Environmental parameter self-adaptive Raman spectrum explosive identification instrument calibration method

Abstract

The invention belongs to the technical field of Raman spectrum, and particularly relates to an environmental parameter self-adaptive Raman spectrum explosive identification instrument calibration method which comprises the following steps: constructing a Raman spectrum detection system; executing calibration, and controlling the laser source to irradiate the built-in standard substance module; a stable built-in standard substance module is integrated in the identification instrument, dependence on a real explosive sample is eliminated, a light source irradiates a harmless built-in standard substance during calibration, safety risks and regulatory obstacles caused by storage and use of dangerous goods in public places are fundamentally eliminated, and in addition, the identification instrument is convenient to use and popularize. The calibration process can be automatically triggered when external conditions change, drift of core parameters of the instrument caused by temperature drift, mechanical stress and the like can be quantified and compensated in real time through a wavelength-pixel coordinate correction relation established through polynomial fitting, it is ensured that wavelength coordinates output by the instrument are accurate in the on-site environment, and the calibration accuracy is improved. And a solid foundation is laid for subsequent accurate identification.

Description

Technical Field

[0001] This invention belongs to the field of Raman spectroscopy technology, specifically a calibration method for an environmental parameter adaptive Raman spectroscopy explosives identifier. Background Technology

[0002] Raman spectrometers used to identify explosives may experience data drift during prolonged use and in different environments. This can cause the output spectral data of the Raman spectrometer to deviate from the reference spectral data of the explosive, leading to false alarms and missed alarms. A Chinese patent with publication number CN104458701A discloses an automatic calibration method for a Raman spectrometer for explosives identification. The method includes: initializing the Raman spectrometer for explosives identification according to requirements; establishing analysis curves and saving data according to different analysis stages, where the analysis stages include a first analysis stage and other analysis stages; preparing the Raman spectrometer for explosives identification on-site; and analyzing samples using the Raman spectrometer for explosives identification. This method improves the measurement accuracy and precision of the Raman spectrometer for explosives identification, reduces the influence of measured explosives, especially infrared explosives with rich infrared spectra, on the spectral signal of the Raman spectrum, and thus enables the Raman spectrometer for explosives identification to be automatically calibrated under various temperature conditions.

[0003] In existing technologies, the core principle is to measure various known explosive standards in a controlled laboratory environment by changing parameters such as temperature, establishing a data set containing the correspondence between "environmental parameters and reference spectra". When the application scenario is on-site, unknown samples are measured based on the identification instrument, and then, according to the current environmental parameters, pre-stored reference spectra are retrieved from the database for matching and calibration. However, in actual security inspection and security scenarios, it is impossible to store or use real explosive samples for on-site calibration in places such as airports and train stations. Furthermore, due to the wide variety of explosives and the possibility of new unknown types, it is impossible to exhaust all possible standards. In addition, calibration needs to be completed quickly, and the process of measuring standards on-site will introduce unacceptable delays.

[0004] Therefore, the present invention provides a method for calibrating an explosives identifier based on environmental parameters using Raman spectroscopy. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve its technical problem is: a calibration method for an environmental parameter adaptive Raman spectroscopy explosives identifier, comprising the following steps: S1: Construct a Raman spectroscopy detection system, including a laser source, spectrometer, detector, built-in standard module, storage unit storing a standard Raman spectrum database and calibration history database, processing unit, and environmental sensor module; S2: Perform calibration, including: S21: Control the laser source to irradiate the built-in standard module, wherein the built-in standard module is loaded with calibration material that has known and stable Raman peak positions; S22: Acquire the first Raman spectrum generated by the calibration material through the detector and spectrometer; S23: Identify the measured pixel coordinates of the calibration material in the first Raman spectrum; S24: Compare the measured pixel coordinates with the standard Raman shift in the standard Raman spectroscopy database, calculate the wavelength-pixel coordinate correction relationship under the current instrument operating conditions, and save the key parameters of this calibration to the calibration history database; S3: Perform testing on the sample to be tested, including: S31: Control the laser source to irradiate the sample to be tested; S32: Acquire the second Raman spectrum generated by the sample to be tested; S33: Using the wavelength-pixel coordinate correction relationship established in S24, the pixel index of the second Raman spectrum is corrected to generate wavelength spectral data and obtain the corrected second Raman spectrum. S34: Perform a similarity calculation between the corrected second Raman spectrum and the reference spectrum in the standard Raman spectrum database to obtain the similarity score; S35: Based on similarity and real-time environmental parameters from the environmental sensor module, calculate the reliability score of this detection result based on the data reliability assessment model; S36: Output the detection results and the corresponding reliability score.

[0007] In one embodiment, in step S2, the calibration material is a silicon wafer with a standard Raman shift of 520.7. .

[0008] In one embodiment, the method for establishing the wavelength-pixel coordinate correction relationship in step S24 includes: S241: Data acquisition, based on S23, identifies multiple measured pixel coordinates of the calibration material. ,in No. The pixel positions in the spectrum output by the instrument under the current operating conditions where a characteristic peak appears are determined based on a peak lookup algorithm. S242: Curve fitting, using the measured pixel coordinates With the standard characteristic wavelength The data pairs By performing polynomial fitting, the corrected polynomial function is obtained: ; in, For the first The standard characteristic wavelengths of each characteristic peak are stored in a standard Raman spectroscopy database; Let be the fitting function, representing the pixel index. The corresponding wavelength; The input variable is the pixel index number in the original spectrum; This is a constant term, representing the offset of the wavelength axis intercept; The coefficient of the first term represents the linear scaling factor of the wavelength axis, corresponding to the dispersion of the spectrometer; These are the coefficients of higher-order terms, used to correct nonlinear responses; S243: Calculation based on least squares fitting, solution , , .

[0009] In one embodiment, in step S33, the method for correcting the second Raman spectrum is a resampling interpolation method, specifically including: S331: Define new wavelength axes with equal spacing

[0010] S332: The inverse function based on the aforementioned corrected polynomial function Calculate the new wavelength axis wavelength point on The corresponding original pixel coordinates ; S333: Using a linear interpolation algorithm, based on the original pixel coordinates and its corresponding spectral intensity Calculate the wavelength point spectral intensity ; S334: For the new wavelength axis Each wavelength point in Repeat steps S332-S333 to obtain a set of intensity values. ; S335: Will The data pair is defined as the corrected second Raman spectrum.

[0011] In one embodiment, the environmental sensor module includes a temperature sensor and a humidity sensor; the real-time environmental parameters in S35 include ambient temperature and ambient humidity; the data reliability assessment model is used to output a reliability score based on the ambient temperature, ambient humidity, and the signal-to-noise ratio of the second Raman spectrum.

[0012] In one embodiment, the method for outputting the detection result and the corresponding reliability score in step S36 includes: Data acquisition and input, including ambient temperature, ambient humidity, and signal-to-noise ratio; Data preprocessing and feature vector components: The processing unit calculates the reliability score based on ambient temperature, ambient humidity, and signal-to-noise ratio. The output includes the detection results and reliability score.

[0013] In one embodiment, the key parameters in S24 include the coefficients of the first-order term of the polynomial fitting. and quadratic coefficient ; Also includes S4: Predictive Maintenance: S41: The processing unit retrieves historical continuous data from the calibration history database. Next, the key parameters recorded are calibrated to form a parameter time series; S42: Perform trend analysis on the time series of the parameters. If it is determined that the drift rate of any key parameter exceeds a preset threshold, generate an early warning message and prompt maintenance.

[0014] In one embodiment, S33 further includes: S336: Preprocess the second Raman spectrum, including algorithms based on wavelet transform or principal component analysis to reduce noise and enhance the signal-to-noise ratio of characteristic peaks.

[0015] In one embodiment, in S34, the matching calculation uses a cosine similarity algorithm or the least squares method.

[0016] In one embodiment, the calibration phase in S2 is automatically triggered under any one or more of the following conditions: when the identifier is powered on, when a preset timing period is reached, and when the ambient temperature changes beyond a predetermined range.

[0017] The beneficial effects of this invention are as follows: 1. The present invention discloses an environmental parameter adaptive Raman spectroscopy explosive identification instrument calibration method. By integrating a stable built-in standard module within the identification instrument, it eliminates the dependence on actual explosive samples. During calibration, the light source illuminates the harmless built-in standard, eliminating the safety risks and regulatory obstacles associated with storing and using hazardous materials in public places. Furthermore, the calibration process can be automatically triggered when external conditions change, avoiding the operational delays and interruptions caused by manual intervention, gas cylinder replacement, or standard sample replacement required in traditional methods. The wavelength-pixel coordinate correction relationship established through polynomial fitting can quantify and compensate for the drift of the instrument's core parameters caused by temperature drift, mechanical stress, etc., in real time, ensuring that the wavelength coordinates output by the instrument are accurate under any field environment, laying a solid foundation for subsequent accurate identification.

[0018] 2. The environmental parameter adaptive Raman spectroscopy explosives identification instrument calibration method of the present invention reduces the risk of false alarms and missed alarms caused by environmental interference by quantifying and evaluating the quality of the current measurement environment and the measurement process itself through reliability score, thereby improving the intelligence and reliability of security inspection decisions. Attached Figure Description

[0019] The invention will now be further described with reference to the accompanying drawings.

[0020] Figure 1 This is a flowchart of the present invention; Figure 2 This is a flowchart of the reliability score calculation in this invention. Detailed Implementation

[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0022] like Figures 1 to 2 As shown in the embodiment of the present invention, a method for calibrating an environmental parameter adaptive Raman spectrometer for explosives identification includes the following steps: S1: Construct a Raman spectroscopy detection system, including a laser source, spectrometer, detector, built-in standard module, storage unit storing a standard Raman spectrum database and calibration history database, processing unit, and environmental sensor module; S2: Perform calibration, including: S21: Control the laser source to irradiate the built-in standard module, wherein the built-in standard module is loaded with calibration material that has known and stable Raman peak positions; S22: Acquire the first Raman spectrum generated by the calibration material through the detector and spectrometer; S23: Identify the measured pixel coordinates of the calibration material in the first Raman spectrum; S24: Compare the measured pixel coordinates with the standard Raman shift in the standard Raman spectroscopy database, calculate the wavelength-pixel coordinate correction relationship under the current instrument operating conditions, and save the key parameters of this calibration to the calibration history database; S3: Perform testing on the sample to be tested, including: S31: Control the laser source to irradiate the sample to be tested; S32: Acquire the second Raman spectrum generated by the sample to be tested; S33: Using the wavelength-pixel coordinate correction relationship established in S24, the pixel index of the second Raman spectrum is corrected to generate wavelength spectral data and obtain the corrected second Raman spectrum. S34: Perform a similarity calculation between the corrected second Raman spectrum and the reference spectrum in the standard Raman spectrum database to obtain the similarity score; S35: Based on similarity and real-time environmental parameters from the environmental sensor module, calculate the reliability score of this detection result based on the data reliability assessment model; S36: Output the detection results and the corresponding reliability score.

[0023] When the application scenario is on-site, the instrument measures unknown samples and then calls up pre-stored reference spectra from the database for matching and calibration based on the current environmental parameters. However, in actual security inspection and security scenarios, it is impossible to store or use real explosive samples for on-site calibration in places such as airports and train stations. Furthermore, due to the wide variety of explosives and the possibility of new unknown types, it is impossible to exhaust all possible standard samples. In addition, calibration needs to be completed quickly, and the process of measuring standard samples on-site will introduce unacceptable delays.

[0024] Therefore, in one embodiment of the present invention, a Raman spectroscopy detection system is first constructed. Notably, this detection system includes a built-in standard module. When calibration is required, the calibration material stored in the built-in standard module can be irradiated by the light path emitted from the laser source, generating a first Raman spectrum corresponding to the calibration material. Then, by identifying and analyzing the first Raman spectrum, the measured pixel coordinates of the calibration material are obtained. Based on the standard Raman shift of the calibration material pre-stored in the standard Raman spectrum database, a wavelength-pixel coordinate correction relationship corresponding to the calibration material under the current operating conditions of the identifier is established. Specifically, this relationship is a function of wavelength-pixel coordinates under the current operating conditions. Based on the wavelength-pixel coordinate correction relationship established using the calibration material, the sample to be tested is then detected. The sample is irradiated by the light path, and a second Raman spectrum corresponding to the sample is obtained. Based on the established wavelength-pixel coordinate correction relationship, the pixel index of the second Raman spectrum can be corrected, generating spectral data corresponding to the wavelength, thereby generating calibration data. The second Raman spectrum after correction is understood to be subject to drift due to accumulated usage time and environmental influences under current operating conditions. Therefore, the Raman spectrum generated by the light path illuminating the sample may deviate from the spectral data recorded in the standard Raman spectral database, leading to false alarms and missed alarms. In this embodiment, based on the detection of the built-in calibration material, the measured pixel coordinates in the first Raman spectrum of the known and stable calibration material under current operating conditions are generated and compared with the standard Raman shift recorded in the standard Raman spectral database to generate a wavelength-pixel coordinate correction relationship. This wavelength-pixel coordinate correction relationship can quantify the drift of the standard Raman shift in the Raman spectral data under current operating conditions and correct the measured pixel coordinates. The corrected second Raman spectrum of the sample can be directly compared with the reference spectrum in the standard Raman spectral database for similarity calculation, thereby determining whether the sample is characterized as a known explosive type, thus assisting in determining the presence and type of explosive. Furthermore, it should be noted that although the corrected second Raman spectrum and the calculated similarity are used to calibrate the identifier, enabling the corrected Raman spectrum output by the identifier to be directly compared with the reference spectrum in the standard Raman spectrum database for similarity calculation, the influence of the external environment on the Raman spectrum output by the identifier needs to be considered under the background of environmental changes. Therefore, this embodiment also includes calculating the reliability score of the detection result based on the real-time environmental parameters output by the environmental sensor module and combined with the data reliability assessment model. It can be understood that the above-mentioned second Raman spectrum of the test sample corrected based on the built-in standard can only be considered as calibrating the accuracy of the identifier, ensuring that the measured spectral peak position is accurate and can be directly compared with the standard spectral peak position for similarity calculation. However, under different environmental backgrounds, such as in a quiet and well-lit experimental environment or in a complex background environment, the same test sample may not be able to achieve the same level of similarity. The detection results output by the product detection cannot be directly determined. Specifically, low temperatures lead to a decrease in laser source efficiency, and detector sensitivity also changes with temperature. In addition, lens contamination can also lead to a decrease in light throughput. Although the above corrections are made to ensure the accuracy of the measured pixel coordinates, the weakened signal can cause the characteristic peaks to be masked by noise, resulting in a decrease in signal-to-noise ratio. Therefore, it can be understood that under low signal-to-noise ratio conditions, the output high similarity detection results may be based on coincidence and noise and are not entirely reliable. In addition, since explosives often adhere to substrates such as cloth and plastic, these foreign materials will produce strong fluorescent backgrounds. This phenomenon will be exacerbated under high temperatures. The strong fluorescent background will drown out the weak Raman signal. Even if the spectrum is calibrated, it is difficult to accurately extract the characteristic peaks for similarity calculation. Therefore, in this embodiment, a reliability score is introduced to quantify the degree of influence of the current environment on the detection results.

[0025] In one embodiment, in step S2, the calibration material is a silicon wafer with a standard Raman shift of 520.7. .

[0026] In this embodiment, the calibration material is set to a silicon wafer, and the standard Raman shift is at 520.7. Based on the core principle that Raman spectroscopy must be calibrated by wavelength, the standard materials used for calibrating Raman spectrometers are clearly specified, including silicon, with a wavelength of 520.7 nm. The characteristic peaks are taken as the primary criteria.

[0027] In one embodiment, the method for establishing the wavelength-pixel coordinate correction relationship in step S24 includes: S241: Data acquisition, based on S23, identifies multiple measured pixel coordinates of the calibration material. ,in No. The pixel positions in the spectrum output by the instrument under the current operating conditions where a characteristic peak appears are determined based on a peak lookup algorithm. S242: Curve fitting, using the measured pixel coordinates With the standard characteristic wavelength The data pairs By performing polynomial fitting, the corrected polynomial function is obtained: ; in, For the first The standard characteristic wavelengths of each characteristic peak are stored in a standard Raman spectroscopy database; Let be the fitting function, representing the pixel index. The corresponding wavelength; The input variable is the pixel index number in the original spectrum; This is a constant term, representing the offset of the wavelength axis intercept; The coefficient of the first term represents the linear scaling factor of the wavelength axis, corresponding to the dispersion of the spectrometer; These are the coefficients of higher-order terms, used to correct nonlinear responses; S243: Calculation based on least squares fitting, solution , , .

[0028] When the identification instrument is used for a long time or the environment changes, it is necessary to illuminate the built-in standard based on the light path output by the laser source and output the first Raman spectrum. Based on the analysis of the first Raman spectrum, the measured pixel coordinates of the calibration material are identified, and a wavelength-pixel correction relationship is established based on the measured pixel coordinates and the corresponding standard characteristic wavelength. The second Raman spectrum generated by the detection of the sample to be tested is corrected based on the wavelength-pixel correction relationship. Specifically, before establishing the wavelength-pixel coordinate correction relationship, the standard wavelength of the laser source is used as the basis for the correction. and the known standard Raman displacements of the built-in standard Through the formula:

[0029] The theoretical scattered light wavelength value of the built-in standard under the current laser conditions was calculated. The theoretical scattered wavelength value of the built-in standard is recorded in the standard Raman spectroscopy database as a standard characteristic wavelength, and Represented as the first The standard characteristic wavelength of each characteristic peak; In this embodiment, the theoretical scattered light wavelength value of the laser is... Taking silicon wafers as the calibration material as an example, the standard Raman shift of the main peak of silicon wafers Calculated as follows:

[0030] Therefore, it can be concluded that the theoretical wavelength of the Raman scattered light from the main peak of the silicon wafer under 785nm laser excitation should be 818.4nm. Find the standard Raman shift of the silicon wafer by referring to the table. As shown below:

[0031] Under the demonstration ambient temperature (25℃), the built-in silicon wafer was measured by the recognition instrument. Due to temperature drift, the spectrum shifted. After using the peak finding algorithm, the pixel coordinates of the four characteristic peaks in the current spectrum were obtained. As shown in the table below:

[0032] Based on the above, four sets of data pairs were obtained for fitting. : (480.3, 795.5), (495.2, 804.1), (520.5, 818.4), (562.8, 849.2) Substitute the above data into the polynomial And the least squares method is used for fitting and calculation; The fitting results are obtained:

[0033]

[0034]

[0035] Therefore, the polynomial can be expressed as:

[0036] In , , Once the solution is obtained, application verification can be performed. At this point, samples suspected of containing explosives are tested. Taking TNT as an example, the standard Raman shift corresponding to the characteristic peak of TNT is 900°. Before calibration, the sample to be tested is measured based on the recognition instrument, and the characteristic peak of the sample appears at the pixel coordinates. Location; At this point, the characteristic peaks of the sample to be tested, i.e., pixel coordinates, are... Substitute this into the polynomial above;

[0037] Based on calculations, we obtain ; Then, based on the calculated wavelength, it is converted into a Raman shift for comparison with a standard database:

[0038] Analysis revealed that before correction, due to the drift in the spectral data of the identifier, the theoretical standard Raman shift corresponding to the characteristic peak was 900°. However, calculations show that the actual Raman shift of the characteristic peak of the sample to be tested is... Therefore, the Raman shifts corresponding to the characteristic peaks of the two are basically the same. This indicates that, based on the correction relationship obtained above, and then comparing the actual Raman shift with the standard Raman shift, it was found that the Raman shift of the characteristic peak of the sample to be tested is basically consistent with the standard Raman shift recorded in the standard database. Therefore, it can be determined that the sample to be tested is suspected of containing TNT explosives, reducing false alarms and missed alarms caused by instrument spectral data drift. In practical applications, the identifier is first calibrated, and a wavelength-pixel coordinate correction relationship is established. Then, by identifying the position of the pixel coordinates corresponding to the characteristic peak of the second Raman spectrum, the pixel coordinates are input into the correction relationship. Based on the calculated actual Raman shift, it can be compared with all the reference spectra pre-stored in the standard Raman spectral database to determine whether it belongs to a known explosive.

[0039] In one embodiment, in step S33, the method for correcting the second Raman spectrum is a resampling interpolation method, specifically including: S331: Define new wavelength axes with equal spacing ; S332: The inverse function based on the aforementioned corrected polynomial function Calculate the new wavelength axis wavelength point on The corresponding original pixel coordinates ; S333: Using a linear interpolation algorithm, based on the original pixel coordinates and its corresponding spectral intensity Calculate the wavelength point spectral intensity ;

[0040] in, Original pixel coordinates The value rounded down; Original pixel coordinates The value rounded up; S334: For the new wavelength axis wavelength point in Repeat steps S332-S333 to obtain a set of intensity values. ; S335: Will The data pair is defined as the corrected second Raman spectrum.

[0041] In the above embodiments, after correction, only the corrected actual Raman shift is obtained. Similarity calculations between this corrected shift and reference data in the standard Raman spectral database are then performed based on spectral intensity. Calculations are performed, and a normalized intensity vector is established based on the spectral intensity. Compared with the reference spectrum in the standard Raman spectroscopy database Similarity calculation is performed; cosine similarity calculation relies entirely on spectral intensity values, but its effectiveness is absolutely contingent on the wavelength coordinates being precisely aligned through calibration.

[0042] In this embodiment, the raw intensity containing at least one characteristic peak is read from the detector. :

[0043] Based on the above, the target new wavelength axis parameters are: Starting wavelength:

[0044] End wavelength:

[0045] Wavelength spacing

[0046] Therefore, the new wavelength axis for:

[0047] Based on the resampling method, for the new wavelength axis Each wavelength point defined above Find these wavelength points The coordinates corresponding to the original pixel coordinates Then, the interpolation method is used to calculate the value corresponding to each wavelength point. intensity ; by Taking this wavelength point as an example: First, solve the equation based on the established wavelength-pixel coordinate correction relationship:

[0048] Then convert the equation into a standard quadratic equation:

[0049] According to the quadratic formula:

[0050] The pixel positions corresponding to each new wavelength point on the new wavelength axis are calculated. : As can be seen from the above, after calculating the new wavelength point... The corresponding pixel coordinates After that, it is also necessary to determine the pixel coordinates. Solving for the new wavelength point The corresponding new intensity value, generally speaking, is the new wavelength point. The intensity can be determined based on the pixel coordinates. Directly look up the table, but due to wavelength points... Falling on pixel index The position, and Since it is not an integer, the corresponding raw strength cannot be obtained directly by looking up a table. In this case, linear interpolation is required to calculate the interpolation intensity. because It is located between pixels 177 and 178; Therefore, by definition , ; Based on the formula:

[0051] Based on the above, the new wavelength axis Each wavelength point The calibrated spectral data were obtained through calculation, as shown in the table below:

[0052] Based on the above, multiple new wavelength points on the new wavelength axis are obtained. The corresponding interpolation intensity The interpolation intensity is obtained based on this calculation. In subsequent calculations, based on each wavelength point Corresponding interpolation intensity An intensity vector corresponding to the sample to be tested is constructed, and then a similarity calculation is performed with the intensity vectors of suspected explosive types recorded in the standard Raman spectral database. It is understood that, in the above embodiment, the matching of the actual Raman spectral shift and the standard Raman spectral shift alone is insufficient to indicate that the sample to be tested is a certain explosive. In specific application scenarios, it is also necessary to perform similarity calculation based on intensity in order to quantify whether the sample to be tested belongs to a certain type of explosive.

[0053] In one embodiment, the environmental sensor module includes a temperature sensor and a humidity sensor; the real-time environmental parameters in S35 include ambient temperature and ambient humidity; the data reliability assessment model is used to output a reliability score based on the ambient temperature, ambient humidity, and the signal-to-noise ratio of the second Raman spectrum.

[0054] In step S36, the method for outputting the detection result and the corresponding reliability score includes: Data acquisition and input, including ambient temperature, ambient humidity, and signal-to-noise ratio; Data preprocessing and feature vector components: The processing unit calculates the reliability score based on ambient temperature, ambient humidity, and signal-to-noise ratio. The output includes the detection results and reliability score.

[0055] In the aforementioned prior art, due to the influence of temperature and fluorescence background, even if the identifier is calibrated and the identification result shows that the spectrum of the sample is highly similar to that of at least one known explosive and exceeds a set threshold, it is impossible to directly output the conclusion that the sample contains an explosive. It is understandable that while there is a preprocessing step (noise reduction and signal-to-noise ratio enhancement) when the identifier measures the sample, this preprocessing only affects the output spectrum of the identifier, serving as a mandatory data purification function before similarity calculation. This maximizes the extraction of effective signals and suppresses noise, providing a clean and reliable data foundation for subsequent calculation steps. In this invention, considering the above factors, the influence of environment and fluorescence background must also be considered when outputting the results. Therefore, combining the reliability score can provide a reference for explosive identification. It can be understood that the output based on the reliability score enables quality assessment and decision support after preprocessing and spectral similarity calculation. The reliability score calculation in this embodiment is based on environmental temperature, environmental humidity, and the signal-to-noise ratio of the spectrum, specifically including: First, the ambient temperature, ambient humidity, and signal-to-noise ratio are input. Then, the processing unit calculates a reliability score based on the ambient temperature, ambient humidity, and signal-to-noise ratio. Among these parameters, the ambient temperature... The temperature sensor in the environmental sensor module reads data in real time, in degrees Celsius (°C), such as 25.3°C; ambient humidity... The humidity sensor in the environmental sensor module reads data in real time, in percentages (%), such as 45.6%. The spectral signal-to-noise ratio is calculated in real time by the processing unit after the second Raman spectrum is acquired. The calculation method is to divide the average signal intensity of the characteristic peak region by the standard deviation of the signal intensity of the region without characteristic peaks. For example... ; Processing unit according to , , , The reliability score is calculated. According to the formula:

[0056] in, Temperature reliability factor The real-time ambient temperature output by the environmental sensor module; The ideal operating temperature for the identification device; Temperature tolerance is a parameter used to control the curve width. It can be understood that when... The larger the value, the wider the curve, indicating that the identifier is less sensitive to temperature changes. In this embodiment, the control is based on an ideal temperature range, for example, it is desired that the operating temperature of the identifier be controlled between 15°C and 25°C, and the reliability factor... Through reverse derivation, then ;

[0057] in, Humidity reliability factor, The real-time ambient humidity output by the environmental sensor module; The humidity impact threshold is derived from the ideal operating conditions of the identifier. For example, in this embodiment, the ideal humidity for the Raman spectroscopy identifier is below 60%, and at a humidity of 60%, the humidity reliability factor should be approximately equal to 1. This is the attenuation coefficient, used to control real-time ambient humidity. Below the humidity impact threshold The rate of reliability degradation can be derived in reverse, based on the expectation that the humidity reliability factor should be approximately equal to 1 at 60% humidity, in this embodiment. This is used as the attenuation coefficient in the calculation of the humidity reliability factor of the identification instrument;

[0058] in, The signal-to-noise ratio reliability factor. For the actual measured signal-to-noise ratio, This is a half-saturation value, representing the reliability factor when it reaches 0.5. value; , where is the Hill coefficient, used to control the steepness of the curve. The larger the value, the steeper the curve and the faster the transition. In this embodiment, two signal-to-noise ratio (SNR) calibration points are set. Calibration point 1 is used to determine that the signal quality is poor and the reliability is extremely low when the SNR is below a certain value, for example... Calibration point 2 is used to determine that the signal quality is excellent when the signal-to-noise ratio is higher than a certain value, for example... Based on this, the equations established by the two calibration points are solved, and the results are calculated. and The value;

[0059] in, The overall credibility score; Furthermore, it is worth noting that in this embodiment, the measured data is substituted into the aforementioned data reliability assessment model to output an overall credibility score. When the recognition instrument measures the sample and outputs the result, it combines the similarity score to output the similarity score. and overall reliability score The overall conclusion is that, understandably, the similarity score... Used to represent the degree of matching of spectral shapes, measuring the properties of the material itself, while the overall reliability score Used to indicate the reliability of the measurement environment and measure the quality of the measurement process, the alarm decision made based on the comprehensive conclusion can be set as follows: Only when similarity score The similarity score is greater than the similarity threshold, and the reliability score is also higher. An alarm is triggered only when the similarity score exceeds a reliability threshold. The similarity score is greater than the similarity threshold, but the reliability score is higher. When the similarity score is less than or equal to the reliability threshold, a decision to recommend review is output, but an alarm is not directly triggered. Based on the above, in this embodiment, the data reliability assessment model constructed above can directly quantify the impact of environmental parameters on measurement quality. Therefore, during the detection process of the recognition instrument, the output detection results can be directly intervened, and an auxiliary verification of environmental impact can be added to the detection results. It is foreseeable that when the similarity score... Extremely high, but the reliability score is... A very low similarity score indicates that environmental parameters have a significant impact on measurement quality; therefore, even at this level, the similarity score will be low. The overall decision is highly positive and includes recommendations for retesting.

[0060] In one embodiment, the key parameters in S24 include the coefficients of the first-order term of the polynomial fitting. and quadratic coefficient ; Also includes S4: Predictive Maintenance: S41: The processing unit retrieves historical continuous data from the calibration history database. Next, among them The key parameters recorded are calibrated to form a parameter time series; S42: Perform trend analysis on the time series of the parameters. If it is determined that the drift rate of any key parameter exceeds a preset threshold, generate an early warning message and prompt maintenance.

[0061] After prolonged use of the identification instrument, the key parameters in the wavelength-pixel coordinate correction relationship constructed based on the built-in standard will inevitably change. Therefore, it is necessary to correct the spectrum of the sample under test based on the new wavelength-pixel coordinate correction relationship generated by the timed calibration. It should also be noted that the trend of the key parameter changes can reflect the degree of data drift of the identification instrument to a certain extent. When the degree of data drift of the identification instrument is too large, calibration alone will not be able to correct the spectrum of the sample under test. In this embodiment, after each calibration, in addition to updating the calibration relationship, the key parameters obtained from fitting the calibration relationship are also saved in the calibration history database to form a record. The key parameters include the coefficients of the first-order term. and quadratic coefficient Each of the records mentioned above includes a coefficient for a linear term. and quadratic coefficient and timestamp; Based on the stored records, when the number of records accumulates to a certain level, in this embodiment, the processing unit processes the records in a historical, continuous manner. Next, in this embodiment, a setting is provided. The key parameters recorded are calibrated to form a parameter time series, and the coefficients of the linear term are analyzed based on a trend analysis algorithm. and quadratic coefficient Linear fitting is performed on the time series data to output a straight line that represents the trend of parameter changes, and the slope of the corresponding line is obtained; where the horizontal axis of the time series data is time / calibration number, and the vertical axis is the parameter value; The average drift rate of the parameters can be represented by the straight line and its slope obtained from the linear fitting, expressed as the coefficient of the first-order term. For example, if The slope is -0.0005 This means that the coefficient is calculated once a week. The average decrease was 0.0005; understandably, the slope of the key parameters was too large, and the cumulative drift exceeded the initial value. If the wavelength accuracy of the identifier cannot be guaranteed, then the accuracy of the identifier's wavelength cannot be guaranteed; correspondingly, if the drift rate is within the initial value... If the readings are within the range, the identification device is considered to be in good condition; otherwise, it indicates that the device's condition has deteriorated and maintenance needs to be arranged as soon as possible. In the prior art, the maintenance of the identification device usually occurs when the performance has obviously deteriorated or the explosive has not been identified. However, in this embodiment, based on the recording of key parameters and slope calculation, the status of the identification device can be predictively identified and monitored. This allows for early warnings and maintenance reminders to be issued before the device's performance becomes significantly inaccurate, thus preventing the identification device from malfunctioning at critical moments and causing missed detections.

[0062] In one embodiment, S33 further includes: S336: Preprocess the second Raman spectrum, including algorithms based on wavelet transform or principal component analysis to reduce noise and enhance the signal-to-noise ratio of characteristic peaks.

[0063] In one embodiment, in S34, the matching calculation uses a cosine similarity algorithm or the least squares method.

[0064] In this embodiment, the similarity calculation uses a cosine similarity algorithm. Based on the interpolation intensity obtained by the above interpolation method, the similarity calculation is performed between the sample spectral data and the spectral data stored in the cosine similarity database of the explosive. The calibrated spectral data obtained from the above calculations are shown in the table below:

[0065] According to the cosine similarity formula:

[0066] in, For the sample at wavelength Strength at the location; For the standard spectrum of TNT at wavelength Strength at the location; vector The model, For vectors The model; For vectors with vector The angle between them; It is understandable that when two spectral shapes are exactly the same, it means that the directions are exactly the same and the included angle is the same. 0°, cosine value The greater the difference in shape, the larger the included angle, and the closer the cosine value is to 0; The data in the table above can be viewed as two 6-dimensional vectors:

[0067]

[0068]

[0069]

[0070]

[0071]

[0072] As can be seen from the above, based on the calculated similarity... A value of 0.9999, close to 1, indicates that after wavelength calibration, the spectrum of the sample under test has an almost identical shape to the standard TNT spectrum. If the result is positive, it means that the sample to be tested contains at least TNT.

[0073] In one embodiment, the calibration phase in S2 is automatically triggered under any one or more of the following conditions: when the identifier is powered on, when a preset timing period is reached, and when the ambient temperature changes beyond a predetermined range.

[0074] A Raman spectrometer for identifying explosives includes a laser source, a spectrometer, a detector, a processing unit, a storage unit, and a built-in standard module. The built-in standard module is located in the optical path of the laser source and can be periodically irradiated for instrument calibration. The built-in standard module is a silicon wafer integrated on an optical path switching device. Through the optical path switching device, the laser can selectively irradiate the silicon wafer or an external sample to be tested.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for calibrating an environmental parameter-adaptive Raman spectroscopy explosives identifier, characterized in that: Includes the following steps: S1: Construct a Raman spectroscopy detection system, including a laser source, spectrometer, detector, built-in standard module, storage unit storing a standard Raman spectrum database and calibration history database, processing unit, and environmental sensor module; S2: Perform calibration, including: S21: Control the laser source to irradiate the built-in standard module, wherein the built-in standard module is loaded with calibration material that has known and stable Raman peak positions; S22: Acquire the first Raman spectrum generated by the calibration material through the detector and spectrometer; S23: Identify the measured pixel coordinates of the calibration material in the first Raman spectrum; S24: Compare the measured pixel coordinates with the standard Raman shift in the standard Raman spectroscopy database, calculate the wavelength-pixel coordinate correction relationship under the current instrument operating conditions, and save the key parameters of this calibration to the calibration history database; S3: Perform testing on the sample to be tested, including: S31: Control the laser source to irradiate the sample to be tested; S32: Acquire the second Raman spectrum generated by the sample to be tested; S33: Using the wavelength-pixel coordinate correction relationship established in S24, the pixel index of the second Raman spectrum is corrected to generate wavelength spectral data and obtain the corrected second Raman spectrum. S34: Perform a similarity calculation between the corrected second Raman spectrum and the reference spectrum in the standard Raman spectrum database to obtain the similarity score; S35: Based on similarity and real-time environmental parameters from the environmental sensor module, calculate the reliability score of this detection result based on the data reliability assessment model; S36: Output the detection results and the corresponding reliability score.

2. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 1, characterized in that: In step S2, the calibration material is a silicon wafer with a standard Raman shift of 520.

7. .

3. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 2, characterized in that: In step S24, the method for establishing the wavelength-pixel coordinate correction relationship includes: S241: Data acquisition, based on S23, identifies multiple measured pixel coordinates of the calibration material. ,in For the first The pixel positions in the spectrum output by the instrument under the current operating conditions where a characteristic peak appears are determined based on a peak lookup algorithm. S242: Curve fitting, using the measured pixel coordinates With the standard characteristic wavelength Composition of data pairs By performing polynomial fitting, the corrected polynomial function is obtained: ; in, For the first The standard characteristic wavelengths of each characteristic peak are stored in a standard Raman spectroscopy database; Let be the fitting function, representing the pixel index. The corresponding wavelength; The input variable is the pixel index number in the original spectrum; This is a constant term, representing the offset of the wavelength axis intercept; The coefficient of the first term represents the linear scaling factor of the wavelength axis, corresponding to the dispersion of the spectrometer; These are the coefficients of higher-order terms, used to correct nonlinear responses; S243: Calculation based on least squares fitting, solution , , .

4. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 3, characterized in that: In S33, the method for correcting the second Raman spectrum is a resampling interpolation method, specifically including: S331: Define new wavelength axes with equal spacing ; S332: The inverse function based on the aforementioned corrected polynomial function Calculate the new wavelength axis wavelength point on The corresponding original pixel coordinates ; S333: Using a linear interpolation algorithm, based on the original pixel coordinates and its corresponding spectral intensity Calculate the wavelength point spectral intensity ; S334: For the new wavelength axis Each wavelength point in Repeat steps S332-S333 to obtain a set of intensity values. ; S335: Will and The data pair is defined as the corrected second Raman spectrum.

5. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 4, characterized in that: The environmental sensor module includes a temperature sensor and a humidity sensor; the real-time environmental parameters in S35 include ambient temperature and ambient humidity; the data reliability assessment model is used to output a reliability score based on the ambient temperature, ambient humidity, and the signal-to-noise ratio of the second Raman spectrum.

6. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 5, characterized in that: In step S36, the method for outputting the detection result and the corresponding reliability score includes: Data acquisition and input, including ambient temperature, ambient humidity, and signal-to-noise ratio; Data preprocessing and feature vector components: The processing unit calculates the reliability score based on ambient temperature, ambient humidity, and signal-to-noise ratio. The output includes the detection results and reliability score.

7. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 6, characterized in that: The key parameters in S24 include the coefficients of the first-order terms in the polynomial fitting. and quadratic coefficient ; Also includes S4: Predictive Maintenance: S41: The processing unit retrieves historical continuous data from the calibration history database. Next, the key parameters recorded are calibrated to form a parameter time series; S42: Perform trend analysis on the time series of the parameters. If it is determined that the drift rate of any key parameter exceeds a preset threshold, generate an early warning message and prompt maintenance.

8. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 7, characterized in that: S33 further includes: S336: Preprocess the second Raman spectrum, including algorithms based on wavelet transform or principal component analysis to reduce noise and enhance the signal-to-noise ratio of characteristic peaks.

9. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 7, characterized in that: In step S34, the similarity calculation employs a cosine similarity algorithm or the least squares method.

10. The method for calibrating an environmental parameter adaptive Raman spectroscopy explosives identifier according to claim 7, characterized in that: In S2, the calibration phase is automatically triggered under any one or more of the following conditions: when the identifier is powered on, when the preset timing period is reached, and when the ambient temperature changes beyond a predetermined range.

Images