A rapid detection method for rice quality based on double-pulse differential LIBS

Abstract

The application belongs to the technical field of nondestructive detection of agricultural products, and discloses a rice quality rapid detection method based on double-pulse differential LIBS, which comprises the following steps: a detection system including a double-laser differential detection module, an AOTF, a gated PMT and an FPGA is configured for a sample; full-index detection is decomposed into multiple continuous detection periods, and in each detection period, a series of operations such as setting a target wavelength-triggering a first laser pulse-a first predetermined delay-collecting a reference background signal-a predetermined inter-pulse delay-triggering a second laser pulse-a second predetermined delay-collecting an analysis mixed signal-calculating a net characteristic signal of the current detection period are sequentially performed; and comprehensive evaluation of rice quality is jointly performed in combination with all net characteristic signals. Through the application, continuous background interference in LIBS detection can be effectively inhibited, and detection results including heavy metal element content and other physical defect information in rice can be accurately obtained, so that rapid and integrated discrimination of rice quality is realized.

Description

Technical Field

[0001] This invention belongs to the technical field of non-destructive testing of agricultural products, and more specifically, relates to a rapid detection method for rice quality based on dual-pulse differential LIBS. Background Technology

[0002] Currently, rice quality testing mainly covers two aspects: first, chemical contamination, such as heavy metals like cadmium (Cd) and lead (Pb); and second, physical defects, such as mold, insect damage, and yellowing. Traditional testing methods typically rely on different analytical equipment: for example, using atomic absorption spectrometry (AAS) or inductively coupled plasma mass spectrometry (ICP-MS) for elemental analysis, while simultaneously using machine vision or near-infrared spectroscopy (NIRS) for identifying external and internal defects. This multi-device cascade testing mode suffers from drawbacks such as cumbersome procedures, low efficiency, high costs, and large sample consumption, and it also cannot perform multi-dimensional correlation analysis on the same sample point.

[0003] Laser-induced breakdown spectroscopy (LIBS), as an emerging atomic emission spectrometry technique, has the potential for simple sample pretreatment, fast analysis speed, simultaneous multi-element analysis, and minimally invasive or even non-destructive testing, and has been explored for application in agricultural product testing. However, directly applying conventional LIBS technology to the comprehensive grade testing of rice still faces several key technical bottlenecks:

[0004] I. Limited Functionality and Insufficient Information Dimensions: Existing LIBS technologies and applications mostly focus on elemental composition analysis, making it difficult to simultaneously and effectively acquire information closely related to physical defects such as mold, insect damage, and yellowing. Therefore, there is a lack of a one-stop solution that can simultaneously address the challenges of elemental and physical defect detection within a single system. For example, patent document CN106546575A only detects a single heavy metal element (copper) in rice, without addressing physical defects; patent document CN104374752B focuses on crop nutrients and similarly does not integrate physical defect identification functionality.

[0005] II. Strong Continuous Background Radiation Interference: Studies have found that LIBS plasma exhibits extremely strong bremsstrahlung continuous background radiation within the first 1-2 microseconds of its generation. This continuous background severely obscures the weak atomic / ionic characteristic spectral lines, especially those of trace heavy metals, resulting in a very poor signal-to-noise ratio (SNR) and a detection limit (LOD) that fails to meet food safety standards. Although patent document CN102798625A proposes using a polarizer to filter the continuous background, its effectiveness depends on fine adjustment of the polarization angle and cannot achieve dynamic background subtraction over time, limiting its applicability and flexibility.

[0006] III. Limitations in Detection Efficiency: The study also found that while spectrometers equipped with CCD / ICCD can acquire full-spectrum information across a wide wavelength range in a single operation, their inherent charge accumulation and readout processes result in millisecond-level readout times. This makes it impossible to complete two independent, high-speed, and time-precise signal acquisitions of the same spatial point and the same characteristic wavelength within a single laser cycle. Consequently, it cannot achieve the stringent time-series differential detection required by this invention to physically subtract early continuous background. Furthermore, while the combination of fixed filters and photomultiplier tubes (PMTs) provides high-speed time response capabilities, this approach typically relies on a pre-defined set of fixed-wavelength filters and corresponding PMT detectors. This structure cannot flexibly and quickly switch to any selected characteristic wavelength, making it difficult to adapt to the complex detection process requirements of multiple indicators and reconfigurability.

[0007] Accordingly, there is an urgent need in this field to conduct research and improvements in order to better solve the aforementioned technical problems. Summary of the Invention

[0008] In view of the above-mentioned defects or needs of the existing technology, the purpose of this invention is to provide a rapid rice quality detection method based on dual-pulse differential LIBS. The method uses dual lasers to perform differential detection and redesigns the entire process and working mechanism to effectively suppress continuous background interference in LIBS detection and accurately obtain detection results of rice, including the content of heavy metal elements and other physical defects, so as to achieve rapid and integrated judgment of rice quality.

[0009] Based on this, the present invention also specifically introduces acousto-optic tunable filters (AOTF), gated photomultiplier tubes (PMT) and field-programmable gate arrays (FPGA) to work together with the above-mentioned dual lasers to perform the entire differential detection process. This ensures the execution of simultaneous multi-wavelength sequence detection at the same point, realizing the synchronous and homogeneous strategies for multiple preparations at a single physical point, which significantly improves the consistency and reliability of the data.

[0010] To achieve the above objectives, according to the present invention, a rapid rice quality detection method based on dual-pulse differential LIBS is provided, characterized in that the method includes the following steps:

[0011] S1. For the rice sample being tested, a detection system is configured including a dual-laser differential detection module, an acousto-optic tunable filter, a gated photomultiplier tube, and a field-programmable gate array, wherein:

[0012] The dual-laser differential detection module includes a first pulse laser and a second pulse laser that are independent of each other and have different laser energies. They are used to output laser pulses to act on the surface of the rice sample or its surrounding gas to generate different plasmas. The gated photomultiplier tube is used to collect the analysis signals corresponding to the plasmas generated by the first pulse laser and the second pulse laser respectively. The acousto-optic tunable filter is used to set multiple different transmission wavelengths. The field-programmable gate array is used to provide unified timing control for the entire detection system.

[0013] S2. The full-index detection of the rice sample is decomposed into multiple consecutive detection cycles, and each detection cycle corresponds to a preset characteristic wavelength; wherein, within each detection cycle, the following operations are performed:

[0014] S21. The acousto-optic tunable filter sequentially selects one of a plurality of different transmission wavelengths and sets the transmission wavelength as the target wavelength of the current detection cycle.

[0015] S22. The field-programmable gate array triggers the first pulse laser in the dual-laser differential detection module to output a first laser pulse, which ablates the ambient gas near a detection point on the surface of the rice sample and generates a first plasma.

[0016] S23. At a first predetermined delay ΔT1 after the first laser pulse, the first gate τ1 of the gated photomultiplier tube is activated, and the reference background signal I_ corresponding to the first plasma is acquired under the filtering effect of the target wavelength. ref ;

[0017] S24. A predetermined inter-pulse delay ΔT_ after triggering the first laser pulse. inter At the location, the field-programmable gate array triggers the second pulse laser in the dual-laser differential detection module to output a second laser pulse. The laser energy of the second laser pulse is greater than that of the first laser pulse and fully ablates and vaporizes the detection point on the surface of the rice sample, generating a second plasma.

[0018] S25. At the second predetermined delay ΔT2 after the second laser pulse, the second gate τ2 of the gated photomultiplier tube is activated, and the analytical mixed signal I_ corresponding to the second plasma is acquired under the filtering effect of the target wavelength. ana ;

[0019] S26, regarding the reference background signal I_ ref and the analysis of the mixed signal I_ ana The net feature signal S_ for the current detection period is obtained by real-time calculation using the following difference formula. net:

[0020] S_ net =I_ ana -k×I_ ref , where k represents the preset background subtraction coefficient;

[0021] S3, the net feature signal S_ obtained by combining all detection cycles. net Together, we will conduct a comprehensive assessment of rice quality, thereby completing a rapid testing process for rice quality.

[0022] As a further preferred embodiment of the present invention, in step S1, the first pulsed laser is an Nd:YAG laser with a wavelength of 532nm and an energy range of 5mJ to 15mJ, and a repetition frequency of 10Hz to 100Hz; the second pulsed laser is an Nd:YAG laser with a wavelength of 1064nm and an energy range of 50mJ to 150mJ, and a repetition frequency of 10Hz to 100Hz.

[0023] As a further preferred embodiment of the present invention, in step S1, the spectral response range of the gated photomultiplier tube covers 200nm to 900nm, and the minimum gate width is ≤100 nanoseconds; the operating wavelength of the acousto-optic tunable filter is 400nm to 1400nm, and the wavelength switching time is <1 millisecond, and the diffraction efficiency is >70%; the internal clock frequency of the field-programmable gate array is ≥200MHz.

[0024] As a further preferred embodiment of the present invention, step S2 further includes processing the net feature signal S_ net Perform signal quality assessment operations:

[0025] This operation includes processing each of the net feature signals S_ net The signal-to-noise ratio (SNR) is calculated separately and then compared with a preset signal quality threshold. When the SNR is greater than or equal to the signal quality threshold, the detection result of the current detection cycle is deemed qualified, the data is stored, and the next detection cycle is automatically started. When the SNR is less than the signal quality threshold, the detection result of the current detection cycle is deemed unqualified, and the operation of the current detection cycle is re-executed.

[0026] As a further preferred embodiment of the present invention, in step S2, for each detection cycle, the detection point of the rice sample remains stationary; after the current detection cycle is completed, a three-dimensional electric motion platform is used to make micro-motion, so that the laser focus acts on another new detection point on the surface of the rice sample or the ambient gas near it.

[0027] As a further preferred embodiment of the present invention, in step S2, the target wavelength is selected from at least four of the following wavelengths: 643.8 nm, 405.8 nm, 766.5 nm, 769.9 nm and 589.0 nm.

[0028] As a further preferred embodiment of the present invention, in step S2, the first predetermined delay ΔT1 and the second predetermined delay ΔT2 are set to 1.5 microseconds, and the first gate τ1 and τ2 are set to 500 nanoseconds; the predetermined inter-pulse delay ΔT_ inter Set to 50 microseconds.

[0029] As a further preferred embodiment of the present invention, in step S3, the net feature signal S_ of all detection cycles is... net After all these are calculated, these net characteristic signals S_ net The feature vector is directly input into the trained machine learning model to obtain the final detection result; the machine learning model includes random forest, support vector machine or artificial neural network.

[0030] In summary, the technical solutions conceived by this invention have the following main technical advantages compared with the prior art:

[0031] (1) By employing dual lasers to perform differential detection and redesigning the entire process and working mechanism, this invention not only directly and effectively suppresses continuous background interference in LIBS detection from a physical perspective, making it possible to perform high-sensitivity detection of trace heavy metal elements (such as Cd and Pb) in the early stage of plasma, but also significantly reduces the detection limit. It can accurately obtain the detection results of rice, including the content of heavy metal elements and other physical defects, without being limited by the polarization angle, and realize rapid and integrated judgment of rice quality.

[0032] (2) The present invention further introduces AOTF, gated PMT and FPGA to work together with the above dual lasers to perform the entire differential detection process. Among them, through the fast wavelength switching of AOTF, flexible and comprehensive acquisition is realized, and the information coverage is wide. Correspondingly, on a single LIBS hardware platform, it can still solve the detection problems from trace element contamination to various physical defects (such as mold, insect holes, yellowing, etc.) synchronously and from the same source. The gated PMT has high-speed gating capability and high time resolution. Compared with components such as CCD / ICCD, it can ensure the high-speed signal acquisition required within the same detection cycle. In addition, the collaborative work of FPGA and the above functional components further ensures the speed, accuracy and automation of the entire detection process.

[0033] (3) The overall detection method of the present invention is easy to operate, has high working efficiency, can effectively suppress the initial background noise of LIBS, and realize high-speed differential acquisition at the same point and wavelength. Accordingly, it can simultaneously detect the content of multiple key elements, realize the integrated and high-accuracy detection of rice quality, and is therefore particularly suitable for complex detection process requirements with multiple indicators and reconfigurability. Attached Figure Description

[0034] Figure 1 This is the overall process flow diagram of the rapid rice quality detection method based on dual-pulse differential LIBS according to the present invention;

[0035] Figure 2 This is a schematic diagram of the overall structure of a supporting detection system designed according to a preferred embodiment of the present invention;

[0036] Figure 3 This is a detailed timing diagram illustrating the execution of dual-pulse differential detection within a single detection cycle of the present invention;

[0037] Figure 4 This is a flowchart used to exemplify the explanation of the multi-detection period sequence and data fusion decision according to the present invention;

[0038] In all the accompanying drawings, the same reference numerals are used to denote the same elements or structures, wherein:

[0039] 1a-First pulse laser; 1b-Second pulse laser; 2-Laser reflector; 3-Dichroic mirror; 4-Focusing lens; 5-Rice sample; 6-Three-dimensional electric motion platform; 7-Acquisition probe; 8-Fiber optic; 9-Acousto-optic tunable filter; 10-Gated photomultiplier tube; 11-Field programmable gate array; 12-Data acquisition card; 13-Computer. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0041] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" 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 this invention and 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 this invention.

[0042] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0043] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; 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; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0044] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0045] Figure 1 This is a flowchart illustrating the overall process flow of the rapid rice quality detection method based on dual-pulse differential LIBS according to the present invention. See below for further details. Figure 1 To explain in more detail.

[0046] like Figure 1 As shown, the rapid rice quality detection method of the present invention mainly includes the following steps:

[0047] Step 1: Configuration of the dual-laser differential detection module and other functional modules.

[0048] In this step, a detection system is configured for the rice sample being tested, including a dual-laser differential detection module, an acousto-optic tunable filter, a gated photomultiplier tube, and a field-programmable gate array.

[0049] The dual-laser differential detection module includes a first pulse laser and a second pulse laser that are independent of each other and have different laser energies. They are used to output laser pulses to act on the surface of the rice sample or its surrounding gas to generate different plasmas. The gated photomultiplier tube is used to collect the analysis signals corresponding to the plasmas generated by the first pulse laser and the second pulse laser respectively. The acousto-optic tunable filter is used to set multiple different transmission wavelengths. The field-programmable gate array is used to provide unified timing control for the entire detection system.

[0050] Step 2: Execution process for multiple detection cycles.

[0051] First, the full-index detection of the rice sample is broken down into multiple consecutive detection cycles, with each detection cycle corresponding to a preset characteristic wavelength; within each detection cycle, the following operations are performed:

[0052] Next, the acousto-optic tunable filter sequentially selects one of a plurality of different transmission wavelengths and sets that transmission wavelength as the target wavelength for the current detection cycle;

[0053] Next, the field-programmable gate array triggers the first pulse laser in the dual-laser differential detection module to output a first laser pulse. This first laser pulse ablates the ambient gas near a detection point on the surface of the rice sample and generates a first plasma.

[0054] Next, at a first predetermined delay ΔT1 after the first laser pulse, the first gate τ1 of the gated photomultiplier tube is activated, and the reference background signal I_ corresponding to the first plasma is acquired under the filtering effect of the target wavelength. ref ;

[0055] Next, a predetermined inter-pulse delay ΔT_ is applied after triggering the first laser pulse. inter At the location, the field-programmable gate array triggers the second pulse laser in the dual-laser differential detection module to output a second laser pulse. The laser energy of the second laser pulse is greater than that of the first laser pulse and fully ablates and vaporizes the detection point on the surface of the rice sample, generating a second plasma.

[0056] Next, at a second predetermined delay ΔT2 following the second laser pulse, the second gate τ2 of the gated photomultiplier tube is activated, and the analytical mixed signal I_ corresponding to the second plasma is acquired under the filtering effect of the target wavelength. ana ;

[0057] Finally, the reference background signal I_ ref and the analysis of the mixed signal I_ ana The net feature signal S_ for the current detection period is obtained by real-time calculation using the following difference formula. net :

[0058] S_ net =I_ ana -k×I_ ref , where k represents the preset background subtraction coefficient.

[0059] Step 3: Results of rice quality testing.

[0060] In this step, the net feature signal S_ obtained from all detection cycles can be combined. net Together, we will conduct a comprehensive assessment of rice quality, thereby completing a rapid testing process for rice quality.

[0061] Based on the above concepts, this invention constructs a "time-driven, wavelength-scanning" detection process, whose basic working unit is defined as a detection cycle. A detection cycle refers to a complete double-pulse laser excitation and signal acquisition process performed to obtain a high signal-to-noise ratio differential signal at a specific wavelength. Accordingly, the full-index detection of rice can be achieved by continuously executing multiple detection cycle sequences targeting different characteristic wavelengths.

[0062] More specifically, the working mechanism of the dual-laser differential detection and background suppression involved in this invention can be explained as follows: The first laser pulse serves as a pre-ablation, for example, an Nd:YAG laser (with a frequency-doubled output wavelength of 532 nm) can be used to output a low-energy (5-15 mJ) laser pulse. This pulse can act on the ambient gas near a detection point on the surface of the rice sample (e.g., directly above), generating an initial plasma. In the early stages of evolution (approximately 1-2 μs), this plasma mainly consists of surface contaminants, spectral lines of N and O elements in the air, and a strong continuous bremsstrahlung background. Due to the extremely small amount of ablation, the characteristic spectral signals of heavy metal elements (such as Cd and Pb) inside the rice matrix are so weak as to be negligible. Therefore, the signal acquired at a specific wavelength during this stage can be considered as a nearly pure reference background signal (I_ ref It should be noted that, as an alternative, the first laser pulse can also be used to slightly ablate the surface of the rice sample, and a similar reference background signal can be obtained.

[0063] The second laser pulse is the primary ablation laser, such as an Nd:YAG laser (fundamental output wavelength 1064nm), which outputs a high-energy (50-150 mJ) laser pulse. This pulse is emitted when the first plasma has fully expanded and cooled but has not yet completely dissipated (inter-pulse delay ΔT_ inter The high-energy pulse (preferably around 10-50 μs) is precisely applied to the same detection point. This high-energy pulse thoroughly ablates and vaporizes the rice sample, producing strong characteristic spectral lines containing rice matrix elements (such as K) and heavy metal contaminants (Cd, Pb), while also including a strong continuous background in the analytical plasma. The signal acquired during this stage is an analytical mixed signal (I_ ana ), i.e. I_ ana = I_ element + I_ background .

[0064] The reason for the difference in energy between the two laser beams is as follows: the first pulse uses low energy to minimize sample consumption and create a reference signal source dominated by background radiation; the second pulse uses high energy to ensure that the sample is fully excited, thereby obtaining elemental characteristic spectral line signals of sufficient intensity. The energy difference between the two is the physical premise of differential detection. If the energies are the same, the elemental signals will be subtracted in subsequent differential operations.

[0065] In this context, the present invention further utilizes the fast time response and gating capability of a gated photomultiplier tube (PMT) to acquire I_ under the precise control of a field-programmable gate array (FPGA) at the same detection period and the same characteristic wavelength point, respectively after the first pulse and the second pulse, for example, by using the same acquisition delay and gating width. ref and I_ ana Subsequently, through real-time difference calculation: S_ net = I_ ana -k×I_ ref (where k is the background subtraction coefficient determined by system calibration, usually close to 1), directly subtracting the continuous background from the original mixed signal to obtain a net feature signal (S_net) with high signal-to-noise ratio.

[0066] Accordingly, compared with existing technologies that use polarizers or dual-pulse heating without temporal differential, the background suppression mechanism of this invention is based on physical subtraction of temporal differential and energy difference, rather than relying on the polarization characteristics of optical elements or simple signal enhancement, thereby achieving accurate and flexible continuous background removal on the time scale.

[0067] Figure 2 This is a schematic diagram of the overall structure of a supporting detection system designed according to a preferred embodiment of the present invention. The following will be combined with... Figure 2To explain in more detail the basic structure and components of the detection system.

[0068] like Figure 2 As shown, the first pulsed laser 1a is preferably an Nd:YAG laser with a wavelength of 532 nm and an energy range of 5 mJ to 15 mJ, with a repetition frequency of 10 Hz to 100 Hz. The second pulsed laser 1b is preferably an Nd:YAG laser with a wavelength of 1064 nm and an energy range of 50 mJ to 150 mJ, with a repetition frequency of 10 Hz to 100 Hz. More specifically, the two lasers are arranged in optical path order and preferably collinear through a laser reflector 2 and a dichroic mirror 3. After collinearity, the laser beam is precisely focused onto the rice sample 5 placed on the three-dimensional electric motion platform 6 after passing through a focusing lens 4. Furthermore, the plasma emission light is preferably designed to be collected by the acquisition probe 7 and coupled into the optical fiber 8.

[0069] like Figure 2 As shown, the optical signal output from fiber 8 can be designed to be incident on an acousto-optic tunable filter 9 equipped with a driver. The acousto-optic tunable filter 9 receives wavelength setting instructions from the field-programmable gate array 11, and the monochromatic light diffracted by the acousto-optic tunable filter 9 is detected by a gated photomultiplier tube 10 through fiber 8. The gated photomultiplier tube 10 can be powered by a programmable high-voltage power supply, and its gating signal is directly provided by the field-programmable gate array 11.

[0070] Similarly, Figure 2 As shown, the field-programmable gate array 11 can send all synchronization signals through digital I / O ports. Preferably, a data acquisition card 12 can be used to acquire the analog voltage signal output by the gated photomultiplier tube 10 and digitize it. The computer 13 is responsible for communicating with the field-programmable gate array 11, processing the acquired data, executing machine learning models, and displaying the final results.

[0071] This invention utilizes the rapid wavelength switching of an optically oriented transducer (AOTF), with an operating band covering 400-1400 nm, to ensure the transmission of spectral lines containing key signals in rice. The wavelength switching time is <1 ms, the diffraction efficiency is >70%, and it can be equipped with an independent AOTF driver. Correspondingly, compared to existing solutions using spectrometers, the AOTF provides rapid and flexible wavelength selection capabilities, allowing switching to any characteristic wavelength in a short time without mechanical moving parts, making it particularly suitable for sequential detection of multiple indicators.

[0072] This invention also employs a gated PMT, which possesses high-speed gating capability and high time resolution. Compared to components like CCDs / ICCDs, it ensures the necessary high-speed signal acquisition within the same detection cycle. Furthermore, as the synchronization and control core of the entire detection system, a Xilinx Artix-7 or Altera Cyclone V-series FPGA is preferably used. Its internal clock frequency is ≥ 200 MHz, equipped with multi-bit high-speed counters, and outputs multiple synchronous trigger signals through LVDS or LVTTL interfaces, thereby further ensuring the speed, accuracy, and automation of the entire detection process.

[0073] According to a preferred embodiment of the present invention, the full-index detection process is performed on the same micro-region of the sample. The system first moves a specific point on the sample to be tested to the laser focal point using a three-dimensional motion platform and keeps it fixed. Then, while the point remains stationary, the timing control system (FPGA) sequentially executes multiple detection cycles. Each cycle switches to a different characteristic wavelength via an AOTF and completes a dual-pulse differential detection at the same location. This ensures that the net signals of all characteristic wavelengths originate from the exact same sample volume, thereby ensuring physical consistency and reliability for subsequent data fusion and comprehensive grade determination. Only after completing the full-wavelength detection of one point will the motion platform move to the next test point.

[0074] Correspondingly, compared with the existing technologies that use single-point single-wavelength or fixed multi-wavelength acquisition, or use mobile platforms to acquire multiple points, the present invention achieves synchronous and homogeneous measurement of multiple indicators (elements and physical defects) at a single physical point through FPGA-controlled "same point-multi-wavelength" sequence detection, further improving the consistency and reliability of the data.

[0075] According to another preferred embodiment of the present invention, the detection process further includes processing the net feature signal S_ net The operation of performing signal quality assessment includes evaluating each of the net feature signals S_ net Calculate the signal-to-noise ratio (SNR) of each device and compare it with a preset signal quality threshold (e.g., 10). When the SNR is greater than or equal to the signal quality threshold, the detection result of the current detection cycle is deemed acceptable, the data is stored, and the device automatically enters the next detection cycle. When the SNR is less than the signal quality threshold, the detection result of the current detection cycle is deemed unacceptable, and the operation of the current detection cycle is re-executed.

[0076] More specifically, when the test results for the current testing cycle are deemed unreasonable, one of the following instructions may be taken, depending on the specific circumstances:

[0077] The three-dimensional electric motion platform is driven to make micro-motion, so that the laser focus acts on a new position on the sample surface;

[0078] Fine-tune the PMT gate delay ΔT1 for subsequent acquisitions (e.g., scan in 100 ns increments) to find the optimal signal acquisition window for plasma evolution;

[0079] Alternatively, the acquisition process for the current detection cycle can be re-executed at the current detection point (with a maximum number of retries, such as 3 times).

[0080] It should be noted that the above-mentioned actions of the displacement platform only occur during the retry process of a single detection cycle. Specifically: when the net feature signal (signal-to-noise ratio) obtained within the detection cycle of the current target wavelength does not meet the requirements, the system will instruct the three-dimensional motion platform to move a tiny distance, thereby re-executing the detection for wavelength λ at a fresh position on the sample. x The detection cycle is specified. This mechanism aims to overcome interference caused by sample surface inhomogeneities and does not mean that each wavelength is sampled at a different point. Once the signal quality of the current target wavelength meets the standard, the system will continue to perform the detection cycle for subsequent wavelengths at the original point. The above process constitutes a simple closed-loop feedback system, realizing the system's self-verification and adaptive optimization, and improving the overall reliability and robustness.

[0081] According to another preferred embodiment of the present invention, after the net feature signal S_net for all detection cycles has been calculated, these net feature signals S_net are preferably directly input as feature vectors into the trained machine learning model to obtain the final detection result; the machine learning model includes random forest, support vector machine or artificial neural network.

[0082] More specifically, the algorithm first collects net elemental signals from all detection cycles. Then, feature extraction and normalization are performed, and a mold characteristic index is calculated. Finally, the constructed comprehensive feature vector is input into a pre-trained machine learning model (e.g., random forest, support vector machine, or neural network). This model outputs a detailed comprehensive report, including: heavy metal element content (and whether it meets standards), probability of mold, probability of insect damage, degree of yellowing, and the final comprehensive grade determination result. Accordingly, compared with existing technologies, this invention, for the first time, integrates multi-dimensional information on elemental content and physical defects into the LIBS system, and can further employ machine learning models for comprehensive judgment, achieving a one-stop, intelligent assessment of rice quality.

[0083] Figure 3 This is a detailed timing diagram illustrating the execution of dual-pulse differential detection within a single detection cycle of the present invention.

[0084] like Figure 3As shown in the figure, taking the detection of Cd in rice as an example, the detailed steps of a single detection cycle are illustrated.

[0085] Assuming this is the first detection cycle, the target is to detect Cd, and the selected spectral line is Cd I 643.8 nm.

[0086] First, cycle initialization and wavelength setting. The FPGA sends a command to the AOTF module through its I / O port to set the transmission wavelength of the AOTF to 643.8 nm. After completion, the FPGA waits for the laser ready signal.

[0087] Next, the background signal is acquired. At time T0, the FPGA emits a Laser1_Fire TTL pulse, triggering the first pulse laser 1a to emit a 532 nm laser pulse with an energy of 15 mJ. This pulse acts on the surface of the rice sample, generating a reference plasma signal. At time T0 + ΔT1, the FPGA's internal counter starts counting from T0. When it reaches the count value corresponding to ΔT1 = 1.5 μs, the FPGA immediately emits a Gate1_... Enable A pulse is applied to activate the PMT's gating mechanism, with a gating width τ = 500 ns. During this 500 ns gating period, the electrical signal strength detected by the PMT is integrated and recorded by the data acquisition card; this value is I_ ref(Cd) It primarily reflects the continuous background intensity at that wavelength.

[0088] Next, the sample signal acquisition is analyzed. Time T0 + ΔT_ inter Meanwhile, the FPGA continues counting, and when it reaches the count value corresponding to ΔT_inter = 50 μs, the FPGA issues Laser2_ Fire The pulse triggers the second pulsed laser 1b to emit a 1064 nm laser pulse with an energy of 100 mJ. This pulse is precisely applied to the same location, generating the analytical plasma. Time T0 + ΔT_ inter When +ΔT1, the FPGA waits the same ΔT1 = 1.5 μs after the second pulse before issuing Gate2 again. Enable A pulse is applied, enabling the PMT gate τ = 500 ns. The acquisition card records the signal strength during this period, which is I_ ana(Cd) The signal contains the superposition of the characteristic spectral line intensities of Cd and the continuous background.

[0089] Finally, real-time differential calculation is performed. The computer reads I_ in real time. ref(Cd) and I_ ana(Cd) And perform the operation: S_ net(Cd) =I_ana(Cd) -0.98× I_ ref(Cd)(Here, k=0.98 is a pre-calibrated value), and the net cadmium signal S_ is obtained. net(Cd) The system then calculates S_ net(Cd) The signal-to-noise ratio (SNR). If SNR > 12, the signal is considered acceptable and stored. _net(Cd) If SNR ≤ 12, the motion platform is moved by 0.2 mm and the cycle is re-executed.

[0090] Figure 4 This is a flowchart used to exemplify and explain the multi-detection period sequence and data fusion decision according to the present invention. For example... Figure 4 As shown, a complete detection task consists of five detection cycles in sequence: Cd 643.8 nm, Pb 405.8 nm, K 766.5 nm, K 769.9 nm, and Na 589.0 nm, which are used for heavy metal pollution detection, mold identification, insect eye and yellowing identification, respectively.

[0091] See Figure 4 The full-index detection process is as follows: First, control the three-dimensional electric motion platform to move a new point of the rice sample to the laser focus (step S301).

[0092] Subsequently, the platform remained stationary, and the system executed the following steps in sequence: setting AOTF to Cd 643.8 nm and executing the detection cycle (S302).

[0093] Next, the AOTF was set to Pb 405.8 nm and the detection cycle was executed (S303).

[0094] ...

[0095] This process continues until all preset wavelengths have been detected (S306). During this period, the laser is always applied to the same fixed point.

[0096] Finally, data fusion and intelligent decision-making are performed (S307). If the next point needs to be detected, the process starting from S301 is repeated.

[0097] More specifically, according to a preferred embodiment of the invention, after all cycles are completed, the directly measured net signal S_ is preferably... net(Cd) , S_ net(Pb) And the calculated mold index R _m and S _net(Na)This creates a feature vector. This feature vector is then input into a machine learning model trained with rice samples of known quality grades. The model learns the complex non-linear relationship between these elemental features and the final quality (e.g., "Cadmium Exceeds Standard," "Lead Exceeds Standard," "Mold," "Insect Damage / Yellowing," and "Normal"), and outputs a fast and objective judgment result. For example: Cadmium content: Exceeds standard, >50 mg / kg; Lead content: Not detected; Mold: Yes; Insect Damage: No; Overall grade: Grade 2.

[0098] In summary, this invention utilizes dual lasers for differential detection and redesigns the entire process and working mechanism. Compared with existing technologies, it effectively suppresses continuous background interference in LIBS detection and accurately acquires detection results in rice, including heavy metal content and other physical defect information, achieving rapid and integrated quality assessment of rice. Furthermore, this invention specifically introduces AOTF, gated PMT, and FPGA to work in conjunction with the dual lasers to execute the entire differential detection process. This ensures simultaneous multi-wavelength sequential detection at a single point, enabling synchronous and homogeneous strategies for multiple preparations at a single physical point. This significantly improves data consistency and reliability, making it particularly suitable for complex detection processes requiring multiple indicators and reconfigurability, and possessing broad application prospects.

[0099] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A rapid rice quality detection method based on dual-pulse differential LIBS, characterized in that, The method includes the following steps: S1. For the rice sample being tested, a detection system is configured including a dual-laser differential detection module, an acousto-optic tunable filter, a gated photomultiplier tube, and a field-programmable gate array, wherein: The dual-laser differential detection module includes a first pulse laser and a second pulse laser that are independent of each other and have different laser energies. They are used to output laser pulses to act on the surface of the rice sample or its surrounding gas to generate different plasmas. The gated photomultiplier tube is used to collect the analysis signals corresponding to the plasmas generated by the first pulse laser and the second pulse laser respectively. The acousto-optic tunable filter is used to set multiple different transmission wavelengths. The field-programmable gate array is used to provide unified timing control for the entire detection system. S2. The full-index detection of the rice sample is decomposed into multiple consecutive detection cycles, and each detection cycle corresponds to a preset characteristic wavelength; wherein, within each detection cycle, the following operations are performed: S21. The acousto-optic tunable filter sequentially selects one of a plurality of different transmission wavelengths and sets the transmission wavelength as the target wavelength of the current detection cycle. S22. The field-programmable gate array triggers the first pulse laser in the dual-laser differential detection module to output a first laser pulse, which ablates the ambient gas near a detection point on the surface of the rice sample and generates a first plasma. S23. At a first predetermined delay ΔT1 after the first laser pulse, the first gate τ1 of the gated photomultiplier tube is activated, and the reference background signal I_ corresponding to the first plasma is acquired under the filtering effect of the target wavelength. ref ; S24. A predetermined inter-pulse delay ΔT_ after triggering the first laser pulse. inter At the location, the field-programmable gate array triggers the second pulse laser in the dual-laser differential detection module to output a second laser pulse. The laser energy of the second laser pulse is greater than that of the first laser pulse and fully ablates and vaporizes the detection point on the surface of the rice sample, generating a second plasma. S25. At the second predetermined delay ΔT2 after the second laser pulse, the second gate τ2 of the gated photomultiplier tube is activated, and the analytical mixed signal I_ corresponding to the second plasma is acquired under the filtering effect of the target wavelength. ana ; S26, regarding the reference background signal I_ ref and the analysis of the mixed signal I_ ana The net feature signal S_ for the current detection period is obtained by real-time calculation using the following difference formula. net : S_ net =I_ ana -k×I_ ref , where k represents the preset background subtraction coefficient; S3, the net feature signal S_ obtained by combining all detection cycles. net Together, we will conduct a comprehensive assessment of rice quality, thereby completing a rapid testing process for rice quality.

2. The method as described in claim 1, characterized in that, In step S1, the first pulsed laser is a Nd:YAG laser with a wavelength of 532nm and an energy range of 5mJ to 15mJ, and a repetition frequency of 10Hz to 100Hz; the second pulsed laser is a Nd:YAG laser with a wavelength of 1064nm and an energy range of 50mJ to 150mJ, and a repetition frequency of 10Hz to 100Hz.

3. The method as described in claim 2, characterized in that, In step S1, the spectral response range of the gated photomultiplier tube covers 200nm to 900nm, and the minimum gate width is ≤100 nanoseconds; the operating wavelength of the acousto-optic tunable filter is 400nm to 1400nm, and the wavelength switching time is <1 millisecond, and the diffraction efficiency is >70%; the internal clock frequency of the field programmable gate array is ≥200MHz.

4. The method according to any one of claims 1 to 3, characterized in that, Step S2 also includes processing the net feature signal S_ net Perform signal quality assessment operations: This operation includes processing each of the net feature signals S_ net Calculate the signal-to-noise ratio (SNR) of each device and compare it with a preset signal quality threshold. When the SNR is greater than or equal to the signal quality threshold, the detection result of the current detection cycle is deemed qualified, the data is stored, and the next detection cycle is automatically started. When the signal-to-noise ratio is less than the signal quality threshold, the detection result of the current detection cycle is deemed unqualified, and the operation of the current detection cycle is re-executed.

5. The method according to any one of claims 1 to 3, characterized in that, In step S2, for each detection cycle, the detection point of the rice sample remains stationary; after the current detection cycle is completed, a three-dimensional electric motion platform is used to make micro-motion so that the laser focus acts on another new detection point on the surface of the rice sample or the ambient gas nearby.

6. The method according to any one of claims 1 to 3, characterized in that, In step S2, the target wavelength is selected from at least four of the following wavelengths: 643.8 nm, 405.8 nm, 766.5 nm, 769.9 nm, and 589.0 nm.

7. The method as described in claim 6, characterized in that, In step S2, the first predetermined delay ΔT1 and the second predetermined delay ΔT2 are set to 1.5 microseconds, and the first gate τ1 and τ2 are set to 500 nanoseconds; the predetermined inter-pulse delay ΔT_ inter Set to 50 microseconds.

8. The method according to any one of claims 1 to 3, characterized in that, In step S3, the net feature signal S_ of all detection cycles is... net After all these are calculated, these net characteristic signals S_ net The feature vector is directly input into the trained machine learning model to obtain the final detection result; the machine learning model includes random forest, support vector machine or artificial neural network.

Images