Efficient refrigeration type infrared detector gas leakage detection system and method

Abstract

The invention discloses an efficient refrigeration type infrared detector gas leakage detection system and method, the system comprises a refrigeration type detector module, a gas enhancement algorithm module and an intelligent operation and maintenance module, the refrigeration type detector module is used for collecting an infrared image of leaked gas and transmitting the infrared image to the gas enhancement algorithm module; the gas enhancement algorithm module comprises a pretreatment unit and a detection unit; and the intelligent operation and maintenance module is used for performing health assessment on the refrigeration type detector module. According to the high-efficiency refrigeration type infrared detector gas leakage detection system and method disclosed by the invention, by optimizing the material, algorithm and system integration of the refrigeration type infrared detector, the bottlenecks of traditional equipment in sensitivity, environmental adaptability and multi-scene generalization ability are broken through, and high-efficiency and accurate detection of trace gas leakage in industrial scenes is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of gas detection, and particularly relates to a high-efficiency refrigeration type infrared detector gas leakage detection system and method. BACKGROUND

[0002] In the field of industrial gas leakage detection, there are many technical problems at present, which seriously restrict the efficiency and accuracy of detection, as follows: Sensitivity and detection distance contradiction: non-refrigeration type infrared detectors perform poorly in detecting trace gas leakage (such as <0.001 mL / s) at a long distance (such as more than 20 meters) due to the limitation of thermal noise, and are easily disturbed by environmental temperature fluctuations.

[0003] Refrigeration system life limitation: traditional refrigeration type detectors rely on mercury cadmium telluride (MCT) materials, and the service life of the refrigerator is usually less than 8000 hours, with high maintenance cost, which greatly restricts its long-term deployment.

[0004] Poor adaptability to complex scenes: most existing algorithms rely on a single technology, such as frame difference method or optical flow method, which is difficult to deal with dynamic background interference, such as light changes, equipment vibration, etc., resulting in a high false alarm rate.

[0005] Therefore, the above problems are further improved. SUMMARY

[0006] The main purpose of the application is to provide a high-efficiency refrigeration type infrared detector gas leakage detection system and method, which breaks through the bottleneck of traditional equipment in sensitivity, environmental adaptability and multi-scene generalization ability by optimizing the material, algorithm and system integration of the refrigeration type infrared detector, and realizes efficient and accurate detection of trace gas leakage in industrial scenes.

[0007] To achieve the above purpose, the application provides a high-efficiency refrigeration type infrared detector gas leakage detection system, which comprises a refrigeration type detector module, a gas enhancement algorithm module and an intelligent operation and maintenance module, wherein: The refrigeration type detector module is used for collecting infrared images of the leaked gas and transmitting the infrared images to the gas enhancement algorithm module; The gas enhancement algorithm module comprises a preprocessing unit and a detection unit, wherein: The preprocessing unit is used for background noise suppression and temperature drift compensation of the infrared images; The detection unit outputs a leakage area heat map and a concentration curve through optical flow field analysis and CNN feature extraction; The intelligent operation and maintenance module is used for health assessment of the refrigeration type detector module.

[0008] As a further preferred technical solution of the above technical solution, the background noise suppression is specifically implemented as: The image is decomposed by wavelet transform; The high-frequency coefficients are thresholded; The image is reconstructed.

[0009] As a further preferred technical solution of the above technical solution, the temperature drift compensation is specifically implemented as: ; Wherein, is the temperature coefficient, is the temperature change, is the original output value, is the compensated output value.

[0010] As a further preferred technical solution of the above technical solution, the detection unit is specifically implemented as: First, calculate the continuous frame optical flow field, and the optical flow field is calculated by Horn-Schunck method, and the formula is: ; Wherein, and represent the spatial gradient of the infrared image; represent the time gradient, and represent the optical flow vector; and combine the constraint condition of the gas diffusion model: , represent the attenuation coefficient; Second, extract the abnormal motion area, model the background, and the foreground is the area deviating from the model, specifically: Gaussian mixture model: ; Wherein, is the current pixel value, is the weight of the kth Gaussian distribution, is the Gaussian probability density function, is the sum of K Gaussian distributions; Foreground detection: If does not match any Gaussian distribution, it is determined as abnormal; Third, CNN feature classification, CNN extracts hierarchical features of input image through multi-layer convolution and pooling operation, and finally classifies through full connection layer, specifically: Convolution layer: ; Wherein, is the feature map of the llth layer, is a convolution kernel weight, is a bias term, is an activation function; a pooling layer: ; is a local neighborhood centered at ; Fourth, concentration inversion, specifically: ; wherein C is a gas concentration value, is a light intensity attenuation, A and B are adaptive parameters; by pixel-by-pixel inversion of the concentration , generate a spatial distribution heat map, intuitively display the leakage location and range, the formula is: , is an image pixel coordinate; for a fixed monitoring point continuous inversion , output the concentration-time curve, evaluate the leakage diffusion trend; The detection unit finally outputs data including heat map, curve and alarm threshold.

[0011] As a further preferred technical solution of the above technical solution, it further comprises a multi-sensor fusion module integrated with a TDLAS module to realize qualitative and quantitative dual-mode detection, wherein the multi-sensor fusion algorithm is: ; ; ; wherein, denotes the fused concentration, denotes the TDLAS measured concentration, denotes the infrared imaging inverted concentration, is the weight of TDLAS data, is the weight of infrared imaging data, and SNR represents the signal-to-noise ratio.

[0012] As a further preferred technical solution of the above technical solution, for the intelligent operation and maintenance module, the health degree index formula is: ; wherein, denotes the health degree index, denotes the weight of the i-th parameter, denotes the current parameter value, denotes the initial parameter value.

[0013] To achieve the above object, the application further provides a high-efficiency refrigeration type infrared detector gas leakage detection method applied to the high-efficiency refrigeration type infrared detector gas leakage detection system. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 is a schematic diagram of the system of the application. DETAILED DESCRIPTION

[0015] The following description is provided to enable any person skilled in the art to practice the application. The preferred embodiments in the following description are only examples of the application and other obvious variants can be conceived by those skilled in the art. The basic principles of the application defined in the following description can be applied to other embodiments, variants, improvements, equivalents and other technical solutions without departing from the spirit and scope of the application.

[0016] In the preferred embodiments of the application, those skilled in the art should note that the gas and the like involved in the application can be regarded as prior art.

[0017] Preferred embodiments.

[0018] As Figure 1 shown, the application discloses a high-efficiency refrigeration type infrared detector gas leakage detection system, a refrigeration type detector module, a gas enhancement algorithm module and an intelligent operation and maintenance module, wherein: The refrigeration type detector module is used for collecting infrared images of the leaked gas (when the gas leaks, the leaked gas cloud will absorb the infrared radiation of the background (such as the ground and equipment), which is shown as "dark spots" or "shadows" in the detector imaging) and transmitting the infrared images to the gas enhancement algorithm module; It is worth mentioning that for the refrigeration type detector module, a HOT type II superlattice (T2SL) refrigeration type core is adopted, covering short, medium and long wave infrared bands, improving the response sensitivity to multi-gas absorption peaks; and a dynamic temperature control system is integrated, which adjusts the refrigeration power in real time through a PID algorithm, reduces thermal noise interference, prolongs the service life of the refrigeration machine, and the response model of the HOT type II superlattice detector is: ; Wherein, is the responsivity (A / W); is the quantum efficiency; is the electronic charge (1.6×10⁻¹ 9 C); is the detection wavelength (μm); is the Planck constant (6.63×10⁻³ 4 J·s); is the speed of light (3×10 8 m / s); For photoelectric gain.

[0019] The gas enhancement algorithm module includes a preprocessing unit and a detection unit, wherein: The preprocessing unit is used to suppress background noise (based on wavelet transform) and compensate for temperature drift in infrared images. The detection unit outputs a heat map and concentration curve of the leakage area through optical flow field analysis and CNN feature extraction. The intelligent operation and maintenance module is used to perform health assessments on the cooled detector module.

[0020] Specifically, background noise suppression is implemented as follows: The image is decomposed using wavelet transform; Threshold processing of high-frequency coefficients; Reconstruct the image.

[0021] More specifically, temperature drift compensation is implemented as follows: ; in, For temperature coefficient, Temperature change This is the original output value. This is the output value after compensation.

[0022] Furthermore, the detection unit is specifically implemented as follows: First, calculate the optical flow field of consecutive frames. The Horn-Schunck method is used for optical flow field calculation, and the formula is as follows: ; in, and Represents the spatial gradient of an infrared image; Represents the time gradient, and This represents the optical flow vector; and it incorporates the constraints of the gas diffusion model: , Indicates the attenuation coefficient; Gas diffusion models describe the motion of gas molecules using partial differential equations (such as convection-diffusion equations). Introducing these models as constraints into optical flow calculations offers the following advantages: (1) Enhance dynamic adaptability: Physically driven smoothness: The continuity of gas diffusion is more consistent with real fluid motion (such as smoke, heat radiation, etc.) than pure mathematical smoothing, avoiding blurring of motion edges caused by over-smoothing.

[0023] Motion direction constraint: the velocity field (e.g. wind speed) in the diffusion model can guide the optical flow direction, making it more consistent with the actual gas diffusion path.

[0024] (2) Handling complex motion patterns: Turbulence and vortex modeling: the gas diffusion model can describe non-linear, multi-scale motion (e.g. vortex flow), making up for the shortcomings of Horn-Schunck in handling complex motion.

[0025] Time continuity reinforcement: the time derivative term of the diffusion equation can constrain the evolution of optical flow in time sequence, reducing inter-frame jumps.

[0026] (3) Anti-noise and occlusion: Robustness of diffusion term: the viscosity coefficient in the diffusion model can suppress high-frequency noise (similar to regularization), while preserving low-frequency motion information.

[0027] Occluded area inference: through the filling effect of the diffusion process, the optical flow of the occluded area can be reasonably inferred.

[0028] Second, extract the abnormal motion area, model the background, and the foreground (abnormal motion) is the area deviating from the model, specifically: Gaussian Mixture Model (GMM): ; where, is the current pixel value, is the weight of the kth Gaussian distribution, is the Gaussian probability density function, is the sum of K Gaussian distributions; Foreground detection: If does not match any Gaussian distribution (e.g. Mahalanobis distance > threshold > threshold), it is determined to be abnormal; Third, CNN feature classification, CNN extracts hierarchical features of input images through multiple convolution and pooling operations, and finally classifies through fully connected layers, specifically: Convolutional Layer (Convolutional Layer): ; where, is the feature map of the llth layer, is the convolution kernel weight (size k x k), is the bias term, is the activation function; Pooling layer (take maximum pooling as an example): ; is the size of The local neighborhood (such as a 2x2 window) centered on the pixel; Fourth, concentration inversion, specifically: ; Where C is the gas concentration value, is the light intensity attenuation, A and B are adaptive parameters; By pixel-by-pixel inversion of the concentration , generate a spatial distribution heat map, intuitively display the leakage location and range, the formula is: , is the image pixel coordinate; Continuous inversion of the fixed monitoring point , output the concentration-time curve, evaluate the leakage diffusion trend; The detection unit finally outputs data including heat map (can view the specific location of the leak), curve (can judge the severity of the leak and whether it is worsening) and alarm threshold.

[0029] The role of concentration inversion is: 1. Quantitative detection: Convert the optical signal (light intensity attenuation I0 / I) into a gas concentration value C, and realize accurate quantification of the leakage amount.

[0030] 2. Eliminate environmental interference: Logarithmic operation (ln(I0 / I)) can suppress nonlinear interference such as light source fluctuations and dust scattering.

[0031] 3. Calibration adaptability: Adjust A and B to adapt to the detection needs of different gases (such as methane, CO2).

[0032] Furthermore, it also includes a multi-sensor fusion module (this module does not work with the gas enhancement algorithm module, but is independent, giving users more choices), integrating a TDLAS (tunable diode laser absorption spectroscopy) module to realize qualitative and quantitative dual-mode detection. The multi-sensor fusion algorithm is: ; ; ; Where, represents the fused concentration, represents the TDLAS measured concentration, represents the infrared imaging inverted concentration, is the weight of TDLAS data (when the signal-to-noise ratio (SNR) of TDLAS is significantly higher than that of infrared imaging (such as high-concentration gas detection), w1→1, the fusion result depends more on the accurate measurement of TDLAS, For the weight of infrared imaging data (when the SNR of infrared imaging is higher (such as large-scale leakage visualization), w2→1, and the fusion result focuses on the spatial distribution information of infrared), SNR represents signal-to-noise ratio.

[0033] Preferably, for the intelligent operation and maintenance module, the health degree index formula is: ; Among them, represents the health degree index (0-1), represents the weight of the ith parameter, represents the current parameter value (including the amplitude value, temperature deviation, and efficiency attenuation rate), represents the initial parameter value.

[0034] The application further discloses a high-efficiency refrigeration type infrared detector gas leakage detection method applied to the high-efficiency refrigeration type infrared detector gas leakage detection system.

[0035] It is worth mentioning that the gas and other technical features involved in the present application should be regarded as prior art, and the specific structure, working principle, and possible control mode and spatial arrangement mode of these technical features can be selected according to the prior art, and should not be regarded as the invention point of the present application, and the present application will not be further described in detail.

[0036] For those skilled in the art, the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced, and any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A high-efficiency refrigeration-type infrared detector gas leak detection system, characterized by, The system comprises a refrigeration-type detector module, a gas enhancement algorithm module and an intelligent operation and maintenance module, wherein: The refrigeration-type detector module is used for collecting infrared images of the leaked gas and transmitting the infrared images to the gas enhancement algorithm module; The gas enhancement algorithm module comprises a preprocessing unit and a detection unit, wherein: The preprocessing unit is used for background noise suppression and temperature drift compensation of the infrared images; The detection unit outputs a leaked area heat map and a concentration curve through optical flow field analysis and CNN feature extraction; The intelligent operation and maintenance module is used for health assessment of the refrigeration-type detector module.

2. The high-efficiency refrigeration-type infrared detector gas leak detection system according to claim 1, wherein The background noise suppression is implemented as follows: Wavelet transform is used to decompose the image; Threshold processing is performed on the high-frequency coefficients; The image is reconstructed.

3. The high-efficiency refrigeration-type infrared detector gas leak detection system of claim 2, wherein, The temperature drift compensation is implemented as follows: ； wherein, is a temperature coefficient, is a temperature change amount, is an original output value, is a compensated output value.

4. The high-efficiency refrigeration-type infrared detector gas leak detection system of claim 3, wherein, The detection unit is implemented as follows: First, the continuous frame optical flow field is calculated, and the Horn-Schunck method is used for optical flow field calculation, and the formula is as follows: ； wherein, and denotes a spatial gradient of the infrared image; denotes a temporal gradient, and denotes an optical flow vector; and in combination with a constraint of a gas diffusion model: , denotes the attenuation coefficient; Second, the abnormal motion area is extracted, and the background is modeled, and the foreground is the area deviating from the model, and the specific implementation is as follows: Gaussian mixture model: ； wherein, is the current pixel value, is the weight of the k-th Gaussian distribution, is the Gaussian probability density function, is the sum over K Gaussian distributions; Foreground detection: If If not matching any Gaussian distribution, then it is determined as an anomaly; Third, CNN feature classification, CNN extracts hierarchical features of the input image through multiple convolution and pooling operations, and finally classifies through a fully connected layer, and the specific implementation is as follows: Convolution layer: ； wherein, is a feature map of the llth layer, is a convolution kernel weight, is a bias term, is an activation function; Pooling layer: ； a local neighborhood centered at a local neighborhood centered at Fourth, concentration inversion, specifically: ; wherein C is a gas concentration value, is the light intensity decay, A, B are adaptive parameters; By pixel by pixel inversion concentration , generate spatial distribution heat map, intuitive display leakage location and range, formula is: , For image pixel coordinates; Continuous inversion of fixed monitoring points Output concentration-time curve, evaluate the leakage diffusion trend The detection unit finally outputs data including a heat map, a curve and an alarm threshold.

5. The high-efficiency refrigeration-type infrared detector gas leak detection system of claim 4, wherein, The system further comprises a multi-sensor fusion module integrated with a TDLAS module to realize qualitative and quantitative dual-mode detection, wherein the multi-sensor fusion algorithm is as follows: ； ； ； wherein, represents the concentration after fusion, represents the TDLAS measured concentration, represents the infrared imaging inversion concentration, is the weight of the TDLAS data, is the weight of the infrared imaging data, and SNR represents the signal-to-noise ratio.

6. The high-efficiency refrigeration-type infrared detector gas leak detection system of claim 5, wherein, For the intelligent operation and maintenance module, the health degree index formula is as follows: ； wherein, is represented as a health index, represents the weight of the i-th parameter, represents the current parameter value, represents the initial parameter value.

7. A method for detecting gas leaks in a high-efficiency refrigeration type infrared detector, the method comprising: providing a high-efficiency refrigeration type infrared detector; and detecting a gas leak in the high-efficiency refrigeration type infrared detector. The system is applied to the high-efficiency refrigeration-type infrared detector gas leakage detection system of any one of claims 1-6.

Images