An exhaled breath synchronous detection device and method based on an aliasing spectral model

Abstract

The application provides an exhaled gas synchronous detection device and method based on a superimposed spectrum model, and relates to the technical field of exhaled gas detection. The device comprises an LED broadband light source, a band-pass filter, a collimating lens combination, an optical resonant cavity, a converging lens, an optical fiber, a grating spectrometer, a detector and an upper computer. Infrared light emitted by the LED broadband light source passes through the band-pass filter, the collimating lens combination, the optical resonant cavity, the converging lens, the optical fiber and the grating spectrometer in sequence to form a composite light signal. The detector converts the composite light signal into a superimposed spectrum signal. The upper computer is integrated with an exhaled gas synchronous detection method based on the superimposed spectrum model, that is, the superimposed spectrum model is used for decoupling analysis of the superimposed spectrum signal. The technical scheme of the application overcomes the problems in the prior art, such as the superimposed interference of adjacent spectrum lines and high detection cost when synchronously detecting ethanol and n-pentane.

Description

Technical Field

[0001] This invention relates to the field of exhaled gas detection technology, specifically to a device and method for simultaneous exhaled gas detection based on an aliasing spectral model. Background Technology

[0002] During normal human metabolism, intestinal bacterial fermentation may produce trace amounts of ethanol (C2H5OH), but its concentration is usually below 1 ppm. When a person produces a high concentration of ethanol without having consumed alcohol, it may be related to abnormal intestinal bacterial fermentation, which is a pathological condition. Similarly, the normal production of n-pentane (C5H5OH) by the human body... 12 The concentration should be below 10 ppb. Concentrations above this level may be associated with lipid peroxidation, kidney disease, schizophrenia, and certain cancers. Therefore, simultaneous detection of the concentrations of these two gases can help doctors identify potential health problems early, enabling more precise disease screening and a more comprehensive assessment of the patient's health status.

[0003] Early screening and diagnosis methods for human diseases include X-ray examination, CT, MRI, and cell biopsy, but these methods suffer from drawbacks such as high cost and invasiveness. Exhaled breath testing, as a non-invasive method, detects the composition and concentration of exhaled gases to assess human health status. It is applicable to the early screening and diagnosis of various diseases and has broad application prospects. Currently, the main technologies for exhaled breath detection both domestically and internationally include gas chromatography, gas chromatography-mass spectrometry, electrochemical technology, and infrared absorption spectroscopy. Compared with other technologies, infrared absorption spectroscopy not only has higher detection sensitivity and accuracy but also a shorter response time, enabling real-time online analysis of the gas being tested without frequent replacement of detection elements.

[0004] To prevent misdiagnosis during breath analysis, simultaneous analysis of multiple gases and volatile compounds is often necessary. Therefore, achieving high sensitivity for multi-component gas detection is crucial. Incoherent Broadband Cavity-Enhanced Absorption Spectroscopy (IBB-CEAS) is a high-sensitivity gas detection method based on an optical resonant cavity. It achieves a high effective absorption path length through multiple reflections within a short optical resonant cavity and enables simultaneous detection of multiple component gas concentrations over a wide spectral range using a single broadband light source. This reduces costs, system complexity, and optical path coupling difficulties.

[0005] Different gas molecules have different specific absorption lines. When detecting the concentration of a gas to be measured, a relatively independent specific absorption peak is often selected as the target absorption line. However, due to the complexity of absorption lines in the mid-infrared region, it is more susceptible to aliasing interference from adjacent spectral lines.

[0006] Therefore, there is a need for a low-cost device and method for simultaneous detection of exhaled gases that is not easily affected by the aliasing of adjacent spectral lines when simultaneously detecting ethanol and n-pentane. Summary of the Invention

[0007] The main objective of this invention is to provide a device and method for simultaneous detection of exhaled gases based on an aliasing spectral model, in order to solve the problems of easy interference from aliasing of adjacent spectral lines and high detection costs when simultaneously detecting ethanol and n-pentane in the prior art.

[0008] To achieve the above objectives, this invention provides a synchronous exhaled gas detection device based on an aliasing spectral model, comprising: an LED broadband light source, a bandpass filter, a collimating lens assembly, an optical resonant cavity, a converging lens, an optical fiber, a grating spectrometer, a detector, and a host computer; the infrared light emitted by the LED broadband light source passes sequentially through the bandpass filter, the collimating lens assembly, the optical resonant cavity, the converging lens, the optical fiber, and the grating spectrometer to form a composite optical signal; the detector converts the composite optical signal into an aliasing spectral signal; and the host computer integrates a synchronous exhaled gas detection method based on an aliasing spectral model, i.e., the aliasing spectral model performs decoupling analysis on the aliasing spectral signal.

[0009] Furthermore, the collimating lens assembly includes a biconvex lens and a biconcave lens, with light rays illuminating the biconcave lens from the biconvex lens.

[0010] Furthermore, the optical resonant cavity includes: a cavity body and mirrors located at both ends of the cavity body. The mirrors are fixed to the inner wall of the cavity body by mirror frames located on the inner walls at both ends of the cavity body. One mirror is placed at each end of the cavity body. When a light beam shines on a mirror at one end of the optical resonant cavity, the light is continuously reflected within the mirror, and the formula for calculating the final transmitted light intensity I is:

[0011]

[0012] Among them, I in The input incoherent light intensity is represented by L, the normalized absorption loss of the gas being measured is represented by R1 and R2, which are the reflectivities of the two mirrors, respectively.

[0013] Furthermore, PID fuzzy control is used to control the temperature of the LED broadband light source. The parameter ΔK of the PID fuzzy control is expressed as:

[0014]

[0015] Where, μ i c is the membership degree of the fuzzy set i. i This is the output of the corresponding control rule. The parameters of PID fuzzy control include: proportional ΔK. p Integral ΔK i Differential ΔK d .

[0016] Furthermore, the LED broadband light source is driven by a constant current source driving circuit, which includes: a first operational amplifier AOL1, a second operational amplifier AOL2, a first transistor BG1, a second transistor BG2, a load resistor RL, a sampling resistor RS, a sliding rheostat RW, and other resistors. The base of the first transistor BG1 is connected to the collector of the second transistor BG2, and then connected to the first operational amplifier AOL1 through the first resistor R1. The non-inverting input terminal of the first operational amplifier AOL1 is grounded through the eighth resistor R8 and the first capacitor C1, and connected to the adjustable terminal of the sliding rheostat RW through the sixth resistor R6. The fifth resistor R5 is connected to the inverting input and output of the first operational amplifier AOL1. The inverting input of the first operational amplifier AOL1 is connected to the output of the second operational amplifier AOL2 through the fourth resistor R4. The second resistor R2 is connected to the inverting input and output of the second operational amplifier AOL2. The inverting input of the second operational amplifier AOL2 is grounded through the third resistor R3. The non-inverting input of the second operational amplifier AOL2 is connected to the sampling resistor RS through the seventh resistor R7. The collector of the first transistor BG1 is connected to the emitter of the second transistor BG2, and then grounded through the load resistor RL and the sampling resistor RS.

[0017] The magnitude of the output current of the constant current source drive circuit is expressed as:

[0018] I L =R2×U ref [(R2+R3)×R S ];

[0019] Among them, U ref This is the reference voltage.

[0020] Furthermore, the grating in the grating spectrometer satisfies the grating equation:

[0021] mλ=d(sinα+sinβ);

[0022] Where m is the diffraction order, λ is the incident light wavelength, d is the grating constant, α is the incident angle, and β is the diffraction angle.

[0023] This invention also provides a method for simultaneous detection of exhaled gases based on an aliasing spectral model, specifically including the following steps:

[0024] S1, using the LDA algorithm, extracts the characterizations of C2H5OH and C5H from the original aliased absorption spectrum signal. 12 Feature values ​​of gas concentration information.

[0025] S2, use KELM to establish an aliasing spectral model.

[0026] S3. Divide the feature values ​​extracted in step S1 into a training set and a test set. Use the training set samples to train the constructed aliasing spectral model, and use the test set to test and evaluate the aliasing spectral model to obtain the optimal aliasing spectral model.

[0027] Furthermore, step S1 specifically includes the following steps:

[0028] S1.1, Calculate the mean of each type of spectral data and the overall mean, and let N be the number of spectral data in the k-th type. k Then the mean vector of the k-th class Where, x i For the i-th spectral data, C k Let k be the set of spectral data of the kth class, and let the population mean vector be... Where N is the number of all spectral data.

[0029] S1.2, Define the within-class scatter matrix S W and the inter-class scatter matrix S B :

[0030]

[0031] Where K′ is the number of categories.

[0032] S1.3, Calculation Eigenvalues ​​and eigenvectors:

[0033] For a spectral dataset with K′ categories, each spectral data point x∈R d Where d is the dimension, find a projection direction w∈R d The projected between-class variance and within-class variance are expressed as follows:

[0034] w T S B w、w T S W w; The optimal projection direction w is obtained by maximizing the ratio of the two, and the objective function J(w) is then:

[0035]

[0036] First, construct the Lagrangian function l(w,λ):

[0037] l(W,λ)=wT S B w-λ(w T S W w-1);

[0038] Where λ is a Lagrange multiplier;

[0039] Then, taking the derivative with respect to w and setting the derivative to 0:

[0040]

[0041] S B w=λS W w;

[0042] If S W If it is invertible, then the optimal solution is:

[0043]

[0044] At this point, the eigenvector corresponding to the largest eigenvalue is taken as the projection direction w.

[0045] When the number of categories K′>2, find a set of orthogonal projection directions W=[w1,w2,…,w K-1 This ensures that the projected data retains the maximum inter-class difference and minimum intra-class difference in the low-dimensional space. The objective function J(W) is then:

[0046]

[0047] Where tr() is the trace of the matrix;

[0048] At this point, the optimal solution is The eigenvectors corresponding to the first K′-1 largest eigenvalues.

[0049] S1.4 Project the original spectral data into a low-dimensional space Y = XW.

[0050] Furthermore, step S2 specifically includes the following steps:

[0051] S2.1, use KELM to solve for the output weight β:

[0052]

[0053] Where H is the kernel matrix, and the elements H ij =K(y i ,y j K is the kernel function type, C is the regularization parameter, T is the concentration label matrix, the kernel function is the radial basis function (RBF), and the kernel parameter is γ.

[0054] K(y i ,y j)=exp(-γ||y i -y j || 2 );

[0055] Here, exp() is the exponential function.

[0056] S2.2 uses the ALO algorithm to optimize parameters, specifically the regularization parameter C and kernel parameter γ in KELM.

[0057] The present invention has the following beneficial effects:

[0058] This invention employs a single incoherent broadband light source combined with a collimating lens group and a long optical path optical resonator to achieve the desired optical path for C2H5OH and C5H 12 Simultaneous detection of two-component gas concentrations, and the construction of an aliased spectral model using dimensionality reduction and machine learning algorithms, to complete the decoupling analysis of the aliased spectra, thereby providing accurate detection results.

[0059] This invention employs a single broadband infrared light source combined with a long optical path resonant cavity to achieve the absorption of C2H5OH and C5H in the fundamental frequency absorption band. 12 Simultaneous detection of two-component gas concentrations reduces system complexity while increasing the detection limit, providing a non-invasive and harmless detection technology for early disease screening.

[0060] This invention takes into account that incoherent broadband light sources have a large divergence angle, which can lead to problems such as difficulty in optical path coupling and low light source utilization. Therefore, a combination of biconvex and biconcave lenses is designed to converge and collimate the diverging beam, so that the beam can be coupled into the optical resonant cavity in the form of parallel light, thereby improving the coupling efficiency of the beam and the effective utilization of the light source.

[0061] This invention aims to achieve C2H5OH and C5H 12 Accurate inversion of gas concentrations led to the construction of an aliasing spectral model. The LDA algorithm was used to extract features from the aliasing absorption spectra, specifically extracting characterizing C2H5OH and C5H... 12 The feature values ​​of single gas concentration information were then used to perform concentration inversion on the extracted feature values ​​using the KELM algorithm. An ALO optimization algorithm was introduced to find the optimal parameters of the KELM algorithm, thereby further improving the concentration inversion accuracy. This is because C2H5OH and C5H... 12 A new solution is provided to address the problem of reduced concentration inversion accuracy caused by aliasing of gas absorption spectra. Attached Figure Description

[0062] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0063] Figure 1 A diagram of a synchronous exhaled gas detection device based on an aliasing spectral model according to the present invention is shown.

[0064] Figure 2 A schematic diagram of a biconvex lens is shown.

[0065] Figure 3 A simulation diagram based on the Zemax lens combination theory is shown.

[0066] Figure 4 A cross-sectional view of the optical resonator is shown.

[0067] Figure 5 A schematic diagram illustrating the principle of beam transmission in an optical resonant cavity is shown.

[0068] Figure 6 The constant current source drive circuit is shown.

[0069] The reference numerals in the above figures are:

[0070] 10. LED broadband light source; 20. Bandpass filter; 30. Collimating lens assembly; 31. Biconvex lens; 32. Biconcave lens; 40. Optical resonant cavity; 41. Mirror; 50. Converging lens; 60. Optical fiber; 70. Grating spectrometer; 80. Detector; 90. Host computer. Detailed Implementation

[0071] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] Example 1

[0073] like Figure 1The device for synchronous detection of exhaled gases based on an aliasing spectral model, as shown, includes: an LED broadband light source 10, a bandpass filter 20, a collimating lens assembly 30, an optical resonant cavity 40, a converging lens 50, an optical fiber 60, a grating spectrometer 70, a detector 80, and a host computer 90. The infrared light emitted by the LED broadband light source passes sequentially through the bandpass filter, the collimating lens assembly, the optical resonant cavity, the converging lens, the optical fiber, and the grating spectrometer to form a composite optical signal. The detector converts the composite optical signal into an aliasing spectral signal. The host computer integrates a method for synchronous detection of exhaled gases based on an aliasing spectral model, i.e., the aliasing spectral model performs decoupling analysis on the aliasing spectral signal.

[0074] LED broadband light sources emit infrared light with a wide wavelength range, enabling them to cover selected C2H5OH and C5H 12 The target absorption peak is determined; a bandpass filter removes wavelengths outside the high reflectivity range of the mirror to prevent high transmittance in the low reflectivity region from saturating the spectrometer. A collimating lens assembly converges and collimates the diverging light, coupling the outgoing light into the optical resonant cavity as parallel light. The optical resonant cavity, as the site of light-gas reaction, increases the effective optical path by causing multiple reflections of the incident light between two highly reflective mirrors, thus meeting the detection limit requirement. A converging lens collimates and converges the transmitted light signal from the optical resonant cavity, which is then coupled into the grating spectrometer via fiber optic cable. The grating spectrometer analyzes the intensity of the light signal emitted from the optical resonant cavity in the specified wavelength range. The host computer performs subsequent spectral data processing. An aliasing spectral model completes the decoupling analysis of the aliased absorption spectra, enabling simultaneous inversion of the concentration of the analyte gas.

[0075] The basic working principle of the device provided by this invention is as follows: the light emitted by the LED broadband light source is converged by a collimating lens combination and coupled into a high-precision optical resonant cavity in the form of parallel light. The light then undergoes multiple reflections within the cavity and is affected by C2H5OH and C5H... 12 Gas absorption occurs when a converging lens focuses the transmitted light and couples it to a grating spectrometer via fiber optic coupling for spectral analysis. The detector converts the composite optical signal into an electrical signal, which is then processed by the DSC300PA data acquisition system and transmitted to the host computer, forming a spectral signal reflecting the concentration of the analyte gas. An aliasing spectral model decouples the aliased spectral signals to determine the concentrations of C2H5OH and C5H... 12 Synchronous inversion of gas concentration.

[0076] The broadband LED light source is an important component of the IBB-CEAS system. In this invention, a broadband infrared light source (HIS2000R-0WC) is selected as the radiation source for this system. This light source is a hot emitter based on a nickel-chromium filament, packaged in a TO-8 package, and equipped with a gold-plated reflector and a gold-plated Winston cone collimator. When the reflector guides the radiation emitted from the rear of the filament to the front, the Winston cone collimator focuses and concentrates the beam to achieve optimal optical performance. Its scannable wavelength range is 2-20 μm, and the light source can achieve a power output of up to 830 mW, simultaneously covering the mid-infrared bands of C2H5OH and C5H... 12 The target absorption peak can not only improve the detection limit of the system, but also has a lower cost.

[0077] In this invention, considering the high transmittance in the low reflectivity band corresponding to the high-reflectivity mirror, to prevent the high amount of transmitted light generated by the high-reflectivity mirror in the low reflectivity band from saturating the spectrometer and affecting the measurement results, a bandpass filter is placed before the collimating lens assembly to filter out wavelength components outside the high reflectivity range, so that the emission spectrum matches the high reflectivity region of the optical resonant cavity and eliminates stray light interference. Combined with the selected C2H5OH and C5H... 12 Based on the target absorption spectral band range and the band range corresponding to the high reflectivity of the collimating lens combination, a bandpass filter with a center wavelength of 3400nm (±15nm) and a full width at half maximum (FWHM) of 155nm (±15nm) was finally selected.

[0078] Specifically, the collimating lens assembly includes a biconvex lens 31 and a biconcave lens 32, with light rays illuminating the biconcave lens from the biconvex lens.

[0079] Selecting a suitable lens can increase the coupling efficiency of the light beam and improve the effective utilization rate of the light source. In this invention, a combination of a biconvex lens and a biconcave lens is used to converge and collimate the diverging light beam radiated by the light source. By rationally designing the position and focal length of the biconcave lens, the light beam converged by the biconvex lens can be readjusted into parallel light, thus allowing the beam to couple into the optical resonant cavity in the form of parallel light and be effectively reflected within the cavity. In addition, a converging lens is used to effectively converge the light transmitted through the optical resonant cavity, enabling it to be effectively received by the grating spectrometer. In the design process of the collimating lens combination, the focal lengths of the biconvex and biconcave lenses need to be matched to ensure that the light beam, after being converged by the biconvex lens, diverges through the biconcave lens, but through appropriate position adjustment, the final output is parallel light, thereby achieving the collimation effect. Combining parameters such as the position of the light source, the focal length of the lens, and the divergence angle of the beam, the best collimation effect can be achieved by optimizing the distance between the two lenses. The principle of beam convergence by the biconvex lens is as follows: Figure 2 As shown.

[0080] For a biconvex lens with thickness d, radii of curvature R1 and R2, and refractive index n, the effective focal length f can be calculated using the following formula:

[0081]

[0082] Using a light source with a selected divergence angle of approximately 20°, a theoretical simulation of the aligned lens combination was performed using Zemax ray tracing software. The simulation results are as follows: Figure 3 As shown, the final parameters of the collimating lens assembly were determined. Specifically, the biconvex lens has a radius of curvature of 24.5 mm and an outer diameter of 25.4 mm; the biconcave lens has a radius of curvature of 69.6 mm and an outer diameter of 25.4 mm, with a distance of 2 cm between them. In addition, when determining the lens positions, it was necessary to ensure that the centers of both lenses were coaxial with the light source to avoid tilting and introducing aberrations, and that the lens apertures covered the beam diameter to prevent vignetting.

[0083] Specifically, the optical resonant cavity includes: a cavity body and mirrors located at both ends of the cavity body. The mirrors are fixed to the inner wall of the cavity body by mirror frames located on the inner walls at both ends of the cavity body. A mirror 41 is placed at each end of the cavity body. When a light beam shines on a mirror at one end of the optical resonant cavity, the light is continuously reflected within the mirror, and the formula for calculating the final transmitted light intensity I is:

[0084]

[0085] Among them, I in The input incoherent light intensity is represented by L, the normalized absorption loss of the gas being measured is represented by R1 and R2, which are the reflectivities of the two mirrors, respectively.

[0086] As the site of light-gas reaction, the optical resonant cavity demands extremely high stability. In its design, two mirror mounts are fixed to both ends of the optical cavity, and plano-concave mirrors are secured to the mounts via threads, allowing the two mirrors to be placed coaxially at a certain distance. The mounts and cavity are tightly secured using hexagonal screws and sealing rings, satisfying both the coaxial requirement and ensuring the cavity's sealing performance. The cavity is constructed entirely of magnesium-aluminum alloy, ensuring rigidity while effectively reducing weight and improving the overall system's portability. To enhance the integration of the overall optical system, four 30*30mm M3 threaded holes are designed on the exterior of the two mirror mounts, allowing components such as the light source and collimating lens group to be connected to the cavity via cage-like rods. A cross-sectional view of the overall mechanical structure is shown below. Figure 4 As shown, the incident incoherent beam undergoes multiple reflections within the optical resonant cavity to increase the effective absorption path of the gas, thereby further improving the system's sensitivity. Its beam transmission principle is as follows: Figure 5As shown in the diagram, when light is reflected inside the cavity, some light is transmitted through each reflection. The intensity of the transmitted light decreases exponentially with the number of reflections. The total transmitted light is the sum of the transmitted light from each reflection, as shown in the following formula.

[0087]

[0088] Since R1, R2, and L are all less than 1, the above equation can be simplified to: Subsequently, a grating spectrometer was used to detect the light intensity signal of the cavity in the frequency domain, thereby determining the relationship between the transmitted light intensity and the wavelength.

[0089] Specifically, incoherent broadband light sources are sensitive to temperature fluctuations. If heat dissipation performance is poor, thermal drift may occur, leading to changes in light intensity. To ensure the stability and reliability of the light source, temperature control is implemented, combining a temperature control system (PTC-10K) with a PID fuzzy control algorithm to jointly control the light source's temperature. The PID fuzzy control algorithm utilizes fuzzy set theory to transform precise input quantities into fuzzy quantities. Through reasoning using a predefined fuzzy rule base, it ultimately defuzzifies the fuzzy output quantity into a precise control quantity. This control method not only retains the advantages of PID controllers—simple structure and high reliability—but also effectively solves the limitations of traditional PID controllers in handling nonlinear, time-varying, and uncertain systems. Fuzzification involves converting the error e and the rate of change of error... Mapping to membership functions in the fuzzy set, typically using triangular or trapezoidal membership functions, followed by fuzzy inference based on the fuzzy rule base and the fuzzified input, determines the fuzzy set of control actions. The fuzzy control actions are then converted back to precise numerical values ​​to dynamically adjust the PID controller parameters to adapt to temperature changes in the light source. In this case, PID fuzzy control is used for temperature control of a broadband LED light source. The parameter ΔK of the PID fuzzy control is expressed as:

[0090]

[0091] Where, μ i c is the membership degree of the fuzzy set i. i This is the output of the corresponding control rule. The parameters of PID fuzzy control include: proportional ΔK. p Integral ΔK i Differential ΔK d .

[0092] When the temperature of the light source changes, the resistance of the thermistor also changes. Since the circuit formed by the temperature controller and the thermistor is a constant current source circuit with a known current, a feedback voltage can be obtained. The controller then uses a PID fuzzy control algorithm to intelligently adjust the operating state of the thermoelectric cooler (TEC1-6308) according to the preset temperature threshold, thereby achieving the purpose of temperature control. The basic principle of the thermoelectric cooler is the Peltier effect, which controls the temperature of the light source by adjusting the direction of current flow. The relationship between voltage and temperature is related to the thermistor model; the thermistor model selected in this invention is NTC 10K. Based on the relationship between the emitted light intensity and temperature of the light source, the setpoint of the temperature controller is established to ensure the stability of the emitted light power of the light source.

[0093] Specifically, because the driving current of incoherent broadband light sources is large and easily affected by heat generation, in order to protect the safety of the light source and provide a highly stable driving current, a design such as... Figure 6 The constant current source drive circuit shown is shown.

[0094] LED broadband light sources are driven by constant current source drive circuits, such as Figure 6 As shown, the constant current source drive circuit includes: a first operational amplifier AOL1, a second operational amplifier AOL2, a first transistor BG1, a second transistor BG2, a load resistor RL, a sampling resistor RS, a sliding rheostat RW, and other resistors; wherein, the base of the first transistor BG1 is connected to the collector of the second transistor BG2, and then connected to the first operational amplifier AOL1 through the first resistor R1. The non-inverting input terminal of the first operational amplifier AOL1 is grounded through the eighth resistor R8 and the first capacitor C1, and connected to the adjustable terminal of the sliding rheostat RW through the sixth resistor R6. The fifth resistor R5 is connected to the first operational amplifier AOL1. The inverting input and output of the first operational amplifier AOL1 are connected. The inverting input of the first operational amplifier AOL1 is connected to the output of the second operational amplifier AOL2 through the fourth resistor R4. The second resistor R2 is connected to the inverting input and output of the second operational amplifier AOL2. The inverting input of the second operational amplifier AOL2 is grounded through the third resistor R3. The non-inverting input of the second operational amplifier AOL2 is connected to the sampling resistor RS through the seventh resistor R7. The collector of the first transistor BG1 is connected to the emitter of the second transistor BG2, and then grounded through the load resistor RL and the sampling resistor RS.

[0095] The magnitude of the output current of the constant current source drive circuit is expressed as:

[0096] I L =R2×U ref [(R2+R3)×R S ];

[0097] Among them, U ref This is the reference voltage.

[0098] The constant current source drive circuit consists of two operational amplifiers forming the comparison and amplification stage, transistors BG1 and BG2 forming the adjustment stage, and R... L R is the load resistance. S R is the sampling resistor. W Provides a reference voltage for the circuit. If U is affected by power supply fluctuations... in When the voltage is reduced, thus decreasing the load current, the sampling voltage U... S This will inevitably decrease, thus reducing the difference between the sampled voltage and the reference voltage (U). S -U tef ) will inevitably decrease. Because U 1A It is an inverting amplifier, therefore its output voltage U b = (R5·R4)×U a It will inevitably rise, thus adjusting the process to make U S The increase returned to its original stable value, ensuring that U... S The voltage is stable, which in turn stabilizes the current. When U in When the voltage increases, the principle is the same as before; the circuit uses a closed-loop feedback system to make U... S The current drops back to its original stable value, thus maintaining a constant current. Adjust R. W Then change U ref It allows the current value to be continuously adjusted between 0 and 4A.

[0099] Specifically, the core of a grating spectrometer is the spectral dispersion function of the grating, which is mainly used to resolve the intensity of transmitted light at various wavelengths. The transmitted light from the resonant cavity, after being converged and collimated, is transmitted through optical fiber to the entrance slit of the grating spectrometer (Omni-λ500, Zolix). It is then projected onto the collimating objective lens to form a parallel beam, which is then projected onto the grating. The grating disperses the light to different angles according to the wavelength, and the beam is then imaged onto the exit slit by a focusing lens. The intensity of the emitted light is then detected using a liquid nitrogen-cooled detector. The detector outputs the converted spectral data to the DSC300PA data acquisition system via a USB interface. After processing the data, the DSC300PA data acquisition system transmits it to a computer. The computer then uses an aliasing spectral decoupling model to analyze the obtained transmission spectral data, thereby retrieving C2H5OH and C5H. 12 Gas concentration information. During this process, the slit width affects resolution and luminous flux; a narrower slit results in higher resolution but lower signal strength. Secondly, the number of grating lines and the blaze wavelength determine the operating range and resolution; selecting a suitable grating is crucial. The grating diffracts and disperses the incident light through a periodic groove structure, satisfying the grating equation:

[0100] mλ=d(sinα+sinβ)

[0101] Where m is the diffraction order, λ is the incident wavelength, d is the grating constant, α is the incident angle, and β is the diffraction angle. The existing grating spectrometers in the laboratory have grating line counts of 300 lines / mm, 600 lines / mm, and 1200 lines / mm. The grating line count of 300 lines / mm corresponds to a wavelength range of 2000-6000 nm, thus enabling the determination of wavelengths from selected C2H5OH and C5H... 12 Analysis of the absorption spectrum of the gas target. Since the grating line count N = 300 lines / mm, therefore... For a target wavelength λ = 3.3 μm, if the incident angle α = 0°, then the diffraction angle must satisfy:

[0102]

[0103] At this point, rotating the grating to the corresponding angle allows selection of a 3.3μm wavelength, thus enabling precise analysis of the gas absorption spectrum at 3.3μm.

[0104] Example 2

[0105] This invention also provides a method for simultaneous detection of exhaled gases based on an aliasing spectral model. Since aliasing of the selected target absorption lines severely affects the system's detection accuracy, this invention constructs an aliasing spectral model to separate the concentration information of different gas components from the aliased spectral signals. This model integrates feature engineering, machine learning, and intelligent optimization algorithms. It uses the LDA algorithm for dimensionality reduction to extract discriminative features, combines the powerful nonlinear modeling capabilities of the kernel KELM algorithm, and utilizes the ALO algorithm to optimize key parameters, collectively achieving high-precision inversion of the concentration of aliased absorption spectra. Specifically, it includes the following steps:

[0106] S1, using the LDA algorithm, extracts the characterizations of C2H5OH and C5H from the original aliased absorption spectrum signal. 12 Feature values ​​of gas concentration information. The LDA algorithm is used to extract low-dimensional discriminative features from high-dimensional aliased spectral data, reducing redundant information and enhancing the separability between components. LDA, a classic supervised learning algorithm, is mainly used for classification and feature dimensionality reduction. Its core idea is to maximize inter-class differences and minimize intra-class differences by projecting data into a low-dimensional space, thereby extracting the most discriminative features and achieving optimal classification or separation. Therefore, this algorithm can be used to extract features that can approximately represent C2H5OH and C5H from the original aliased absorption spectral signal. 12 The characteristic values ​​of gas concentration information are used to separate the gas concentration information of different components in the aliased absorption spectrum.

[0107] S2, using KELM to establish an aliased spectral model. As a kernel method variant of the extreme learning machine, KELM has the ability to handle nonlinear problems and is suitable for establishing complex mapping relationships between spectral features and concentrations.

[0108] S3. Divide the feature values ​​extracted in step S1 into a training set and a test set. Use the training set samples to train the constructed aliasing spectral model, and use the test set to test and evaluate the aliasing spectral model to obtain the optimal aliasing spectral model.

[0109] Specifically, step S1 includes the following steps:

[0110] S1.1, Calculate the mean of each type of spectral data and the overall mean, and let N be the number of spectral data in the k-th type. k Then the mean vector of the k-th class Where, x i For the i-th spectral data, C k Let k be the set of spectral data of the kth class, and let the population mean vector be... Where N is the number of all spectral data.

[0111] S1.2, Define the within-class scatter matrix S W and the inter-class scatter matrix S B :

[0112]

[0113]

[0114] Where K′ is the number of categories.

[0115] S1.3, Calculation Eigenvalues ​​and eigenvectors:

[0116] For a spectral dataset with K′ categories, each spectral data point x∈R d Where d is the dimension, find a projection direction w∈R d The projected between-class variance and within-class variance are expressed as follows:

[0117] w T S B w、w T S W w; The optimal projection direction w is obtained by maximizing the ratio of the two, and the objective function J(w) is then:

[0118]

[0119] First, construct the Lagrangian function l(W,λ):

[0120] l(w,λ)=wT s B W-λ(W T s W W-1);

[0121] Where λ is a Lagrange multiplier;

[0122] Then, taking the derivative with respect to W and setting the derivative to 0:

[0123]

[0124] S B w=λS W w;

[0125] If S W If it is invertible, then the optimal solution is:

[0126]

[0127] At this point, the eigenvector corresponding to the largest eigenvalue is taken as the projection direction w.

[0128] When the number of categories K′>2, find a set of orthogonal projection directions W=[w1,w2,…,w K-1 This ensures that the projected data retains the maximum inter-class difference and minimum intra-class difference in the low-dimensional space. The objective function J(W) is then:

[0129]

[0130] Where tr() is the trace of the matrix;

[0131] At this point, the optimal solution is The eigenvectors corresponding to the first K′-1 largest eigenvalues.

[0132] S1.4 Project the original spectral data into a low-dimensional space Y = XW.

[0133] Specifically, step S2 includes the following steps:

[0134] S2.1, use KELM to solve for the output weight β:

[0135]

[0136] Where H is the kernel matrix, and the elements H ij =K(y i ,y j K is the kernel function type, C is the regularization parameter, T is the concentration label matrix, the kernel function is the radial basis function (RBF), and the kernel parameter is γ.

[0137] K(y i ,y j)=exp(-γ||y i -y j || 2 );

[0138] Here, exp() is the exponential function.

[0139] S2.2 uses the ALO algorithm to optimize parameters, specifically the regularization parameter C and kernel parameter γ in KELM.

[0140] In KELM, the regularization parameter C controls model complexity and overfitting, while the kernel parameter γ affects the nonlinear mapping capability of the feature space. To optimize the regularization parameter C and kernel parameter γ in KELM, the ALO algorithm is used for parameter optimization, thereby improving the model's prediction accuracy. The ALO algorithm is a swarm intelligence optimization algorithm that simulates the hunting behavior of antlions and is suitable for continuous space optimization problems. The model performance under different parameter combinations is evaluated using mean squared error as the evaluation metric, and the optimal parameter combination is found through iterative optimization.

[0141] After constructing the aliased spectra, the stability and accuracy of the model were evaluated using the five-fold cross-validation method. The extracted feature values ​​were divided into training and test sets, with the training set comprising 80% of the total dataset and the test set comprising 20%. The constructed aliased spectra decoupling model was trained using the training set samples, and the model was tested and evaluated using the test set. After establishing the parameters of the optimal aliased spectra decoupling model, the properties of C2H5OH and C5H... 12 The aliased spectral data of the two-component gas concentration information are imported into this model to achieve decoupling analysis of the aliased absorption spectra, thereby completing the analysis of C2H5OH and C5H 12 Simultaneous detection of two-component gas concentrations.

[0142] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A device for simultaneous detection of exhaled gases based on an aliasing spectral model, characterized in that, include: The system consists of an LED broadband light source, a bandpass filter, a collimating lens assembly, an optical resonant cavity, a converging lens, an optical fiber, a grating spectrometer, a detector, and a host computer. The infrared light emitted by the LED broadband light source passes sequentially through the bandpass filter, the collimating lens assembly, the optical resonant cavity, the converging lens, the optical fiber, and the grating spectrometer to form a composite optical signal. The detector converts the composite optical signal into an aliased spectral signal. The host computer integrates a method for synchronous detection of exhaled gas based on an aliased spectral model, i.e., the aliased spectral model performs decoupling analysis on the aliased spectral signal. The collimating lens assembly includes a biconvex lens and a biconcave lens, with light rays passing from the biconvex lens onto the biconcave lens; An optical resonant cavity includes a cavity body and mirrors located at both ends of the cavity body. The mirrors are fixed to the inner wall of the cavity body by mirror holders located on the inner walls at both ends of the cavity body. A mirror is placed at each end of the cavity body. When a light beam shines on a mirror at one end of the optical resonant cavity, the light is continuously reflected within the mirror, and the intensity of the ultimately transmitted light is... The calculation formula is: ； in, This represents the intensity of the input incoherent light. This represents the normalized absorption loss of the measured gas. , These are the reflectivities of the two mirrors, respectively. The LED broadband light source is driven by a constant current source driver circuit, which includes: a first operational amplifier AOL1, a second operational amplifier AOL2, a first transistor BG1, a second transistor BG2, a load resistor RL, a sampling resistor RS, a sliding rheostat RW, and other resistors. The base of the first transistor BG1 is connected to the collector of the second transistor BG2, and then connected to the first operational amplifier AOL1 through the first resistor R1. The non-inverting input of the first operational amplifier AOL1 is grounded through the eighth resistor R8 and the first capacitor C1, and then connected to the adjustable terminal of the sliding rheostat RW through the sixth resistor R6. The five resistors R5 are connected to the inverting input and output of the first operational amplifier AOL1. The inverting input of the first operational amplifier AOL1 is connected to the output of the second operational amplifier AOL2 through the fourth resistor R4. The second resistor R2 is connected to the inverting input and output of the second operational amplifier AOL2. The inverting input of the second operational amplifier AOL2 is grounded through the third resistor R3. The non-inverting input of the second operational amplifier AOL2 is connected to the sampling resistor RS through the seventh resistor R7. The collector of the first transistor BG1 is connected to the emitter of the second transistor BG2, and then grounded through the load resistor RL and the sampling resistor RS. The magnitude of the output current of the constant current source drive circuit is expressed as: ； in, This is the reference voltage.

2. The exhaled gas synchronous detection device based on an aliasing spectral model according to claim 1, characterized in that, Temperature control of an LED broadband light source is achieved using PID fuzzy control. The parameters of the PID fuzzy control are... Represented as: ； in, It is a fuzzy set membership degree This is the corresponding control rule output. The parameters of PID fuzzy control include: proportional... ,integral ,differential .

3. The exhaled gas synchronous detection device based on an aliasing spectral model according to claim 1, characterized in that, The grating in a grating spectrometer satisfies the grating equation: ； in, For diffraction orders, The incident light wavelength, The grating constant is Angle of incidence It is the diffraction angle.

4. A method for simultaneous detection of exhaled gases based on an aliasing spectral model, applied to the device described in any one of claims 1-3, characterized in that, Specifically, the steps include the following: S1, using the LDA algorithm, extracts the characterizations of C2H5OH and C5H from the original aliased absorption spectrum signal. 12 Feature values ​​of gas concentration information; S2, Use KELM to establish an aliasing spectral model; S3. Divide the feature values ​​extracted in step S1 into a training set and a test set. Use the training set samples to train the constructed aliasing spectral model, and use the test set to test and evaluate the aliasing spectral model to obtain the optimal aliasing spectral model.

5. The method for simultaneous detection of exhaled gases based on an aliasing spectral model according to claim 4, characterized in that, Step S1 specifically includes the following steps: S1.1, calculate the mean of each type of spectral data and the overall mean, let the... The number of spectral data is Then the first mean vector of the class ,in, For the first Spectral data, for A collection of spectral-like data, population mean vector ,in, The number of all spectral data; S1.2, Define the within-class scatter matrix and inter-class scatter matrix : ； ； in, Number of categories; S1.3, Calculation Eigenvalues ​​and eigenvectors: For a person with Spectral data of each category, each spectral data ,in, For a dimension, find a projection direction. The projected between-class variance and within-class variance are expressed as follows: , The optimal projection direction is obtained by maximizing the ratio of the two. At this time, the objective function for: ； First, construct the Lagrange function. : ； in, For Lagrange multipliers; Subsequently Find the derivative and set it to 0: ； ； like If it is invertible, then the optimal solution is: ； At this point, the eigenvector corresponding to the largest eigenvalue is taken as the projection direction. ; When the number of categories At that time, find a set of orthogonal projection directions This ensures that the projected data retains the maximum inter-class difference and minimum intra-class difference in the low-dimensional space, at which point the objective function... for: ； in, The trace of the matrix; At this point, the optimal solution is The former The eigenvectors corresponding to the largest eigenvalues; S1.4 Project the raw spectral data into a low-dimensional space. .

6. The method for simultaneous detection of exhaled gases based on an aliasing spectral model according to claim 4, characterized in that, Step S2 specifically includes the following steps: S2.1, Solve for the output weights using KELM. : ； in, For the kernel matrix, elements ; It is a kernel function type. For regularization parameters; For the concentration label matrix, the radial basis function (RBF) is chosen as the kernel function, and the kernel parameters are as follows: : ； in, It is an exponential function; S2.2, Use the ALO algorithm to optimize parameters and improve the regularization parameters in KELM. and kernel parameters .

Images