A mixed gas detection and early warning device in a rescue environment

Abstract

The application discloses a mixed gas detection and early warning device in a rescue environment, and relates to the technical field of gas detection, spectrum analysis and emergency rescue equipment. In the device, a gas spectrum detection module collects original infrared spectrum data of a to-be-detected mixed gas and three standard mixed gases on a rescue environment site; an air parameter measurement module collects environmental parameter measurement values; a central processor corrects and pre-processes the original infrared spectrum data of the to-be-detected mixed gas based on temperature and humidity measurement values and the original infrared spectrum data of the three standard mixed gases, performs spectrum unmixing on the pre-processed infrared spectrum data of the to-be-detected mixed gas under multiple iterations by using a target function and a constraint condition, obtains an optimal pure component spectrum matrix and an optimal abundance matrix, identifies a gas type, determines the predicted concentration of each type of gas, and performs hierarchical alarm. The application improves the detection accuracy and real-time performance of mixed gas in a rescue site, and realizes portable independent hierarchical early warning.

Description

Technical Field

[0001] This application relates to the fields of gas detection, spectral analysis and emergency rescue equipment technology, and in particular to a mixed gas detection and early warning device for rescue environments. Background Technology

[0002] In the event of an accident within a large setting such as an existing building or underground tunnel, the sooner rescuers enter the scene, the more likely they are to begin rescue operations. However, rescue sites may present various secondary risks, including power outages and dim lighting, high dust concentrations, and high concentrations of mixed gases. Real-time, on-site monitoring of the concentration of mixed gases in the ambient air is crucial to quickly identify potential secondary hazards and alert rescuers to take appropriate measures to mitigate these risks. Post-disaster air often contains common mixed gas components, such as high concentrations of carbon dioxide. carbon monoxide And methane gas (mainly composed of gas) which poses a risk of secondary explosion. ).in, , At a certain concentration, it can cause rapid asphyxiation, leading to unconsciousness and incapacitation of rescuers; in localized environments... When the concentration reaches 5%–16%, it may explode at any time upon contact with a spark, causing secondary disasters and endangering the lives of rescue personnel. Therefore, it is necessary to propose a micro-electro-mechanical systems (MEMS) mixed gas spectroscopic detection device suitable for rescue personnel to carry. This device should feature high integration, low cost, and low power consumption, and can directly detect the air at the rescue site and analyze... , , It is important to determine whether the concentration has reached a dangerous level and to promptly warn rescue personnel to take appropriate protective measures.

[0003] Currently, mixed gas detection is a routine procedure in normal production processes, both underground and inside buildings. The usual practice is to install detectors at structural change points within building corridors and tunnels. , Gas sensors, including those for gases such as methane, can be of various types, such as resistive chemical detectors, catalytic combustion detectors, and optical detectors. These sensors can detect gases in real time and upload data to a monitoring center, which then interprets the data and issues warnings. This approach has two problems: ① Currently, gas sensors / detectors mainly target single gas components, and most are designed for specific applications. , , These are chemical sensors that detect single gases. For mixed gas compositions in ambient air, these sensors often lack sufficient detection accuracy. ② These gas sensors typically rely on network transmission for early warning and are mainly used for routine environmental monitoring. They cannot independently provide effective early warning of multi-component mixed gases under conditions of power outages or network interruptions in buildings or underground tunnels.

[0004] Therefore, there is an urgent need for a device suitable for rescue personnel to carry, that does not rely on external power supply and communication networks at the rescue site, and that can detect airborne particles in real time and in situ. , , A portable MEMS spectroscopic detection device that can quickly classify and warn of mixed gas concentrations is proposed to solve the above problems. Summary of the Invention

[0005] The purpose of this application is to provide a mixed gas detection and early warning device for rescue environments, thereby improving the detection and early warning capabilities at rescue sites. , , The improved accuracy and real-time performance of mixed gas detection solves the problems of existing equipment relying on external power supply and communication networks, being unable to provide independent early warnings during power outages and network interruptions, and the insufficient accuracy of single gas sensors in mixed gas detection. It enables portable, independent, and tiered early warning systems, improving rescue safety and efficiency.

[0006] To achieve the above objectives, this application provides the following solution: This application provides a mixed gas detection and early warning device for rescue environments. The device includes: a gas spectrum detection module, an air parameter measurement module, a central processor, an audible and visual alarm module, and a power supply module. The gas spectrum detection module includes a gas chamber to be tested and three standard gas chambers. Each standard gas chamber is filled with a standard mixed gas with a preset composition, and the composition of the standard mixed gas in the three standard gas chambers is different from each other. The gas chamber to be tested is introduced into the mixed gas to be tested in the rescue environment. The gas spectral detection module is used to collect the raw infrared spectral data of the gas mixture to be tested and the raw infrared spectral data of three standard gas mixtures at the rescue site. The air parameter measurement module is used to collect real-time environmental parameter measurements at the rescue site; these parameters include: temperature, humidity, wind speed, and air pressure. The central processor is used for: Based on the temperature and humidity measurements from the environmental parameter measurements and the original infrared spectral data of three standard mixed gases, the original infrared spectral data of the mixed gas to be tested is corrected to obtain the corrected infrared spectral data of the mixed gas to be tested. The corrected infrared spectral data of the gas mixture to be tested are preprocessed to obtain the preprocessed infrared spectral data of the gas mixture to be tested. The preprocessed infrared spectral data of the gas mixture to be tested were subjected to spectral unmixing in multiple iterations using an objective function and constraints to obtain the optimal pure component spectral matrix and the optimal abundance matrix. The objective function includes an adaptive weighting term for residuals, an abundance sparsity constraint term, and an endmember independence constraint term. Based on the optimal pure component spectral matrix, identify the types of gases contained in the gas mixture to be tested at the rescue environment site; Based on the gas types and optimal abundance matrix of the gas mixture to be tested, the unmixing abundance of each type of gas is extracted, and the concentration inversion regression model is used to determine the predicted concentration of each type of gas based on the wind speed and air pressure measurements in the environmental parameter measurements. Determine whether the predicted concentration of each type of gas exceeds the corresponding preset warning threshold, and generate a graded alarm control signal when the predicted concentration of any type of gas exceeds the corresponding preset warning threshold; The audible and visual alarm module is used to issue corresponding audible and visual alarm signals based on the tiered alarm control signal; The power supply module is used to power the gas spectrum detection module, air parameter measurement module, central processor, and audible and visual alarm module.

[0007] According to the specific embodiments provided in this application, this application has the following technical effects: This application discloses a mixed gas detection and early warning device for rescue environments. It collects raw infrared spectral data of the target gas mixture and the three standard gas mixtures in the rescue environment using one test gas chamber and three standard gas chambers, respectively. A central processor corrects the raw infrared spectral data of the target gas mixture based on temperature and humidity measurements and the raw infrared spectral data of the three standard gas mixtures, eliminating the influence of temperature and humidity fluctuations and baseline drift on the spectral measurements. The corrected infrared spectral data of the target gas mixture is preprocessed, and spectral demixing is performed through multiple iterations using an objective function and constraints including adaptive residual weighting, abundance sparsity constraints, and endmember independence constraints. The adaptive residual weighting suppresses noise and baseline drift, the abundance sparsity constraints enhance the characteristics of weakly absorbing components, and the endmember independence constraints reduce... , By eliminating cross-interference of gas absorption peaks, the types of gases contained in the gas mixture to be tested can be accurately identified, and the optimal pure component spectral matrix and optimal abundance matrix can be obtained, solving the problem of insufficient accuracy in mixed gas detection by a single gas sensor. Furthermore, based on the gas types contained in the gas mixture and the optimal abundance matrix, the unmixing abundance of each type of gas is extracted. A concentration inversion regression model is then used, and the predicted concentration of each type of gas is determined based on wind speed and air pressure measurements, further improving accuracy. , , The system improves the accuracy of detecting mixed gases in complex rescue environments. A central processor determines whether the concentration of each type of gas exceeds its preset warning threshold. When the concentration of any type of gas exceeds the limit, a tiered alarm control signal is generated, and the audible and visual alarm module issues the tiered alarm signal on-site. This achieves portable, independent tiered warning, improving rescue safety and efficiency. Furthermore, the power module supplies power to the gas spectrum detection module, air parameter measurement module, central processor, and audible and visual alarm module. The audible and visual alarm module directly issues tiered alarm signals based on the tiered alarm control signal generated by the central processor. The entire detection and warning process requires no external power supply or communication network, solving the problem of existing equipment relying on external power supply and communication networks and being unable to issue independent warnings in the event of power outages or network interruptions. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A schematic diagram of the functional modules of a mixed gas detection and early warning device in a rescue environment provided in an embodiment of this application; Figure 2 A schematic flowchart illustrating the execution steps of a central processor provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a gas spectral detection module provided in one embodiment of this application; Figure 4 A schematic diagram of the channel structure of a four-channel microelectromechanical system infrared detector provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a multi-chamber integrated component provided in an embodiment of this application.

[0010] Figure label: Gas Spectroscopy Detection Module-1, Air Parameter Measurement Module-2, Central Processor-3, Audible and Visual Alarm Module-4, Power Supply Module-5, Storage and Communication Module-6. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0013] In one exemplary embodiment, such as Figure 1 As shown, a mixed gas detection and early warning device for rescue environments is provided, including: a gas spectrum detection module 1, an air parameter measurement module 2, a central processor 3, an audible and visual alarm module 4, and a power supply module 5; the gas spectrum detection module 1 includes a gas chamber to be tested and three standard gas chambers, each standard gas chamber is filled with a standard mixed gas with a preset composition, and the composition of the standard mixed gas in the three standard gas chambers is different from each other, and the gas chamber to be tested is introduced into the gas chamber to be tested in the rescue environment.

[0014] The gas spectral detection module 1 is used to collect the raw infrared spectral data of the gas mixture to be tested and the raw infrared spectral data of three standard gas mixtures at the rescue site.

[0015] The air parameter measurement module 2 is used to collect environmental parameter measurements at the rescue site in real time; environmental parameters include: temperature, humidity, wind speed and air pressure.

[0016] like Figure 2 As shown, the central processor 3 is used to perform the following steps: Step S1: Based on the temperature and humidity measurements from the environmental parameter measurements and the original infrared spectral data of the three standard mixed gases, the original infrared spectral data of the mixed gas to be tested is corrected to obtain the corrected infrared spectral data of the mixed gas to be tested. Step S2: Preprocess the corrected infrared spectral data of the gas mixture to be tested to obtain the preprocessed infrared spectral data of the gas mixture to be tested. Step S3: The preprocessed infrared spectral data of the gas mixture to be tested is subjected to spectral unmixing in multiple rounds of iteration using the objective function and constraints to obtain the optimal pure component spectral matrix and the optimal abundance matrix; the objective function includes residual adaptive weighting term, abundance sparsity constraint term and endmember independence constraint term; Step S4: Based on the optimal pure component spectral matrix, identify the types of gases contained in the gas mixture to be tested at the rescue environment site; Step S5: Based on the gas types and optimal abundance matrix of the gas mixture to be tested, extract the unmixing abundance of each type of gas, and use the concentration inversion regression model to determine the predicted concentration of each type of gas based on the wind speed measurement and air pressure measurement values ​​in the environmental parameter measurements. Step S6: Determine whether the predicted concentration of each type of gas exceeds the corresponding preset warning threshold, and generate a graded alarm control signal when the predicted concentration of any type of gas exceeds the corresponding preset warning threshold.

[0017] The audible and visual alarm module 4 is used to issue corresponding hierarchical audible and visual alarm signals based on the hierarchical alarm control signals.

[0018] Power module 5 is used to supply power to gas spectrum detection module 1, air parameter measurement module 2, central processor 3, and audible and visual alarm module 4.

[0019] Specifically, in the gas spectral detection module 1, the gas chamber to be tested and three standard gas chambers constitute a multi-chamber integrated component. The three standard gas chambers are designated C1, C2, and C3, and the gas chamber to be tested is designated C4. Each standard gas chamber is filled with a standard gas of a preset composition, and the gas chamber to be tested is introduced with a mixed gas from the rescue environment, as shown in Table 1. Each chamber contains multiple high-reflectivity mirrors with multilayer films, enabling the light beam to form a stable multi-reflection optical path within the cavity; compressing the optical path length of the chamber to less than 20mm ensures that the mixed gas to be tested interacts fully with infrared radiation.

[0020] Table 1. Comparison of the filling gas conditions in each gas chamber

[0021] The air parameter measurement module 2 mainly measures the environmental parameters of the environment in which the rescuers are located. It integrates sensors for temperature, humidity, wind speed, and air pressure, and collects the corresponding measurement values.

[0022] The central processor 3 encapsulates the execution program (corresponding to steps S1-S6), which pre-stores the standard mixed gas type, composition ratio and corresponding standard spectral curve of the multi-chamber gas integrated component.

[0023] The audible and visual alarm module 4 includes a MEMS buzzer alarm unit and different colored light signal alarm units. It employs a tiered, progressive warning mechanism, emitting sound and light signals to alert personnel to take action and prevent secondary injuries.

[0024] Power module 5 includes a 3.7V battery and a display.

[0025] As an optional implementation method, such as Figure 3 As shown, the gas spectral detection module 1 also includes: an infrared light source emitter, a spectrometer (not shown in the figure), an FP cavity filter, and a four-channel microelectromechanical system infrared detector (including detector one, detector two, detector three, and a reference detector). An infrared light source emitter is used to emit infrared light in a preset wavelength band; The beam splitter is used to split the infrared light emitted by the infrared light source emitter into four parallel infrared beams, and input the four parallel infrared beams into the gas chamber to be tested and three standard gas chambers respectively through the FP cavity filter; wherein, the FP cavity filter is set in the optical path between the beam splitter and each gas chamber, and is used to adjust the center wavelength of each parallel infrared beam by voltage. The four-channel microelectromechanical system infrared detector is integrated on the same silicon-based chip, and the four channels are respectively connected to the optical paths of the gas chamber to be tested and the three standard gas chambers to output the raw infrared spectral data of the gas mixture to be tested and the raw infrared spectral data of the three standard gas mixtures.

[0026] Specifically, the preset wavelength range of the infrared light emitted by the infrared source emitter corresponds to 2.5µm to 25µm. The spectral measurement range of the four-channel microelectromechanical system (MEMS) infrared detector is within the wavelength range of 2.5µm to 25µm. The four-channel MEMS infrared detector is integrated on the same silicon-based chip, such as... Figure 4 As shown. The gas chamber for the test gas and three standard gas chambers are directly connected to a four-channel microelectromechanical system (MEMS) infrared detector, which can directly detect the raw infrared spectral data of the four chambers (the raw infrared spectral data includes absorbance data at each wavelength), as shown below. Figure 5 DataC1, DataC2, DataC3, and DataC4 are shown; where DataC1, DataC2, and DataC3 correspond to the raw infrared spectral data of the standard mixed gas in the three standard gas chambers, and DataC4 corresponds to the raw infrared spectral data of the mixed gas to be tested in the gas chamber to be tested.

[0027] As an optional implementation, step S1 specifically includes: Step S11: Based on the temperature and humidity measurements in the environmental parameter measurements, perform temperature and humidity absorbance ratio correction on the original infrared spectral data of the three standard mixed gases and the original infrared spectral data of the mixed gas to be tested, respectively, to obtain the temperature and humidity corrected infrared spectral data of the three standard mixed gases and the temperature and humidity corrected infrared spectral data of the mixed gas to be tested. Step S12: Obtain the standard spectral absorbance data of the three standard gas chambers calibrated by a Fourier spectrometer under standard laboratory conditions. Based on the temperature and humidity corrected infrared spectral data of the three standard mixed gases and the corresponding standard spectral absorbance data, perform baseline correction on the temperature and humidity corrected infrared spectral data of the mixed gas to be tested to obtain the corrected infrared spectral data of the mixed gas to be tested.

[0028] As an optional implementation, step S11 specifically includes: Step S111: Obtain the reference temperature and reference relative humidity of the three standard gas chambers during calibration; Step S112: Calculate the temperature scaling factor based on the temperature measurement value and the reference temperature, and calculate the humidity scaling factor based on the humidity measurement value and the reference relative humidity. Step S113: Based on the temperature proportionality factor and humidity proportionality factor, determine the comprehensive proportionality correction coefficient of temperature and humidity absorbance according to the preset mapping relationship; Step S114: Using the temperature, humidity and absorbance comprehensive proportional correction coefficient, the absorbance data at each wavelength of the original infrared spectral data of the three standard mixed gases and the original infrared spectral data of the mixed gas to be tested are proportionally corrected to obtain the temperature and humidity corrected infrared spectral data of the three standard mixed gases and the temperature and humidity corrected infrared spectral data of the mixed gas to be tested.

[0029] As an optional implementation, step S12 specifically includes: Step S121: For each standard gas chamber, calculate the ratio of the absorbance data of the corresponding standard spectral spectrum at each wavelength to the absorbance data in the infrared spectrum of the corresponding standard mixed gas after temperature and humidity correction, and obtain the correction ratio factor of each standard gas chamber at each wavelength. Step S122: Take the average value of the correction scale factor of the three standard gas chambers at the same wavelength to obtain the comprehensive correction scale at each wavelength. Step S123: Use the comprehensive correction scale at each wavelength to correct the infrared spectral data of the mixed gas to be tested after temperature and humidity correction, and obtain the corrected infrared spectral data of the mixed gas to be tested.

[0030] Specifically, before performing baseline correction and other preprocessing on the raw infrared spectral data of the gas mixture to be tested, the ratios are first calculated based on the temperature and humidity measurements and the reference temperatures and relative humidity corresponding to the three standard gas chambers during calibration. Then, the absorbance data in the raw infrared spectral data of the gas mixture to be tested and the three standard gas chambers are corrected using a proportionality coefficient, as follows: The reference temperature corresponding to the three standard gas chambers during calibration is set as follows: (Kelvin temperature), reference relative humidity is The air parameter measurement module 2 collects the real-time ambient temperature at the rescue site as follows: (Kelvin temperature), relative humidity: Calculate the temperature scaling factor between the measured environment and the standard gas chamber calibration conditions. Humidity ratio factor The specific formula is as follows: (1) (2) According to the temperature proportionality factor and humidity ratio factor The comprehensive proportional correction coefficient for temperature, humidity, and absorbance is determined according to the preset mapping relationship. for: (3) In equation (3), This indicates a preset mapping relationship for determining the comprehensive proportional correction coefficient based on the temperature proportional factor and the humidity proportional factor. Using the combined proportional correction coefficient of temperature, humidity, and absorbance, the absorbance of the gas chamber to be tested and the three standard gas chambers at various wavelengths are... Raw absorbance data measured at [location] Each absorbance data was obtained after proportional correction to obtain its own temperature and humidity corrected values. As shown in the following formula: (4) Then, after correcting the raw infrared spectral data of the gas mixture under test for temperature and humidity, the infrared spectral data of the standard gas mixture, which were obtained on-site using three standard gas chambers and corrected for temperature and humidity, were used. , , , The absorbance data of the corresponding standard spectrum obtained by Fourier transform spectrometer calibration under laboratory standard conditions. , , The comprehensive correction scale is calculated using the ratio method, and then the infrared spectral data of the gas mixture to be tested are adjusted according to temperature and humidity correction. Baseline correction was performed to obtain the corrected infrared spectral data of the gas mixture under test. .

[0031] The collected spectral data can be represented as: , (5) In the formula, For the first The incident light wavelength of each air cell is the spectral absorbance value measured in the i-th air cell.

[0032] For C1, C2, and C3 gas cells with known standard spectra, at each wavelength Correction scaling factor at the location The calculation is as follows: (6) in, This represents the standard spectral absorbance value of the i-th standard gas chamber obtained by Fourier transform spectrometer calibration under standard laboratory conditions. This represents the spectral absorbance value measured in the i-th standard gas chamber after temperature and humidity correction.

[0033] The average of the correction scale factors for the three standard gas chambers is used to obtain the comprehensive correction scale. : (7) Using a comprehensive calibration scale Infrared spectral data of the gas mixture to be tested after temperature and humidity correction. Baseline correction was performed to obtain the corrected infrared spectral data of the gas mixture under test. : (8) in, For wavelength, This is the absorbance after baseline correction.

[0034] Furthermore, in step S2, the corrected infrared spectral data of the gas mixture to be tested are preprocessed, including Savitzky-Golay smoothing filtering and normalization.

[0035] Specifically, after baseline correction, the infrared spectral data of the measured gas mixture are denoised to eliminate noise, baseline drift, and intensity fluctuations in the measured spectrum. The denoised infrared spectral data of the measured gas mixture is filtered using a Savitzky-Golay smoothing algorithm with a window width of 10 points and a polynomial order of 2 to suppress high-frequency random noise, resulting in the denoised absorbance. : (9) in, This represents the Savitzky-Golay smoothing operator. For window width, The order of the polynomial; The final denoised spectral data is represented as follows: (10) Finally, the absorbance after noise reduction is... A standard normal transformation is performed to correct for errors caused by optical path differences. Normalized absorbance. It can be represented as: (11) in, For the first Normalized absorbance of each air cell For the first The absorbance of each air chamber after noise reduction For the first The absorbance after noise reduction at each spectral sampling point Current spectral absorbance The arithmetic mean, This represents the total number of spectral sampling points. Index of spectral sampling points; After normalization, the preprocessed infrared spectral data of the gas mixture to be tested, used for subsequent spectral demixing, are obtained: (12).

[0036] Furthermore, in actual emergency rescue scenarios, the gas mixture to be tested often contains multiple hazardous gases (such as...). , , Because the characteristic absorption peak positions and absorption intensities of different gases vary greatly, traditional nonnegative matrix decomposition tends to overfit the signals of components with strong and wide absorption ranges, thereby weakening the characteristics of low-concentration and weak-absorbing components, ultimately affecting the accuracy of spectral unmixing. , The characteristic absorption peaks exhibit a clear spatial misalignment distribution on the wavenumber axis (i.e., their respective strong absorption peak intervals do not interfere with each other), indicating that the infrared characteristic spectral vectors of different component gases possess natural quasi-orthogonality and spatial sparsity. To address this issue, this application simultaneously introduces residual adaptive weighting, abundance sparsity constraints, and endmember independence constraints into the nonnegative matrix factorization algorithm, proposing a residual-adaptive weighted sparse nonnegative matrix factorization (RAW-SNMF) blind source unmixing model for the infrared spectra of multi-component gas mixtures, performing blind source unmixing (i.e., spectral unmixing) on ​​the preprocessed infrared spectral data of the gas mixtures.

[0037] As an optional implementation, in step S3, the expression for the objective function is: (13) in, The objective function value; The spectral matrix of the pure components; Abundance matrix; This is the residual weight matrix; The infrared spectral data of the gas mixture to be tested after preprocessing; It is the Frobenius norm; The parameter is the sparse regularization penalty parameter; It is a sparse weight matrix; For Hadamard's element-wise product; It is an element-wise 1-norm; This is the endmember independence regularization parameter; This is the transpose of the matrix.

[0038] As an optional implementation, the constraint is that both the pure component spectral matrix and the abundance matrix are non-negative matrices; During step S3, the spectral unmixing process at any current iteration number specifically includes: Step S31: Based on the pure component spectral matrix and abundance matrix at the current iteration number, calculate the reconstruction residual between the preprocessed infrared spectral data of the mixed gas to be tested and the reconstructed spectral data at the current iteration number, and calculate the residual weight matrix at the current iteration number based on the reconstruction residual; wherein, the reconstructed spectral data is obtained by multiplying the pure component spectral matrix and abundance matrix at the current iteration number; Step S32: Calculate the sparse weight matrix for the next iteration number based on the abundance matrix for the current iteration number. Step S33: Substitute the residual weight matrix under the current iteration number and the sparse weight matrix under the next iteration number into the objective function. Under the constraint that both the pure component spectral matrix and the abundance matrix are non-negative matrices, solve the objective function to obtain the pure component spectral matrix and abundance matrix under the next iteration number. Step S34: Determine whether the pure component spectral matrix and abundance matrix in the next iteration satisfy the preset convergence condition; Step S35: If not, update the pure component spectral matrix and abundance matrix of the next iteration to the pure component spectral matrix and abundance matrix of the current iteration, and return to execute "Calculate the reconstruction residual between the preprocessed infrared spectral data of the mixed gas to be tested and the reconstructed spectral data of the current iteration based on the pure component spectral matrix and abundance matrix of the current iteration", and continue to the next iteration; Step S36: If yes, then the pure component spectral matrix and abundance matrix for the next iteration number are determined as the optimal pure component spectral matrix and optimal abundance matrix.

[0039] Specifically, the infrared spectral data of the preprocessed gas mixture to be tested... RAW-SNMF blind source unmixing was performed to obtain the pure component spectral matrix A and abundance matrix S.

[0040] (14) in, This represents the pure component spectral matrix, where m is the number of spectral sampling points contained in each spectrum; the v-th column represents the characteristic spectrum of the v-th pure component gas obtained by demixing. For abundance matrix, This indicates the number of independent gas components in the gas mixture being tested.

[0041] In order to accurately calculate the pure component spectral matrix A and abundance matrix S from the preprocessed infrared spectral data X of the gas mixture to be tested, it is necessary to construct a suitable objective function and constraints; The constraints and objective function for unmixing the blind source of the gas mixture to be tested are constructed. This application introduces adaptive residual weighting, abundance sparsity constraints, and endmember independence constraints to define the unmixing objective function. The objective function includes an adaptive residual weighting term, an abundance sparsity constraint term, and an endmember independence constraint term, as shown in equation (13) above. Wherein, Dr of the residual weight matrix... The diagonal matrix representing the average residual weights of the bands, where m is the number of spectral sampling points in each spectrum, and the j-th diagonal element of this matrix... for: , (15) Where n is the number of mixed gas spectral samples, u is the mixed gas spectral sample number, and j is the spectral sampling point index; This represents the reconstruction residual of the u-th mixed gas spectral sample at the j-th spectral sampling point, which is used to dynamically adjust the role of different spectral bands in the reconstruction error. In the objective function, the first term is the residual weighted reconstruction error (i.e., the residual adaptive weighting term): The residual weight matrix Dr is used to suppress high-frequency spectral noise and baseline drift; the second term is the abundance sparsity constraint term. , The term represents the relative concentration weight (unmixing abundance) of the v-th component in the u-th mixed gas spectral sample. By applying an adaptive L1 norm constraint to the abundance matrix S, the sparsity of the abundance vector is enhanced, which conforms to the sparse physical characteristics of actual gas distribution; the third term is the endmember independence constraint term. By applying orthogonal constraints to the endmember matrix (i.e., the pure component spectral matrix) A, the reduction Cross-interference caused by spectral absorption peaks.

[0042] When solving the objective function, a residual weight matrix R and a sparse weight matrix W are constructed to achieve dynamic updates of the weight matrix. The larger the reconstructed residual, the smaller the corresponding weight, and the less impact strong noise and baseline drift have on model parameter estimation, as shown in the equation: (16) (17) During mixed spectral resolution, the iterative sparse weight matrix algorithm calculates the weights for the next iteration based on the current abundance matrix: (18) in, A positive number is set to prevent the denominator from being zero. ; This represents the residual weight of the u-th mixed gas spectral sample at the j-th spectral sampling point during the t-th iteration; This represents the spectral value of the u-th mixed gas spectral sample at the j-th spectral sampling point; This represents the pure component spectral matrix obtained in the t-th iteration; This represents the abundance matrix obtained in the t-th iteration; This represents the sparse weight of the v-th gas component in the u-th mixed gas spectral sample at the (t+1)-th iteration. Let represent the unmixing abundance of the v-th gas component in the u-th mixed gas spectral sample at the t-th iteration.

[0043] As an optional implementation, step S4 specifically includes: Step S41: Extract multiple pure component spectra from the optimal pure component spectral matrix; Step S42: Compare the spectra of each pure component with the corresponding standard gas spectral data to determine the gas type corresponding to each pure component spectrum; Step S43: Determine the gas types contained in the mixed gas to be tested at the rescue environment site based on the gas types corresponding to the spectra of each pure component.

[0044] As an optional implementation, step S5 specifically includes: Step S51: Extract the unmixing abundance corresponding to each type of gas from the optimal abundance matrix according to the type of mixed gas. Step S52: Substitute the unmixing abundance corresponding to each type of gas into the concentration inversion regression model to calculate the initial predicted concentration of each type of gas. Step S53: Using the wind speed measurement and air pressure measurement values ​​in the environmental parameter measurements as correction variables, the initial predicted concentration is dynamically compensated to obtain the final predicted concentration of each type of gas. The expression for the concentration inversion regression model is as follows: (19) in, For the first In the first mixed gas spectral sample The initial predicted concentration of the gas; For the first The intercept term for each gas reflects the baseline response of the model when the unmixing abundance is zero; For the first The slope coefficient of a gas indicates the degree of influence of unmixing abundance on concentration; For the first In the first mixed gas spectral sample The unmixing abundance of the gas; For the first In the first mixed gas spectral sample The random error term for each gas reflects the unexplained random bias during the model fitting process; Number the spectral samples of the mixed gas. Number the gas components.

[0045] Specifically, based on the gas types contained in the gas mixture obtained from spectral unmixing, the unmixing abundance of the corresponding gas components is extracted. Combined with the wind speed and air pressure measurements collected in real time by the air parameter measurement module 2, dynamic compensation and concentration inversion of the multi-component hazardous gases are completed. After completing the unmixing of the gas mixture and gas type identification, a univariate linear regression model based on the unmixing abundance matrix is ​​constructed as the concentration inversion regression model to realize the concentration inversion of each type of gas, as shown in equation (19) above. Among them, the regression coefficients... The least squares method is used for optimization to minimize the sum of squared residuals between the predicted and actual concentrations. The optimization objective of the least squares method is: (20) In the formula, n is the total number of mixed gas spectral samples used to establish the concentration inversion regression model. For the u-th mixed spectral sample, the th The actual calibrated concentrations of the gas components This represents the initial predicted concentration of the v-th gas in the u-th mixed gas spectral sample calculated based on the unmixing abundance. The formula for calculating the regression coefficients is as follows: (twenty one) (twenty two) (twenty three) (twenty four) in, Let be the sample mean of the actual calibrated concentration of the v-th gas. Let be the sample mean of the unmixing abundance of the vth gas.

[0046] After obtaining the initial predicted concentration Then, the ambient air pressure collected in real time by the air parameter measurement module 2 is further analyzed. With wind speed Dynamic compensation is performed to obtain the final output of the true gas concentration. : (25) in, Let be the final predicted concentration of the v-th gas in the u-th mixed gas spectral sample; This is an environmental dynamic compensation function, which uses air pressure and wind speed measurements from environmental parameters as correction variables.

[0047] As an optional implementation, each preset warning threshold corresponds to a warning level; the warning levels include Level 1 warning, Level 2 warning, and Level 3 warning, with Level 3 warning being higher than Level 2 warning, and Level 2 warning being higher than Level 1 warning; then step S6 specifically includes: Step S61: Determine whether the predicted concentration of each type of gas exceeds the preset warning threshold corresponding to each warning level. Step S62: If the predicted concentration of at least one type of gas exceeds the preset warning threshold under the corresponding warning level, then determine the warning level reached by the predicted concentration of each type of gas, and determine the highest warning level reached as the final warning level. Step S63: Generate a corresponding graded alarm control signal based on the final warning level.

[0048] Specifically, based on the gas types and concentrations identified in the aforementioned gas mixture, the predicted concentrations of each gas type are inverted and compared with... , , The system compares the values ​​of the gases with preset warning thresholds, and the audible and visual alarm module 4 issues a buzzer alarm and illuminates a warning light. The regression coefficients obtained through the above solution determine the concentration inversion regression model. Subsequently, simply substituting the detected unmixing abundance of the corresponding gas into the concentration inversion regression model allows for the rapid acquisition of the corresponding predicted concentration without complex iterative calculations, thus meeting the real-time requirements for rapid detection at rescue sites.

[0049] The predicted concentrations of various types of gases obtained by the inversion method are compared with the preset warning thresholds. Based on the relationship between the two, a comprehensive judgment is made on whether there are potential safety risks or pollution exceeding the standards in the current environment, thereby providing a scientific basis and decision support for environmental monitoring, risk warning and subsequent response measures.

[0050] To achieve accurate and efficient early warning, the central processor 3 adopts a hierarchical early warning mechanism. As shown in Table 2, the central processor 3 has multiple preset early warning thresholds, each of which corresponds to an early warning level.

[0051] The preset warning thresholds corresponding to Level 1 warnings include: Concentration ≥1.0%, Concentration ≥0.5% or Concentration ≥ 24 ppm. When the predicted concentration of any type of gas reaches the preset warning threshold corresponding to the Level 1 warning, the central processor 3 generates a Level 1 alarm control signal, and the audible and visual alarm module 4 emits a slow-interval buzzer and flashes a yellow light, once every 3 seconds. At the same time, it pushes the warning information to the on-site monitoring terminal to remind staff to pay more attention and check the local ventilation.

[0052] Table 2. Correspondence between graded early warning thresholds and audible / visual alarm modes

[0053] The preset warning thresholds corresponding to a Level 2 warning include: Concentration ≥1.5%, Concentration ≥1.5% or Concentration ≥ 100 ppm. When the predicted concentration of any type of gas reaches the preset warning threshold corresponding to the Level II warning, the central processor 3 generates a Level II alarm control signal, the audible and visual alarm module 4 emits a rapid interval beep, once every 1 second, and the warning light flashes red. If the gas exceeds the limit, the central processor 3 immediately activates the automatic power-off mechanism, and the warning information is simultaneously sent to the on-site command center and emergency management platform, and the on-site personnel evacuation plan is activated.

[0054] The preset warning thresholds for Level 3 warning include: 16% ≥ Concentration ≥ 5.0%, Concentration ≥3.0% or Concentration ≥ 1600 ppm. When the predicted concentration of any type of gas reaches the preset warning threshold corresponding to the Level 3 warning, the central processor 3 generates a Level 3 alarm control signal, the audible and visual alarm module 4 emits a continuous sharp buzzer warning light and a red high-frequency flashing light, and the emergency rescue team quickly assembles to carry out disposal work after receiving the instruction.

[0055] Furthermore, the central processor 3 also features multi-dimensional early warning and linkage functions. In addition to numerical early warnings based on predicted gas concentrations, it integrates multi-source data such as ambient temperature and humidity, oxygen content, wind speed and air pressure, and the location of on-site personnel for comprehensive analysis. When the predicted concentration of toxic gas is detected to be close to the preset early warning threshold and the on-site oxygen content is below 19.5%, the central processor 3 upgrades the early warning level in advance to ensure personnel safety. When the wind speed or air pressure measurement is detected to be below the safety threshold, the central processor 3 immediately triggers an equipment fault early warning, with a buzzer emitting a special rhythmic alert, a warning light flashing purple, and simultaneously pushing equipment maintenance instructions to the operation and maintenance management platform. At the same time, early warning information can be simultaneously sent through various methods such as sound and light, SMS, and APP push notifications, enabling real-time information sharing between the on-site and remote command centers and providing comprehensive data support for emergency decision-making.

[0056] Furthermore, the preset early warning threshold can be set according to the downhole safety limits for each gas well shown in Table 3. As shown in Table 3, The safety limit for intake airflow in working faces is ≤0.5%, the safety limit for return airflow in mining faces is ≤1.0%, and the safety limit for total return airflow in mines is ≤0.75%, with explosion limits of 5% to 16%. The maximum permissible concentration is ≤0.0024% (24ppm), and its explosive range is 13% to 75%. The safety limit for the intake airflow at the working face is ≤0.5%, the safety limit for the return airflow at the mining face is ≤1.5%, and the safety limit for the total return airflow in the mine is ≤0.75%. Operations must cease when the concentration reaches 1.5%. The central processor 3 can set preset warning thresholds for each type of gas based on these safety limits, and generate a tiered alarm control signal when the predicted concentration of any type of gas exceeds the corresponding preset warning threshold.

[0057] The integrated mixed gas detection and early warning device for rescue environments provided in this application has a size not exceeding 160mm×70mm×70mm, allowing it to be carried by the person or placed on top of a safety helmet. The entire working mechanism is as follows: When rescuers enter a rescue scene carrying the device, the airflow within the scene automatically enters the gas chamber to be tested; a four-channel microelectromechanical system infrared detector records the raw infrared spectral data of the gas chamber to be tested and three standard gas chambers according to a set sampling frequency; simultaneously, the air parameter measurement module 2 collects real-time measurements of environmental parameters such as temperature, humidity, wind speed, and air pressure at the rescue environment. The raw infrared spectral data and environmental parameter measurements are automatically transmitted to the central processor 3. After performing corresponding data processing, the central processor 3 outputs the gas types contained in the mixed gas to be tested in the environment where the rescuers are located at the current moment and the predicted concentration of each type of gas; when the predicted concentration of any type of gas exceeds the corresponding preset early warning threshold, the audible and visual alarm module 4 is triggered to issue a graded audible and visual alarm signal, reminding the rescuers to avoid the risk in time.

[0058] Furthermore, the mixed gas detection and early warning device for rescue environments provided in this application also includes a storage and communication module 6, which can list and record information such as the types of gases contained in the mixed gas to be tested at the rescue environment site, the predicted concentration of each type of gas, and whether an alarm has been triggered, according to a preset time frequency. It can also establish a wireless communication connection with other rescue personnel or the communication center of the emergency command vehicle through its internal LoRa wireless communication unit. The storage and communication module 6 can also receive corresponding instructions sent externally and send the real-time records of the central processor 3 to the external command vehicle.

[0059] 1) The gas spectrum detection module, air parameter measurement module, central processor and audible and visual alarm module are powered by the power module; the audible and visual alarm module directly emits graded audible and visual alarm signals based on the graded alarm control signals generated by the central processor. The entire detection and early warning process does not require external power supply and communication network, enabling the device to work independently under extreme rescue conditions without external power supply and communication network support. This solves the problem that existing equipment relies on external power supply and communication network and cannot provide independent early warning under power outage and network interruption.

[0060] 2) The gas spectral detection module includes one test gas chamber and three standard gas chambers, which collect raw infrared spectral data of the test gas mixture and the three standard gas mixtures, respectively. The central processor performs calibration based on temperature and humidity measurements and the raw infrared spectral data of the three standard gas mixtures. It then preprocesses the calibrated infrared spectral data of the test gas mixture and performs spectral demixing through multiple iterations to obtain the optimal pure component spectral matrix and the optimal abundance matrix, thereby identifying the types of gases contained in the test gas mixture. This solves the problem of insufficient accuracy in mixed gas detection by a single gas sensor and improves the accuracy of on-site rescue operations. , , Accuracy and real-time performance of detecting mixed gases.

[0061] 3) Based on the temperature and humidity measurements from the environmental parameters, the raw infrared spectral data of the three standard mixed gases and the test mixed gas are respectively corrected for temperature and humidity absorbance ratios. Then, based on the temperature and humidity corrected infrared spectral data of the three standard mixed gases, the infrared spectral data of the test mixed gas after temperature and humidity correction is baseline corrected. This eliminates the influence of on-site temperature and humidity fluctuations and baseline drift on spectral measurements, improves the accuracy of spectral data, and provides reliable input for subsequent unmixing and concentration inversion.

[0062] 4) Using an objective function and constraints that include adaptive residual weighting, abundance sparsity constraints, and endmember independence constraints, the preprocessed infrared spectral data of the target mixed gas are subjected to multi-round iterative spectral demixing. The adaptive residual weighting suppresses noise and baseline drift, the abundance sparsity constraints enhance the characteristics of weakly absorbing components, and the endmember independence constraints reduce... and By eliminating cross-interference of gas absorption peaks, the types of gases contained in the gas mixture to be tested can be accurately identified, and the optimal pure component spectral matrix and the optimal abundance matrix can be obtained, thus solving the problem of insufficient accuracy of traditional unmixing methods.

[0063] 5) Based on the gas types and optimal abundance matrix of the gas mixture to be tested, the unmixing abundance of each gas type is extracted. An initial predicted concentration is calculated using a concentration inversion regression model. Dynamic compensation is then applied to the initial predicted concentration based on wind speed and air pressure measurements from environmental parameters to obtain the final predicted concentration of each gas type, further improving the accuracy of the prediction. , , The accuracy of detecting mixed gases in complex rescue environments can be improved, and the hazard level of gases in the on-site microenvironment can be quickly predicted.

[0064] 6) By issuing corresponding graded audible and visual alarm signals based on graded alarm control signals, portable independent graded early warning is realized, reminding rescue personnel to take appropriate protective measures in a timely manner, thereby improving the safety and efficiency of rescue.

[0065] 7) A microelectromechanical system (MEMS) spectral mixed gas detection device is adopted, with high integration, low cost, low power consumption, and long service life as its core development goals. The integrated device is small in size and can be carried by the person or placed on the top of a safety helmet, thus achieving portable detection with small size, high sensitivity, and high selectivity. It can quickly and accurately identify gases in the air at the rescue site under extreme rescue conditions such as power outages and network outages. , , By identifying the types of mixed gases and determining the predicted concentrations of each type, the safety and scientific rigor of the rescue process can be ensured.

[0066] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0067] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the device and its core ideas. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A mixed gas detection and early warning device for rescue environments, characterized in that, The device includes: a gas spectrum detection module, an air parameter measurement module, a central processor, an audible and visual alarm module, and a power supply module; the gas spectrum detection module includes a gas chamber to be tested and three standard gas chambers, each standard gas chamber is filled with a standard mixed gas with a preset composition, and the composition of the standard mixed gas in the three standard gas chambers is different from each other, and the gas chamber to be tested is introduced with the gas mixed gas to be tested in the rescue environment. The gas spectral detection module is used to collect the raw infrared spectral data of the gas mixture to be tested and the raw infrared spectral data of three standard gas mixtures at the rescue site. The air parameter measurement module is used to collect real-time environmental parameter measurements at the rescue site; these parameters include: temperature, humidity, wind speed, and air pressure. The central processor is used for: Based on the temperature and humidity measurements from the environmental parameter measurements and the original infrared spectral data of three standard mixed gases, the original infrared spectral data of the mixed gas to be tested is corrected to obtain the corrected infrared spectral data of the mixed gas to be tested. The corrected infrared spectral data of the gas mixture to be tested are preprocessed to obtain the preprocessed infrared spectral data of the gas mixture to be tested. The preprocessed infrared spectral data of the gas mixture to be tested were subjected to spectral unmixing in multiple iterations using an objective function and constraints to obtain the optimal pure component spectral matrix and the optimal abundance matrix. The objective function includes an adaptive weighting term for residuals, an abundance sparsity constraint term, and an endmember independence constraint term. Based on the optimal pure component spectral matrix, identify the types of gases contained in the gas mixture to be tested at the rescue environment site; Based on the gas types and optimal abundance matrix of the gas mixture to be tested, the unmixing abundance of each type of gas is extracted, and the concentration inversion regression model is used to determine the predicted concentration of each type of gas based on the wind speed and air pressure measurements in the environmental parameter measurements. Determine whether the predicted concentration of each type of gas exceeds the corresponding preset warning threshold, and generate a graded alarm control signal when the predicted concentration of any type of gas exceeds the corresponding preset warning threshold; The audible and visual alarm module is used to issue corresponding audible and visual alarm signals based on the tiered alarm control signal; The power supply module is used to power the gas spectrum detection module, air parameter measurement module, central processor, and audible and visual alarm module.

2. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, The gas spectral detection module also includes: an infrared light source emitter, a spectrometer, an FP cavity filter, and a four-channel microelectromechanical system infrared detector; An infrared light source emitter is used to emit infrared light in a preset wavelength band; The beam splitter is used to split the infrared light emitted by the infrared light source emitter into four parallel infrared beams, and input the four parallel infrared beams into the gas chamber to be tested and three standard gas chambers respectively through the FP cavity filter; wherein, the FP cavity filter is set in the optical path between the beam splitter and each gas chamber, and is used to adjust the center wavelength of each parallel infrared beam by voltage. The four-channel microelectromechanical system infrared detector is integrated on the same silicon-based chip, and the four channels are respectively connected to the optical paths of the gas chamber to be tested and the three standard gas chambers to output the raw infrared spectral data of the gas mixture to be tested and the raw infrared spectral data of the three standard gas mixtures.

3. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, Based on the temperature and humidity measurements from environmental parameter measurements and the raw infrared spectral data of three standard gas mixtures, the raw infrared spectral data of the gas mixture to be tested is corrected to obtain the corrected infrared spectral data of the gas mixture to be tested, specifically including: Based on the temperature and humidity measurements from the environmental parameters, the original infrared spectral data of the three standard mixed gases and the original infrared spectral data of the gas to be tested were corrected for temperature and humidity absorbance ratios to obtain the temperature and humidity corrected infrared spectral data of the three standard mixed gases and the temperature and humidity corrected infrared spectral data of the gas to be tested. Standard spectral absorbance data of three standard gas chambers were obtained by Fourier spectrometer calibration under standard laboratory conditions. Based on the temperature and humidity corrected infrared spectral data of the three standard mixed gases and the corresponding standard spectral absorbance data, baseline correction was performed on the infrared spectral data of the mixed gas to be tested after temperature and humidity correction, and the corrected infrared spectral data of the mixed gas to be tested were obtained.

4. The mixed gas detection and early warning device in a rescue environment according to claim 3, characterized in that, Based on the temperature and humidity measurements from the environmental parameters, the original infrared spectral data of the three standard gas mixtures and the original infrared spectral data of the gas mixture to be tested were respectively corrected for temperature and humidity absorbance ratios. This yielded the temperature and humidity corrected infrared spectral data of the three standard gas mixtures and the temperature and humidity corrected infrared spectral data of the gas mixture to be tested, specifically including: Obtain the reference temperature and reference relative humidity of the three standard gas chambers during calibration; The temperature scaling factor is calculated based on the temperature measurement and the reference temperature, and the humidity scaling factor is calculated based on the humidity measurement and the reference relative humidity. Based on the temperature proportionality factor and humidity proportionality factor, the comprehensive proportionality correction coefficient of temperature and humidity absorbance is determined according to the preset mapping relationship. Using the temperature, humidity and absorbance comprehensive proportional correction coefficient, the absorbance data at each wavelength of the original infrared spectral data of the three standard mixed gases and the original infrared spectral data of the gas to be tested are proportionally corrected to obtain the temperature and humidity corrected infrared spectral data of the three standard mixed gases and the temperature and humidity corrected infrared spectral data of the gas to be tested.

5. The mixed gas detection and early warning device in a rescue environment according to claim 3, characterized in that, Based on the temperature and humidity corrected infrared spectral data of three standard gas mixtures and their corresponding standard absorbance data, baseline correction was performed on the temperature and humidity corrected infrared spectral data of the gas mixture to be tested, resulting in the corrected infrared spectral data of the gas mixture to be tested, specifically including: For each standard gas chamber, the ratio of the absorbance data of the corresponding standard spectral spectrum at each wavelength to the absorbance data in the infrared spectrum of the corresponding standard mixed gas after temperature and humidity correction is calculated to obtain the correction ratio factor of each standard gas chamber at each wavelength. The average value of the correction scale factor of the three standard gas chambers at the same wavelength is used to obtain the comprehensive correction scale at each wavelength. The infrared spectral data of the gas mixture to be tested, after temperature and humidity correction, are corrected using a comprehensive correction scale at each wavelength to obtain the corrected infrared spectral data of the gas mixture to be tested.

6. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, The expression for the objective function is: ； in, The objective function value; The spectral matrix of the pure components; Abundance matrix; This is the residual weight matrix; The infrared spectral data of the mixed gas to be tested are after preprocessing. It is the Frobenius norm; The parameter is the sparse regularization penalty parameter; It is a sparse weight matrix; For Hadamard's element-wise product; It is an element-wise 1-norm; This is the endmember independence regularization parameter; This is the transpose of the matrix.

7. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, The constraint is that both the spectral matrix and abundance matrix of the pure components are non-negative matrices; In the process of "using the objective function and constraints to perform multi-round iterative spectral unmixing on the preprocessed infrared spectral data of the gas mixture to be tested, and obtaining the optimal pure component spectral matrix and the optimal abundance matrix", the spectral unmixing process at any current iteration number specifically includes: Based on the pure component spectral matrix and abundance matrix at the current iteration number, the reconstruction residual between the preprocessed infrared spectral data of the gas mixture to be tested and the reconstructed spectral data at the current iteration number is calculated, and the residual weight matrix at the current iteration number is calculated based on the reconstruction residual; wherein, the reconstructed spectral data is obtained by multiplying the pure component spectral matrix and abundance matrix at the current iteration number; Based on the abundance matrix at the current iteration number, the sparse weight matrix for the next iteration number is calculated. Substitute the residual weight matrix of the current iteration number and the sparse weight matrix of the next iteration number into the objective function, and solve the objective function under the constraint that both the pure component spectral matrix and the abundance matrix are non-negative matrices, to obtain the pure component spectral matrix and abundance matrix of the next iteration number. Determine whether the pure component spectral matrix and abundance matrix in the next iteration satisfy the preset convergence condition; If not, update the pure component spectral matrix and abundance matrix of the next iteration to the pure component spectral matrix and abundance matrix of the current iteration, and return to execute "Calculate the reconstruction residual between the preprocessed infrared spectral data of the mixed gas to be tested and the reconstructed spectral data of the current iteration based on the pure component spectral matrix and abundance matrix of the current iteration", and continue to the next iteration; If so, the pure component spectral matrix and abundance matrix in the next iteration will be determined as the optimal pure component spectral matrix and optimal abundance matrix.

8. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, Based on the optimal pure component spectral matrix, the types of gases contained in the gas mixture to be tested at the rescue environment are identified, specifically including: Extract multiple pure component spectra from the optimal pure component spectral matrix; The spectra of each pure component are compared with the corresponding standard gas spectra to determine the gas type corresponding to each pure component spectrum; Based on the gas types corresponding to the spectra of each pure component, the types of gases contained in the mixed gas to be tested at the rescue environment site are determined.

9. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, Based on the gas types and optimal abundance matrix of the gas mixture to be tested, the unmixing abundance of each gas type is extracted. Then, using a concentration inversion regression model, based on wind speed and air pressure measurements from environmental parameters, the predicted concentrations of each gas type are determined, specifically including: Based on the type of mixed gas, extract the unmixing abundance corresponding to each type of gas from the optimal abundance matrix; Substitute the unmixing abundance corresponding to each type of gas into the concentration inversion regression model to calculate the initial predicted concentration of each type of gas. Using wind speed and air pressure measurements from environmental parameters as correction variables, the initial predicted concentration is dynamically compensated to obtain the final predicted concentration of each type of gas. The expression for the concentration inversion regression model is as follows: ； in, For the first In the first mixed gas spectral sample The initial predicted concentration of the gas; For the first The intercept term for a gas; For the first The slope coefficient of the gas; For the first In the first mixed gas spectral sample The unmixing abundance of the gas; For the first In the first mixed gas spectral sample The random error term for the gas; Number the spectral samples of the mixed gas. Number the gas components.

10. The mixed gas detection and early warning device in a rescue environment according to claim 1, characterized in that, Each preset warning threshold corresponds to a warning level; The warning levels include Level 1, Level 2, and Level 3. Level 3 is higher than Level 2, and Level 2 is higher than Level 1. Determine whether the predicted concentration of each type of gas exceeds the corresponding preset warning threshold, and generate a graded alarm control signal when the predicted concentration of any type of gas exceeds the corresponding preset warning threshold, specifically including: Determine whether the predicted concentration of each type of gas exceeds the preset warning threshold corresponding to each warning level; If the predicted concentration of at least one type of gas exceeds the preset warning threshold under the corresponding warning level, then the warning level reached by the predicted concentration of each type of gas is determined, and the highest warning level reached is determined as the final warning level. Based on the final warning level, a corresponding graded alarm control signal is generated.

Images