Spectral measurement method of two-dimensional spectrometer

Abstract

The invention provides a spectral measurement method of a two-dimensional spectrometer, which comprises the following steps of: forming a two-dimensional spectrogram by adopting a cross dispersion mode, and calculating the spectrogram to obtain spectral information of a measured light source; the method comprises the following steps: S1, calculating an ideal spectrogram model: obtaining a corresponding relation between a dispersion wavelength and a spatial position of incident light under an ideal condition by utilizing a dispersion equation of a cross dispersion element and structural parameters of an optical system; s2, obtaining a corrected spectrogram model: measuring a light source with known characteristic spectral lines to obtain a two-dimensional spectrogram, and performing parameter inversion on the structural parameters of the optical system involved in the step S1 to obtain the corrected spectrogram model; s3, establishing wavelength-pixel mapping: performing coordinate value fitting on wavelength data in the corrected spectrogram model to complete mapping from the corrected spectrogram model to actual pixel coordinates; and S4, measuring the spectral information of the measured light source: obtaining the spectral information of the measured light source by using the wavelength-pixel mapping relationship established in S3, thereby effectively improving the accuracy, stability and efficiency of the spectral measurement of the two-dimensional spectrometer.

Description

Technical Field

[0001] The invention relates to the field of optical radiation measurement, and in particular to a spectrum measurement method of a two-dimensional spectrometer. Background Art

[0002] Spectrometers are optical instruments that use the properties of light such as refraction, scattering, and interference when interacting with matter to decompose complex light into spectral components with different wavelengths according to specific rules and detect them. A two-dimensional spectrometer decomposes the measured light by wavelength in two-dimensional space through the cross-dispersion of two dispersive elements, and uses an array detector to receive the spatially dispersed light to obtain a two-dimensional spectrum. By analyzing and calculating the two-dimensional spectrum, the spectral information of the measured light in a wide band can be obtained, including the wavelength of the spectral line and the change of spectral power with wavelength. However, there are challenges in the analysis and calculation of the spectral information of the measured light from its two-dimensional spectrum. In the prior art, whether it is direct calculation using optical parameters through dispersion models or ray tracing through optical design software, there are major defects. This is mainly due to the large number of optomechanical components in a two-dimensional spectrometer and the complex optical parameters involved. Not only is the analysis model computationally intensive, but more importantly: (1) the actual values ​​of the spatial positions and optical parameters of each component deviate from the theoretical design, and this deviation will greatly affect the spectrum analysis results; (2) cross-dispersion components are greatly affected by the surrounding environment, especially temperature changes, mechanical vibrations, etc., which will change their characteristic optical parameters and spatial position relationships, resulting in large deviations, further exacerbating the difficulty of analysis and calculation; in addition, for continuous spectrum measurement, theoretically, the size of the spot formed on the image plane will have a dispersion effect, and crosstalk between spots of different orders will also affect the accuracy of the final spectrum analysis. Summary of the invention

[0003] In view of the deficiencies of the prior art, the present invention provides a two-dimensional spectrometer spectrum measurement method, which effectively solves the problems in the prior art solutions, such as the complex optical parameters of the two-dimensional spectrometer, the deviation between the actual value and the theoretical value of the position and parameters of each component, resulting in poor universality and affected accuracy. A two-dimensional spectrometer is used to form a two-dimensional spectrum on the array detector by cross-dispersion. The corresponding relationship between the spectrum pixel coordinates and the wavelength is accurately obtained through the establishment, correction and function mapping of the spectrum model, and then the spectrum information of the measured light source is accurately obtained, thereby improving the accuracy, stability and efficiency of the spectrum measurement of the two-dimensional spectrometer, and realizing accurate measurement of a wider spectral range and higher spectral resolution. This solution effectively improves the universality and restoration accuracy of the algorithm, and reduces the influence of environmental factors and instrument component deviations on the measurement results.

[0004] To achieve the above technical objectives, the present invention provides a two-dimensional spectrometer measuring device, comprising a slit, a collimation unit, a cross-dispersion element, a focusing unit and a planar array detector arranged in the order of incidence of the light source to be measured; the slit is arranged on or near the object plane focus of the collimation unit, the incident light is secondary dispersed by the cross-dispersion element, and then focused by the focusing unit to be imaged onto the receiving surface of the planar array detector, the main sections of the cross-dispersion element are perpendicular to each other, the light beam emitted by the light source to be measured is dispersed in two directions perpendicular to each other, and the obtained two-dimensional spectrum presents multiple levels of spectral wavelengths.

[0005] Take the cross-dispersion element as a grating and a prism as an example, where the grating is the main dispersion element and the prism is arranged behind the grating as an auxiliary dispersion element. The grating satisfies the grating equation, see formula (1):

[0006] mλ=d(sini+sinθ)cosγ……………………(1)

[0007] Where m is the diffraction order, λ is the wavelength, d is the grating constant, i is the incident angle, θ is the diffraction angle, and γ is the deflection angle.

[0008] For a fixed wavelength incident light, the diffraction angle is affected by the incident angle i and the deflection angle γ. 0 ) is incident on the main dispersion element grating, the grating satisfies the quasi-Littrow condition. In a specific order spectrum, only a certain wavelength of light has a diffraction angle equal to the blaze angle, that is, i = θ = θ 0 , for other wavelengths, i = θ 0 , θ=θ 0 +Δθ, Δθ is the change of diffraction angle with wavelength. In different orders, the wavelengths where the incident angle is equal to the diffraction angle and the blaze angle are called central wavelengths. The wavelength range of each order can be determined based on the central wavelength and free spectral range.

[0009] The free spectrum region is defined as the region of the blaze wavelength corresponding to each order in the grating light splitting direction, which has the characteristic that the wavelengths of the adjacent order free spectrum regions are connected end to end, that is, the interval where the spectra of two adjacent orders do not overlap. When the adjacent order spectra do not overlap, the m-order wavelength (λ+Δλ) and the (m+1)-order wavelength λ have the same diffraction angle, see formula (2).

[0010]

[0011] Therefore, the free spectral range can be expressed as: Δλ = λ / m. The free spectral range of a grating working at high diffraction orders (tens to hundreds of orders) is very narrow, and multiple diffraction orders need to be used simultaneously during operation to cover the entire spectral range.

[0012] After the light passes through the main dispersion element grating, the use of multiple orders at the same time will cause serious order overlap problems. The diffraction orders overlap and spectrum identification and analysis cannot be performed. The auxiliary dispersion element is used to perform cross dispersion in the vertical direction of the main dispersion. Its main function is to expand the overlapping spectrum after passing through the grating in the horizontal direction. The auxiliary dispersion element can be a grating or a prism. If the auxiliary dispersion element is a grating, it also satisfies the grating equation. When the auxiliary dispersion element is a prism, the prism dispersion follows the law of refraction, see formula (3).

[0013] n 0 ·sini p =n·sinβ……………………(3)

[0014] Where n 0 is the refractive index of air, n is the refractive index of the prism material, i p is the incident angle, and β is the refraction angle. When different light rays have different deflection angles after passing through the main dispersion element, the outgoing light rays will bend along the axial direction.

[0015] The free spectrum after cross dispersion is usually fan-shaped, and the wavelengths of the adjacent free spectrums are continuously transitioned, or the entire wavelength range is composed of several diffraction orders, and each order corresponds to a certain width of the free spectrum range, such as Figure 2 As shown, the short-wave free spectrum region is narrower, and the longer-wave free spectrum region becomes wider.

[0016] According to the imaging characteristics of a two-dimensional spectrometer, the original image is a two-dimensional array image, and the two-dimensional spectrum needs to be converted into a one-dimensional spectrum that can be used for spectrum analysis by spectrum restoration. During the spectrum restoration process, the actual wavelength value obtained may not be consistent with the theoretical value due to factors such as device processing errors and installation errors. Therefore, the present invention provides an analysis method for a two-dimensional spectrum.

[0017] The present invention provides a spectrum measurement method of a two-dimensional spectrometer. The two-dimensional spectrometer uses a cross-dispersion method to form a two-dimensional spectrum of a measured light source on a planar array detector. The spectrum information of the measured light source is obtained by spectrum calculation, that is, a spectrum distribution curve with the abscissa being the wavelength and the ordinate being the intensity is obtained. The specific steps are as follows:

[0018] S1: Calculate the ideal spectrum model: Use the dispersion equation of the cross-dispersion element and the structural parameters of the optical system to establish the ideal spectrum model, and obtain the correspondence between the dispersion wavelength and spatial position of the incident light under ideal conditions;

[0019] S2: Obtaining a corrected spectrum model: measuring a light source with known characteristic spectrum lines to obtain a two-dimensional spectrum, using the wavelength of the characteristic spectrum lines and the corresponding image spot position coordinates, performing parameter inversion on the optical system structural parameters involved in step S1, obtaining corrected optical system structural parameters, and then obtaining a corrected spectrum model;

[0020] S3: Establish wavelength-pixel mapping: Using the two-dimensional spectrum of the known characteristic spectral line, according to the wavelength of its characteristic spectral line and the corresponding pixel coordinate position, the wavelength data in the modified spectrum model is fitted with coordinate values ​​to complete the mapping of the modified spectrum model to the actual pixel coordinates, that is, the corresponding relationship between any pixel coordinate on the array detector and the actual wavelength is obtained;

[0021] S4: measuring the spectrum information of the measured light source: using the two-dimensional spectrometer to measure the measured light source to obtain a two-dimensional spectrum; using the wavelength-pixel mapping relationship established in S3, obtaining the spectrum information of the measured light source.

[0022] The intensity of spatial scattered light received by the area array detector is directly related to the power of the measured light source at the corresponding wavelength. Therefore, in addition to obtaining the wavelength of the spectral component of the measured light source, the spectral information can further include the spectral power distribution at each wavelength.

[0023] In the above step S1, the two cross-dispersion elements are grating and prism as an example:

[0024] (1) In the grating dispersion (vertical Y) direction, the grating satisfies the quasi-Littrow condition, see formula (4):

[0025]

[0026] In the formula, m i Corresponding to different orders, d is the grating constant, θ 0 is the incident angle and diffraction angle that satisfy the quasi-Littrow condition, γ is the deflection angle, (θ 0 +Δθ) satisfies the diffraction angle. For different wavelengths at the same diffraction order, the diffraction angle changes accordingly.

[0027] The diffraction angle θ of the central wavelength 0 As the reference angle, if the system focal length f is known, the relative position relationship in the vertical Y direction can be constructed, see formula (5):

[0028]

[0029] In the formula, Δy represents the diffraction direction of the grating, λ i The distance between the wavelength and the central wavelength in the y direction of the image plane.

[0030] Based on the above grating equation and optical system structural parameters, the wavelength matrix M related to the diffraction distance y in the vertical direction and the order m can be obtained: λ , see formula (6):

[0031]

[0032] (2) Prism dispersion (horizontal X) direction. Taking the reflective prism material as an example, the reflective prism dispersion angle follows formula (7):

[0033]

[0034] In the formula, A p is the prism vertex angle, i p is the incident angle of the prism, β 0 is the prism dispersion angle corresponding to the initial wavelength, n(λ i ) is the refractive index curve relationship of the prism with respect to wavelength.

[0035] The prism dispersion angle β corresponding to a fixed wavelength 0 As the reference angle, if the system focal length f is known, the relative position relationship in the horizontal X direction can be constructed, see formula (8):

[0036]

[0037] In the formula, Δx represents λ i The distance between the wavelength and the fixed wavelength in the x direction of the image plane.

[0038] Based on the dispersion formula and the wavelength λ determined by y and m in (1), the horizontal dispersion distance matrix M related to the vertical diffraction distance y and the order m can be constructed: X , see formula (9):

[0039]

[0040] Construct the diffraction distance matrix M corresponding to the diffraction distance y in the vertical direction and the order m Y , see formula (10):

[0041]

[0042] (3) According to the above three matrix formulas (6), (9), and (10), the ideal spectrum model of xy-λ can be constructed to obtain the matrix M, as shown in formula (11):

[0043]

[0044] (4) Complete the construction of an ideal spectrum model that depends on the dispersion equation of the above-mentioned cross-dispersion element and the structural parameters of the optical system. When the cross-dispersion element is changed or the structural parameters of the optical system are changed, a new ideal spectrum model is automatically generated.

[0045] In step S1, a wavelength distribution matrix M corresponding to the image plane of the planar array detector is constructed, that is, an ideal spectrum model is constructed, and the range of the ideal spectrum model is the free spectrum region.

[0046] It should be noted that factors such as ambient temperature, humidity changes, installation deviations, and mechanical vibrations will cause the theoretical optical system structural parameters in the two-dimensional spectrometer to be inconsistent with the actual state. The optical system structural parameters mainly include grating incident angle, grating deflection angle, prism incident angle, system focal length, etc. Taking the grating incident angle as an example, the parameter deviation of the grating incident angle will affect the wavelength extraction accuracy of the two-dimensional spectrometer. Specifically, this deviation will cause changes in the diffraction order of a certain fixed wavelength, making it difficult to determine the position of the wavelength points on both sides of their own order in the two-dimensional spectrum. Up and down misalignment will cause the relative position to change completely, such as Figure 3 As shown, the final wavelength extraction accuracy is affected. Therefore, it is necessary to perform parameter inversion on the optical system structural parameters involved in step S1 to obtain the corrected optical system structural parameters, and then obtain the corrected spectrum model. The specific process is shown in step S2.

[0047] As a technical solution, in step S2, two or more light sources with different characteristic spectral lines are used to obtain their two-dimensional spectra respectively, and the characteristic spectral lines are extracted from each light source and summarized for parameter inversion; in step S3, these characteristic spectral lines are used to perform wavelength-pixel coordinate mapping. Using two or more light sources with different characteristic spectral lines can, on the one hand, broaden the spectral coverage range, improve the accuracy of spectral measurement in a wide band, and make the measurement method more flexible and universal; on the other hand, the characteristic spectral lines of multiple light sources complement and verify each other, which can reduce the impact of spectral line errors or uncertainties that may exist in a single light source on the measurement results. For example, the spectral lines of some light sources in a specific wavelength region may be sharper and easier to accurately identify, while other light sources perform better in other wavelength regions. By comprehensively utilizing the characteristic spectral lines of multiple light sources, the accuracy of the overall measurement can be improved and the measurement error can be reduced.

[0048] As a technical solution, after step S3, it also includes using another one or more light sources with different characteristic spectral lines, iteratively repeating steps S2 and S3, so as to correct the wavelength-pixel mapping relationship. This technical solution can refine the calibration of the established wavelength-pixel mapping relationship and improve the reliability and stability of the measurement by iteratively repeating steps S2 and S3 using another one or more light sources with different characteristic spectral lines. For example, after the mapping is established for the first time using a mercury lamp, a xenon lamp and a krypton lamp are introduced for iteration. The unique characteristic spectral lines of the xenon lamp and the krypton lamp can reveal subtle deviations that have not been calibrated before, making the wavelength-pixel mapping relationship more accurate. Each iteration introduces a new light source with different characteristic spectral lines to cover a wider spectral region. For example, the light source used initially mainly covers the visible light region, and the light source introduced in subsequent iterations covers the near-infrared or ultraviolet region, thereby meeting the needs for measurement of different spectral ranges, expanding the scope of application of spectral measurement, and meeting diverse measurement needs.

[0049] As a technical solution, for a measured light source having discrete spectral lines, step S4 specifically includes: determining the pixel coordinates of the discrete spectral lines on the two-dimensional spectrum by a centroid extraction algorithm, and obtaining the wavelength value of the corresponding discrete spectral line by using the wavelength-pixel mapping relationship established in S3. The spectrum of the measured light source with specific discrete spectral lines in this technical solution can be composed of only discrete spectral lines, or it can be a continuous spectrum superimposed by discrete spectral lines (such as a fluorescent lamp). In view of this characteristic, the centroid extraction algorithm can effectively process discrete signal distributions, and compared with other processing methods suitable for continuous spectra, it can more accurately analyze the spectrum of discrete spectral line light sources, and has good measurement repeatability and reliability, and high measurement accuracy.

[0050] Furthermore, the above-mentioned centroid extraction algorithm includes the following steps: ① subtracting background noise; ② reading pixel values ​​one by one in the image and calculating the maximum pixel value PixelMaxVal; ③ calculating the pixel threshold PixelThreshVal according to the threshold coefficient ThreshCoe set by sensitivity; ④ binarizing the image and creating a new image with the same size as the original image, and the pixels are divided into 0 / 255 according to the threshold; ⑤ image processing, selecting a rectangular template of 5×5 pixels to corrode the binary image, filter out noise interference points, and the remaining data is considered to be a valid light spot; ⑥ calculating the centroid of each light spot to obtain a series of coordinates of the center of mass of the light spot.

[0051] As a technical solution, the cross-dispersion element includes a main dispersion element and an auxiliary dispersion element; the dispersion directions of the two dispersion elements are perpendicular to each other; the planar array detector receives and detects the two-dimensional spectrum formed by the dispersion; the dispersion element includes a grating or a prism or a combination of the two. In a specific embodiment, the main dispersion element is a grating, which has the characteristics of high dispersion, high resolution and full-wave blaze, works at a high diffraction order, and the dispersion of the main dispersion element occurs in a single plane; the auxiliary dispersion element is a grating or a prism with low dispersion ability, and the dispersion direction of the auxiliary dispersion element is perpendicular to the dispersion direction of the main dispersion element.

[0052] When the prism is used as an auxiliary dispersion element, the order in the two-dimensional spectrum will show a bending phenomenon due to the secondary dispersion of the auxiliary dispersion element. This bending phenomenon is caused by the different deflection angles of different light rays after passing through the dispersion element. The spectral line bending direction is the same and symmetrical about the center. The closer to the edge of the free spectrum area, the greater the bending degree. In order to more accurately fit this spectral line bending, if only one mapping function is used to take into account the spectral line bending fitting effect of the entire spectrum range, it is often difficult to achieve the desired effect. Therefore, in step S3, the pixel coordinates are divided into two or more regions, and different functions are used to perform coordinate mapping fitting or the mapping of each region is corrected by using a deviation function in sections. In order to flexibly adapt to the spectral line bending conditions in different regions, targeted adjustments are made according to the actual bending characteristics by using the deviation function in sections, thereby more accurately describing and fitting the spectral line bending phenomenon in the two-dimensional spectrum, thereby improving the precision and accuracy of spectral measurement.

[0053] As a technical solution, in step S2 and / or step S4, the two-dimensional spectrogram is filtered and background value corrected. When measuring, the measured light source, in addition to the characteristic spectral line signal, also includes interference information such as noise and background. The noise data is the response caused by the influence of various factors such as the surrounding environment and the internal instrument system when the two-dimensional spectrometer is working. It has uncertainty itself and will affect the repeatability and accuracy of the data. The two-dimensional spectrogram is filtered to reduce the influence of noise on the measurement result. The background is the signal intensity output by the two-dimensional spectrometer when the light source is not measured, or the response of the two-dimensional spectrometer produced by the non-measured component, which directly affects the analysis result of the two-dimensional spectrometer and may cover up or interfere with the identification of the characteristic spectral line signal. By correcting the background value of the two-dimensional spectrogram, the characteristic spectral line signal can be more accurately extracted to improve the measurement precision and accuracy.

[0054] As a technical solution, in step S2, the light source of the known characteristic spectral line is an atomic spectral line lamp or a laser light source. Atomic spectral line lamps include but are not limited to mercury lamps (Hg lamps), xenon lamps (Xe lamps), krypton lamps (Kr lamps), Xe-Hg hybrid lamps or other hollow cathode lamps, wherein the two-dimensional spectrum of the Hg lamp is as follows Figure 4 shown.

[0055] As a technical solution, in step S2, the value range of the optical system structural parameters is set, the values ​​in the value range are traversed according to a certain step length, and the corresponding two-dimensional spectrum model is calculated, and the optical system structural parameters closest to the actual two-dimensional spectrum model are taken to further improve the measurement accuracy. The optical system structural parameters include: grating incident angle, grating deflection angle, prism incident angle, system focal length, etc. In the calculation of step S2, one or more of the above optical system structural parameters are inverted.

[0056] As a technical solution, in step S3, the light spots of the characteristic spectral lines in the free spectrum region and one or more heterogeneous sub-spectral regions are selected for polynomial fitting, and the rotation or offset of the spectrum is corrected at the same time. On the one hand, the light sources with known characteristic spectral lines, such as Hg lamp, Xe lamp, Kr lamp, etc., have a small number of spectral lines and cannot completely cover a relatively wide wavelength range, which affects the fitting accuracy; on the other hand, the fitting error near the edge of the free spectrum region during the fitting process is relatively large. For a specific single spectral line, there may be multiple light spots on the image plane, such as Figure 5 As shown, due to the limitations of dispersion efficiency and detector size, usually only adjacent orders of light spots can be seen. Therefore, only the adjacent ±1 orders need to be considered during calculation. The specific process is as follows: ① Calculate the different order points of a specific wavelength in the ideal spectrum model. According to the prism dispersion equation, the same wavelength of different orders in the horizontal direction is dispersed the same distance. According to the grating diffraction equation, the order corresponding to the current wavelength is changed to calculate the offset in the vertical direction; ② According to the above steps, all possible different order points of each wavelength can be calculated. Due to the limitations of dispersion efficiency and detector size, only the adjacent ±1 orders need to be calculated; ③ Find the different order points of a certain wavelength in the spectrum obtained by actual measurement. In this process, color filters or based on light can be used. The wavelength of the light spot is judged based on the simulation results. Usually, the points of different orders are closer to the edge than the points of the actual order. The mapping function is calculated through the existing data, and the possible points of different orders are found through the points in ①; ④ The points of different orders confirmed in ③ are paired, and these points iteratively participate in the calculation of the mapping function from the ideal spectrum model to the corrected spectrum model, which effectively improves the wavelength extraction accuracy. On the other hand, it can be used to correct the image rolling error. This error will cause the light spots of different orders and the same wavelength in the image to shift significantly in the horizontal direction. After correction, the complexity of the mapping function can be reduced, and the wavelength extraction accuracy can be further improved.

[0057] As a technical solution, after step S2, step S3a is further included, which specifically includes: using a standard light source with a continuous spectral distribution and a known spectral power distribution to calibrate the amplitude of each pixel response in the two-dimensional spectrometer.

[0058] Since the grating works in multiple diffraction orders at the same time, there is a free spectral region mentioned above in each used order, including the central wavelength and the edge wavelength. The diffraction efficiency is higher at the central wavelength and lower at the edge wavelength. Therefore, when restoring a one-dimensional spectrum, especially a continuous spectrum, each order presents an envelope shape, that is, the spectrum line fluctuates, such as Figure 6 As shown. At the same time, due to the influence of factors such as optical system attenuation, photoelectric imaging conversion efficiency and electronic system, there is a deviation between the measured light signal intensity and the actual light signal intensity, which leads to distortion when the intensity value of the measured light signal is directly used to draw the spectrum. The basic idea of ​​the current quantitative research on spectral analysis is to determine the content of a certain element in the target to be measured by the relative intensity of the spectral line of the element in the target to be measured, or to use a standard light source (such as a halogen tungsten lamp) and its known reference spectral line. Under the same external conditions, a two-dimensional spectrometer is used to collect the standard light source spectrum. By comparing the relationship between the standard light source spectrum and the actual standard light source reference spectral line, the response coefficient corresponding to the wavelength of the two-dimensional spectrometer is obtained, and then the unknown spectrum is calibrated using the obtained response coefficient to complete the radiation calibration work.

[0059] As a technical solution, deconvolution is used for the two-dimensional spectrum. Due to the influence of aberrations, debugging errors, environment and other factors, the size of the image on the image plane and the pinhole size of the light spot will theoretically diffuse 3-4 times the theoretical area. However, the resolution of the two-dimensional spectrometer is high and the wavelength range is wide. The distance between adjacent levels is nonlinear, and the distance between each level is very small, especially near the long-wave band. False peaks are prone to appear in the process of wavelength extraction, such as Figure 7 As shown in the figure, the peak appears at the position without signal, causing misjudgment of order or misidentification of peak, which affects the accuracy of wavelength extraction. Even if the instrument design is optimized for improvement, it cannot be completely avoided. Through deconvolution processing, energy can be effectively gathered, the spot size can be reduced, and crosstalk between orders can be avoided. Different positions of the spectrum theoretically correspond to different convolution kernels. For simplicity and practicality, the adjacent n orders are regarded as sharing the same convolution kernel. The convolution kernels between adjacent orders can be fitted by linear difference, and finally the deconvolution result is obtained.

[0060] As a technical solution, after step S3, step S3b is also included, which specifically includes: establishing a deep learning neural network model, using the wavelength-pixel mapping obtained in step S3 as a physical prior knowledge constraint network, using a two-dimensional spectrometer to measure the two-dimensional spectra of multiple light sources with known spectral information, and inputting the two-dimensional spectra of these light sources and the known spectral information as samples into the deep learning neural network model.

[0061] Furthermore, the deep learning neural network model includes: a multi-layer perceptron (MLP) or a convolutional neural network (CNN). A multi-layer perceptron is an artificial neural network, which consists of an input layer, a hidden layer and an output layer. Each layer of ganglia is connected by weights, and information is transmitted in one direction. During forward propagation, the input is output after weighting, summing, activation and other operations. In reverse propagation, the weights are adjusted according to the loss to optimize the model. Each pixel value of the two-dimensional spectrogram can be input as the feature value of the input layer, and the target one-dimensional spectrogram data can be output as the prediction result of the output layer. Convolutional neural networks are specially designed to process data with a grid structure, including convolutional layers, pooling layers and fully connected layers. The convolutional layer uses a convolution kernel to extract local features of the data, the pooling layer compresses the data dimension to retain the main features, and the fully connected layer is used for tasks such as classification or regression. The two-dimensional spectrogram can be directly input as the input layer of the network, and the target one-dimensional spectrogram data can be output as the prediction result of the output layer by using a fully connected layer and other methods.

[0062] Furthermore, in step S3b, the light source measured by the two-dimensional spectrometer includes a light source with a continuous spectrum and a light source with discrete spectral lines; a data set is constructed based on the two-dimensional spectra of these light sources and the known spectral information, and the data set is divided into: a training set and a test set, wherein the training set is used to train the network model, and the test set is used to verify the performance of the network model. There is crosstalk in the continuous spectrum in space, and the nonlinear dispersion of the optical path characteristics, gratings and prisms causes different response errors at different positions of the sensor. Therefore, when constructing a data set, the collection of continuous light collected should cover all effective image plane areas of the sensor as much as possible. Only by collecting a rich type of two-dimensional spectra of mixed light of different bandwidths and the corresponding one-dimensional spectral data, the constructed data set has sufficient generalization and practicality.

[0063] Furthermore, in step S3b, the two-dimensional spectrum data of the light source is used as input and the corresponding spectral information is used as a label during the training of the network model; one or more two-dimensional spectra with discrete spectral lines and the corresponding spectral line wavelengths are used as inherent inputs of the network model to participate in the calculation of the loss function, and the physical prior knowledge contained therein is used to assist the network in effective learning during the training process.

[0064] Furthermore, in step S3b, verifying the performance of the network model includes inputting the two-dimensional spectrum in the test set into the trained network model, outputting its spectral information and comparing it with the known spectral information to verify the effectiveness of the network model. For the continuous light wavelength region with unsatisfactory effect, increase the relevant data set data, optimize the network structure, learning rate, and loss function until the test results meet expectations.

[0065] The present invention also provides a measuring device using the above-mentioned two-dimensional spectrometer spectrum measurement method, comprising a slit, a collimation unit, a main dispersion element, an auxiliary dispersion element, a focusing unit and a planar array detector arranged according to the incident order of the light source to be measured; the slit is arranged on or near the object plane focus of the collimation unit, and the incident light is focused by the focusing unit after secondary dispersion of the main dispersion element and the auxiliary dispersion element, and is imaged to the receiving surface of the planar array detector, and the main cross-section of the main dispersion element and the main cross-section of the auxiliary dispersion element are perpendicular to each other; the slit is perpendicular to the main cross-section of the main dispersion element.

[0066] Compared with the prior art, the present invention has the beneficial effect of providing a two-dimensional spectrometer spectrum measurement method, using a two-dimensional spectrometer to adopt a cross-dispersion method to form a two-dimensional spectrum of a measured light source on a planar array detector, accurately obtaining the corresponding relationship between the spectrum pixel coordinates and the wavelength through model establishment, correction and function mapping, accurately obtaining the spectrum information of the measured light source, simplifying the optical system assembly process and steps while ensuring the accuracy, stability and other indicators, integrating a dynamic feedback program in the algorithm, automatically calibrating environmental changes, and using heterogeneous points for polynomial fitting, thereby improving the accuracy, stability and efficiency of the two-dimensional spectrometer spectrum measurement, and achieving accurate measurement of a wider spectral range and higher spectral resolution. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Attached Figure 1 It is a schematic diagram of the optical path of the two-dimensional spectrometer of the present invention;

[0068] Attached Figure 2 It is a schematic diagram of the free spectrum region of the present invention;

[0069] Attached Figure 3 Schematic diagram of the relationship between the positions of the light spots in the order;

[0070] Attached Figure 4 A two-dimensional spectrum of a mercury lamp taken for a two-dimensional spectrometer;

[0071] Attached Figure 5 It is a schematic diagram of the order of the monochromatic light spot and the corresponding points of different orders of the present invention;

[0072] Attached Figure 6 is the one-dimensional spectrum of a continuous spectrum light source;

[0073] Attached Figure 7 Schematic diagram of false peaks when extracting adjacent order wavelengths;

[0074] Attached Figure 8 It is the optical simulation two-dimensional spectrum of the Hg lamp in Example 1;

[0075] Attached Fig. 9 Schematic diagram of the process of the spectrum measurement method of the two-dimensional spectrometer in Example 1;

[0076] Attached Fig.10 Schematic diagram of the process of the spectrum measurement method of the two-dimensional spectrometer in Example 2;

[0077] Attached Fig.11 It is a one-dimensional spectrum after the Kr lamp spectrum in Example 2 is restored;

[0078] Attached Fig.12 Schematic diagram of the process of the spectrum measurement method of the two-dimensional spectrometer in Example 3;

[0079] Attached Fig.13 Schematic diagram of the process of the spectrum measurement method of the two-dimensional spectrometer in the fourth embodiment;

[0080] Attached Fig.14 This is a one-dimensional spectrum diagram obtained by testing the standard A light source and spectrometer in Example 4. DETAILED DESCRIPTION

[0081] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0082] Embodiment 1

[0083] This embodiment designs and builds a two-dimensional spectrometer spectrum measurement device, such as Figure 1 As shown, it includes a slit 7, a collimating unit 2, a main dispersion element 3, an auxiliary dispersion element 4, a focusing unit 5 and a planar array detector 6 arranged in the order of the incident light source 1 to be measured; the slit 7 is arranged on or near the object plane focus of the collimating unit 2, and the incident light is imaged to the receiving surface of the planar array detector 6 after secondary dispersion by the main dispersion element 3 and the auxiliary dispersion element 4, and the main cross section of the main dispersion element 3 and the auxiliary dispersion element 4 are perpendicular to each other, and the light beam emitted by the measured light source is dispersed in two directions perpendicular to each other, and the obtained two-dimensional spectrum shows multiple orders of spectral wavelengths. In this embodiment, the main dispersion element 3 is a grating, the auxiliary dispersion element 4 is a prism, and the measurement wavelength range of the measuring device covers 400nm-1000nm. It can be seen from the spot distribution corresponding to the low-order range, that is, the adjacent spots are clearly spaced and the spots are small, and adjacent orders will not interfere with each other. Optical simulation is performed on a mercury lamp, as shown in Figure 8 As shown, there are many wavelengths, and it is not easy to judge the wavelength corresponding to each light spot when it is presented on the entire image plane. In addition, this embodiment also provides a spectrum measurement method of a two-dimensional spectrometer, which uses a cross-dispersion method to form a two-dimensional spectrum of the measured light source on the array detector. Through the establishment, correction and function mapping of the spectrum model, the corresponding relationship between the spectrum pixel coordinates and the wavelength is accurately obtained, and then the spectrum information of the measured light source is accurately obtained; Fig. 9 As shown, the specific steps are as follows:

[0084] A1: Calculate the ideal spectrum model: Use the cross-dispersion element grating and prism dispersion equations and optical system structural parameters to establish an ideal spectrum model to obtain the correspondence between the dispersion wavelength of the incident light and the spatial position coordinates under ideal conditions.

[0085] A2: Obtain a modified spectrum model: Measure the Hg lamp with known characteristic spectrum lines to obtain a two-dimensional spectrum, filter the two-dimensional spectrum and correct the background value; interpret its characteristic spectrum lines in combination with the optical simulation results and narrow-band color filters, and use the wavelength of the characteristic spectrum lines and the corresponding light spots in the free spectrum area of ​​the image plane and the position coordinates of the different-order light spots outside the free spectrum area to perform parameter inversion on the optical system structural parameters involved in step A1, including the grating incident angle, grating deflection angle, prism incident angle, system focal length, etc., set the value range of the optical system structural parameters, traverse the values ​​in the value range according to a certain step length, calculate the final real optical system structural parameters, and then obtain the modified spectrum model. Taking the grating incident angle as an example, based on the ideal grating incident angle θ 0 , it can be assumed that the actual grating incident angle value should be θ 0 Within the range of ±2°, the step size Δθ' is set to construct a possible grating incident angle set: {θ all |θ 0 -2,θ 0 -2+Δθ',…,θ 0 ,…,θ 0 -Δθ'+2,θ 0 +2}; Knowing the centroid position of the spot in the two-dimensional spectrum image data collected by the Hg lamp and the wavelength information corresponding to the wavelength, construct the vertical position information {y i} and wavelength information {λ i}One-to-one matching data table; traverse the values ​​in the grating incident angle set and calculate the grating incident angle in a very small area to meet the requirements.

[0086] A3: Establish wavelength-pixel mapping: Using the Hg lamp two-dimensional spectrum with known characteristic spectral lines, according to the wavelength of the characteristic spectral lines and the corresponding pixel coordinate position, the wavelength data in the modified spectrum model is fitted with coordinate values ​​to complete the mapping of the modified spectrum model to the actual pixel coordinates, that is, the corresponding relationship between any pixel coordinates on the array detector and the actual wavelength is obtained. Considering that the bending of the spectrum line may cause deviations in the wavelengths of the same level of the edge of the free spectrum region, a single mapping function cannot take into account the fitting effect within the entire image range. Specifically, the pixel coordinates are divided into two regions, and different functions are used for mapping coordinate mapping fitting or the deviation function is used in segments to correct the mapping of each region. i}According to the vertical position relationship, it is divided into two subsets: upper and lower: {y i0},{yi1}, and polynomial fitting is performed respectively. In this process, according to the prism dispersion formula and the grating diffraction equation, the offsets of different wavelengths in the horizontal and vertical directions are calculated respectively, and finally all possible heterogeneous points of each wavelength are obtained, and the heterogeneous points of a certain wavelength in the ideal spectrum are calculated. The light spots of the characteristic spectrum lines in the free spectrum region and one or more heterogeneous spectrum regions are selected for polynomial fitting, which effectively improves the accuracy and corrects the rotation or offset of the spectrum.

[0087] A4: Measure the two-dimensional spectrum information of Kr lamp: Use the two-dimensional spectrometer to measure the Kr lamp to obtain a two-dimensional spectrum, filter the two-dimensional spectrum and correct the background value, determine the image plane coordinates of any light spot on the two-dimensional spectrum through the centroid extraction algorithm, and use the wavelength-pixel mapping relationship established in A3 to restore the two-dimensional spectrum corresponding to the Kr lamp to obtain the spectrum information of the Kr lamp, that is, obtain a spectral distribution curve with the horizontal axis as wavelength and the vertical axis as luminous intensity.

[0088] Using the spectrum measurement method of the two-dimensional spectrometer provided in this embodiment, the spectrum information of the Kr lamp measured is substantially consistent with the spectrum information of the standard Kr lamp.

[0089] Embodiment 2

[0090] This embodiment provides another two-dimensional spectrometer spectrum measurement method, which tests a discrete spectrum light source, uses a two-dimensional spectrometer to form a two-dimensional spectrum on a planar array detector by cross-dispersion, and obtains spectrum information of the light source by spectrum calculation; Fig.10 As shown, the specific steps are as follows:

[0091] B1: Calculate the ideal spectrum model: Use the cross-dispersion element grating and prism dispersion equations and optical system structural parameters to establish an ideal spectrum model to obtain the correspondence between the dispersion wavelength of the incident light and the spatial position coordinates under ideal conditions.

[0092] B2: Obtain the corrected spectrum model: measure the Hg lamp with known characteristic spectrum lines to obtain a two-dimensional spectrum, filter the two-dimensional spectrum and correct the background value; use the wavelength of the characteristic spectrum line and the corresponding spot in the free spectrum area of ​​the image plane and the position coordinates of the different-order spot outside the free spectrum area to perform parameter inversion on the optical system structure parameters involved in step B1, set the value range of the optical system structure parameters, traverse the values ​​in the value range according to a certain step length, calculate the final real optical system structure parameters, and then obtain the corrected spectrum model. Taking the grating incident angle as an example, based on the ideal grating incident angle θ 0 , it can be assumed that the actual grating incident angle value should be θ 0 Within the range of ±2°, the step size Δθ' is set to construct a possible grating incident angle set: {θ all |θ 0-2,θ 0 -2+Δθ',…,θ 0 ,…,θ 0 -Δθ'+2,θ 0 +2}; Knowing the centroid position of the spot in the two-dimensional spectrum image data collected by the Hg lamp and the wavelength information corresponding to the wavelength, construct the vertical position information {y i} and wavelength information {λ i}One-to-one matching data table; traverse the values ​​in the grating incident angle set and calculate the grating incident angle in a very small area to meet the requirements.

[0093] B3: Establish wavelength-pixel mapping: Using the two-dimensional spectrum of known characteristic spectral lines, according to the wavelength of its characteristic spectral lines and the corresponding pixel coordinate position, the wavelength data in the modified spectrum model is fitted with coordinate values ​​to complete the mapping of the modified spectrum model to the actual pixel coordinates, that is, the corresponding relationship between any pixel coordinates on the array detector and the actual wavelength is obtained. Considering that the bending of the spectrum line may cause deviations in the wavelengths of the same level of the edge of the free spectrum region, a single mapping function cannot take into account the fitting effect within the entire image range. Specifically, the pixel coordinates are divided into three regions, and different functions are used for mapping coordinate mapping fitting or the deviation function is used in sections to correct the mapping of each region. i}According to the vertical position relationship, it is divided into three subsets: upper, middle and lower: {y i0},{y i1},{y i2}, and polynomial fitting is performed respectively. In this process, according to the prism dispersion formula and the grating diffraction equation, the offsets of different wavelengths in the horizontal and vertical directions are calculated respectively, and finally all possible heterogeneous points of each wavelength are obtained, and the heterogeneous points of a certain wavelength in the ideal spectrum are calculated. The light spots of the characteristic spectrum lines in the free spectrum region and one or more heterogeneous spectrum regions are selected for polynomial fitting, which effectively improves the accuracy and corrects the rotation or offset of the spectrum.

[0094] B4: Use another or more light sources with different characteristic spectral lines, such as a Xe lamp, and iteratively repeat steps B2 and B3 to correct the wavelength-pixel mapping relationship.

[0095] B5: Measure the two-dimensional spectrum information of Kr lamp: Use the two-dimensional spectrometer to measure Kr lamp to obtain a two-dimensional spectrum, filter the two-dimensional spectrum and correct the background value, determine the image plane coordinates of any light spot on the two-dimensional spectrum by the centroid extraction algorithm, and use the wavelength-pixel mapping relationship established in B3 to obtain the spectrum information of Kr lamp, that is, obtain the spectrum distribution curve with the horizontal axis as wavelength and the vertical axis as luminous intensity, as shown in Fig.11 shown.

[0096] The obtained spectral distribution curve is subjected to spectral analysis. Generally, the half-width of the spectrum line is used to represent the spectral resolution at the spectrum line. The half-width is calculated for the wavelength of the Kr lamp. As shown in Table 1, the resolution in the entire wavelength range is less than 0.1nm. Another important indicator is the wavelength accuracy, that is, the deviation between the indication of the two-dimensional spectrometer and the true value. In this embodiment, the Hg lamp is selected to calibrate the pixel response in the two-dimensional spectrometer, and the Kr lamp is used as the test light source. The wavelength accuracy obtained is shown in Table 1, which is better than 0.02nm.

[0097] Table 1 Kr lamp wavelength resolution and wavelength extraction accuracy results

[0098]

[0099] Embodiment 3

[0100] This embodiment provides another two-dimensional spectrometer spectrum measurement method, which tests a continuous light source, uses a two-dimensional spectrometer to form a two-dimensional spectrum on a planar array detector by cross-dispersion, and obtains spectrum information of the light source by spectrum calculation; Fig.12 As shown, the specific steps are as follows:

[0101] C1: Calculate the ideal spectrum model: Use the cross-dispersion element grating and prism dispersion equations and optical system structural parameters to establish an ideal spectrum model to obtain the correspondence between the dispersion wavelength of the incident light and the spatial position coordinates under ideal conditions.

[0102] C2: Obtaining a corrected spectrum model: Measure Hg lamp, Xe lamp, and Kr lamp with known characteristic spectral lines to obtain two-dimensional spectra respectively, interpret their characteristic spectral lines in combination with optical simulation results and narrow-band filters, extract the wavelengths of the characteristic spectral lines respectively, and use the light spot in the free spectrum region of the corresponding image plane and the position coordinates of the different-order light spots outside the free spectrum region to perform parameter inversion on the optical system structural parameters involved in step C1, calculate the final real optical system structural parameters (including grating incidence angle, grating deflection angle, prism incidence angle, system focal length, etc.), and then obtain the corrected spectrum model.

[0103] C3: Establish wavelength-pixel mapping: Utilize the two-dimensional spectrum of known characteristic spectral lines, and fit the coordinate values ​​of the wavelength data in the modified spectrum model according to the wavelength of each characteristic spectral line and the corresponding pixel coordinate position, and complete the mapping of the modified spectrum model to the actual pixel coordinates, that is, obtain the correspondence between the coordinates of any pixel on the array detector and the actual wavelength; select the light spots of the characteristic spectral lines in the free spectrum region and one or more heterogeneous sub-spectral regions for polynomial fitting, and correct the rotation or offset of the spectrum at the same time.

[0104] C4: Deconvolution is performed on the obtained two-dimensional spectrum to avoid crosstalk between orders, and the amplitude of each pixel response in the two-dimensional spectrometer is calibrated using a standard light source (halogen tungsten lamp, deuterium lamp or a combination of the two) with continuous spectral distribution and known spectral power distribution.

[0105] C5: Measure the spectral information of the measured light source: Use the two-dimensional spectrometer to measure the measured light source to obtain a two-dimensional spectrum; use the wavelength-pixel mapping relationship established in C3 and the calibration coefficient obtained in C4 to obtain the absolute spectral information of the measured light source.

[0106] Embodiment 4

[0107] This embodiment provides another two-dimensional spectrometer spectrum measurement method, which tests a continuous light source, uses a two-dimensional spectrometer to form a two-dimensional spectrum on a planar array detector by cross-dispersion, and obtains spectrum information of the light source by spectrum calculation; Fig.13 As shown, the specific steps are as follows:

[0108] D1: Calculate the ideal spectrum model: Use the cross-dispersion element grating and prism dispersion equations and optical system structural parameters to establish an ideal spectrum model to obtain the correspondence between the dispersion wavelength of the incident light and the spatial position coordinates under ideal conditions.

[0109] D2: Obtaining a corrected spectrum model: Measure the Hg lamp, Xe lamp, and Kr lamp with known characteristic spectral lines to obtain two-dimensional spectra respectively, interpret their characteristic spectral lines in combination with optical simulation results and narrow-band filters, extract the wavelengths of the characteristic spectral lines respectively, and use the light spot in the free spectrum region of the corresponding image plane and the position coordinates of the different-order light spots outside the free spectrum region to perform parameter inversion on the optical system structural parameters involved in step D1, calculate the final real optical system structural parameters (including grating incidence angle, grating deflection angle, prism incidence angle, system focal length, etc.), and then obtain the corrected spectrum model.

[0110] D3: Establish wavelength-pixel mapping: Utilize the two-dimensional spectrum of known characteristic spectral lines, and fit the coordinate values ​​of the wavelength data in the modified spectrum model according to the wavelength of each characteristic spectral line and the corresponding pixel coordinate position, and complete the mapping of the modified spectrum model to the actual pixel coordinates, that is, obtain the correspondence between the coordinates of any pixel on the array detector and the actual wavelength; select the light spots of the characteristic spectral lines in the free spectrum region and one or more heterogeneous sub-spectral regions for polynomial fitting, and correct the rotation or offset of the spectrum at the same time.

[0111] D4: Deconvolution is performed on the obtained two-dimensional spectrum to avoid crosstalk between levels, and the amplitude of each pixel response in the two-dimensional spectrometer is calibrated using a standard light source with a continuous spectral distribution and a known spectral power distribution.

[0112] D5: Establish a convolutional neural network model, use the wavelength-pixel mapping obtained in step D3 as the physical prior knowledge to constrain the network, use a two-dimensional spectrometer to measure the two-dimensional spectra of multiple light sources with known spectral information, the light sources include light sources with continuous spectra and light sources with discrete spectral lines, and construct a data set including a training set and a test set based on the two-dimensional spectra of these light sources and the known spectral information, and input this data set as a sample into the convolutional neural network model. The training set is used to train the network model. During the training process, the two-dimensional spectrum data of the light source is used as input, and the corresponding spectral information is used as a label. At the same time, one or more two-dimensional spectra with discrete spectral lines and the corresponding spectral line wavelengths are used as the inherent input of the network model to participate in the calculation of the loss function. In the output layer, a fully connected layer is used to output the target one-dimensional spectrum data as the prediction result of the output layer. The two-dimensional spectrum in the test set is input into the trained training network model, and its spectral information is output and compared with the known spectral information to verify the effectiveness of the network model.

[0113] D6: Verify using standard light source A and measure the two-dimensional spectrum information of standard light source A: Use the two-dimensional spectrometer to measure the standard light source A to obtain a two-dimensional spectrum. Use the above spectrum measurement method to restore the two-dimensional spectrum corresponding to the standard light source A to obtain the spectrum information of the standard light source A, that is, to obtain a spectrum distribution curve with the wavelength as the horizontal axis and the luminous intensity as the vertical axis; Fig.14 It can be seen that the measured spectrum of the standard A light source is basically consistent with the spectrum of the standard A light source.

[0114] The spectrum measurement method of the two-dimensional spectrometer provided in this embodiment can meet the needs of more precise quantification work.

[0115] The specific embodiments of the present invention are described above in conjunction with the accompanying drawings, but it should be understood by those skilled in the art that the above embodiments are only for illustration and not for limiting the scope of the present invention. It should be understood by those skilled in the art that the above embodiments may be modified without departing from the scope and spirit of the present invention. The protection scope of the present invention is defined by the appended claims.

Claims

1. A two-dimensional spectrometer spectrum measurement method, characterized in that: The two-dimensional spectrometer uses cross-dispersion to form a two-dimensional spectrum of the measured light source on the array detector. The spectrum information of the measured light source is obtained through spectrum calculation. The specific steps are as follows: S1: Calculate the ideal spectrum model: Use the dispersion equation of the cross-dispersion element and the structural parameters of the optical system to establish the ideal spectrum model, and obtain the correspondence between the dispersion wavelength and spatial position of the incident light under ideal conditions; S2: Obtaining a corrected spectrum model: measuring a light source with known characteristic spectrum lines to obtain a two-dimensional spectrum, using the wavelength of the characteristic spectrum lines and the corresponding image spot position coordinates, performing parameter inversion on the optical system structural parameters involved in step S1, obtaining corrected optical system structural parameters, and then obtaining a corrected spectrum model; S3: Establish wavelength-pixel mapping: Using the two-dimensional spectrum of the known characteristic spectral line, according to the wavelength of its characteristic spectral line and the corresponding pixel coordinate position, the wavelength data in the modified spectrum model is fitted with coordinate values ​​to complete the mapping of the modified spectrum model to the actual pixel coordinates, that is, the corresponding relationship between any pixel coordinate on the array detector and the actual wavelength is obtained; S4: measuring the spectrum information of the measured light source: using the two-dimensional spectrometer to measure the measured light source to obtain a two-dimensional spectrum; using the wavelength-pixel mapping relationship established in S3, obtaining the spectrum information of the measured light source.

2. A two-dimensional spectrometer spectrum measurement method according to claim 1, characterized in that: In step In S2, two or more light sources with different characteristic spectral lines are used to obtain their two-dimensional spectra respectively, and the characteristic spectral lines are extracted from each light source and summarized for parameter inversion; in step S3, these characteristic spectral lines are used for wavelength-pixel coordinate mapping.

3. The spectrum measurement method of a two-dimensional spectrometer according to claim 1, characterized in that: In step After S3, the method further includes using another or more light sources with different characteristic spectral lines, and iteratively repeating steps S2 and S3, so as to correct the wavelength-pixel mapping relationship.

4. A two-dimensional spectrometer spectrum measurement method according to claim 1, 2 or 3, characterized in that: For a measured light source having discrete spectral lines, step S4 specifically includes: determining the pixel coordinates of the discrete spectral lines on the two-dimensional spectrum using a centroid extraction algorithm, and obtaining the wavelength value of the corresponding discrete spectral line using the wavelength-pixel mapping relationship established in S3.

5. The spectrum measurement method of a two-dimensional spectrometer according to claim 1, characterized in that: The cross dispersion element comprises a main dispersion element and an auxiliary dispersion element; the dispersion directions of the two dispersion elements are perpendicular to each other; the planar array detector receives and detects a two-dimensional spectrum formed by dispersion; the dispersion element comprises a grating or a prism or a combination of the two.

6. A two-dimensional spectrometer spectrum measurement method according to claim 1, 2 or 3, characterized in that: In step S3, the pixel coordinates are divided into two or more regions, and different functions are used to map the coordinates for fitting or the mapping of each region is corrected by using a deviation function in sections.

7. A two-dimensional spectrometer spectrum measurement method according to claim 1, 2 or 3, characterized in that: In step S2 and / or step S4, the two-dimensional spectrum is filtered and background value corrected.

8. A two-dimensional spectrometer spectrum measurement method according to claim 1, 2 or 3, characterized in that: In step S2, the light source of the known characteristic spectral line is an atomic spectral line lamp or a laser light source.

9. A two-dimensional spectrometer spectrum measurement method according to claim 1 or 2, characterized in that: The optical system structural parameters include: grating incident angle, grating deflection angle, prism incident angle, and system focal length. In the calculation of step S2, inversion calculation is performed on one or more of the optical system structural parameters.

10. A two-dimensional spectrometer spectrum measurement method according to claim 1, 2 or 3, characterized in that: In step S3, a polynomial fitting is performed on the light spots of the characteristic spectral lines in the free spectral region and one or more heterogeneous sub-spectral regions, and the rotation or offset of the spectrum is corrected at the same time.

11. A two-dimensional spectrometer spectrum measurement method according to claim 1, 2 or 3, characterized in that: After step S2, step S3a is further included, which specifically includes: using a standard light source with a continuous spectral distribution and a known spectral power distribution to calibrate the amplitude of each pixel response in the two-dimensional spectrometer.

12. A two-dimensional spectrometer spectrum measurement method according to claim 11, characterized in that: Deconvolution is performed on the two-dimensional spectra.

13. A two-dimensional spectrometer spectrum measurement method according to claim 1 or 3, characterized in that: After step S3, step S3b is also included, which specifically includes: establishing a deep learning neural network model, using the wavelength-pixel mapping obtained in step S3 as a physical priori knowledge constraint network, using a two-dimensional spectrometer to measure the two-dimensional spectra of multiple light sources with known spectral information, and inputting the two-dimensional spectra of these light sources and the known spectral information as samples into the deep learning neural network model.

14. A two-dimensional spectrometer spectrum measurement method according to claim 13, characterized in that: The deep learning neural network model includes: a multi-layer perceptron (MLP) or a convolutional neural network (CNN).

15. The spectrum measurement method of a two-dimensional spectrometer according to claim 13, characterized in that: In step S3b, the light sources measured using the two-dimensional spectrometer include light sources with continuous spectra and light sources with discrete spectral lines; a data set is constructed based on the two-dimensional spectra of these light sources and known spectral information, and the data set is divided into: a training set and a test set, wherein the training set is used to train the network model, and the test set is used to verify the performance of the network model.

16. A two-dimensional spectrometer spectrum measurement method according to claim 15, characterized in that: In step S3b, the two-dimensional spectrum data of the light source is used as input and the corresponding spectral information is used as a label during the training of the network model; one or more two-dimensional spectra with discrete spectral lines and the corresponding spectral line wavelengths are used as inherent inputs of the network model to participate in the calculation of the loss function, and the physical prior knowledge contained therein is used to assist the network in effective learning during the training process.

17. The spectrum measurement method of a two-dimensional spectrometer according to claim 15, characterized in that: In step S3b, verifying the performance of the network model includes inputting the two-dimensional spectrum in the test set into the trained network model, outputting its spectral information and comparing it with the known spectral information to verify the validity of the network model.

18. A measuring device using the spectrum measuring method of a two-dimensional spectrometer according to claim 1, characterized in that: The invention comprises a slit, a collimating unit, a main dispersion element, an auxiliary dispersion element, a focusing unit and a planar array detector which are arranged in the order of the incident light source to be measured; the slit is arranged on or near the object plane focus of the collimating unit; the incident light is focused by the focusing unit after secondary dispersion by the main dispersion element and the auxiliary dispersion element and is imaged to the receiving surface of the planar array detector; the main cross section of the main dispersion element and the main cross section of the auxiliary dispersion element are perpendicular to each other.

Images