A mica content detection system and method based on multi-light source imaging

Abstract

The application discloses a mica content detection system and method based on multi-light source imaging, and belongs to the field of fine aggregate performance detection. The system comprises a sampling device, a pretreatment device and a multi-light source imaging and laser-induced breakdown spectroscopy detection device. The sampling device automatically samples from the production line based on the partition sampling theory and stirs and mixes; the pretreatment device separates the sample from impurities through backflushing and kneading; the multi-light source imaging and laser-induced breakdown spectroscopy detection device irradiates the falling sample with multiple wavelength light sources and collects images in real time, and simultaneously excites the sample to generate plasma through laser and collects the emission spectrum. The data processing unit fuses and calculates the mica content after processing the image and spectrum data respectively. The application realizes rapid online detection of the mica content, the single analysis time is less than 1 minute, the detection accuracy is more than 95%, and the system can be embedded into the fine aggregate production line to realize full-process automation.

Description

Technical Field

[0001] This invention relates to the field of fine aggregate performance testing, specifically to a mica content detection system and method based on multi-source imaging, which is suitable for online rapid detection of mica content in fine aggregates during sand and gravel processing. Background Technology

[0002] Fine aggregates, as the basic raw materials for concrete and mortar, directly determine key performance indicators such as strength and durability. Mica, a typical harmful impurity in fine aggregates, significantly reduces the bonding force between fine aggregate particles due to its flaky structure, low strength, and weak binding properties, leading to serious quality problems such as cracking and strength reduction in concrete and mortar. Therefore, accurately controlling the mica content in fine aggregates has become a crucial aspect of quality control, requiring efficient and reliable testing technologies.

[0003] Currently, mica content detection mainly relies on laboratory analysis, requiring procedures such as sample drying, sieving, chemical digestion, or manual microscopic examination. This method has the following drawbacks: First, the testing cycle is long. A single test can take several hours to several days, and the results are significantly delayed compared to the production process. This makes real-time quality control impossible, and substandard raw materials may have already entered subsequent processes, resulting in resource waste and quality risks.

[0004] Second, the pretreatment is complex. Chemical digestion requires the use of strong acid and strong alkali reagents, which is not only cumbersome and time-consuming, but also prone to secondary pollution, posing a hazard to the environment and operators. Manual microscopic examination relies on the operator's experience, resulting in highly subjective and erroneous results, and poor comparability of test results between different operators.

[0005] Third, limitations in application scenarios. Existing testing methods are limited to laboratory environments and cannot be used for online testing in sand and gravel processing systems. This makes it difficult to cover the entire fine aggregate production process and is seriously out of step with the current trend of industrial intelligence and automation.

[0006] For the detection of solid-phase materials, some studies have explored laser spectroscopy technology. For example, patent application CN113252577A discloses a system and method for synergistic spectral detection of solid-phase materials. However, it focuses only on single spectral data acquisition and lacks the ability to simultaneously process image data. It cannot verify mica content through a dual-dimensional "spectrum + image" approach, making it unsuitable for high-efficiency detection requirements. Another example is patent application CN114047174A, which discloses a two-dimensional plasma lattice grating-enhanced laser-induced breakdown spectroscopy detection sensitivity device. However, it does not design a light source and spectral collection module specifically for the flow characteristics of fine aggregates, resulting in a limited spectral signal collection range and insufficient accuracy, making it difficult to distinguish mica from other minerals with similar spectral characteristics. While these technologies focus on optimizing spectral detection, they do not address the practical pain points of online, rapid, and accurate detection of mica in fine aggregates.

[0007] Furthermore, in the field of mica testing for fine aggregates, existing technologies remain at the traditional testing stage, showing a significant disconnect from the demands of industrial intelligence. On the one hand, the continuous and dynamic nature of fine aggregate transport on production lines necessitates real-time feedback on mica content to adjust production parameters. On the other hand, traditional testing methods are unsuitable for assembly line operations, leading to lagging quality control and the easy influx of substandard raw materials into subsequent processes, increasing production costs and quality risks for enterprises. Existing patented technologies do not integrate the entire process of "intelligent sampling—sample pretreatment—online testing—data feedback," still requiring manual intervention in key stages and unable to be directly embedded into fine aggregate production lines. Their level of intelligence and automation is far below actual industrial needs.

[0008] In summary, current fine aggregate mica testing technology cannot meet the demands of "fast, accurate, online, and intelligent" in industrial production. There is an urgent need to optimize the design of the testing system based on the physical properties of fine aggregates and break through key technical bottlenecks such as preprocessing, sampling, and spectral-image fusion analysis. Summary of the Invention

[0009] To address the shortcomings of existing technologies, such as long detection cycles, complex preprocessing, reliance on manual operation, and inability to perform online detection of mica, this invention provides a mica content detection system and method based on multi-source imaging. The aim is to achieve rapid, accurate, and stable online detection of mica content through the integration of multiple technologies and structural optimization, while improving operational convenience, adapting to the actual needs of sand and gravel processing and production systems, and realizing full-process automation from sampling to result output.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention proposes a mica content detection system based on multi-source imaging, comprising: The sampling device is used to collect fine aggregate samples from each section of the aggregate conveyor belt in the sand and gravel processing production line. After sampling of all sub-sections is completed, the samples collected from each sampling point are stirred and mixed to form the sample to be tested. The pretreatment device is used to rinse and rub the sample to be tested with backwash water to remove impurities adhering to the sample. The multi-source imaging and laser-induced breakdown spectroscopy detection device is used to irradiate the sample during its fall using light sources of multiple wavelengths and acquire image data of the sample in real time; and to generate plasma by exciting the sample with a laser and collect the spectrum emitted by the plasma to obtain spectral data.

[0011] Furthermore, the sampling device includes an electric sampling shovel, a pneumatic suction device, a stirrer, partitioned hoppers, and a track; the track is arranged on one side of the aggregate conveyor belt; the electric sampling shovel and the pneumatic suction device are mounted on the track; multiple partitioned hoppers are arranged in a straight line and positioned close to the track, and the partitioned hoppers are used to temporarily store the samples collected by the electric sampling shovel and the pneumatic suction device; a conveyor belt is arranged directly below all the partitioned hoppers; the discharge end of the conveyor belt is located directly above the stirrer.

[0012] Furthermore, the pretreatment device includes a backwash water flow device, a kneading drum, and a feeding hopper; the outlet of the agitator is located directly above the feeding hopper, and the feeding hopper is used to receive the sample to be tested after mixing by the agitator; the outlet end of the feeding hopper is aligned with the kneading drum; the water outlet pipe of the backwash water flow device is aligned with the kneading drum; a screen is provided below the outlet of the kneading drum to receive the test sample after being rinsed and kneaded by the backwash water flow, and to filter the test sample.

[0013] Furthermore, the multi-source imaging and laser-induced breakdown spectroscopy detection device includes a constant temperature chamber, a feeding tray, a vibrating feeding mechanism, and a LIBS detection unit. The feeding tray is positioned below the screen, and its discharge end is positioned above the feeding end of the vibrating feeding mechanism. The discharge end of the vibrating feeding mechanism is positioned directly above the top inlet of the constant temperature chamber. An imaging sensor and multiple light source components are installed inside the constant temperature chamber. Each light source component emits light alternately at a preset frequency to irradiate the sample during its descent. The imaging sensor acquires sample images in real time at a frame rate of 100Hz. A laser window probe is installed on the left side wall of the constant temperature chamber. A spectrometer is installed on the right side wall of the constant temperature chamber, and a spectral fiber is installed at the entrance pupil of the spectrometer. The laser output from the laser in the LIBS detection unit is beam shaped, combined, expanded, and focused, and then focused onto the surface of the falling sample through the laser window probe, exciting the sample to generate plasma. The spectral fiber is used to collect the plasma emission spectrum in real time and transmit it to the spectrometer.

[0014] Furthermore, the system also includes a vortex booster pump, which is connected to the laser window probe via an air pipe; the outlet end of the laser window probe is configured as a conical opening; the vortex booster pump is used to pump high-pressure gas into the laser window probe, and the high-pressure gas is ejected from the conical opening of the laser window probe to purge steam and dust from the laser transmission path.

[0015] Furthermore, a collection trough is provided at the bottom of the constant temperature chamber for collecting falling samples; a buffer pad is provided in the collection trough.

[0016] Furthermore, the light source assembly includes lasers with wavelengths of 632.8 nm, 355 nm, 532 nm, and 1064 nm.

[0017] Secondly, the present invention also provides a method for detecting mica content using the above-described system, comprising the following steps: S1. Preliminary preparation: Inspect and debug each unit of the system, and locate and calibrate the focusing area; S2. Intelligent Sampling: Based on the theory of zonal sampling, equal intervals are divided laterally along the aggregate conveyor belt to generate sampling point coordinates. Sampling is completed at each sampling point. The samples collected at each sampling point are temporarily stored in the zonal hopper. After all sub-zone sampling is completed, the samples in the zonal hopper are sent to the mixer by the conveyor belt to be mixed to form the sample to be tested. Step S3, Sample Pretreatment: The mixed sample to be tested is sent into the pretreatment device, and the sample is rinsed by the backwash water flow device. At the same time, the rubbing drum is started to separate the sample from the impurities by rotating and rubbing. The separated sample is collected on the screen to remove excess water. S4. Detection: After pretreatment, the sample enters the vibrating feeding mechanism through the feeding tray, allowing the sample to fall evenly into the constant temperature chamber. The light source components of the multi-source imaging and laser-induced breakdown spectroscopy detection device are activated to irradiate the sample during the falling process. The imaging sensor acquires the sample image in real time and obtains image data. At the same time, the laser of the LIBS detection unit is triggered. The laser is focused onto the surface of the falling sample through the laser window probe to generate micro-plasma. The spectral fiber collects the plasma emission spectrum and transmits it to the spectrometer to obtain spectral data. S5. Data Processing: Process the image data and calculate the area ratio of mica particles; process the spectral data and extract the relative content of mica characteristic elements; fuse the image analysis results with the spectral analysis results to calculate the mica content.

[0018] Further, the image data processing method in step S5 is as follows: based on the HSV color space algorithm, the image is first converted to grayscale, and then the mica particles are separated from the fine aggregate background by threshold segmentation. The threshold is set according to the color difference between the fine aggregate and the mica. The H channel threshold is 30~60, the S channel threshold is 40~80, and the V channel threshold is 50~90. Finally, morphological processing is performed to remove noise points after segmentation, and the number, area and pixel ratio of mica particles are counted. The ratio of the total area of ​​mica particles to the total imaging area of ​​the sample is recorded. If the sample coverage rate in a single frame image is less than 70% or the number of mica particles identified is less than 5, the frame image is deemed invalid and discarded. The area ratio of mica particles in the valid images is then recalculated.

[0019] Furthermore, the method for spectral data processing in step S5 is as follows: using a multivariate correction algorithm based on polynomial basis functions, the spectrum is first corrected for baseline, the baseline is fitted using a 5th-order polynomial to deduct the influence of baseline drift, and then normalization is performed to standardize the spectral intensity to the range of 0~1. Subsequently, the spectral peak intensities of mica characteristic elements are extracted and compared with the built-in mica standard spectral library to calculate the relative content of mica characteristic elements in the sample. If the intensity of the spectral signal is below the threshold or the peak shape of the characteristic spectral line is abnormal, the set of spectral data is determined to be invalid, and adjacent valid data is used for interpolation to supplement it.

[0020] Due to the adoption of the above technical solution, the beneficial effects of the present invention are as follows: (1) This invention integrates laser-induced breakdown spectroscopy detection technology, multi-source imaging technology and physical separation technology to achieve rapid online detection of mica content, which greatly improves detection efficiency and accuracy, and is significantly better than traditional detection methods. It effectively solves the technical problems of long detection cycle and delayed results.

[0021] (2) This invention uses high-energy pulsed laser to excite micro-plasma and combines it with multivariate correction algorithm for spectral analysis, and uses multi-source imaging for dual-dimensional verification. This solves the problems of complex preprocessing and strong subjectivity in traditional detection, and eliminates the need for strong acid and strong alkali reagents, thus avoiding secondary pollution.

[0022] (3) This invention utilizes a vibration feeding mechanism and a multi-light source configuration to perform real-time imaging and excitation spectroscopy during sample descent, thereby improving the efficiency and accuracy of spectral acquisition. Through the synergistic pretreatment of backwash water flow and kneading action, mica particles are effectively separated from other impurities, reducing the complexity of imaging and spectral analysis.

[0023] (4) The present invention designs a sampling device based on the theory of partitioned sampling, realizes the function of automatic sampling from the production line, ensures the representativeness of the sample, can be directly embedded in the fine aggregate production line, and realizes the full-process automation of "intelligent sampling - sample pretreatment - online detection - data feedback". Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the mica content detection system provided by the present invention.

[0026] Figure 2 This is a schematic diagram of the laser unit arrangement of the mica content detection system provided by the present invention.

[0027] Figure 3 This is a schematic diagram of the laser window probe in this invention.

[0028] Reference numerals in the attached diagrams are as follows: 1. Sampling device; 101. Electric sampling shovel; 102. Pneumatic suction device; 103. Stirrer; 104. Zoned hopper; 105. Track; 106. Conveyor belt; 107. Aggregate conveyor belt; 2. Pretreatment device; 201. Backwash water device; 202. Kneading drum; 203. Screen; 204. Feed hopper; 3. Multi-source imaging and laser-induced breakdown spectroscopy detection device; 301. Feeding tray; 302. Vibrating feeding mechanism; 4. Constant temperature chamber; 501. LIBS detection unit; 502. Spectrometer; 503. Spectroscopic fiber; 601. First laser; 602. Second laser; 603. Third laser; 604. Fourth laser; 605. Fifth laser; 606. Sixth laser; 607. Imaging sensor; 608. Imaging excitation coordination area; 7. Laser window probe. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0030] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0031] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0032] This embodiment provides a complete implementation process for a mica content detection system and method based on multi-source imaging. For example... Figure 1As shown, the overall system structure mainly includes a sampling device 1, a pretreatment device 2, and a multi-source imaging and laser-induced breakdown spectroscopy detection device 3. The sampling device 1 is located on one side of the aggregate conveyor belt 107 in the sand and gravel processing production line. It is used to collect fine aggregate samples from various sections of the aggregate conveyor belt 107. After sampling from all sub-sections, the samples collected from each sampling point are stirred and mixed to form the sample to be tested. The pretreatment device 2 is connected after the sampling device 1 and is used to rinse and rub the sample to remove impurities. The multi-source imaging and laser-induced breakdown spectroscopy detection device 3 is connected after the pretreatment device 2. It is used to irradiate the sample during its descent using multiple wavelengths of light and acquire image data of the sample in real time; and to generate plasma by laser excitation of the sample and collect the spectrum emitted by the plasma to obtain spectral data. The following will provide a detailed description of each device in the system: I. Sampling Device The sampling device 1 includes an electric sampling shovel 101, a pneumatic suction device 102, a mixer 103, a partitioned hopper 104, and a track 105. The track 105 is located on one side of the aggregate conveyor belt 107, and the length of the track 105 is not less than 2 / 3 of the length of the aggregate conveyor belt 107. The electric sampling shovel 101 and the pneumatic suction device 102 are mounted on the track 105 and can slide along the track 105 to facilitate sampling at different points.

[0033] The electric sampling shovel 101 is used to collect coarse particle samples with a particle size greater than or equal to 0.613 mm. Its model number is SQ-300-AC220, and its sampling volume is 50g~75g. When it is adapted to a vibrating plate, the sampling volume is 75g and the stroke is 100mm~250mm; when it is adapted to a linear vibrating table, the sampling volume is 50g and the stroke is 50mm~300mm.

[0034] The pneumatic sampler 102 is used to collect fine particulate samples with a particle size of less than 0.613 mm. Its model is XY-500-0.6MPa, the sampling rate is 50 g / s, the single sampling volume is 75 g to 100 g, the lateral adjustment range is ±80 mm, and the backflush frequency is 5 Hz.

[0035] The electric sampling shovel 101 and the pneumatic suction device 102 work together to ensure that sample particles of different sizes can be effectively collected.

[0036] Multiple partitioned hoppers 104 are arranged in a straight line near the track 105. The partitioned hoppers 104 are used to temporarily store samples collected by the electric sampling shovel 101 and the pneumatic aspirator 102. Each partitioned hopper 104 is compatible with 6 partitions, with a single partition volume of 300mL, a vibration anti-clogging frequency of 50Hz, a discharge valve response time ≤20ms, and can accommodate 5-12 sampling points with a total sample weight ≥300g and a residue ≤0.5g. A conveyor belt 106 is installed directly below all the partitioned hoppers 104, with the discharge end of the conveyor belt 106 positioned directly above the agitator 103.

[0037] In this embodiment, the sampling device 1 is controlled by an embedded controller, model STM32H743VIT6, which is equipped with a sampling algorithm. Based on the theory of partitioned sampling, it can automatically calculate the optimal number and distribution of sampling points according to the size of the material batch to ensure the representativeness of the sample.

[0038] II. Pretreatment Device The pretreatment device 2 includes a backwash water flow device 201, a kneading drum 202, and a feed hopper 204. The outlet of the agitator 103 is located directly above the feed hopper 204, which receives the sample to be tested after mixing by the agitator 103. The outlet end of the feed hopper 204 is aligned with the kneading drum 202. The outlet pipe of the backwash water flow device 201 is aligned with the kneading drum 202. A screen 203 is installed below the outlet of the kneading drum 202 to receive the test sample after backwashing and kneading, and to filter the water from the test sample.

[0039] The backwash water flow device 201 uses a compressed air tank controlled by an electromagnetic valve, with a pressure range of 0.2MPa~0.6MPa and a pulse frequency of 5Hz~10Hz, to output a high-pressure pulsed water flow to rinse the sample. The kneading cylinder 202 adopts a cylindrical design with a diameter of 300mm or 400mm and a rotation speed of 1000rpm~1200rpm, and is used to separate the sample from impurities through rotational kneading action to improve sample purity.

[0040] The parameter selection principles for the backwash water flow device 201 and the kneading drum 202 are as follows: For fine particle samples with a particle size of less than 0.613 mm, a pressure of 0.3 MPa to 0.5 MPa and a rotation speed of 1200 rpm are used to avoid mica particle breakage; for coarse particle samples with a particle size between 0.613 mm and 4.75 mm, a pressure of 0.4 MPa to 0.6 MPa and a rotation speed of 1000 rpm are used to ensure the removal of deep impurities.

[0041] The working process of pretreatment device 2 is as follows: The mixed fine aggregate sample enters the feed hopper 204 through the agitator 103, and then enters the kneading drum 202. The backwash water flow device 201 is activated, outputting a high-pressure pulsed water flow to initially rinse the sample. The impact force of the water flow first removes loose impurities such as mud and dust adhering to the sample surface, and at the same time, the pulsed impact force breaks the adhesion between fine aggregate particles, causing the agglomerated particles to loosen and disperse, and the flaky mica particles to initially separate with the water flow.

[0042] After the backwash water flow is started for 1-2 minutes, the kneading drum 202 is activated. The rotating kneading drum 202 applies shear and compressive forces to the loosened sample, simulating a "kneading" motion. The friction generated by kneading further removes residual impurities from the mica surface and simultaneously dissociates mica fragments encased within other particles, preventing impurities from obscuring or affecting subsequent detection. The kneading pretreatment time is set to 5-8 minutes, during which the separation effect is continuously observed. If there are many residual impurities, the treatment time can be appropriately extended by 1-2 minutes.

[0043] After pretreatment, turn off the backwash water flow and the kneading drum 202, collect the separated sample onto the sieve 203, remove excess water, until the sample has no obvious free water.

[0044] III. Multi-source imaging and laser-induced breakdown spectroscopy detection device The multi-source imaging and laser-induced breakdown spectroscopy detection device 3 includes a constant temperature chamber 4, a feed tray 301, a vibrating feeding mechanism 302, a LIBS detection unit 501, a spectrometer 502, a spectral fiber 503, a light source assembly, an imaging sensor 607, and a laser window probe 7. The feed tray 301 is positioned below the sieve 203, and its outlet end is located above the inlet end of the vibrating feeding mechanism 302. The feed tray 301 is a conveyor belt with a 15° inclination angle to facilitate smooth sample drop. The outlet end of the vibrating feeding mechanism 302 is located directly above the top inlet of the constant temperature chamber 4. The vibrating feeding mechanism 302 is used to ensure uniform sample drop.

[0045] The constant temperature chamber 4 employs a double-layer structure, with an inner and outer heat-absorbing layer and a outer reflective layer, respectively, providing a stable temperature environment for the samples. Chamber 4 is equipped with a temperature control system, a humidity control system, and a pressure control system. The temperature control system has a control accuracy of ±1℃ or ±0.5℃, maintaining a stable ambient temperature through the coordinated operation of heating and cooling modules. The humidity control system uses a combination of a humidifier and a dehumidifier, with a relative humidity control range of 30%~90%. The pressure control system uses a pressure regulating valve controlled by an electromagnetic valve, with a pressure adjustment range of 0MPa~0.6MPa.

[0046] An imaging sensor 607 and multiple light source components are installed inside the constant temperature chamber 4. Each light source component is used to emit light alternately at a preset frequency to irradiate the sample from all directions during the falling process. The imaging sensor 607 acquires sample images in real time at a frame rate of 100Hz.

[0047] A laser window probe 7 is installed on the left side wall of the constant temperature chamber 4; a spectrometer 502 is installed on the right side wall of the constant temperature chamber 4, and a spectral fiber 503 is installed at the entrance pupil position of the spectrometer 502; the laser output from the laser in the LIBS detection unit 501 is focused onto the surface of the falling sample through the laser window probe 7 after beam shaping, beam combining, beam expanding and focusing, thereby exciting the sample to generate plasma; the spectral fiber 503 is used to collect the plasma emission spectrum in real time and transmit it to the spectrometer 502.

[0048] The multi-source imaging and laser-induced breakdown spectroscopy detection device 3 also includes a vortex booster pump, which is connected to the laser window probe 7 via an air pipe. The outlet end of the laser window probe 7 is configured as a conical opening 702. The vortex booster pump is used to pump high-pressure gas into the laser window probe 7, and the high-pressure gas is ejected from the conical opening 702 of the laser window probe 7 to purge steam and dust from the laser transmission path, ensuring the stability of laser transmission. The vortex booster pump is model HG-250, with a rated power of 250W and a maximum air pressure of 0.8Mpa.

[0049] In this embodiment, the sample falls in a constant temperature chamber 4 with an approximately parabolic trajectory, and the vertical drop height ranges from 80mm to 120mm, with a standard accuracy of 80mm and a high accuracy of 120mm. Different specifications of constant temperature chambers 4 are set according to the detection accuracy. The central region of the constant temperature chamber 4 is the imaging excitation coordination region 608.

[0050] In the normal precision mode, the dimensions of the constant temperature chamber 4 are length × width × height = 500 × 400 × 600 mm. The coordinate reference of the center point P of the imaging excitation coordination area 608 inside the constant temperature chamber 4 is X = 250 mm, Y = 200 mm, Z = 300 mm. The vibration feeding mechanism 302 adopts a vibratory plate (frequency 50 Hz, amplitude 2 mm), and the imaging sensor 607 adopts 1024 × 1024 pixels / 100 Hz frame rate.

[0051] In high-precision mode, the dimensions of the constant temperature chamber 4 are 600×500×700mm (length×width×height). The coordinate reference of the center point P of the imaging excitation coordination area 608 inside the constant temperature chamber 4 is X=300mm, Y=250mm, Z=350mm. The vibration feeding mechanism 302 adopts a linear vibration table (frequency 80Hz, amplitude 3mm), and the imaging sensor 607 adopts 2048×2048 pixels / 120Hz frame rate.

[0052] A collection trough is provided at the bottom of the constant temperature chamber 4 to collect falling samples; a buffer pad is provided in the collection trough.

[0053] In this embodiment, the vibratory feeding mechanism 302 ensures uniform sample drop. A vibratory feeder can be used for standard precision; a linear vibration table can be used for high precision. The vibratory feeder is designed with a frequency of 50Hz and an amplitude of 2mm; the linear vibration table is designed with a frequency of 80Hz and an amplitude of 3mm, selected based on sample shape and detection efficiency requirements. The imaging sensor 607 uses a high-resolution charge-coupled device with 1024×1024 pixels and a frame rate of 100Hz (adapted to the vibratory feeder), or 2048×2048 pixels and a frame rate of 120Hz (adapted to the linear vibration table), for real-time sample image acquisition.

[0054] In this embodiment, the spectrometer 502 employs an achromatic design with a wavelength range of 200nm to 2500nm. It can be a dual-pass optical system (spectral resolution 4cm²). -1 (Compatible with nitrogen lasers) or single-pass optical systems (spectral resolution 2cm) -1 (Compatible with neodymium glass lasers). Optical fiber 503 is a SIM-6102-0220 type optical fiber manufactured by Hangzhou Boyuan Optoelectronics Technology Co., Ltd., used to collect plasma emission spectra in real time and transmit them to spectrometer 502.

[0055] In this embodiment, the light source assembly employs a combination of lasers with multiple wavelengths to provide multi-band illumination, highlighting the optical characteristics of mica particles and laying the foundation for image recognition. The light source assembly includes lasers with wavelengths of 632.8 nm, 355 nm, 532 nm, and 1064 nm. Specifically, the light source assembly includes a first laser 601, a second laser 602, a third laser 603, a fourth laser 604, a fifth laser 605, and a sixth laser 606.

[0056] The first laser 601 and the second laser 602 are nitrogen lasers with a wavelength of 355nm.

[0057] The third laser 603 and the fourth laser 604 are nitrogen lasers with a wavelength of 532nm.

[0058] The fifth laser 605 is a He-Ne laser with a wavelength of 632.8 nm.

[0059] The sixth laser 606 is a nitrogen laser with a wavelength of 1064nm.

[0060] In this embodiment, the laser configured within the LIBS detection unit 501 serves as the excitation source, outputting high-energy, short-pulse laser pulses. The laser within the LIBS detection unit 501 is either a nitrogen laser or a neodymium glass laser. The nitrogen laser has an output wavelength of 1064 nm and a maximum output power of 200 W, while the neodymium glass laser has an output wavelength of 1064 nm and a maximum output power of 150 W. The LIBS detection unit 501 also includes a beam shaping mirror group, a laser beam combining mirror group, a laser shaping lens group, and a focusing lens group.

[0061] The beam shaping mirror assembly uses a combination of concave mirrors to shape the laser beam: if a nitrogen laser is used, a combination of double concave mirrors with a radius of 50 mm and a focal length of 500 mm is selected; if a neodymium glass laser is used, a combination of double concave mirrors with a radius of 75 mm and a focal length of 600 mm is selected.

[0062] The laser beam combiner consists of two plane mirrors, positioned 30mm (for nitrogen lasers) or 40mm (for neodymium glass lasers) from the laser output port.

[0063] The laser shaping lens assembly consists of two concave mirrors with a focal length of 100mm (for nitrogen lasers, expanding the laser beam angle from 5° to 20°) or 150mm (for neodymium glass lasers, expanding the laser beam angle from 8° to 25°).

[0064] The focusing lens group consists of two plane mirrors with a focal length of 200mm (for nitrogen lasers) or 250mm (for neodymium glass lasers) and is used to focus the laser beam.

[0065] Combination Figure 2 As shown, the coordinate "0" is set to the top left corner of the bottom of the constant temperature chamber 4. The specific spatial positions of each laser in the light source assembly are shown in Table 1: Table 1. Spatial Position Statistics of Each Laser in the Light Source Assembly

[0066] The spatial locations of various components in the LIBS detection unit 501 are shown in Table 2.

[0067] Table 2 Spatial Locations of Various Components in the LIBS Detection Unit

[0068] Note: ①100 / 120mm means that the standard accuracy is located 100mm to the left of point P, and the high accuracy is located 120mm to the left of point P; ②300 / 370mm indicates that the standard precision is located 300mm to the right of point P, and the high precision is located 370mm to the right of point P.

[0069] The lasers of each laser 601 to 606 and the lasers in the LIBS detection unit 501, as well as the focal points of the imaging sensor 607, all converge at the imaging excitation coordination region 608 (i.e., point P). The laser window probe 7, the focusing lens group, and the center of point P are aligned in a straight line with a deviation of ≤ ±1mm.

[0070] IV. Specific Implementation Methods of the Detection Method The detection method of the present invention will be described in detail below with reference to specific operating steps.

[0071] A method for detecting mica content based on multi-source imaging, employing the aforementioned multi-source imaging-based mica content detection system, includes the following steps: S1. Preliminary preparation: Check and debug the equipment of each unit of the system, and locate and calibrate the focus area.

[0072] Inspection and debugging of LIBS detection unit 501: Check the power connection and cooling system of the laser inside LIBS detection unit 501 to ensure there are no abnormalities; debug the beam shaping mirror group, laser beam combining mirror group, laser shaping mirror group and focusing lens group to ensure the laser beam path is accurate and finally focused on the entrance pupil of spectrometer 502.

[0073] Laser trigger timing adjustment: Place a small amount of sample and allow it to fall with the vibrating feeder 302. Record the time it takes for the particles to travel from the outlet to the focusing area using a high-speed camera. Adjust the trigger timing based on the vibration frequency of the vibrating feeder 302. For standard precision testing, if the time it takes for the particles to travel from the outlet to the focusing area is 20ms, adjust the laser trigger timing to 20ms~24ms after discharge, triggering it along the rising edge of the vibrating plate. For high precision testing, if the time is 16ms, adjust the trigger timing to 16ms~20ms after discharge, using a periodic triggering mode of 3 triggers every 4 cycles of a linear vibrating table.

[0074] Pulse frequency adjustment: Gradually increase the sample feed rate to bring the feeding speed to the rated range, and observe the particle falling density using a high-speed camera. For standard precision detection, if an average of one particle passes through the focusing area every 50ms, maintain a laser pulse frequency of 10Hz, matched with a vibration frequency of 1:5, to ensure that each pulse hits one effective particle. For high precision detection, if an average of one particle passes through every 32ms, maintain a pulse frequency of 15Hz, matched with a vibration frequency of 3:16, to avoid missed detections or simultaneous excitation of multiple particles.

[0075] Multi-source imaging unit debugging: Check the emission status of each laser 601~606 in the light source assembly to ensure that the wavelength and power of each light source are stable; debug the vibration feeding mechanism 302 to ensure smooth operation without jamming; calibrate the imaging sensor 607 and adjust the lens focal length and angle to ensure that the acquired image is clear and without distortion.

[0076] Debugging of pretreatment device 2: Check the pressure gauge and solenoid valve of backwash water flow device 201 to ensure stable output of high-pressure pulse water flow; debug kneading drum 202 and check the lubrication of rotating parts to ensure smooth kneading action.

[0077] Control and environmental adaptive module debugging: Start the temperature control system, humidity control system, and pressure control system to stabilize the environmental parameters inside the constant temperature chamber 4 within the target range. Check the laser window probe 7 and its matching vortex booster pump to ensure that the airflow can effectively blow away steam and dust in the laser transmission path.

[0078] Start the timing controller and set the delay parameters for laser emission and spectral collection; open the spectral analysis software and image processing software, import the standard parameter template for fine aggregate mica detection, and complete the software initialization.

[0079] Spectral acquisition timing optimization: Start the spectrometer 502 to acquire test data, observe the spectral signal intensity and peak shape, and adjust the spectral acquisition delay time and duration. For standard precision testing, if the spectral peak is strongest and noise is lowest with an initial delay of 1μs, maintain this parameter; if the peak shape is messy, the delay can be fine-tuned to 0.8μs~1.2μs. For high-precision testing, the default delay is 0.8μs and the duration is 15μs. If the characteristic spectral lines are not clear, the duration can be shortened to 12μs~18μs to reduce background noise interference.

[0080] Focusing on the area for positioning calibration. Start the vibratory feed mechanism 302 and run it in an unloaded (no sample) state. Turn on the laser in the LIBS detection unit 501 (reduce the output power to a safe threshold) and observe the laser spot position using a high-speed camera. For standard precision testing, adjust the focusing lens group to position the laser spot 150mm from the vibratory feeder outlet, with a vertical coverage of 10mm above and below. For high-precision testing, position the laser spot 200mm from the linear vibration table outlet, with a vertical coverage of 8mm above and below, ensuring the center of the laser spot is aligned with the imaging sensor 607's image capture center.

[0081] S2. Intelligent Sampling: Based on the theory of zonal sampling, the aggregate conveyor belt 107 is laterally divided into equal intervals, and sampling point coordinates are generated. Sampling is completed at each sampling point, and the samples collected at each sampling point are temporarily stored in the zonal hopper 104. After sampling of all sub-regions is completed, the samples in the zonal hopper 104 are sent to the mixer 103 via the conveyor belt 106 for mixing to form the sample to be tested. Specifically, the intelligent sampling control module of the sampling device 1 is activated. This module is designed based on the theory of zonal sampling, and its core hardware is an embedded controller (model STM32H743VIT6). It is equipped with a dedicated zonal sampling algorithm, which can automatically calculate the optimal number and distribution of sampling points according to the batch weight of fine aggregate to ensure the representativeness of the samples.

[0082] The sampling procedure is as follows: First, the sand and gravel processing production line control system automatically and synchronously receives basic parameters such as batch weight of fine aggregate, estimated particle size distribution range, and conveying speed. The embedded controller (model STM32H743VIT6) calls the preset algorithm to divide the sampling level based on the batch weight, and initializes the sampling area boundary according to the flow characteristics of fine aggregate (such as conveyor belt width, particle bulk density, etc.), dividing equal intervals laterally along the aggregate conveyor belt 107.

[0083] Then, based on the theory of partitioned sampling and the dispersion of mica distribution in historical detection data, the algorithm automatically calculates the minimum number of sampling points for each sub-region to ensure that the sample coverage meets the statistical confidence level. Sampling point coordinates are generated using a spatial grid algorithm, avoiding the edges of the transport zone laterally and evenly distributing them longitudinally according to sub-region intervals, with the distance between adjacent sampling points not less than 5 times the maximum particle size to prevent particle overlap interference.

[0084] Next, the embedded controller calculates the real-time position of the sampling point based on the speed of the aggregate conveyor belt 107, triggering the drive structure to operate according to a preset trajectory. Within each sub-area, the drive structure drives the electric sampling shovel 101 and the pneumatic suction sampler 102 to sequentially complete sampling at the calculated coordinate points. During the sampling process, the electric sampling shovel 101 is used to collect coarse particle samples with a particle size greater than or equal to 0.613 mm, and the pneumatic suction sampler 102 is used to collect fine particle samples with a particle size less than 0.613 mm, ensuring that all particles in the sample are collected. During the sampling process, the position is fed back in real time by a displacement sensor, and the controller corrects the action error in a closed loop.

[0085] Samples collected from each sampling point are temporarily stored in the partitioned hopper 104. After sampling is completed in all sub-regions of this batch, the discharge valves of each partitioned hopper 104 are opened, and the samples fall onto the conveyor belt 106 below. The conveyor belt 106 then transports the samples to the agitator 103 for 30 seconds of uniform mixing (mixing uniformity ≥95%, motor power 120W), forming the samples to be tested. After mixing is completed, the controller sends a signal to the subsequent pretreatment device 2 to synchronously start the pretreatment process, achieving a seamless connection between "sampling and pretreatment". At the same time, based on the subsequent test results, the number of sampling points and distribution weights for the next batch are dynamically adjusted to gradually improve the representativeness of the samples.

[0086] Step S3, Sample Pretreatment: The mixed sample to be tested is sent into the pretreatment device 2. The sample is rinsed by the backwash water flow device 201. At the same time, the rubbing drum 202 is started to separate the sample from the impurities by rotating and rubbing. The separated sample is collected on the sieve 203 to remove excess water.

[0087] In this embodiment, the air pressure range is set to 0.2MPa~0.6MPa, the pulse frequency range is 5Hz~10Hz, and the high-pressure pulsed water flow is output to preliminarily rinse the sample. The impact force of the water flow first removes loose impurities such as mud and dust attached to the sample surface, and at the same time, the pulse impact force breaks the adhesion between fine aggregate particles, so that the agglomerated particles are loosened and dispersed, and the flaky mica particles are initially separated by the water flow.

[0088] After the backwash water flow has been started for 1-2 minutes, the kneading drum 202 is activated. The rotation speed range is 1000rpm-1200rpm. The rotating kneading drum 202 applies shear and compressive forces to the loosened sample, simulating a "kneading" motion. The friction generated by kneading further removes residual impurities from the mica surface and simultaneously dissociates mica fragments encased within other particles, preventing impurities from obscuring or affecting subsequent detection. The kneading pretreatment time is set to 5-8 minutes, during which the separation effect is continuously observed. If there are many residual impurities, the treatment time can be appropriately extended by 1-2 minutes.

[0089] After pretreatment, turn off the backwash water flow and the kneading drum 202, collect the separated sample onto the sieve 203, remove excess water, until the sample has no obvious free water.

[0090] S4. Detection: After pretreatment, the sample enters the vibrating feeding mechanism 302 through the feeding tray 301, allowing the sample to fall evenly into the constant temperature chamber 4. The light source component of the multi-source imaging and laser-induced breakdown spectroscopy detection device 3 is activated to irradiate the sample during the falling process. The imaging sensor 607 acquires the sample image in real time, obtaining image data. Simultaneously, the laser of the LIBS detection unit 501 is triggered. The laser is focused onto the surface of the falling sample through the laser window probe 7, exciting the generation of micro-plasma. The spectral fiber 503 collects the plasma emission spectrum and transmits it to the spectrometer 502 to obtain spectral data. The specific process is as follows: The sieve 203 tilts in one direction, allowing the pretreated fine aggregate sample to fall onto the feed tray 301. The feed tray 301 is a conveyor belt with a 15° inclination. Under gravity, the sample slides down the feed tray 301 to the vibrating feeding mechanism 302. The vibrating feeding mechanism 302 (in this embodiment, a vibrating plate with a frequency of 50Hz and an amplitude of 2mm) is activated, and the feeding speed is set to allow the sample to fall evenly and freely into the constant temperature chamber 4.

[0091] Simultaneously, the lasers 601-606 of the light source assembly are activated, and each laser emits light alternately at a preset frequency to irradiate the sample from all directions during its descent. The imaging sensor 607 acquires sample images in real time at a frame rate of 100Hz, and each frame is saved in BMP format. The storage path is set to the "imaging data folder" specified by the system. The acquisition time continues until the sample has completely passed through the imaging area.

[0092] Simultaneously with the start of imaging acquisition, the laser in the LIBS detection unit 501 is triggered via a timing controller. The laser output, after beam shaping, combining, expanding, and focusing, is focused onto the surface of the falling fine aggregate sample through the laser window probe 7, exciting the sample to generate micro-plasma. The laser window probe 7 and its accompanying vortex booster pump operate continuously, outputting a strong airflow to promptly purge steam and dust from the laser transmission path, ensuring laser transmission stability. The spectrometer 502 collects the plasma emission spectrum in real time via optical fiber 503, converting the spectral data into electrical signals and transmitting them to the data processing unit. The acquisition interval is synchronized with the laser emission frequency, and the acquisition duration is consistent with the imaging acquisition. The spectral data is stored in CSV format in the "Spectral Data Folder".

[0093] During the testing process, the temperature, humidity and internal pressure of the constant temperature chamber 4 are continuously monitored. If the temperature fluctuation exceeds ±0.5℃, the humidity deviates from the target range of ±5%, or the pressure is abnormal, the corresponding adjustment system module will be automatically activated to make corrections.

[0094] S5. Data Processing: Process the image data and calculate the area ratio of mica particles; process the spectral data and extract the relative content of mica characteristic elements; fuse the image analysis results with the spectral analysis results to calculate the mica content.

[0095] The image data processing method is as follows: Open the image processing software and import all sample images from the "Imaging Data Folder". The software, based on the HSV color space algorithm, first performs grayscale processing on the images to eliminate the influence of uneven illumination; then, it separates mica particles from the fine aggregate background through threshold segmentation. The threshold is set according to the color difference between the fine aggregate and mica. In this embodiment, the H channel threshold is set to 30-60, the S channel threshold to 40-80, and the V channel threshold to 50-90. Finally, morphological processing is performed, including dilation, erosion, opening, and closing operations, to remove noise points after segmentation. The number, area, and pixel ratio of mica particles are counted, and an "Image Analysis Report" is generated, recording the ratio of the total area S1 of the mica particles to the total imaging area S0 of the sample. If the sample coverage in a single frame is less than 70% or the number of identified mica particles is less than 5, the frame is deemed invalid, discarded, and the area ratio of mica particles in the valid images is recalculated.

[0096] The spectral data processing method is as follows: The spectral analysis software is launched, and the spectral data in the "Spectral Data Folder" is imported. The software employs a multivariate correction algorithm based on polynomial basis functions. First, baseline correction is performed on the spectrum using a 5th-order polynomial to fit the baseline, subtracting the influence of baseline drift and eliminating background noise. Then, normalization is performed to standardize the spectral intensity to the range of 0-1, eliminating the influence of laser power fluctuations on the spectral signal. Subsequently, the spectral peak intensities of mica characteristic elements are extracted and compared with the software's built-in mica standard spectral library to calculate the relative content (g) of mica characteristic elements in the sample, generating a "Spectral Analysis Report." If the spectral signal intensity is below a threshold or the characteristic spectral line peak shape is abnormal, the set of spectral data is deemed invalid, and interpolation with adjacent valid data is used to supplement it.

[0097] The data fusion and computation methods are as follows: The system integrates the mica particle area ratio S1 / S0 from the "Image Analysis Report" with the relative content g of mica characteristic elements from the "Spectral Analysis Report," and calculates the mica content in the fine aggregate using a weighted average algorithm. The formula is as follows: In the formula: Q is the mica content, %; S0 is the total imaging area of ​​the sample, mm. 2 S1 represents the total area of ​​the mica particles, in mm. 2 ; g represents the relative content of mica characteristic elements; a1 represents the image data weight value; a2 represents the spectral data weight value.

[0098] The weight values ​​a1 and a2 are determined as follows: LIBS spectral detection and multi-source imaging identification are performed on each group of standard samples, and the detection is repeated 5 times. The relative standard deviations RSD1 and RSD2 are then calculated. In the formula: x i y represents the LIBS spectral detection value; i Values ​​for multi-source imaging identification.

[0099] The image data weight value a1 and the spectral data weight value a2 are calculated using the stability-first method, as shown in the following formula: In addition, the weight values ​​need to be adjusted if the following situations occur, as follows: (1) If the mica particles are easily dispersed, the imaging recognition accuracy will be improved, and a2 will increase by 5%~10%; otherwise, a2 will decrease by 5%~10%.

[0100] (2) LIBS spectral signals are weak but highly specific, and imaging is prone to missing detection, so a1 increases by 10%~15%.

[0101] (3) If the sample contains a lot of dust and impurities, the imaging is prone to misjudgment, then a2 decreases by 5% and a1 increases by 5%.

[0102] If the deviation between a single test result and historical data exceeds 5%, the test should be repeated 3 times, and the average of the 3 results should be taken as the final mica content test result.

[0103] In addition, the system needs to be calibrated using standard mica samples before testing. Take several groups of standard samples with known mica content and test them according to the above testing steps, obtaining the image area ratio and spectral characteristic intensity data for each group of standard samples. Establish a calibration curve with mica content as the dependent variable and image area ratio and spectral characteristic intensity as independent variables, and update the calibration parameters in the software. The calibration result must meet the requirement of an error ≤2%; otherwise, the equipment needs to be re-checked and calibrated again.

[0104] In this embodiment, the timing controller employs a high-precision delay generator via a PCI bus, with a selectable time resolution of 10 ns or 1 μs. Its function is to precisely control the delay between laser emission and spectral collection, ensuring the timeliness of spectral data acquisition. The spectral analysis software, equipped with a multivariate correction algorithm based on polynomial basis functions, performs baseline correction and normalization on the spectral data, extracting spectral information of mica characteristic elements to achieve qualitative and quantitative analysis. The image processing software, based on a mica recognition algorithm in the HSV color space, analyzes the imaging data through threshold segmentation and morphological processing to identify the morphology, size, and other features of mica particles, providing complementary verification to the spectral analysis results.

[0105] Industrial applicability: The mica content detection system and method based on multi-source imaging provided by this invention can be widely applied to the quality testing of fine aggregates in fields such as sand and gravel aggregate processing, concrete production, and mortar preparation. The system can be directly embedded into fine aggregate production lines, achieving fully automated testing from sampling to result output. A single analysis takes less than one minute, and the overall detection accuracy exceeds 95%, significantly improving testing efficiency and precision. It meets the industrial production requirements for "fast, accurate, online, and intelligent" detection of mica content in fine aggregates, and has good industrial practical value and promising prospects for widespread application.

[0106] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A mica content detection system based on multi-source imaging, characterized in that, include: The sampling device (1) is used to collect fine aggregate samples from each section of the aggregate conveyor belt (107) of the sand and gravel processing production line. After all sub-sections are sampled, the samples collected from each sampling point are stirred and mixed to form the sample to be tested. Pretreatment device (2) is used to rinse and rub the sample to be tested with backwash water to remove impurities adhering to the sample; The multi-source imaging and laser-induced breakdown spectroscopy detection device (3) is used to irradiate the sample during the falling process with light sources of multiple wavelengths and collect the image data of the sample in real time; and to generate plasma by laser excitation of the sample and collect the spectrum of plasma emission to obtain spectral data.

2. The mica content detection system based on multi-source imaging according to claim 1, characterized in that, The sampling device (1) includes an electric sampling shovel (101), a pneumatic suction device (102), a stirrer (103), a partitioned hopper (104), and a track (105). The track (105) is arranged on one side of the aggregate conveyor belt (107); The electric sampling shovel (101) and the pneumatic sampler (102) are mounted on the guide rail (105); Multiple partitioned hoppers (104) are arranged in a straight line and positioned close to the track (105). The partitioned hoppers (104) are used to temporarily store samples collected by the electric sampling shovel (101) and the pneumatic suction device (102). A conveyor belt (106) is installed directly below all the partitioned hoppers (104); the discharge end of the conveyor belt (106) is located directly above the agitator (103).

3. The mica content detection system based on multi-source imaging according to claim 2, characterized in that, The pretreatment device includes a backwash water flow device (201), a kneading drum (202), and a feed hopper (204). The outlet of the stirrer (103) is located directly above the feed hopper (204), which is used to receive the sample to be tested after being mixed by the stirrer (103). The outlet end of the feed hopper (204) is aligned with the kneading drum (202). The outlet pipe of the backwash water flow device (201) is aligned with the kneading drum (202). A screen (203) is provided below the outlet of the kneading drum (202) to receive the test sample after it has been rinsed and kneaded by the backwash water flow, and to filter the test sample.

4. The mica content detection system based on multi-source imaging according to claim 3, characterized in that, The multi-source imaging and laser-induced breakdown spectroscopy detection device (3) includes a constant temperature chamber (4), a feeding tray (301), a vibrating feeding mechanism (302), and a LIBS detection unit (501). The feed tray (301) is located below the screen (203), and the discharge end of the feed tray (301) is located above the feed end of the vibrating feeding mechanism (302). The discharge end of the vibrating feeding mechanism (302) is located directly above the top feed inlet of the constant temperature chamber (4); An imaging sensor (607) and multiple light source components are set in the constant temperature chamber (4). Each light source component is used to emit light alternately at a preset frequency to irradiate the sample in all directions during the falling process. The imaging sensor (607) acquires sample images in real time at a frame rate of 100Hz. A laser window probe (7) is installed on the left side wall of the constant temperature chamber (4); a spectrometer (502) is installed on the right side wall of the constant temperature chamber (4), and a spectral fiber (503) is installed at the entrance pupil position of the spectrometer (502). The laser output from the laser in the LIBS detection unit (501) is shaped, combined, expanded and focused, and then focused onto the surface of the falling sample through the laser window probe (7) to excite the sample to generate plasma. The spectral fiber (503) is used to collect plasma emission spectra in real time and transmit them to the spectrometer (502).

5. The mica content detection system based on multi-source imaging according to claim 4, characterized in that, It also includes a vortex booster pump, which is connected to the laser window probe (7) via an air pipe; the outlet end of the laser window probe (7) is set as a conical port (702); the vortex booster pump is used to pump high-pressure gas into the laser window probe (7), and the high-pressure gas is ejected from the conical port (702) of the laser window probe (7) to purge steam and dust on the laser transmission path.

6. The mica content detection system based on multi-source imaging according to claim 4, characterized in that, A collection trough is provided at the bottom of the constant temperature chamber (4) for collecting falling samples; a buffer pad is provided in the collection trough.

7. The mica content detection system based on multi-source imaging according to claim 4, characterized in that, The light source assembly includes lasers with wavelengths of 632.8 nm, 355 nm, 532 nm, and 1064 nm.

8. A method for detecting mica content based on multi-source imaging, employing the mica content detection system based on multi-source imaging as described in claim 5, characterized in that, Includes the following steps: S1. Preliminary preparation: Inspect and debug each unit of the system, and locate and calibrate the focusing area; S2. Intelligent sampling: Based on the theory of zonal sampling, equal intervals are divided laterally along the aggregate conveyor belt (107) to generate sampling point coordinates. Sampling is completed at each sampling point. The samples collected at each sampling point are temporarily stored in the zonal hopper (104). After all sub-zone sampling is completed, the samples in the zonal hopper (104) are sent to the mixer (103) via the conveyor belt (106) for mixing to form the sample to be tested. Step S3, Sample pretreatment: The mixed sample to be tested is sent into the pretreatment device (2), and the sample is rinsed by the backwash water flow device (201). At the same time, the rubbing drum (202) is started to separate the sample from the impurities by rotating and rubbing. The separated sample is collected on the sieve (203) to remove excess water. S4. Detection: The pre-treated sample enters the vibrating feeding mechanism (302) through the feeding tray (301) and falls evenly into the constant temperature chamber (4). The light source component of the multi-source imaging and laser-induced breakdown spectroscopy detection device (3) is activated to irradiate the sample during the falling process. The imaging sensor (607) collects the sample image in real time and obtains the image data. At the same time, the laser of the LIBS detection unit (501) is triggered. The laser is focused on the surface of the falling sample through the laser window probe (7) to generate micro-plasma. The spectral fiber (503) collects the plasma emission spectrum and transmits it to the spectrometer (502) to obtain the spectral data. S5. Data Processing: Process the image data and calculate the area ratio of mica particles; process the spectral data and extract the relative content of mica characteristic elements; fuse the image analysis results with the spectral analysis results to calculate the mica content.

9. The method for detecting mica content according to claim 8, characterized in that, The image data processing method in step S5 is as follows: Based on the HSV color space algorithm, the image is first converted to grayscale, and then the mica particles are separated from the fine aggregate background by threshold segmentation. The threshold is set according to the color difference between the fine aggregate and the mica. The H channel threshold is 30~60, the S channel threshold is 40~80, and the V channel threshold is 50~90. Finally, morphological processing is performed to remove noise points after segmentation, and the number, area and pixel ratio of mica particles are counted. The ratio of the total area of ​​mica particles to the total imaging area of ​​the sample is recorded. If the sample coverage rate in a single frame image is less than 70% or the number of mica particles identified is less than 5, the frame image is deemed invalid and discarded. The area ratio of mica particles in the valid images is then recalculated.

10. The method for detecting mica content according to claim 8, characterized in that, The method for spectral data processing in step S5 is as follows: a multivariate correction algorithm based on polynomial basis functions is used to first perform baseline correction on the spectrum, then use a 5th-order polynomial to fit the baseline to subtract the influence of baseline drift, and then perform normalization to standardize the spectral intensity to the range of 0~1. Subsequently, the spectral peak intensity of mica characteristic elements is extracted and compared with the built-in mica standard spectral library to calculate the relative content of mica characteristic elements in the sample. If the intensity of the spectral signal is below the threshold or the peak shape of the characteristic spectral line is abnormal, the set of spectral data is determined to be invalid, and adjacent valid data is used for interpolation to supplement it.

Images