Non-contact on-line near infrared spectrometer and analysis method

Abstract

The present application relates to the technical field of non-contact near-infrared spectrum analysis, and discloses a non-contact online near-infrared spectrum analyzer and an analysis method, wherein a main light source and a plurality of auxiliary light sources with characteristic absorption peaks are adopted in a light source module, so that enhanced analysis based on direct response of characteristic wavelengths can be realized, inherent limitations of absorption peak overlap and baseline drift in traditional diffuse reflection spectrum are overcome, prediction accuracy is improved, and measurement effect is optimized. A plurality of optical fibers are adopted in a receiving module, sampling ends of the plurality of optical fibers are uniformly distributed in a circumferential direction, the sampling ends of the plurality of optical fibers are distributed in a high-low manner in a direction perpendicular to a sample, and detection ends of the optical fibers are uniformly arranged in a length direction of a slit of a spectrometer, so that light collection efficiency is greatly improved, utilization efficiency of light source energy is maximized, diffuse reflection / scattering light signals returned from the sample are efficiently captured, and interference of environmental stray light is effectively isolated and weakened. In addition, linear and nonlinear methods are mixedly used in the present application, and higher prediction accuracy is achieved.

Description

Technical Field

[0001] This invention relates to the field of non-contact near-infrared spectroscopy analysis technology, and in particular to a non-contact online near-infrared spectroscopy analyzer and analysis method. Background Technology

[0002] Near-infrared spectroscopy has been widely used in biomedicine, pharmaceuticals, agriculture, and food due to its advantages such as speed and non-destructive nature. However, for the growing demand in areas such as in vivo tissue monitoring, online drug quality control, and industrial online analysis, non-contact online near-infrared spectroscopy measurement still has its limitations.

[0003] In non-contact mode, the light energy emitted by the light source dissipates rapidly in free space. The light flux reaching the sample surface is limited, and the resulting effective diffuse reflection / scattering signal is even weaker. Traditional parallel light illumination or simple focusing optical paths are insufficient to efficiently collect these weak scattered lights, resulting in a low signal-to-noise ratio in the final spectrum, which fails to meet the requirements of high-precision quantitative analysis, leading to low light energy utilization and insufficient signal-to-noise ratio.

[0004] Furthermore, in non-contact measurements, there is a certain distance between the optical fiber and the sample surface. In practical applications, this can lead to drastic changes in the area of ​​the illumination spot and the efficiency of the optical collection path, causing a significant drift in signal intensity and seriously affecting the repeatability and reliability of the measurement results.

[0005] Open, non-contact measurement environments are susceptible to ambient light interference, which further degrades the already weak effective signal, and the measurement results are severely affected by stray light from the environment. For example, in the detection of tobacco samples, which are placed on a conveyor belt and the analyzer is positioned above the conveyor belt for detection, the environment is easily affected by interference. Summary of the Invention

[0006] The purpose of this invention is to provide a non-contact online near-infrared spectroscopy analyzer and analysis method that can effectively suppress stray light and improve the signal-to-noise ratio, thereby improving measurement reliability and repeatability.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] In a first aspect, the present invention provides a non-contact online near-infrared spectroscopy analyzer, comprising:

[0009] The light source module includes a main light source, several auxiliary light sources, a wavelength calibration light source, and a light source switching mirror. The main light source provides near-infrared light to illuminate the sample. The auxiliary light sources emit light within a specific wavelength range to the sample to enhance the signal intensity at the characteristic absorption peaks of the components in the sample. The wavelength calibration light source emits reference light with known characteristic wavelengths for wavelength calibration. The light source switching mirror guides the auxiliary light sources and / or the wavelength calibration light sources to the optical path module.

[0010] An optical path module is used to guide the light emitted by the main light source and / or the auxiliary light source and / or the wavelength calibration light onto the sample, and / or to shape the light emitted by the main light source and / or the auxiliary light source and / or the wavelength calibration light source.

[0011] A calibration plate, located between the optical path module and the sample, is used for background calibration and wavelength calibration;

[0012] The receiving module includes multiple optical fibers, each optical fiber having a sampling end and a detection end. The sampling end is used to collect optical signals, and the detection end is coupled to a spectrometer to transmit the optical signals to the spectrometer. The sampling ends of the multiple optical fibers are uniformly distributed circumferentially, and the sampling end of each optical fiber has a different height position relative to the sampling ends of other optical fibers in a direction perpendicular to the sample. The detection ends of the multiple optical fibers are uniformly arranged along the slit length direction of the spectrometer.

[0013] A spectrometer, which is used to receive the optical signal transmitted by the optical fiber and generate corresponding spectral data;

[0014] A data processing and analysis module is used to process and analyze the spectral data.

[0015] Secondly, the present invention provides a non-contact online near-infrared spectroscopy analysis method, based on the above-mentioned analyzer, comprising:

[0016] S100. Only the main light source is turned on. The light emitted by the main light source passes through the optical path module to form an emitting cone-shaped optical path that illuminates the sample. The sampling end of the optical fiber receives the light signal generated by the sample under the illumination of the light source module, which carries its composition information, and forms a receiving cone-shaped optical path. The receiving cone-shaped optical paths of multiple optical fibers form a three-dimensional cross-sampling area in the vertical direction of the sample. The emitting cone-shaped optical path and the receiving cone-shaped optical path intersect in the vertical direction of the sample, so that the light signal generated after the light emitted by the light source module interacts with the sample is captured by the optical fiber and transmitted to the spectrometer, which generates basic light intensity data.

[0017] S200: Only turn on the selected auxiliary light source to illuminate the sample and obtain the corresponding supplementary light intensity data;

[0018] S300: Place the calibration plate in the emission cone-shaped light path to block the sample, turn on only the main light source, and the spectrometer receives the light signal reflected by the calibration plate and generates the corresponding first calibration light intensity data; turn off the main light source and turn on only the auxiliary light source used in step S200 to illuminate the calibration plate and obtain the corresponding second calibration light intensity data.

[0019] S400: With only the wavelength calibration light source turned on, the spectrometer receives the light signal reflected by the calibration plate and generates corresponding third calibration light intensity data;

[0020] S500: The data processing and analysis module preprocesses the basic light intensity data, the supplementary light intensity data, the first calibration light intensity data, the second calibration light intensity data, and the third calibration light intensity data generated by the spectrometer to obtain preprocessed sample spectral data. Then, based on a preset algorithm, the preprocessed sample spectral data is calculated and the corresponding component content value is output.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0022] The light source module of this invention employs a main light source and several auxiliary light sources with characteristic absorption peaks, enabling enhanced analysis based on direct response of characteristic wavelengths. This overcomes the inherent limitations of overlapping absorption peaks and baseline drift in traditional diffuse reflectance spectroscopy, significantly improving the prediction accuracy, robustness, and ability to distinguish similar materials of the calibration model, thereby optimizing measurement results. The receiving module of this invention uses multiple optical fibers with sampling ends uniformly distributed circumferentially and at varying heights perpendicular to the sample. The detection ends of the fibers are uniformly arranged along the slit length of the spectrometer. Within a certain axial distance variation range, the sampling ends at different heights receive different ranges of diffuse reflected light, increasing the depth of field of the received light. This allows the instrument to better adapt to samples at different distances. Furthermore, this structure is equivalent to constructing an "optical funnel" in space, capable of collecting scattered light from the sample from multiple angles and over a wide range, greatly improving light collection efficiency, maximizing the utilization efficiency of the light source energy, and efficiently capturing diffuse reflected / scattered light signals returned from the sample. Simultaneously, it achieves "adaptive" tolerance at the physical level, eliminating the need for complex active feedback mechanisms. Furthermore, the circumferentially varying heights of the multiple optical fibers allow them to primarily receive signal light from the sample detection area, while naturally shielding stray ambient light from other angles. This creates a spatial filtering effect, effectively isolating and reducing interference from stray ambient light, forming a stray light suppression mechanism, and effectively improving the signal-to-noise ratio. In addition, this invention employs a hybrid approach of linear and nonlinear methods, resulting in higher prediction accuracy. Attached Figure Description

[0023] Figure 1 This is the optical path diagram of the non-contact online near-infrared spectrometer according to an embodiment of the present invention.

[0024] Figure 2 This is a spectral diagram of a tobacco sample from Embodiment 2 of the present invention.

[0025] Figure 3 This is the spectral diagram of the cellulose acetate sample from Embodiment 2 of the present invention.

[0026] Figure 4 This is a flowchart of a non-contact online near-infrared spectroscopy analysis method according to an embodiment of the present invention.

[0027] In the diagram, 1-main light source; 2-auxiliary light source; 3-wavelength calibration light source; 4-light source switching mirror; 5-calibration plate; 6-fiber optic cable; 7-spectrometer; 8-shaping unit; 801-off-axis parabolic mirror; 802-plano-convex lens; 803-extinction lens tube; 9-reflector; 10-protective window. Detailed Implementation

[0028] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0029] In the description of this invention, it should be noted that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0030] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0031] Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0032] Example 1

[0033] like Figure 1 As shown, a preferred embodiment of the present invention provides a non-contact online near-infrared spectrometer, comprising:

[0034] The light source module includes a main light source 1, several auxiliary light sources 2, a wavelength calibration light source 3, and a light source switching mirror 4. The main light source 1 is used to provide near-infrared light to illuminate the sample. The auxiliary light sources 2 are used to emit light in a specific wavelength range to the sample to enhance the signal intensity at the characteristic absorption peaks of the components in the sample. The wavelength calibration light sources 3 are used to emit reference light with known characteristic wavelengths, which is used for wavelength calibration. The light source switching mirror 4 is used to guide the auxiliary light sources 2 and / or the wavelength calibration light sources 3 to the optical path module.

[0035] The optical path module is used to guide the light emitted by the main light source 1 and / or the auxiliary light source 2 and / or the wavelength calibration light source 3 onto the sample, and / or to shape the light emitted by the main light source 1 and / or the auxiliary light source 2 and / or the wavelength calibration light source 3.

[0036] Calibration plate 5, located between the optical path module and the sample, is used for background calibration and wavelength calibration; calibration plate 5 can be rotated to be located between the optical path module and the sample or moved away from the optical path module and the sample.

[0037] The receiving module includes multiple optical fibers 6, each fiber having a sampling end and a detection end. The sampling end is used to collect optical signals, and the detection end is coupled to a spectrometer 7 to transmit the optical signals to the spectrometer 7. The sampling ends of the multiple optical fibers 6 are uniformly distributed circumferentially, and the sampling end of each optical fiber 6 has a different height position relative to the sampling ends of other optical fibers 6 in the direction perpendicular to the sample. The detection ends of the multiple optical fibers 6 are uniformly arranged along the length of the slit of the spectrometer 7.

[0038] Spectrometer 7 is used to receive the optical signal transmitted by optical fiber 6 and generate the corresponding spectral data;

[0039] The data processing and analysis module is used to process and analyze spectral data.

[0040] The light source module in this embodiment uses a main light source 1 and several auxiliary light sources 2 with characteristic absorption peaks. It can realize enhanced analysis based on the direct response of characteristic wavelengths, overcome the inherent limitations of absorption peak overlap and baseline drift in traditional diffuse reflectance spectroscopy, and thus significantly improve the prediction accuracy, robustness and ability to distinguish similar materials of the calibration model, so as to optimize the measurement effect. The receiving module in this embodiment employs multiple optical fibers 6. The sampling ends of these fibers 6 are uniformly distributed circumferentially, and their heights vary perpendicular to the sample. The detection ends of the fibers 6 are uniformly arranged along the length of the slit in the spectrometer 7. Within a certain axial distance variation range, the sampling ends of the fibers 6 at different heights receive different ranges of diffuse reflection light, increasing the depth of field of the received light. This allows the instrument to better adapt to samples at different distances. This structure essentially creates an "optical funnel" in space, enabling the collection of scattered light from the sample from multiple angles over a wide area, significantly improving light collection efficiency and maximizing the utilization efficiency of the light source energy. It also efficiently captures diffuse reflection / scattered light signals returned from the sample, achieving physical "adaptive" tolerance without the need for complex active feedback mechanisms. Furthermore, the circumferential height variation of the fibers 6 ensures that they primarily receive signal light from the sample detection area, while naturally shielding stray light from other angles, creating a spatial filtering effect. This effectively isolates and reduces interference from stray light, forming a stray light suppression mechanism and effectively improving the signal-to-noise ratio.

[0041] In some embodiments, the optical path module includes: a shaping unit 8 for collimating and / or focusing the light emitted by the light source module; and a reflector 9 located at the geometric center surrounded by the sampling ends of multiple optical fibers 6, which guides the light emitted by the light source module onto the sample. The light emitted by the light source module illuminates the sample after passing through the shaping unit 8 and the reflector 9. The shaping unit 8 converts divergent and disordered light energy into a concentrated and controllable beam, greatly improving light energy utilization and signal quality, fundamentally ensuring high repeatability, stability, and reliability of the analysis model. The use of the reflector 9 makes the overall device smaller. In addition, besides… Figure 1 Besides the positions of the light source module and optical path module shown, the reflector 9 can be used in other positions. The reflector 9 is tilted. In this embodiment, the tilt angle of the reflector 9 is 45°. It should be noted that the protection of this patent is not limited to only 45°. When the tilt angle of the reflector 9 is other angles (such as 30°, 60°, etc.), the angle between the optical fiber 6 and the reflector 9 will also change accordingly. The angle of change is determined by the light spot presented.

[0042] Furthermore, the shaping unit 8 can move towards or away from the light source module. By moving the shaping unit 8, the distance between the shaping unit 8 and the light source module can be changed, thereby controlling the size of the light spot emitted by the light source module. Optionally, the shaping unit 8 includes one or more lenses; therefore, the number and shape of the lenses are not limited. When the shaping unit 8 includes more than one lens, adjacent lenses can move closer or further apart, adjusting the size of the irradiated light spot, energy density, etc. In this embodiment, the shaping unit 8 includes two lenses, namely an aspherical lens and a plano-convex lens 802, with the aspherical lens located between the light source module and the plano-convex lens 802. The combination of the aspherical lens and the extinction lens tube 803 is the core foundation of the analyzer equipment. Its important purpose is to achieve a high-quality, controllable parallel light beam, reduce stray light from the analyzer equipment, improve the shape and uniformity of the light spot irradiated on the sample, and further improve the repeatability and accuracy of the analyzer equipment. Optionally, the aspherical lens can be a biconvex lens.

[0043] In other embodiments, the analyzer further includes a protective window 10, which is disposed between the optical path module and the sample. The protective window 10 isolates the internal and external environments of the analyzer. Light emitted from the light source module is guided by the optical path module, transmitted through the protective window 10, and then illuminates the sample. The protective window 10 is tilted to reduce specular reflection from the light source module. The protective window 10 is a planar lens. The analyzer has a housing with a window. The protective window 10 is located at the window. Near-infrared light emitted from the light source module is transmitted through the protective window 10 and finally illuminates the sample. The protective window 10 acts as a medium to isolate the internal and external environments of the device. The protective window 10 is not horizontal but tilted at a certain angle, mainly because horizontal placement may cause specular reflection, leading to errors in the light collection spot. Tilting it reflects the reflected light outside the collection range.

[0044] This embodiment uses four optical fibers 6, with the reflector 9 located at the geometric center enclosed by the four optical fibers 6. It should be noted that the number of optical fibers 6 is not limited, but should be as large as possible while taking into account both signal collection intensity and the height of the spectrometer slit.

[0045] In this embodiment, the light source module emits near-infrared light, which is focused into a specific range of illumination by the shaping unit 8. After passing through the reflector 9, the reflected light passes through the protective window 10 and then illuminates the sample. This optical path is an emitting cone-shaped optical path. At the same time, the shaping unit 8 can move horizontally left and right linearly to control the size of the light spot on the emitting optical path, so that the light spot illuminating the sample is only slightly larger than the sample, making the light intensity approach its maximum within the range of the sample illuminated by the light source module, thus constructing an independent emitting light cone. The light illuminating the sample is collected by the optical fiber 6, and then transmitted after being aggregated. Sensed by the photosensitive element inside the spectrometer 7, this optical path is a receiving conical optical path. Simultaneously, the optical fiber 6 can move linearly up and down in a direction perpendicular to the sample, controlling the size of the light spot in the receiving optical path. This ensures that the light spot for receiving the sample is only slightly smaller than the light spot in the emitting optical path. Furthermore, the light spots in the receiving and emitting optical paths can act on the sample area at near proximity, causing them to intersect in the sample space, forming a three-dimensional cross-sampling area. This is how the emitting and receiving conical optical paths can work synergistically, thus constructing a dual-conical synergistic optical path device. Optionally, the sample distribution range, the illumination area formed by the light source module, and the receiving area formed by all the optical fibers 6 are infinitely close. Specifically, the sample distribution range occupies less than 90-95% of the illumination area formed by the light source module, and the receiving area formed by all the optical fibers 6 occupies 90-98% of the illumination area formed by the light source module.

[0046] Example 2

[0047] The difference between this embodiment and Embodiment 1 is that the shaping unit 8 in this embodiment is different from that in Embodiment 1, and the analyzer is further described based on Embodiment 1.

[0048] In this embodiment, the shaping unit 8 includes an off-axis parabolic mirror 801, a plano-convex lens 802, and an extinction mirror tube 803. The off-axis parabolic mirror 801 and plano-convex lens 802 are installed inside the extinction mirror tube 803. The light emitted by the light source module is sequentially irradiated onto the sample by the off-axis parabolic mirror 801, the plano-convex lens 802, and the mirror 9. The off-axis parabolic mirror 801 and plano-convex lens 802 achieve reflective aberration-free and refractive flexible control to obtain better light spot quality and light source uniformity. The off-axis parabolic mirror 801 collimates and focuses the beam, first achieving perfect beam control with a wide spectral range and no aberrations, fundamentally eliminating dispersion. The plano-convex lens 802 serves as a subsequent fine-tuning component, performing local shaping, fine focusing, or aberration compensation on the beam. Ultimately, while ensuring full-band spectral fidelity, it achieves higher precision and flexibility in optical path adjustment. The light source module is located at the focal point of the parent parabolic surface of the off-axis parabolic reflector 801. The light emitted by it is reflected by the off-axis mirror area to form an unobstructed parallel beam.

[0049] For the extinction lens barrel 803, an extinction lens barrel 803 with an extinction structure and / or with an extinction material can be adopted. In terms of structure, the extinction structure refers to the geometric structure set in the inner wall of the lens barrel or the internal optical path to suppress the reflection and propagation of stray light. It reduces the possibility of stray light reaching the imaging surface by changing the incident angle of light, increasing the optical path blockage, or extending the path of non-imaging light rays. For example, the matting structure may include, but is not limited to, inner wall grooves, light-blocking rings, multi-level steps, or rough surfaces formed by sandblasting. In terms of materials, matting materials refer to materials with low reflectivity or high light absorption properties used to form the lens barrel body or its surface treatment layer, which suppress stray light by reducing the reflection of light on the inner wall of the lens barrel. Such materials may include, but are not limited to: coating a black matte coating (such as black chrome, black nickel, carbon nanotube coating, or Acktar metallic black film) on the surface of a metal substrate (such as aluminum alloy or stainless steel); directly injection molding using engineering plastics with high light absorption properties (such as polyether ether ketone PEEK or polysulfone PSU with added carbon black); or forming a low reflectivity surface layer on metal and plastic substrates through processes such as anodizing, chemical deposition, and physical vapor deposition (PVD); or spraying with commercially available high-absorption matting paints (such as Nextel Velvet Coating, Aeroglaze Z306, etc.).

[0050] In some embodiments, the light source switching mirror 4 is movably disposed between the main light source 1 and the optical path module, such that the light source switching mirror 4 is located in or removed from the optical path of the main light source 1. When the light source switching mirror 4 is not located in the optical path of the main light source 1, the light emitted by the main light source 1 is guided to the sample through the optical path module. When the main light source 1 is turned off, and the light source switching mirror 4 is located in the optical path of the main light source 1, the light emitted by the auxiliary light source 2 and / or the wavelength calibration light source 3 is guided to the optical path module through the light source switching mirror 4 and then guided to the sample through the optical path module. Specifically, the light source switching mirror 4 is rotated to be placed in or removed from the optical path of the main light source 1.

[0051] In other embodiments, the light source switching mirror 4 is a semi-reflective mirror with a transmission side and a reflection side arranged opposite to each other. The transmission side faces the main light source 1, and the reflection side faces the auxiliary light source 2 and the wavelength calibration light source 3. The light source switching mirror 4 can always be in the optical path, and the switching of multiple light sources can be achieved through electronic control. Anti-reflection coatings and anti-reflection coatings are respectively coated on opposite sides of the light source switching mirror 4 to form a transmission side and a reflection side. The transmission side can effectively reduce the reflection loss of the main light source 1, allowing most of the light energy emitted by the main light source 1 to pass through the mirror, greatly improving the energy utilization efficiency of the optical path. The reflection side can enhance the reflection capability of the auxiliary light source 2 and the wavelength calibration light source 3, causing most of the light energy emitted by the light source to be reflected and redirected, realizing efficient beam splitting or redirection of the optical path, while suppressing unnecessary transmission loss.

[0052] Furthermore, in this embodiment, three auxiliary light sources 2 are provided, and one of the auxiliary light sources 2 is used to enhance the signal intensity at the characteristic absorption peak of moisture. Light source switching is achieved using a light source switching mirror 4, allowing different light sources to enter the shaping unit 8. It should be noted that the auxiliary light source 2 is used to emit light within a specific wavelength range onto the sample to enhance the signal intensity at the characteristic absorption peak of the components in the sample. The selection of the auxiliary light source 2 depends on the components of the sample to be detected. If it is necessary to detect proteins, fats, etc., then an auxiliary light source 2 that can enhance the signal intensity at the characteristic absorption peaks of proteins and fats is selected. This embodiment is used to detect the moisture content of tobacco leaves and acetate fibers; therefore, an auxiliary light source 2 selected to enhance the signal intensity at the characteristic absorption peak of moisture is chosen. Optionally, this embodiment selects an auxiliary light source 2 that enhances the signal intensity near the characteristic absorption peak of moisture at 1450 nm, where 1450 nm is approximately 1450 ± 20 nm.

[0053] In addition, the main light source 1 operates in the 400–2500 nm wavelength range, meaning the wavelength of the light emitted by the main light source 1 covers a wide range of 400–2500 nm. The auxiliary light source 2 can be a narrowband light source such as an LED or LD. The wavelength calibration light source 3 can be a xenon lamp, an LED, or an SLD laser diode.

[0054] In this embodiment, the light emitted by the main light source 1 and / or the auxiliary light source 2 and / or the wavelength calibration light source 3 is collimated by the off-axis parabolic reflector 801 to form a high-quality parallel beam. After being converged by the plano-convex lens 802, it is transmitted to the reflector 9. After being reflected by the reflector 9, the reflected beam passes through the protective window 10 and then illuminates the sample. At the same time, the plano-convex lens 802 can move linearly left and right in the horizontal direction to control the size of the light spot in the emitted light path, so that the light spot illuminating the sample is only slightly larger than the sample, making the light intensity approach its maximum within the sample illumination range under the light source module, thus constructing an independent emitted light cone. The light is collected by optical fiber 6, then aggregated and transmitted to the photosensitive element inside the spectrometer 7. This optical path is a receiving conical optical path. Simultaneously, optical fiber 6 can move linearly up and down perpendicular to the sample, controlling the size of the light spot in the receiving optical path. This ensures the light spot size is only slightly smaller than the light spot in the emitting optical path, and the light spots in the receiving and emitting optical paths can act on the sample area at near proximity, intersecting in the sample space to form a three-dimensional cross-sampling area. This is how the emitting and receiving conical optical paths can work synergistically, thus constructing a dual-conical synergistic optical path device. Optionally, the sample distribution range, the illumination area formed by the light source module, and the receiving area formed by all optical fibers 6 are infinitely close. Specifically, the sample distribution range occupies less than 90-95% of the illumination area formed by the light source module, and the receiving area formed by all optical fibers 6 occupies 90-98% of the illumination area formed by the light source module.

[0055] The data processing and analysis module is used to process and analyze the spectral data generated by the spectrometer 7. During detection, only the main light source 1 is turned on. The light emitted from the main light source 1 passes through the optical path module to form an emission cone-shaped optical path that illuminates the sample. The sampling end of the optical fiber 6 receives the light signal carrying its composition information generated by the sample under the illumination of the light source module, forming a receiving cone-shaped optical path. The receiving cone-shaped optical paths of multiple optical fibers 6 form a three-dimensional cross-sampling area in the vertical direction of the sample. The emission cone-shaped optical path and the receiving cone-shaped optical path intersect in the vertical direction of the sample, so that the light signal generated after the interaction between the light emitted by the light source module and the sample is captured by the optical fiber 6 and transmitted to the spectrometer 7, which generates basic light intensity data. Only the selected auxiliary light source 2 is turned on to illuminate the sample, obtaining the corresponding supplementary light intensity data. The calibration plate 5 is placed in the emission cone-shaped optical path. In the process, the sample is shielded, and only the main light source 1 is turned on to illuminate the calibration plate 5. The spectrometer 7 receives the light signal reflected by the calibration plate 5 and generates the corresponding first calibration light intensity data. The main light source 1 is turned off, and only the selected auxiliary light source 2 is turned on to illuminate the calibration plate 5, obtaining the corresponding second calibration light intensity data. Only the wavelength calibration light source 3 is turned on, and the spectrometer 7 receives the light signal reflected by the calibration plate 5 and generates the corresponding third calibration light intensity data. The data processing and analysis module preprocesses the basic light intensity data, supplementary light intensity data, first calibration light intensity data, second calibration light intensity data, and third calibration light intensity data generated by the spectrometer 7 to obtain the sample spectral data. Then, based on a preset algorithm, the sample spectral data is calculated and the corresponding component content value is output.

[0056] Preprocessing includes background calibration, wavelength calibration, and spectral combination.

[0057] Background calibration includes:

[0058] Background calibration is performed on the basic light intensity data based on the first calibration light intensity data to obtain the background-calibrated basic spectral data; background calibration is performed on the supplementary light intensity data based on the second calibration light intensity data to obtain the background-calibrated supplementary spectral data.

[0059] Wavelength calibration includes:

[0060] A third calibration spectral data is generated based on the third calibration light intensity data. The third calibration spectral data is compared with a preset mapping relationship to determine the current wavelength calibration deviation. The preset mapping relationship is updated based on the wavelength calibration deviation to obtain the updated wavelength calibration parameters. The updated wavelength calibration parameters are used to perform wavelength correction on the background-calibrated basic spectral data to output the wavelength-calibrated basic spectral data. The updated wavelength calibration parameters are used to perform wavelength correction on the background-calibrated supplementary spectral data to output the wavelength-calibrated supplementary spectral data.

[0061] The preset mapping relationship is used to characterize the correlation between each sampling unit of the spectrometer 7 detector and its corresponding physical wavelength, such as the functional relationship between pixel index and wavelength (nm). This mapping relationship can be a linear or nonlinear model, obtained by calibration with a standard light source at the factory, and can be dynamically updated during use by wavelength calibration light source 3.

[0062] Spectral combinations include:

[0063] Sample spectral data is generated based on wavelength-calibrated baseline spectral data and wavelength-calibrated supplementary spectral data. In this embodiment, the spectral combination adds the wavelength-calibrated baseline spectral data and the wavelength-calibrated supplementary spectral data to obtain the final sample spectral data. It should be noted that the wavelength-calibrated baseline spectral data and the wavelength-calibrated supplementary spectral data can also be averaged or weighted summed.

[0064] This embodiment primarily measures the moisture content of the sample. In the aforementioned measurement, the auxiliary light source 2 is used to enhance the signal intensity at the characteristic absorption peak of moisture, specifically, an auxiliary light source 2 that enhances the signal intensity near the characteristic absorption peak at 1450 nm. Preprocessing also includes further processing of the sample spectral data obtained from the spectral combination, including processing the sample spectral data according to the sample's moisture content. Based on the supplementary spectral data after wavelength calibration, the sample moisture content is determined as follows: if the sample moisture content is greater than a threshold, the sample moisture content is considered high; if the sample moisture content is lower than the threshold, the sample moisture content is considered low.

[0065] If the sample has a high water content, Savitzky-Golay smoothing filtering is applied to the sample spectral data to suppress high-frequency noise, resulting in smoothed spectral data. Asymmetric least squares (ALS) is then used to perform baseline correction on the smoothed spectral data, yielding baseline-corrected spectral data. Standard normal variable transformation (SNV) is then applied to the baseline-corrected spectra to simultaneously correct for scattering effects caused by differences in sample physical state and amplitude fluctuations due to moisture, resulting in transformed spectral data. The transformed spectral data is divided into low-band and high-band spectral data, and second-order Savitzky-Golay derivative processing is performed on the high-band to enhance component resolution, yielding high-band spectral data. Savitzky-Golay smoothing filtering is then applied to the low-band spectral data, resulting in low-band spectral data. The high-band and low-band spectral data are then stitched together to obtain stitched spectral data, which is then mean-centered to obtain preprocessed sample spectral data.

[0066] If the sample has a low water content, Savitzky-Golay smoothing filtering is applied to the sample spectral data to suppress high-frequency noise, resulting in smoothed spectral data. Asymmetric least squares (ALS) is then used to perform baseline correction on the smoothed spectral data, yielding baseline-corrected spectral data. Standard normal variable transformation (SNV) is then applied to the baseline-corrected spectra to simultaneously correct for scattering effects caused by differences in sample physical state and amplitude fluctuations due to moisture, resulting in transformed spectral data. Continuum removal (CR) processing is then performed on the transformed spectral data, avoiding the bands containing characteristic absorption peaks during envelope fitting, resulting in CR-removed spectral data. Finally, mean centering is applied to the CR-removed spectral data to obtain preprocessed sample spectral data.

[0067] The data processing and analysis module analyzes the preprocessed sample spectral data to calculate and output the corresponding component content values ​​based on a preset algorithm. The preset algorithm includes a calibration model, which is established as follows: a representative set of calibration samples is prepared, and their spectral data are acquired using the analyzer in this embodiment; the concentration or property reference values ​​of the calibration samples are determined using standard methods; the preprocessed sample spectral data is correlated with the reference values, and a calibration model between the spectral data and the reference values ​​is established using a modeling algorithm. The calibration model in this embodiment includes a linear regression model and a nonlinear regression model. The preprocessed sample spectral data is input into the linear regression model to obtain the first component content value; the preprocessed spectral data is input into the nonlinear regression model to obtain the second component content value; the first and second component content values ​​are weighted and fused to obtain the final component content value.

[0068] In one embodiment, tobacco was selected as the sample to be tested. 210 representative tobacco samples from different sources were collected. Using the analyzer of this embodiment, the distance between the tobacco sample and the protective window 10 was 200-300 mm during detection. The component content of these 210 tobacco samples was determined. Spectral data of these 210 tobacco samples were acquired by the spectrometer 7. The preprocessed spectroscopic data of the tobacco samples were output as follows: Figure 2 As shown in the figure. Standard chemical analysis methods were used to collect samples three times, and the average value was taken as the reference value for that sample.

[0069] The spectral data of the tobacco samples were preprocessed using a method with high sample moisture content to obtain preprocessed spectral data of the tobacco samples. The preprocessed spectral data of the tobacco samples and the reference values ​​were used to train and verify the calibration model established by the multivariate calibration algorithm. The performance of the comprehensive calibration model in this embodiment is shown in Table 1.

[0070] Table 1. Detection performance of component content values ​​in tobacco samples

[0071]

[0072] In another embodiment, acetate fiber was selected as the sample to be tested. Fifty-six representative acetate fiber samples from different sources were collected. Using the analyzer of this embodiment, the distance between the acetate fiber sample and the protective window 10 was 200-300 mm during detection. The component content of these 56 acetate fiber samples was determined. Spectral data of these 56 acetate fiber samples were acquired by spectrometer 7. The preprocessed spectrograms of the acetate fiber sample spectral data are shown below. Figure 3 As shown in the figure. Standard chemical analysis methods were used to collect samples three times for each acetate fiber sample, and the average value was taken as the reference value for that sample.

[0073] The spectral data of acetate fiber samples were preprocessed using a method that minimizes sample moisture content, resulting in preprocessed spectral data. This preprocessed spectral data, along with reference values, was used to train and validate the calibration model established using a multivariate calibration algorithm. The overall performance of the calibration model in this embodiment is shown in Table 2.

[0074] Table 2 Performance of Acetate Fiber Sample Component Content Value Detection

[0075]

[0076] As can be seen from the above, when detecting tobacco samples and acetate fiber samples, the correlation coefficient R is high, the root mean square error of calibration SEC and the root mean square error of cross-validation calibration Rmsecv are small, and there is a very strong linear correlation between the model prediction value and the reference value. Therefore, the analyzer in this embodiment has high accuracy, high reliability and good repeatability.

[0077] The other structures in this embodiment are the same as in Embodiment 1, and will not be described again here.

[0078] Example 3

[0079] like Figure 4 As shown, this embodiment of the invention provides a non-contact online near-infrared spectroscopy analysis method, based on the analyzer of Embodiment 1 or Embodiment 2, comprising:

[0080] S100: Only the main light source 1 is turned on. The light emitted by the main light source 1 passes through the optical path module to form an emission cone-shaped optical path that illuminates the sample. The sampling end of the optical fiber 6 receives the light signal generated by the sample under the illumination of the light source module, which carries its composition information, and forms a receiving cone-shaped optical path. The receiving cone-shaped optical paths of multiple optical fibers 6 form a three-dimensional cross sampling area in the vertical direction of the sample. The emission cone-shaped optical path and the receiving cone-shaped optical path intersect in the vertical direction of the sample, so that the light signal generated after the light emitted by the light source module interacts with the sample is captured by the optical fiber 6 and transmitted to the spectrometer 7, which generates basic light intensity data.

[0081] S200: Only turn on the selected auxiliary light source 2 to illuminate the sample and obtain the corresponding supplementary light intensity data;

[0082] S300: Place the calibration plate 5 in the emission cone-shaped light path to block the sample, turn on only the main light source 1, and the spectrometer 7 receives the light signal reflected by the calibration plate 5 and generates the corresponding first calibration light intensity data; turn off the main light source 1 and turn on only the auxiliary light source 2 used in step S200 to illuminate the calibration plate 5, and obtain the corresponding second calibration light intensity data.

[0083] S400, Only the wavelength calibration light source 3 is turned on, the spectrometer 7 receives the light signal reflected by the calibration plate 5, and generates the corresponding third calibration light intensity data;

[0084] S500: The data processing and analysis module preprocesses the basic light intensity data, the supplementary light intensity data, the first calibration light intensity data, the second calibration light intensity data, and the third calibration light intensity data generated by the spectrometer 7 to obtain preprocessed sample spectral data. Then, based on a preset algorithm, the preprocessed sample spectral data is calculated and the corresponding component content value is output.

[0085] In some embodiments, the preprocessing in step S500 includes background calibration, wavelength calibration, and spectral combination.

[0086] The background calibration includes:

[0087] Based on the first calibration light intensity data, background calibration is performed on the basic light intensity data to obtain the background-calibrated basic spectral data; based on the second calibration light intensity data, background calibration is performed on the supplementary light intensity data to obtain the background-calibrated supplementary spectral data.

[0088] The wavelength calibration includes:

[0089] A third calibration spectral data is generated based on the third calibration light intensity data. The third calibration spectral data is compared with a preset mapping relationship to determine the current wavelength calibration deviation. The preset mapping relationship is updated based on the wavelength calibration deviation to obtain updated wavelength calibration parameters. The updated wavelength calibration parameters are used to perform wavelength correction on the background-calibrated basic spectral data to output wavelength-calibrated basic spectral data. The updated wavelength calibration parameters are used to perform wavelength correction on the background-calibrated supplementary spectral data to output wavelength-calibrated supplementary spectral data.

[0090] The spectral combination includes:

[0091] The sample spectral data is generated based on the wavelength-calibrated basic spectral data and the wavelength-calibrated supplementary spectral data.

[0092] This embodiment primarily measures the moisture content of the sample. The auxiliary light source 2 used in step S100 is used to enhance the signal intensity at the characteristic absorption peak of moisture, specifically, an auxiliary light source 2 that enhances the signal intensity near the characteristic absorption peak at 1450 nm. The preprocessing in this embodiment also includes further processing of the sample spectral data obtained from the spectral combination, including processing the sample spectral data according to the sample's moisture content. Based on the supplementary spectral data after wavelength calibration, if the sample's moisture content is greater than a threshold, the sample has a high moisture content; if the sample's moisture content is lower than the threshold, the sample has a low moisture content.

[0093] Further processing steps for the sample spectral data obtained from spectral combination include:

[0094] S510. Select the pretreatment mode according to the moisture content of the sample. If the moisture content of the sample is greater than the threshold, the moisture content of the sample is high. If the moisture content of the sample is lower than the threshold, the moisture content is low. If the moisture content of the sample is high, proceed to step S520. If the moisture content of the sample is low, proceed to step S530.

[0095] S520. If the sample has a high moisture content, proceed to steps S521-S525:

[0096] S521. Perform Savitzky-Golay smoothing filtering on the spectral data generated by spectrometer 7 to suppress high-frequency noise and obtain smoothed spectral data.

[0097] S522. The smoothed spectral data is baseline corrected using the asymmetric least squares method (ALS) to obtain baseline-corrected spectral data.

[0098] S523. Perform standard normal variable transformation (SNV) on the baseline-corrected spectrum to simultaneously correct the scattering effect caused by differences in the physical state of the sample and the amplitude fluctuation caused by moisture, and obtain the transformed spectral data.

[0099] S524. The transformed spectral data is divided into low-band and high-band, and the high-band is processed by second-order Savitzky-Golay derivative processing to enhance the component resolution, thus obtaining high-band spectral data; the low-band is processed by Savitzky-Golay smoothing filtering to obtain low-band spectral data; in this embodiment, the division of low-band and high-band is based on the effective spectrum according to the median wavelength or a preset ratio.

[0100] S525. The high-band spectral data and low-band spectral data are spliced ​​together to obtain the spliced ​​spectral data. The mean of the spliced ​​spectral data is then centered to obtain the preprocessed spectral data.

[0101] S530. If the sample has a low moisture content, proceed to steps S531-S535:

[0102] S531. Perform Savitzky-Golay smoothing filtering on the spectral data generated by spectrometer 7 to suppress high-frequency noise and obtain smoothed spectral data.

[0103] S532. The smoothed spectral data is baseline corrected using the asymmetric least squares method (ALS) to obtain baseline-corrected spectral data.

[0104] S533. Perform standard normal variable transformation (SNV) on the baseline-corrected spectrum to simultaneously correct the scattering effect caused by differences in the physical state of the sample and the amplitude fluctuation caused by moisture, and obtain the transformed spectral data.

[0105] S534. Perform continuum CR removal processing on the transformed spectral data. When fitting the envelope, avoid the bands where the characteristic absorption peaks are located to obtain the removed spectral data. The characteristic absorption peaks are determined by obtaining the absorption peaks of representative components based on prior chemical knowledge. These representative absorption peaks are the characteristic absorption peaks. For example, for tobacco leaves, the characteristic absorption peaks mainly come from the overtone and combination vibrations of O–H, N–H, and C–H bonds in sugars, nicotine, proteins, and cellulose, and usually appear around 1200 nm, 1450–1500 nm, and 1720–1760 nm. For acetate fiber, the characteristic absorption peaks mainly come from the overtone and combination vibrations of acetyl groups and ester bonds, and usually appear around 1170 nm, 1420 nm, and 1720 nm.

[0106] S535. The mean center is applied to the removed spectral data to obtain the preprocessed spectral data.

[0107] In this embodiment, the preset algorithm includes a calibration model, which is established as follows: a set of representative calibration set samples are prepared, and their spectral data are collected by the analyzer of Example 1 or Example 2; the concentration or property reference value of the calibration set samples is determined by standard method; the preprocessed sample spectral data is correlated with the reference value, and a calibration model between the spectral data and the reference value is established using the modeling algorithm.

[0108] In some embodiments, the calibration model includes a linear regression model and a nonlinear regression model. In step 500, the preprocessed sample spectral data is calculated based on a preset algorithm to output the corresponding component content values. This includes: inputting the preprocessed sample spectral data into a linear regression model to obtain a first component content value; inputting the preprocessed spectral data into a nonlinear regression model to obtain a second component content value; and weighting and fusing the first and second component content values ​​to obtain the final component content value. In this embodiment, the linear regression model uses a partial least squares regression (PLS) model, and the nonlinear regression model uses a support vector machine (SVR) regression model.

[0109] Specifically, the spectral data of the calibration set samples after preprocessing are divided into different bands based on the bands corresponding to the components, forming multiple sub-training sets. Linear regression and nonlinear regression models are trained using these corresponding sub-training sets to obtain the trained linear and nonlinear regression models. In the weighted fusion, a validation set sample, independent of the training set and with known true values, is used to calculate the prediction accuracy index of each model in the model set, and a fixed fusion weight is assigned to each model based on its prediction accuracy index. Let the weight be... ,in, For the first The root mean square error (RMSE) of each model on the validation set is n, where n is the total number of models. The validation set consists of representative samples covering the concentration range of the tested component that were not used in any model training.

[0110] The working process of this invention is as follows: (1) Only the main light source 1 is turned on. The light emitted by the main light source 1 is collimated by the off-axis parabolic reflector 801 to form a high-quality parallel beam. After being converged by the plano-convex lens 802, it is transmitted to the reflector 9. After being reflected by the reflector 9, the reflected beam passes through the protective window 10 and then illuminates the sample. At the same time, the plano-convex lens 802 can move horizontally left and right to control the size of the light spot in the emission path, so that the light spot illuminating the sample is only slightly larger than the sample, so that the light intensity is close to the maximum within the range of the sample illuminated by the light source module, thus constructing an independent emission cone; the light illuminating the sample The light is collected by fiber 6, then aggregated and transmitted to the photosensitive element inside the spectrometer 7. This optical path is a receiving conical optical path. Simultaneously, fiber 6 can move linearly up and down in a direction perpendicular to the sample, controlling the size of the light spot in the receiving optical path. This ensures the light spot size is only slightly smaller than the light spot in the emitting optical path, and the light spots in the receiving and emitting optical paths can act on the sample area almost infinitely close, intersecting in the sample space to form a three-dimensional cross-sampling area. This is how the emitting and receiving conical optical paths can work synergistically, thus constructing a dual-conical synergistic optical path device. Optionally, the sample distribution range, the illumination area formed by the light source module, and the receiving area formed by all fibers 6 are infinitely close. Specifically, the sample distribution range occupies less than 90-95% of the illumination area formed by the light source module, and the receiving area formed by all fibers 6 occupies 90-98% of the illumination area formed by the light source module. The light response signals generated by the main light source 1 and auxiliary light source 2 illuminating the sample are transmitted to the spectrometer 7 via fiber 6 to generate basic light intensity data. (2) Only turn on the selected auxiliary light source 2 to irradiate the sample and obtain the corresponding supplementary light intensity data. (3) Place the calibration plate 5 in the emission cone light path to block the sample, turn on the main light source 1, and the spectrometer 7 receives the light signal reflected by the calibration plate 5 and generates the corresponding first calibration light intensity data; turn off the main light source 1, turn on the selected auxiliary light source 2 to irradiate the calibration plate 5 and obtain the corresponding second calibration light intensity data; (4) Only turn on the wavelength calibration light source 3, and the spectrometer 7 receives the light signal reflected by the calibration plate 5 and generates the corresponding third calibration light intensity data; (5) The data processing and analysis module preprocesses the basic light intensity data, supplementary light intensity data, first calibration light intensity data, second calibration light intensity data and third calibration light intensity data generated by the spectrometer 7 to obtain the preprocessed sample spectral data, and then calculates the preprocessed sample spectral data based on the preset algorithm and outputs the corresponding component content value.

[0111] In summary, this invention provides a non-contact online near-infrared spectrometer, whose light source module uses a main light source 1 and several auxiliary light sources 2 with characteristic absorption peaks. It can realize enhanced analysis based on the direct response of characteristic wavelengths, overcoming the inherent limitations of absorption peak overlap and baseline drift in traditional diffuse reflectance spectra. This significantly improves the prediction accuracy, robustness, and ability to distinguish similar materials of the calibration model, thereby optimizing the measurement effect. The receiving module of this invention employs multiple optical fibers 6. The sampling ends of these fibers 6 are uniformly distributed circumferentially, and their heights vary perpendicular to the sample. The detection ends of the fibers 6 are uniformly arranged along the length of the slit in the spectrometer 7. Within a certain axial distance variation range, the sampling ends of the fibers 6 at different heights receive diffuse reflection light from different ranges, increasing the depth of field of the received light. This allows the instrument to better adapt to samples at different distances. Furthermore, this structure essentially constructs an "optical funnel" in space, enabling the collection of scattered light from the sample from multiple angles over a wide area, significantly improving light collection efficiency and maximizing the utilization efficiency of the light source energy. It also efficiently captures diffuse reflection / scattered light signals returned from the sample, achieving physical "adaptive" tolerance without the need for complex active feedback mechanisms. Moreover, the circumferential height variation of the fibers 6 ensures that they primarily receive signal light from the sample detection area, while naturally shielding stray environmental light from other angles, creating a spatial filtering effect. This effectively isolates and reduces interference from stray environmental light, forming a stray light suppression mechanism and effectively improving the signal-to-noise ratio. This invention also preprocesses the spectral data based on water content, which better preserves the characteristics of the spectral data and thus improves the reliability of the final component content values. Furthermore, this invention employs a hybrid approach of linear and nonlinear methods, resulting in higher prediction accuracy.

[0112] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A non-contact on-line near infrared spectroscopic analyzer characterized by, The non-contact online near-infrared spectrum analyzer comprises a light source module, a light path module, a calibration plate, a receiving module, a spectrometer and a data processing and analysis module. The light source module comprises a main light source, a plurality of auxiliary light sources, a wavelength calibration light source and a light source switching mirror. The light path module is used for guiding the light emitted by the main light source and / or the auxiliary light source and / or the wavelength calibration light source to the sample, and / or shaping the light emitted by the main light source and / or the auxiliary light source and / or the wavelength calibration light source. The calibration plate is located between the light path module and the sample and is used for background calibration and wavelength calibration. The receiving module comprises a plurality of optical fibers. The spectrometer is used for receiving the light signal conducted by the optical fiber and generating corresponding spectrum data. The data processing and analysis module is used for processing and analyzing the spectrum data. The non-contact online near-infrared spectrum analyzer is used to realize the following steps: S100, only the main light source is turned on, the light emitted by the main light source forms a transmission cone-shaped light path on the sample through the light path module, the sampling end of the optical fiber receives the light signal carrying the component information of the sample under the irradiation of the light source module, forms a receiving cone-shaped light path, and the receiving cone-shaped light paths of the plurality of optical fibers form a three-dimensional intersection sampling area in the vertical direction of the sample; the transmission cone-shaped light path and the receiving cone-shaped light path intersect in the vertical direction of the sample, so that the light signal generated after the light emitted by the light source module interacts with the sample is captured by the optical fiber and conducted to the spectrometer, and the spectrometer generates basic light intensity data; S200, only the selected auxiliary light source is turned on to irradiate the sample, and corresponding supplementary light intensity data is obtained; S300, place the calibration plate (5) in the emission cone light path, block the sample, only turn on the main light source (1), the spectrometer (7) receives the light signal reflected by the calibration plate (5), and generates corresponding first calibration light intensity data; only turn on the auxiliary light source (2) used in step S200 to irradiate the calibration plate (5), and obtain corresponding second calibration light intensity data; S400, only turn on the wavelength calibration light source (3), the spectrometer (7) receives the light signal reflected by the calibration plate (5), and generates corresponding third calibration light intensity data; S500, the data processing and analysis module pre-processes the basic light intensity data, the supplementary light intensity data, the first calibration light intensity data, the second calibration light intensity data and the third calibration light intensity data generated by the spectrometer (7), obtains the pre-processed sample spectrum data, and calculates and outputs the corresponding component content value based on the preset algorithm.

2. The non-contact online near infrared spectroscopic analyzer according to claim 1, characterized in that, The light path module comprises: The shaping unit (8) is used for collimating and / or converging the light emitted by the light source module; The mirror (9) is located at the geometric center surrounded by the sampling ends of the plurality of optical fibers (6), and the mirror (9) is used for guiding the light emitted by the light source module to the sample.

3. The non-contact online near infrared spectroscopic analyzer according to claim 2, wherein, The shaping unit (8) comprises an off-axis parabolic mirror (801), a plano-convex lens (802) and an extinction lens barrel (803), the off-axis parabolic mirror (801) and the plano-convex lens (802) are installed in the extinction lens barrel (803), and the light emitted by the light source module is sequentially irradiated on the sample through the off-axis parabolic mirror (801), the plano-convex lens (802) and the mirror (9).

4. The non-contact online near infrared spectroscopic analyzer according to claim 1, wherein, Further comprising a protective window piece (10), the protective window piece (10) is arranged between the light path module and the sample, the protective window piece (10) is used for isolating the internal environment of the detection analyzer from the external environment, and the light emitted by the light source module is sequentially guided by the light path module and transmitted by the protective window piece (10) and then irradiated on the sample, wherein the protective window piece (10) is arranged obliquely to reduce the specular reflection light of the light source module.

5. The non-contact online near infrared spectroscopic analyzer according to claim 1, wherein, The light source switching mirror (4) is placed in or removed from the light path of the main light source (1) by rotating.

6. The non-contact online near infrared spectroscopic analyzer according to claim 1, wherein, One of the auxiliary light sources (2) is used to enhance the signal intensity at the water feature absorption peak.

7. A non-contact on-line near infrared spectroscopy method based on the analyzer of any one of claims 1 to 6, characterized in that, Comprise: S100, only turn on the main light source (1), the light emitted by the main light source (1) forms a emission cone light path to irradiate on the sample, the sampling end of the optical fiber (6) receives the light signal carrying the component information of the sample under the irradiation of the light source module, forms a receiving cone light path, and the receiving cone light paths of a plurality of optical fibers (6) form a three-dimensional cross-sampling area in the vertical direction of the sample; the emission cone light path and the receiving cone light path intersect in the vertical direction of the sample, so that the light signal generated after the light emitted by the light source module interacts with the sample is captured by the optical fiber (6) and conducted to the spectrometer (7), and the spectrometer (7) generates basic light intensity data; S200, only turn on the selected auxiliary light source (2) to irradiate the sample, and obtain corresponding supplementary light intensity data; S300, place the calibration plate (5) in the emission cone light path to shield the sample, only turn on the main light source (1), the spectrometer (7) receives the light signal reflected by the calibration plate (5), and generates corresponding first calibration light intensity data; only turn on the auxiliary light source (2) used in step S200 to irradiate the calibration plate (5), and obtain corresponding second calibration light intensity data; S400, only turn on the wavelength calibration light source (3), the spectrometer (7) receives the light signal reflected by the calibration plate (5), and generates corresponding third calibration light intensity data; S500, the data processing and analysis module pre-processes the basic light intensity data, the supplementary light intensity data, the first calibration light intensity data, the second calibration light intensity data and the third calibration light intensity data generated by the spectrometer (7), obtains pre-processed sample spectrum data, and calculates and outputs corresponding component content value based on a preset algorithm.

8. The non-contact online near infrared spectroscopic analysis method according to claim 7, characterized in that, The preprocessing of step S500 includes background calibration, wavelength calibration and spectrum combination, The background calibration includes: Based on the first calibration light intensity data, the background of the basic light intensity data is calibrated to obtain background-calibrated basic spectrum data; based on the second calibration light intensity data, the background of the supplementary light intensity data is calibrated to obtain background-calibrated supplementary spectrum data; The wavelength calibration includes: Based on the third calibration light intensity data, third calibration spectrum data is generated, the third calibration spectrum data is compared with a preset mapping relationship to determine the current wavelength calibration deviation; based on the wavelength calibration deviation, the preset mapping relationship is updated to obtain updated wavelength calibration parameters; the background-calibrated basic spectrum data is wavelength-corrected by using the updated wavelength calibration parameters to output wavelength-calibrated basic spectrum data; the background-calibrated supplementary spectrum data is wavelength-corrected by using the updated wavelength calibration parameters to output wavelength-calibrated supplementary spectrum data; The spectrum combination includes: Based on the wavelength-calibrated basic spectrum data and the wavelength-calibrated supplementary spectrum data, the sample spectrum data is generated.

9. The non-contact online near infrared spectroscopic analysis method according to claim 7, characterized in that, The preset algorithm comprises a correction model, and the correction model is established as follows: a set of representative correction set samples are prepared, and spectral data of the samples are collected by the analyzer according to any one of claims 1-6; the concentration or property reference value of the correction set samples is determined by using a standard method; the preprocessed sample spectral data is associated with the reference value, and the correction model between the spectral data and the reference value is established by using a modeling algorithm.

10. The non-contact online near infrared spectroscopic analysis method according to claim 9, characterized in that, The correction model comprises a linear regression model and a nonlinear regression model, and in step 500, the preprocessed sample spectral data is calculated based on the preset algorithm and corresponding component content values are output, comprising: inputting the preprocessed sample spectral data into the linear regression model to obtain first component content values; inputting the preprocessed spectral data into the nonlinear regression model to obtain second component content values; and weighting and fusing the first component content values and the second component content values to obtain final component content values.

Images