multi-source data acquisition module: used for synchronous acquisition of four-directional polarized light intensity at 0°, 45°, 90°, and 135°. And wind speed v, humidity The environmental data, including temperature T, is collected by the multi-source data acquisition module, which includes a polarization imaging unit and an environmental parameter sensing unit. The polarization imaging unit includes a polarization light source and a detector. The environmental parameter sensing unit includes a wind speed sensor, a humidity sensor, and a temperature sensor. Data processing module: connected to the multi-source data acquisition module, used to execute steps S2-S6 in any of the above methods, to realize polarization state conversion and two-level environmental correction, polarization entropy feature quantification and analysis, Poincaré sphere spatial clustering and intelligent classification, differential concentration inversion and environmental adaptive leakage detection; The result output and alarm module is connected to the data processing module and is used to fuse gas type, concentration, and leak location data, convert them into real geographic coordinates, and generate a three-dimensional distribution map. It triggers graded alarms based on changes in concentration and leak area. The result output and alarm module includes a display unit and an alarm unit. The display unit is used to display the three-dimensional distribution map and detection data, and the alarm unit is used to implement audible and visual alarms and upload alarm information to the industrial monitoring platform.

[0014] As a further improvement of the present invention, the data processing module adopts an embedded processor or an industrial computer, which has high-speed data operation and real-time processing capabilities, supports parallel processing of Stokes transformation matrix operation, environmental compensation matrix operation, polarization entropy calculation, SVM classification, concentration inversion model operation and leakage detection algorithm, and ensures real-time output of detection results.

[0015] The beneficial effects of this invention are: This invention first establishes a complete technical architecture for data acquisition, environmental correction, feature analysis, and multi-parameter fusion. By integrating polarization optical characteristics and environmental adaptive mechanisms, it overcomes the problems of fragmentation and poor coordination in traditional detection schemes. From a technical framework perspective, it provides an integrated solution for the detection of multiple gases such as NH3, NO2, and CH4, ensuring the integrity of the entire detection process and the coordinated cooperation of each link.

[0016] Supported by this architecture, the core detection function can be accurately implemented: on the one hand, by using the Poincaré sphere to map the Stokes vector and combining it with the SVM classifier, the specific modulation characteristics of different gases on polarized light are accurately captured, effectively solving the identification error caused by the cross-interference of multiple gases and achieving accurate differentiation of gas types; on the other hand, polarization entropy is introduced to quantify the disorder of polarization state, and a differentiated concentration inversion model is constructed for the differences in optical properties of the three gases, which greatly improves the accuracy of concentration detection.

[0017] Meanwhile, this invention effectively suppresses the influence of disturbances such as turbulence and temperature and humidity changes in industrial environments on the polarization state through a two-level dynamic compensation mechanism—aerosol transmittance correction and Mueller matrix phase correction—reducing the false alarm rate and enhancing the system's anti-interference capability. Furthermore, it achieves accurate leak detection based on the change in polarization degree and a dynamic threshold equation, improving leak location accuracy. Ultimately, it realizes simultaneous identification, concentration inversion, and leak detection of three gases, significantly improving the anti-interference capability and overall accuracy of multi-gas detection in industrial scenarios. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a model diagram of the application scenario of the present invention; Figure 2 This is a flowchart of the method for inverting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis according to the present invention. Figure 3 This is a Stokes parameter distribution diagram according to the present invention; Figure 4 This is the polarization entropy concentration inversion diagram of the present invention; Figure 5 This is the leakage boundary diagram of the DoP change in this invention;

[0019] Figure 6 This is a projection diagram of the Poincaré sphere classification according to the present invention. Detailed Implementation

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present application.

[0022] A method for inverting NH3, NO2, and CH4 concentrations and detecting leaks based on polarization field analysis includes the following steps: S1, Multi-source data collaborative acquisition Synchronously acquire the intensity of four-directional polarized light at 0°, 45°, 90°, and 135°. At the same time, wind speed (v) and humidity are obtained through environmental parameter sensors. The ambient temperature T is used as the input data for the four-directional polarized light intensity and the ambient data. Unlike the traditional single light intensity acquisition, the four-directional polarized light intensity and the ambient parameters are acquired in conjunction, providing multi-dimensional input for subsequent environmental adaptive correction and solving the problem of insufficient stability caused by ignoring environmental interference. S2, Polarization State Conversion and Two-Stage Environmental Correction Through Stokes transformation matrix Convert the four-axis polarized light intensity into a Stokes vector. The Stokes transformation matrix The expression is: ; Construct an environmental compensation matrix that integrates turbulence phase correction and aerosol extinction compensation. The expression is: ; pass Complete polarization state correction; among which, turbulent phase delay amount And wind speed hour Increased by 30%, aerosol extinction coefficient and humidity hour Increase by 12%, number of control conditions To ensure computational stability; a "dual-level dynamic correction mechanism" is used to specifically suppress interference from aerosol scattering and turbulent drift, avoiding the polarization distortion that is common in traditional technologies in complex industrial environments; S3, Polarization Entropy Feature Quantization Analysis Constructing polarization state vectors ,in , Illumination fluctuation interference is eliminated through normalization, and the covariance matrix is ​​used. eigenvalues Calculate polarization entropy The polarization entropy, an innovative physical parameter introduced in this invention, can quantify the degree of gas perturbation on polarized light, providing a basis for differential characteristics for subsequent concentration inversion and avoiding the problem of difficulty in distinguishing multiple gases due to reliance on a single optical feature. S4, Poincaré Sphere Spatial Clustering and Intelligent Classification normalized parameters Mapped to a two-dimensional Poincaré sphere, a pre-trained SVM classifier is used to identify gas types. The SVM classifier employs a radial basis function kernel. The classification function is The three types of gases form specific clusters on the Poincaré sphere, with NO2 concentrated in [specific region]. NH3 is distributed in CH4 is located The confidence-weighted algorithm is used to optimize the judgment in areas with ambiguous boundaries; the innovation lies in the "fusion of spatial features and machine learning" to achieve accurate differentiation of multiple gases and overcome the defects of misjudgment in cross-identification of traditional technologies. S5, Differential Concentration Inversion Based on the gas characteristics, the inversion model is adapted. NO2 (strong absorption) adopts the exponential growth curve in the high entropy region, NH3 (weak scattering) adopts the linear relationship, and methane (hydrocarbon) is normalized after baseline compensation. The concentration inversion is achieved through the specific correlation between polarization entropy and gas characteristics, which solves the problem of large inversion error. S6, Environmentally Adaptive Leak Detection Calculate the degree of polarization and spatial variation : ; ; Constructing threshold equations that are dynamically correlated with environmental parameters Among them, the threshold increases by 4% for every 1 m / s increase in wind speed, and decreases by 15% when humidity is >80%. The system marks the leakage area in time, and outputs the leakage coordinates after filtering out false targets by combining connected component analysis, with a coordinate error ≤0.1m; the innovation lies in the "dynamic threshold mechanism", which makes the detection accuracy adapt to environmental changes; S7, Multi-parameter fusion and hierarchical alarm By integrating data on gas type, concentration, and leak location, the system converts the data into real geographic coordinates and generates a 3D distribution map. Based on changes in concentration and leak area, it triggers tiered alarms. For example, a Level I alarm corresponds to a concentration >100ppm and leak expansion, indicating a high concentration risk and escalating leak; a Level II alarm corresponds to methane >5%LEL, indicating an explosion risk. Through multi-dimensional information fusion, it provides intuitive and decision-making detection results for industrial scenarios, meeting the needs of high-precision monitoring.

[0023] In step S1, a polarization imaging acquisition module is used to synchronously acquire the intensity of four-directional polarized light. The polarization imaging acquisition module includes a polarization light source and a detector. The detector is used to capture optical signals in the gas overflow path. The environmental parameter sensor is linked with the polarization imaging acquisition module to realize the synchronous acquisition of polarization light intensity and environmental parameters.

[0024] In step S2, the turbulence phase correction uses a Kalman filter to track the phase delay in real time. The aerosol extinction compensation is based on an aerosol transmittance correction model. ,in,d Optical path length θ The angle of incidence is denoted as .

[0025] In step S3, the covariance matrix The eigenvalues ​​were obtained through statistical analysis of the polarization state vector P. The eigenvalue decomposition algorithm is used to quantify the degree of gas disturbance to polarized light.

[0026] In step S4, the pre-trained SVM classifier is trained using a large amount of labeled gas polarization feature data. During the training process, cross-validation is used to optimize the parameters of the radial basis kernel function. and the weights in the classification function With bias The pre-trained SVM classifier is trained on a large amount of sample data and can accurately capture the clustering characteristics of different gases on the Poincaré sphere, achieving accurate differentiation of multiple gases and overcoming the shortcomings of traditional techniques in cross-identification and misjudgment.

[0027] In step S5, the specific adaptation and inversion model based on gas characteristics is as follows: NO2 (strong absorption) concentration is exponentially inversely proportional to polarization entropy, expressed as follows: The concentration of NH3 (weakly scattering) is linearly proportional to the polarization entropy, as expressed by: The concentration of CH4 (hydrocarbons) is linearly related to the normalized entropy value, as expressed by: Where k1, k2, k3, and k4 are model coefficients, and H is the polarization entropy. For the minimum polarization entropy, This represents the maximum polarization entropy.

[0028] The calibration process for the model coefficients k1, k2, k3, and k4 is as follows: Under constant temperature (25℃) and constant pressure (standard atmospheric pressure) conditions, a series of standard gases of known concentrations of NH3, NO2, and CH4 are prepared, and polarization entropy data corresponding to each concentration are collected. The least squares method is used to fit the data to obtain the model coefficients, ensuring the accuracy of the inversion model.

[0029] In step S6, the connected component analysis uses an eight-neighbor connected region labeling algorithm to analyze the area and shape features of the labeled leakage area, filtering out false targets whose area is smaller than a preset threshold and whose shape does not conform to the gas leakage diffusion characteristics.

[0030] In step S7, the three-dimensional distribution map is generated using three-dimensional visualization technology, which intuitively displays the spatial distribution of gas concentration and the relationship between the leak location. The graded alarm is realized through an audible and visual alarm device, and the alarm information and detection data are uploaded to the industrial monitoring platform.

[0031] A system for retrieving and detecting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis, comprising: Multi-source data acquisition module: used for synchronous acquisition of four-directional polarized light intensity at 0°, 45°, 90°, and 135°. And wind speed v, humidity The environmental data, including temperature T, is collected by the multi-source data acquisition module, which includes a polarization imaging unit and an environmental parameter sensing unit. The polarization imaging unit includes a polarization light source and a detector. The environmental parameter sensing unit includes a wind speed sensor, a humidity sensor, and a temperature sensor. Data processing module: connected to the multi-source data acquisition module, used to execute steps S2-S6 in any of the above methods, to realize polarization state conversion and two-level environmental correction, polarization entropy feature quantification and analysis, Poincaré sphere spatial clustering and intelligent classification, differential concentration inversion and environmental adaptive leakage detection; The result output and alarm module is connected to the data processing module and is used to fuse gas type, concentration, and leak location data, convert them into real geographic coordinates, and generate a three-dimensional distribution map. It triggers graded alarms based on changes in concentration and leak area. The result output and alarm module includes a display unit and an alarm unit. The display unit is used to display the three-dimensional distribution map and detection data, and the alarm unit is used to implement audible and visual alarms and upload alarm information to the industrial monitoring platform.

[0032] The data processing module uses an embedded processor or industrial computer, which has high-speed data computing and real-time processing capabilities. It supports parallel processing of Stokes transformation matrix operations, environmental compensation matrix operations, polarization entropy calculation, SVM classification, concentration inversion model operations, and leakage detection algorithms, ensuring real-time output of detection results.

[0033] In its specific implementation, this invention differs from the single data acquisition mode of traditional optical detection. It synchronously acquires four-directional polarized light intensity (0°, 45°, 90°, 135°) in conjunction with environmental parameters such as wind speed, humidity, and temperature, based on the application scenario model diagram (…). Figure 1 As can be seen in the diagram, the polarized light source and detector work together to capture optical signals in the gas overflow path, while the environmental sensor acquires environmental data around the pipeline in real time. This multi-source data collaborative input method lays the foundation for accurate correction of subsequent environmental interference and avoids the problem of insufficient detection stability caused by ignoring environmental factors.

[0034] In the polarization state processing stage, a "two-stage dynamic correction mechanism" is adopted for optimization: first, the four-directional light intensity is converted into a Stokes vector through the Stokes transformation matrix to accurately extract the polarization characteristics of the light field; then, an environmental compensation matrix integrating turbulence phase correction and aerosol extinction compensation is constructed, in which the turbulence phase delay is dynamically adjusted with wind speed, the aerosol extinction coefficient is adaptively changed with humidity, and the calculation stability is ensured by controlling the condition number of the matrix, effectively suppressing polarization state distortion caused by aerosol scattering and turbulence drift in complex industrial environments. (Flowchart of the method is shown below.) Figure 2 This clearly demonstrates the critical position of this calibration process in the overall technical workflow, and this mechanism improves the application of traditional technologies in harsh environments.

[0035] To achieve accurate differentiation and concentration inversion of multiple gases, the physical parameter "polarization entropy" is introduced: a polarization state vector containing normalized Stokes parameters and polarization entropy is constructed, and the polarization entropy is calculated using the eigenvalues ​​of the covariance matrix to quantify the degree of gas perturbation of polarized light. Based on the differences in the optical properties of different gases, differentiated inversion models are adapted for NH3, NO2, and CH4—NO2, due to its strong absorption, uses an exponential curve in the high-entropy region; NH3, due to its weak scattering, uses a linear relationship; and CH4, after baseline compensation, is inverted using normalized entropy values. The polarization entropy concentration inversion diagram is shown below. Figure 4 It intuitively presents the correlation characteristics between different gas concentrations and polarization entropy, improves the accuracy of concentration inversion, and solves the problem of large inversion errors caused by traditional technology relying on a single optical feature.

[0036] In gas type identification, normalized Stokes parameters are mapped onto a two-dimensional Poincaré sphere, utilizing the specific clustering characteristics of the three gas types (NO2 is concentrated in...). NH3 is distributed in CH4 is located (For regions with ambiguous boundaries, a confidence-weighted algorithm is used to optimize the determination), Poincaré sphere classification projection map ( Figure 6 This clearly demonstrates the clustering distribution, and intelligent classification is achieved by combining a pre-trained SVM classifier (using radial basis kernel function). For regions with ambiguous boundaries, a confidence-weighted algorithm is used to optimize the judgment, improving the misjudgment situation of multi-gas cross-identification in traditional technology.

[0037] The leak detection process employs a "dynamic threshold mechanism": by calculating the degree of polarization and its spatial variation, a threshold equation dynamically correlated with environmental parameters is constructed. Leakage boundary diagram of change ( Figure 5 The data shows that areas where the polarization degree change exceeds the threshold are accurately marked as leak areas. When the polarization degree change exceeds the threshold, the leak area is marked. After filtering out false targets through connected component analysis, the leak-related information is output, which is superior to traditional leak detection.

[0038] Stokes parameter distribution plot ( Figure 3 This demonstrates the distribution differences of processed polarization feature parameters in different gas regions, providing reliable basic data for subsequent species identification and concentration inversion. Finally, through multi-parameter fusion technology, data such as gas type, concentration, and leak location are converted into real geographic coordinates and a 3D distribution map is generated. Based on changes in concentration and leak area, graded alarms are triggered; for example, Level I corresponds to high concentration and leak expansion, while Level II targets excessive methane levels. This provides intuitive and decision-making detection results for industrial scenarios, meeting the high-precision monitoring needs of chemical pipelines, natural gas stations, and other similar environments.

[0039] Through the above design, this invention forms a complete technical system of multi-source data fusion, environmental adaptive correction, polarization feature quantification and analysis, and dynamic threshold detection. It solves the core problems of difficulty in synchronously distinguishing multiple gases, weak anti-interference ability, and low detection accuracy in existing technologies. Its beneficial effects are reflected in: by combining polarization entropy with Poincaré sphere classification, the accuracy of synchronous identification of three gases is improved; the two-stage turbulence compensation mechanism significantly reduces the detection error caused by environmental interference and the false alarm rate of turbulence, which significantly meets the high-precision detection requirements of industrial scenarios such as chemical pipelines and natural gas stations.

[0040] Complete system functions: ; This scheme achieves the fusion of multiple physical quantities through a three-level cascaded model, wherein: ; Leakage detection module: ; Concentration Inversion and Classification Module: ; in: For Stokes vectors, Let H be the environmental compensation operator, P be the polarization state vector, and H be the polarization state vector. p For polarization entropy, F system For system functions, f type For gas classification functions, f conc为 Concentration inversion function, f leak为 Leakage detection function For spatial variation, This is an adaptive threshold.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, component splitting or combination, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for inverting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis, characterized in that, Includes the following steps: S1, Multi-source data collaborative acquisition Synchronously acquire the intensity of four-directional polarized light at 0°, 45°, 90°, and 135°. At the same time, wind speed (v) and humidity are obtained through environmental parameter sensors. The environmental data, including temperature T, are used as input data, along with the four-directional polarized light intensity and environmental data. S2, Polarization State Conversion and Two-Stage Environmental Correction Through Stokes transformation matrix Convert the four-axis polarized light intensity into a Stokes vector. The Stokes transformation matrix The expression is: ； Construct an environmental compensation matrix that integrates turbulence phase correction and aerosol extinction compensation. The expression is: ； pass Complete polarization state correction; among which, turbulent phase delay amount And wind speed hour Increased by 30%, aerosol extinction coefficient and humidity hour Increase by 12%, number of control conditions To ensure computational stability; S3, Polarization Entropy Feature Quantization Analysis Constructing polarization state vectors ,in , Illumination fluctuation interference is eliminated through normalization, and the covariance matrix is ​​used. eigenvalues Calculate polarization entropy To quantify the degree of gas disturbance to polarized light; S4, Poincaré Sphere Spatial Clustering and Intelligent Classification normalized parameters Mapped to a two-dimensional Poincaré sphere, a pre-trained SVM classifier is used to identify gas types. The SVM classifier employs a radial basis function kernel. The classification function is The three types of gases form specific clusters on the Poincaré sphere, with NO2 concentrated in [specific region]. NH3 is distributed in CH4 is located For regions with ambiguous boundaries, a confidence-weighted algorithm is used to optimize the determination. S5, Differential Concentration Inversion Based on the gas characteristic adaptation inversion model, NO2 adopts the exponential growth curve in the high entropy region, NH3 adopts the linear relationship, and methane is normalized after baseline compensation. Concentration inversion is achieved through the specific correlation between polarization entropy and gas characteristics. S6, Environmentally Adaptive Leak Detection Calculate the degree of polarization and spatial variation : ； ； Constructing threshold equations that are dynamically correlated with environmental parameters Among them, the threshold increases by 4% for every 1 m / s increase in wind speed, and decreases by 15% when humidity is >80%. The leakage area is marked in time, and after filtering out false targets by combining connected component analysis, the leakage coordinates are output with a coordinate error ≤ 0.1m; S7, Multi-parameter fusion and hierarchical alarm The system integrates data on gas type, concentration, and leak location, converts it into real geographic coordinates, and generates a 3D distribution map. It then triggers tiered alarms based on changes in concentration and leak area.

2. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S1, a polarization imaging acquisition module is used to synchronously acquire the intensity of four-directional polarized light. The polarization imaging acquisition module includes a polarization light source and a detector. The detector is used to capture optical signals in the gas overflow path. The environmental parameter sensor is linked with the polarization imaging acquisition module to realize the synchronous acquisition of polarization light intensity and environmental parameters.

3. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S2, the turbulence phase correction uses a Kalman filter to track the phase delay in real time. The aerosol extinction compensation is based on an aerosol transmittance correction model. ,in, d Optical path length θ The angle of incidence is denoted as .

4. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S3, the covariance matrix The eigenvalues ​​were obtained through statistical analysis of the polarization state vector P. The eigenvalue decomposition algorithm is used to quantify the degree of gas disturbance to polarized light.

5. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S4, the pre-trained SVM classifier is trained using a large amount of labeled gas polarization feature data. During the training process, cross-validation is used to optimize the parameters of the radial basis kernel function. and the weights in the classification function With bias .

6. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S5, the specific method for adapting the inversion model based on gas characteristics is as follows: NO2 concentration is exponentially inversely proportional to polarization entropy, expressed as follows: The concentration of NH3 is linearly proportional to the polarization entropy, as expressed by: The CH4 concentration is linearly related to the normalized entropy value, as expressed by: Where k1, k2, k3, and k4 are model coefficients, and H is the polarization entropy. For the minimum polarization entropy, This represents the maximum polarization entropy.

7. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S6, the connected component analysis uses an eight-neighbor connected region labeling algorithm to analyze the area and shape features of the labeled leakage area, filtering out false targets whose area is smaller than a preset threshold and whose shape does not conform to the gas leakage diffusion characteristics.

8. The method for inverting the concentrations of NH3, NO2, and CH4 and detecting leakage based on polarization field analysis according to claim 1, characterized in that: In step S7, the three-dimensional distribution map is generated using three-dimensional visualization technology, which intuitively displays the spatial distribution of gas concentration and the relationship between the leak location. The graded alarm is realized through an audible and visual alarm device, and the alarm information and detection data are uploaded to the industrial monitoring platform.

9. A system for inverting the concentrations of NH3, NO2, and CH4 and detecting leaks based on polarization field analysis, characterized in that, include: Multi-source data acquisition module: used for synchronous acquisition of four-directional polarized light intensity at 0°, 45°, 90°, and 135°. And wind speed v, humidity The environmental data, including temperature T, is collected by the multi-source data acquisition module, which includes a polarization imaging unit and an environmental parameter sensing unit. The polarization imaging unit includes a polarization light source and a detector. The environmental parameter sensing unit includes a wind speed sensor, a humidity sensor, and a temperature sensor. Data processing module: connected to the multi-source data acquisition module, used to execute steps S2-S6 in the method of any one of claims 1-8, to realize polarization state conversion and two-level environmental correction, polarization entropy feature quantification and analysis, Poincaré sphere spatial clustering and intelligent classification, differential concentration inversion and environmental adaptive leakage detection; The result output and alarm module is connected to the data processing module and is used to fuse gas type, concentration, and leak location data, convert them into real geographic coordinates, and generate a three-dimensional distribution map. It triggers graded alarms based on changes in concentration and leak area. The result output and alarm module includes a display unit and an alarm unit. The display unit is used to display the three-dimensional distribution map and detection data, and the alarm unit is used to implement audible and visual alarms and upload alarm information to the industrial monitoring platform.

10. The NH3, NO2, and CH4 concentration inversion and leakage detection system based on polarization field analysis as described in claim 9, characterized in that: The data processing module uses an embedded processor or industrial computer, which has high-speed data computing and real-time processing capabilities. It supports parallel processing of Stokes transformation matrix operations, environmental compensation matrix operations, polarization entropy calculation, SVM classification, concentration inversion model operations, and leakage detection algorithms, ensuring real-time output of detection results.

Images