Fourier spectrometer wave number calibration method, storage medium and system

Abstract

The invention discloses a Fourier spectrometer wave number calibration method, a storage medium and a Fourier spectrometer wave number calibration system, and belongs to the technical field of spectral analysis. According to the method, a dynamic threshold grading triggering mechanism is adopted, and non-interrupted real-time calibration is realized through cooperation of software-level calibration and hardware-level calibration. Comprising the steps that a reference instrument linear function ILS benchmark model is constructed, residual error distribution characteristics are counted, and a dynamic threshold initial value is set; target interference data are collected in real time, and software-level calibration is triggered through the atmospheric trace gas characteristic peak offset; if the continuous correction fails, switching to a built-in calibration substance light path to carry out hardware-level calibration; and combining sliding window statistics and forgetting factor iteration to update a dynamic threshold value, and self-adapting to environmental noise change. According to the invention, under aerospace on-orbit and industrial on-line scenes, automatic triggering and non-interrupted real-time calibration are realized, the hardware switching frequency is reduced by more than 60%, optical path health early warning and system self-inspection closed loop are integrated, and the long-term stability and reliability of the spectrometer are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the field of spectral analysis technology, and specifically relates to a wavenumber calibration method and system for a Fourier spectrometer, especially a high-precision real-time calibration method based on the dynamic threshold triggering of the instrument line shape function (ILS) residual in dynamic environments (such as in-orbit aerospace and industrial online). Background Art

[0002] The quantitative analysis accuracy of Fourier infrared spectrometer depends on the accurate calibration of wave number axis. Fourier infrared spectrometer can accurately calibrate the wave number axis by matching the characteristic absorption peak (such as CH4 at 1306cm -1 The error directly affects the reliability of the detection results. In scenarios such as aerospace remote sensing and industrial online detection, wavenumber consistency is a key prerequisite for multi-source data fusion. However, traditional wavenumber calibration methods have the following significant drawbacks:

[0003] (1) Dependence on lasers or monochromators: Existing technologies require the use of lasers or monochromators to generate reference wavenumbers, which results in large equipment size and high power consumption. This makes it impossible to integrate the equipment into aerospace payloads or industrial online systems. Furthermore, the calibration process requires interrupting the measurement, which violates the continuous monitoring requirement.

[0004] (2) Inefficiency: Existing online calibration methods lack a trigger mechanism and either require manual intervention or require full-process calibration for each set of spectra, resulting in a significant waste of computing resources and energy consumption.

[0005] (3) Insufficient adaptability: Using a fixed residual threshold to trigger calibration makes it difficult to cope with complex dynamic environments (such as temperature drift and mechanical vibration). Frequent false triggering occurs in low-noise environments, while missed triggering in high-interference scenarios leads to error accumulation.

[0006] To address the above problems, it is urgent to develop a wavenumber calibration method with intelligent triggering and graded correction to resolve the contradiction between accuracy and efficiency in online on-orbit scenarios. Summary of the Invention

[0007] In order to overcome the defects of traditional wavenumber calibration methods such as reliance on lasers, low calibration efficiency and insufficient environmental adaptability, and to meet the demand for long-term stable operation of spectrometers in dynamic scenarios such as aerospace on-orbit and industrial online, the present invention provides a Fourier spectrometer wavenumber calibration method, storage medium and system. The present invention integrates a dynamic threshold model with a hierarchical trigger mechanism, utilizes the synergy of characteristic peaks of atmospheric components (primary calibration) and calibration materials (secondary calibration), realizes non-interruptive high-precision calibration, reduces the hardware switching frequency by more than 60%, significantly improves the system reliability in complex environments, and meets the demand for long-term stable operation of spectrometers in aerospace, environmental protection and other fields.

[0008] According to a first aspect, the present invention provides a Fourier spectrometer wavenumber calibration method, comprising the following steps:

[0009] Step 1: Reference benchmark data collection and preprocessing;

[0010] Step 1.1: Monochromatic light source interferometry data acquisition: Under stable laboratory conditions, use a monochromatic light source such as a laser or a monochromator to continuously acquire interferometry data of a highly stable monochromatic light source.

[0011] Step 1.2: Interference signal apodization and phase error correction: Apply a window function to weight the interferogram to suppress the sidelobe interference caused by the truncation effect. Then, perform phase correction to reduce the phase deviation caused by the nonlinear motion of the moving mirror and the detector delay.

[0012] Step 1.3: Reference instrument line shape function generation: Perform a fast Fourier transform (FFT) on the pre-processed interference signal to generate the spectrum of the monochromatic light source as the instrument line shape function (ILS) for the reference standard.

[0013] Step 2: Reference benchmark threshold model construction and dynamic parameter initialization;

[0014] Step 2.1: Instrument Linear Shape Function (ILS) model fitting and wavenumber correction; The main reason for the inaccurate wavenumber of the Fourier spectrometer is that the central wavelength of the laser used for interferometer sampling shifts with temperature after a period of operation, resulting in a change in the optical path difference sampling interval △OPD. Based on the reference ILS generated in step 1, a matching mathematical model is selected according to the type of toe-cut function for fitting, and the fitting peak position σ is extracted. fit , and then by comparing the theoretical peak position σ theory , calculate the optical path difference sampling interval correction factor γ, calibrate the spectrometer sampling interval parameter △OPD, and regenerate the wavenumber axis based on the calibrated sampling interval to eliminate the inherent wavenumber deviation of the instrument. The calibrated optical path difference sampling interval and the reconstructed wavenumber axis are:

[0015]

[0016] N is the number of sampling points.

[0017] Step 2.2: Statistics of residual distribution and model fit distribution; Statistics of absolute residual after wave number correction △σ=|σ corrected -σ theory |Distribution characteristics, calculate its mean μ △σ and standard deviation s △σ , and simultaneously evaluate the root mean square error (RMSE) between the measured ILS and the fitted model, and calculate the mean RMSE μ RMSE and standard deviation sRMSE ,in

[0018] Step 2.3: Set the initial value of the dynamic threshold; according to the statistical results of step 2.2, set the initial threshold for triggering the wave number correction T = μ △σ +3s △σ , and stipulate the effective fitting condition RMSE∈[μ RMSE -k×s RMSE ,μ RMSE +k×s RMSE ], where k is an integer between 3 and 5 that is dynamically adjusted based on the actual ambient noise level. If it is not satisfied, the fitting is considered invalid. The specific determination method is shown in step 4.1.

[0019] Step 3: Real-time data acquisition and dynamic preprocessing;

[0020] Step 3.1: Online collection of target interference data: In actual working scenarios such as industrial online or aerospace on-orbit, target interference data is collected in real time.

[0021] Step 3.2: Dynamic preprocessing of the interference signal; To address the baseline drift of the interference signal caused by vibration and temperature drift in actual working scenarios, filtering or polynomial fitting is used to remove the low-frequency trend term, and then the window function consistent with step 1.2 is used for toe-cutting and phase correction.

[0022] Step 3.3: High-speed spectral inversion and wavenumber offset calculation: To meet the online real-time measurement requirements of the Fourier spectrometer, the GPU is used to accelerate the inversion of the measured target spectrum. The characteristic peaks of atmospheric trace gases (such as CO2 and H2O) (derived from the NIST database or HITRAN database) are used for judgment. The offset between the measured peak position and the theoretical peak position determines whether to trigger wavenumber calibration.

[0023] Step 4: Hierarchical trigger calibration and correction;

[0024] Step 4.1: Level 1 (software level) calibration: During real-time data acquisition, if the absolute residual △σ between the measured peak position and the theoretical peak position exceeds the threshold T or the RMSE exceeds the limit, the level 1 (software level) calibration is triggered, and the characteristic peaks of atmospheric trace gases (such as CO2, H2O) are used to iteratively correct until △σ≤T and RMSE∈[μ RMSE -k×s RMSE ,μ RMSE +k×s RMSE ], where k is an integer of 3 to 5 that is dynamically adjusted according to the actual ambient noise level. The iterative correction method is as follows:

[0025]

[0026] N is the number of sampling points.

[0027] If the condition is still not satisfied after exceeding the maximum number of iterations (e.g., 10), activate the secondary (hardware-level) calibration, see step 4.2.

[0028] Step 4.2: Secondary (hardware level) calibration: The piezoelectric ceramic mirror (switching time ≤ 5ms) switches the optical path to a calibration optical path containing a built-in calibration material (such as polystyrene film, PET film, PE film), collects high signal-to-noise ratio characteristic peaks, and performs wavenumber correction until △σ≤T is satisfied and RMSE∈[μ RMSE -3s RMSE ,μ RMSE +3s RMSE ] In theory, the signal-to-noise ratio of the characteristic peak of the built-in calibration substance is high enough to meet the standard after a certain number of iterations. If it is still not met after multiple iterations (such as 5 times), an abnormality is reported.

[0029] Step 4.3: Wavenumber correction and system self-check closed loop; based on the first or second level calibration results, update the wavenumber axis parameters and write them into the non-volatile memory; if the correction is successful, the system returns to the real-time monitoring mode; if the correction fails three times in a row, a hardware fault alarm is triggered.

[0030] Step 5: Threshold dynamic update and system maintenance;

[0031] Step 5.1: Sliding window statistics and threshold adaptive update: Based on the latest N = 100 data sets in the current calibration period (window size is adjustable), the sliding window method is used to calculate the distribution statistics to obtain μ △σ and μ RMSE , s △σ and s RMSE , introduce the forgetting factor α = 0.9 (to suppress the weight of historical data), and iteratively update the statistics using the exponentially weighted moving average (EWMA) method:

[0032] μ update =αμ old +(1-α)μ new

[0033] s update =αs old +(1-α)s new

[0034] This iterative method is applicable to both △σ and RMSE.

[0035] Step 5.2: Health status warning of the optical path switching mechanism; the number of switching times of the piezoelectric ceramic reflector is recorded in real time, and when the cumulative number approaches the design life threshold, a warning is triggered.

[0036] Step 5.3: Periodic calibration and status self-check: automatically perform system status self-check according to the preset period (default 24 hours, adjustable) and report the current system status.

[0037] According to a second aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the Fourier spectrometer wavenumber calibration method of the present invention.

[0038] According to a third aspect, the present invention further provides a Fourier spectrometer wavenumber calibration system, comprising the following modules:

[0039] FTIR main module: This module is based on the Michelson interferometer architecture and consists of a beam splitter, a moving mirror, a fixed mirror and a detector. It is used to collect target interference data and transmit it to the pre-processing module in real time.

[0040] Dynamic preprocessing module: This module integrates preprocessing and spectral inversion functions, removes low-frequency trend items caused by environmental factors, performs toe-cut and phase correction on the interference signal, and uses GPU to accelerate the inversion spectrum to ensure real-time online monitoring.

[0041] Hierarchical trigger engine: This module calculates the characteristic peak offset (△σ) and the model root mean square error (RMSE) in real time based on FPGA hardware logic, and uses the dynamic threshold model (initial threshold T = μ △σ +3s △σ And RMSE∈[μ RMSE -k×s RMSE ,μ RMSE +k×s RMSE ], supports noise adaptive coefficient k = 3 ~ 5) to determine whether to trigger calibration, and optimize the wave number axis iteratively through software. If the correction fails after exceeding the maximum number of iterations (10 times), the secondary (hardware level) calibration is activated through the hardware interrupt signal (delay ≤ 1ms) to switch the optical path to the calibration optical path containing built-in calibration material, achieving millisecond-level response.

[0042] Optical path switching mechanism: This module uses a piezoelectric ceramic mirror array (switching time ≤ 5ms, repeatability accuracy ±0.05°) to quickly switch the optical path to a built-in calibration material (such as polystyrene film, PET film, PE film). The mirror position is fed back in real time via a Hall sensor (accuracy ≤ 0.01°). A redundant mirror group is configured to automatically switch to a backup mirror group when the primary mirror group drive voltage exceeds the limit or the displacement deviation is greater than 0.1°, ensuring continued reliable operation of the system in extreme environments.

[0043] Adaptive maintenance unit: This module uses a sliding window statistic (window size 50 to 200 groups adjustable) to dynamically update the residual mean and standard deviation, combined with the exponentially weighted moving average (EWMA) algorithm with a forgetting factor (α = 0.9) to suppress historical data interference, while recording the number of optical path switching times and correlating it with the life prediction model. When the cumulative number reaches the design life (≥1×10 6 A warning is triggered when 90% of the time (times) are exceeded, and a status code (0x00 normal, 0x01 / 0x02 fault) is generated through a daily self-test protocol (diagnosing light source stability, reflector response, and storage verification) to achieve closed-loop health management.

[0044] Beneficial effects of the present invention:

[0045] The present invention adopts a dynamic threshold hierarchical trigger mechanism, and realizes non-interruptive real-time calibration through the coordination of software-level calibration and hardware-level calibration. The present invention automatically triggers and performs non-interruptive real-time calibration in scenarios such as aerospace on-orbit and industrial online, reducing the frequency of hardware switching by more than 60%, and integrating optical path health warning and system self-test closed loop, significantly improving the long-term stability and reliability of the spectrometer. The system of the present invention realizes non-interruptive high-precision wavenumber calibration in complex industrial environments through the deep coordination of hardware modular design and intelligent algorithms, and significantly improves long-term reliability through redundant design and health warning, meeting the dual needs of aerospace, environmental protection and other fields for high performance and high stability of spectrometers. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a diagram of the main steps of a Fourier spectrometer wavenumber calibration method described in the present invention.

[0047] Figure 2 This is a flow chart of an implementation method for Fourier spectrometer wavenumber calibration according to the present invention.

[0048] Figure 3 The figure is a schematic diagram of a Fourier spectrometer wavenumber calibration system according to the present invention.

[0049] Figure 4 This is a diagram showing the effect of instrument linear shape function (ILS) model fitting and wavenumber correction in an embodiment of the present invention.

[0050] Figure 5 This is a comparison diagram before and after correction of ammonia signals in a chemical park in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to better demonstrate the purpose, content and advantages of the present invention, the following describes in detail the specific implementation of the present invention in conjunction with an example of a self-developed long-wave infrared Fourier spectrometer deployed for online monitoring in a chemical park. Figure 1 The main implementation steps are divided into 5 steps. The specific implementation process can be found in Figure 2 .

[0052] Step 1: Reference benchmark data collection and preprocessing;

[0053] Step 1.1: Monochromatic light source interferometry data acquisition: In a constant temperature (25 ± 0.1 ° C) and vibration-isolated laboratory, a quantum cascade laser (QCL) laser (center wave number: 950.57 cm -1 ) as a monochromatic light source. After the laser was warmed up for 30 minutes, 100 sets of interferometric data were collected.

[0054] Step 1.2: Interference signal apodization and phase error correction: A triangular window function is used to weight the interferogram. This window has a wide main lobe width but low sidelobe attenuation, making it suitable for sidelobe-sensitive spectral measurement scenarios such as industrial online monitoring. Mertz phase correction is then applied to reduce phase deviations caused by nonlinear motion of the moving mirror and detector delay.

[0055] Step 1.3: Generate the reference instrument linear function; perform FFT on the pre-processed interferometric data and intercept the central wave number of the quantum cascade laser (950.57±4cm -1 ) interval data to generate a reference instrument linear function (ILS).

[0056] Step 2: Reference benchmark threshold model construction and dynamic parameter initialization;

[0057] Step 2.1: Instrument Linear Shape Function (ILS) model fitting and wavenumber correction; The main reason for the inaccurate wavenumber of the Fourier spectrometer is that the central wavelength of the laser used for interferometer sampling shifts with temperature after working for a period of time, resulting in a change in the optical path difference sampling interval △OPD. Therefore, the peak position σ can be fitted according to the fit and the theoretical peak position σ theory Calculate the optical path difference sampling interval correction factor γ, and regenerate the wave number axis based on the corrected sampling interval. Since the impulse response of the triangular window is a sinc square function, a Gaussian model is used for fitting, and the Levenberg-Marquardt algorithm is used for iteration. The model fitting and wave number correction effects can be referred to Figure 4 , the corrected optical path difference sampling interval and the reconstructed wavenumber axis are:

[0058]

[0059] N is the number of sampling points.

[0060] Step 2.2: Statistics of residual distribution and model fit distribution; statistics of absolute residuals after wave number correction for 100 sets of data

[0061] △σ, calculate its mean μ△σ and standard deviation s △σ , and simultaneously evaluate the root mean square error (RMSE) between the measured ILS and the fitted model, and calculate their mean μ RMSE and standard deviation s RMSE . After calculation, μ △σ =0.05237,s △σ =0.00614, μ RMSE =0.07325,

[0062] s RMSE =0.009746.

[0063] Step 2.3: Set the initial value of the dynamic threshold; set the initial threshold for triggering the wave number correction: T = μ △σ +3s △σ =0.07079, fitting effective condition: RMSE∈[μ RMSE -k×s RMSE ,μ RMSE +k×s RMSE ], where k is an integer of 3 to 5 that is dynamically adjusted according to the actual ambient noise level.

[0064] Step 3: Real-time data acquisition and dynamic preprocessing;

[0065] Step 3.1: Collect target interference data online; deploy spectrometers in the factory corridor and tank area to collect interference data online.

[0066] Step 3.2: Dynamic preprocessing of the interferometric signal; a Savitzky-Golay filter is used to eliminate the low-frequency baseline drift of the interferometric signal, followed by a triangular window for apodization and Mertz phase correction.

[0067] Step 3.3: High-speed spectral inversion and wavenumber offset calculation; the spectral coverage range of the self-developed long-wave infrared Fourier spectrometer is 650cm -1～ 1350cm -1 After GPU accelerated inversion of the spectrum, CO2 (667.95cm -1 ) or H2O(1339.27cm -1 ) characteristic peaks, and calculate the offset △σ between the measured peak position and the theoretical value in the HITRAN database.

[0068] Step 4: Hierarchical trigger calibration and correction;

[0069] Step 4.1: Level 1 (software level) calibration: During real-time data acquisition, if the offset △σ between the measured peak position and the theoretical peak position exceeds the threshold T three times in a row or the RMSE exceeds the limit (RMSE limit range is adjusted according to the actual signal-to-noise ratio), iterative correction is initiated. The correction method is:

[0070]

[0071] N is the number of sampling points.

[0072] The maximum number of iterations is 10, and the convergence condition is Δσ≤0.15cm -1 The RMSE must be less than 0.25 (the widening of the triangular window main lobe naturally increases the peak positioning error, and this condition is relaxed to avoid frequent false triggering of hardware calibration in extreme industrial environments). If the first-level correction fails (e.g., no convergence after 10 iterations), the second-level (hardware-level) calibration is triggered.

[0073] Step 4.2: Secondary (hardware-level) calibration: If the primary calibration fails, the piezoelectric ceramic mirror switches to the internal PET film calibration optical path within 5ms. Five sets of high-SNR interferometry data are collected, and the optical path difference sampling interval △OPD is repeatedly corrected to reconstruct the wavenumber axis. If the conditions are still not met after multiple iterations (e.g., five), an exception is reported.

[0074] Step 4.3: Wavenumber correction and system self-test closed loop; after successful correction, the wavenumber parameters are updated to non-volatile memory. If the correction fails three times in a row, a fault code is triggered, the system switches to the backup mirror set, and resets the control parameters.

[0075] Step 5: Threshold dynamic update and system maintenance;

[0076] Step 5.1: Sliding window statistics and adaptive threshold update: Use the most recent N = 100 data sets within the current calibration period as the sliding window, and update the statistics every 20 new data sets. Use a forgetting factor of α = 0.9 (to suppress the weight of historical data) and iteratively update the statistics using the exponentially weighted moving average (EWMA) method:

[0077] μ update =αμ old +(1-α)μ new

[0078] s update =αs old +(1-α)s new

[0079] This iterative method is applicable to both △σ and RMSE.

[0080] Step 5.2: Health status warning of the optical path switching mechanism; the number of switching times of the piezoelectric ceramic reflector is recorded in real time. When the cumulative number approaches the design life threshold, a warning signal is triggered and a status code is uploaded.

[0081] Step 5.3: Periodic calibration and status self-test: perform self-test at 0:00 every day and report the current system status.

[0082] Figure 5 Comparison of ammonia signals monitored in the chemical park before and after correction.

[0083] The system architecture of this embodiment is as follows Figure 2 As shown in the figure, the specific module hardware configuration and function implementation are as follows:

[0084] FTIR main module: Based on the Michelson interferometer architecture, it consists of a beam splitter, a moving mirror, a fixed mirror and a detector to collect target interference data.

[0085] Dynamic preprocessing module: Integrates preprocessing functions with spectral inversion functions, supports GPU-accelerated spectral inversion, and ensures real-time online monitoring.

[0086] Hierarchical trigger engine: FPGA-based implementation of characteristic peak offset (△σ) and model root mean square error (RMSE) calculation, with hardware-level interrupt triggering function and trigger delay ≤1ms.

[0087] Optical path switching mechanism: It adopts piezoelectric ceramic reflector array and is equipped with built-in calibration material (such as polystyrene film, PET film, PE film). The redundant mirror group supports automatic switching in case of failure.

[0088] Adaptive maintenance unit: records the number of optical path switching times and associates it with the life prediction model, and periodically generates health status reports.

[0089] Table 1 is a comparison of several typical characteristic peak positions before and after correction of ammonia signal in a chemical park in an embodiment of the present invention.

[0090] Table 1

[0091]

[0092]

[0093] This system embodiment achieves non-interruptible high-precision wavenumber calibration in complex industrial environments through deep collaboration between hardware modular design and intelligent algorithms. At the same time, it significantly improves long-term reliability through redundant design and health warning, meeting the dual requirements of high performance and high stability of spectrometers in aerospace, environmental protection and other fields.

Claims

1. A Fourier spectrometer wavenumber calibration method, characterized in that: include: (1) Reference benchmark data collection and preprocessing In a laboratory environment, interferometric data is collected using a monochromatic light source, and after weighting with an apodization function and phase correction, a reference instrument linear function (ILS) is generated. (2) Reference threshold model construction and dynamic parameter initialization Fit the reference ILS model, calculate the wavenumber correction absolute residual △σ and the model root mean square error RMSE, and statistically analyze their mean μ and standard deviation s, set the initial trigger threshold T and the fitting effective interval RMSE; (3) Real-time data acquisition and dynamic preprocessing Collect target interferometric data, remove low-frequency trend terms, invert the spectrum, and extract the characteristic peak offset of atmospheric trace gases; (4) Hierarchical trigger calibration and correction If △σ>T=μ for several consecutive times △σ +3s △σ Or if RMSE exceeds the limit, software-level calibration is triggered and the wavenumber axis is iteratively corrected; If the software-level calibration fails continuously, the system switches to the built-in calibration material optical path to perform hardware-level calibration. If the hardware-level calibration fails continuously, an exception is reported. (5) Threshold dynamic update and system maintenance The threshold parameters are dynamically updated based on the sliding window and the forgetting factor α, the number of optical path switching is accumulated and an early warning is issued, and the system self-check is performed periodically.

2. The method according to claim 1, characterized in that In step (2), the initial trigger threshold T = μ △σ +3s △σ ; The fitting effective interval coefficient is adjustable, satisfying RMSE∈[μ RMSE -k×s RMSE ,μ RMSE +k×s RMSE ], where k is a constant.

3. The method according to claim 2, characterized in that The constant k is an integer of 3 to 5 and is dynamically configured according to the actual environmental noise level. The noise level is determined by the signal-to-noise ratio (SNR) of the real-time interference signal. When SNR is less than or equal to 20 dB, k is set to 5, and when SNR is greater than 20 dB, k is set to 3.

4. The method according to claim 1, wherein In step (4), the software-level calibration uses characteristic peaks of CO2 and H2O in the atmosphere, and the peak positions are selected from the NIST database or the HITRAN database.

5. The method according to claim 1, wherein In step (4), the hardware level calibration uses the transmission characteristic peak of polystyrene film, polyethylene terephthalate or polyethylene film, and the optical path switching time is ≤5ms, and the switching mechanism life is ≥1×10 6 Second-rate.

6. The method according to claim 1, wherein The method for iteratively correcting the wave number axis in step (4) is as follows: Where N is the number of sampling points, △OPD is the optical path difference sampling interval, σ theor is the theoretical peak position.

7. The method according to claim 1, characterized in that The size of the sliding window in step (5) is adjustable from 50 to 200 sets of data, the forgetting factor α ranges from 0.85 to 0.95, and the statistics are iteratively updated using the following formula: m update =am old +(1-a)m new s update =αs old +(1-a)s new 。 8. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program is executed by a processor to implement the steps of a Fourier spectrometer wavenumber calibration method according to any one of claims 1 to 7.

9. A calibration system for implementing the Fourier spectrometer wavenumber calibration method according to any one of claims 1 to 7, characterized in that: include: FTIR main module: Based on the Michelson interferometer architecture, it consists of a beam splitter, a moving mirror, a fixed mirror, and a detector to collect target interference data; Dynamic preprocessing module: Integrates preprocessing and spectrum inversion functions, supports GPU-accelerated spectrum inversion, and ensures real-time online monitoring. Hierarchical trigger engine: FPGA-based calculation of characteristic peak offset △σ and model root mean square error RMSE, with hardware-level interrupt triggering function and trigger delay ≤1ms; Optical path switching mechanism: It uses a piezoelectric ceramic reflector array and is equipped with built-in calibration material. The redundant mirror group supports automatic switching in the event of failure. Adaptive maintenance unit: records the number of optical path switching times and associates it with the life prediction model, and periodically generates health status reports.

10. The calibration system according to claim 9, characterized in that: The redundant mirror group of the optical path switching mechanism includes at least two groups of spare reflective mirrors. When it is detected that the response time of the current mirror group has timed out or the driving voltage is abnormal, it automatically switches to the spare mirror group and marks a fault code.

Images