Simulated lunar soil particle screening device based on real lunar soil and method of using same

Abstract

The present application relates to a simulated lunar soil particle screening device based on real lunar soil and a method for using the same, comprising a micro-flow rate control system, a flow chamber, a particle detection system, a particle identification system and a particle screening system, the flow chamber comprising a sample channel and a sheath liquid pipe, the particle screening system comprising a control module and at least two collectors; a single lunar soil sample particle suspension is ejected from a nozzle at the front end of the sample channel to a detection window of the particle detection system under the action of liquid flow pressure; the sheath liquid is ejected from the nozzle after flowing from all around to the nozzle and surrounding the periphery of the lunar soil sample particles; the control module is used to sort the simulated lunar soil sample particles into the at least two collectors according to the control instructions of the particle identification system. The present application can realize high-throughput particle screening and recovery, solve the problem of a large gap between the properties of the existing simulated lunar soil sample and real lunar soil, and provide an efficient tool for lunar resource in-situ utilization and extraterrestrial matter research.

Description

Technical Field

[0001] This invention relates to the field of space exploration material analysis technology, and in particular to a simulated lunar soil particle screening device based on real lunar soil and its usage method. Background Technology

[0002] The Moon is humanity's outpost and primary destination for deep space exploration and development. Lunar regolith contains crucial information about the Moon's formation and evolution, as well as abundant metallic, non-metallic, volatile components, and water ice resources. Conducting multi-dimensional tests on lunar regolith, including its mineral composition and physicochemical properties, can not only reveal the Moon's origin and evolution but also provide core evidence for exploring lunar resource development and supporting the construction of future lunar bases.

[0003] However, real lunar soil samples are extremely rare. Since the start of lunar exploration activities, only a very limited number of lunar soil samples have been obtained, and most samples need to be preserved for long-term use in core scientific research missions. This makes it difficult to meet the growing demands of lunar scientific experiments and aerospace equipment testing, thus hindering the in-depth development of lunar research. To address the problem of the scarcity of real lunar soil samples, simulated lunar soil has emerged and is widely used in various fields of lunar research. However, existing simulated lunar soil still differs significantly from real lunar soil in terms of composition, structure, and properties. It is difficult to accurately reproduce the basic physical and mechanical properties of real lunar soil, resulting in a significant reduction in the accuracy of data and the reliability of engineering applications based on simulated lunar soil, thus hindering the development of lunar scientific research and engineering technology.

[0004] Therefore, the development of a device capable of screening simulated lunar soil based on the characteristics of real lunar soil is urgently needed. This invention screens simulated lunar soil based on real lunar soil, achieving high-precision matching between simulated and real lunar soil across multiple dimensions by integrating advanced laser-induced fluorescence analysis, multi-directional optical scattering signal detection, and intelligent comparison technologies. This device can not only provide more reliable simulated materials for lunar geological research, lunar resource development, and in-situ utilization of lunar soil, effectively reducing the loss of real lunar soil samples, but also accelerate innovative breakthroughs in lunar scientific research and engineering technologies, helping my country to achieve a leading position in lunar exploration and propelling human lunar development to new heights. Current methods largely rely on manual analysis using electron microscopy, which suffers from low efficiency and accuracy, making it difficult to meet the demands of modern space science for large sample sizes.

[0005] Existing particle screening methods largely rely on electrostatic separators, scanning electron microscopy (SEM) images, and manual statistics. Electrostatic separators classify particles based on their trajectory in an electric field, depending on the uniformity of particle charge, and cannot separate particles with complex morphologies. Optical classification methods such as SEM only allow observation, relying on two-dimensional image projection for morphology estimation, making it difficult to accurately obtain three-dimensional structural information, lacking automation, and failing to generate statistical data sets. Manual statistics use traditional classification equipment, which is simple in structure but inefficient and has poor resolution when processing micron-sized particles. Furthermore, the classification process introduces mechanical stress, easily causing particle damage, affecting sample integrity and representativeness, and cannot distinguish mineral types based on size. These devices are insufficient to meet the analytical needs of lunar soil particles in the context of "diversity, high precision, and large data" characterization. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a simulated lunar soil particle screening device based on real lunar soil and its usage method.

[0007] This invention is achieved through the following technical solution:

[0008] The present invention provides a simulated lunar soil particle screening device based on real lunar soil, comprising a microflow rate control system for supplying sheath fluid to the sheath fluid tube of the flow chamber and supplying lunar soil sample solution to the sample channel of the flow chamber; a flow chamber; a particle detection system for detecting lunar soil sample particles and transmitting the detection data to a particle identification system; a particle identification system and a particle screening system for identifying lunar soil sample particles based on the detection data and sending control commands to the particle screening system based on the identification results; the flow chamber includes a sample channel, a nozzle disposed at the front end of the sample channel, and a sheath fluid tube concentrically disposed outside the sample channel; the particle screening system includes a control module and at least two collectors; a single lunar soil sample particle suspension is ejected from the nozzle at the front end of the sample channel under the action of liquid flow pressure to the detection window of the particle detection system; the sheath fluid flows from all sides of the sheath fluid tube to the nozzle and surrounds the lunar soil sample particle before being ejected from the nozzle; the control module is used to sort the lunar soil sample particles into at least two collectors according to the control commands of the particle screening system.

[0009] Optionally, the particle detection system includes a laser for irradiating and exciting lunar soil sample particles, a photodetector for acquiring the optical signals generated by the lunar soil sample particles after photoexcitation, an image acquisition device for acquiring image signals of lunar soil sample particles, and a microelectrode measuring device for measuring the surface electrical properties of lunar soil sample particles.

[0010] Optionally, the particle detection system may also include a fluorescence detector for acquiring fluorescence signals.

[0011] Optionally, the particle detection system may also include a multi-channel fluorescence detector for simultaneously acquiring fluorescence signals in different wavelength bands.

[0012] The particle identification system includes a computer and an information storage device.

[0013] Optionally, the microfluidic control system includes a compressed gas source, a sheath fluid container, and a sample container. The compressed gas source is connected to the sheath fluid container and the sample container through a vent pipe. The sample container is used to store lunar soil sample solution, and the sheath fluid container is used to store sheath fluid.

[0014] The method of using the simulated lunar soil particle screening device based on real lunar soil provided by the present invention includes the following steps:

[0015] Real lunar soil sample particles were dissolved in deionized water and dispersed to obtain a lunar soil sample solution.

[0016] The lunar soil sample particle suspension is injected into the sample channel by compressed gas, and the sheath flow liquid is injected into the sheath liquid tube at the same time, so that the lunar soil sample particles are located in the center of the flow channel under the sheath flow and pass through in a single order.

[0017] When lunar soil sample particles pass through the detection window of the particle detection system, the system detects the particles and transmits the data to the particle recognition system, which then stores the data. The process is repeated with simulated lunar soil samples. When the particle detection system transmits the detection data to the particle recognition system...

[0018] The particle identification system compares the particles based on the detection data and controls the control module based on the comparison results;

[0019] The control module operates according to the control instructions of the particle identification system, sorting lunar soil sample particles into at least two collectors.

[0020] A single lunar soil sample particle suspension is ejected from the nozzle at the front end of the sample channel under the action of liquid flow pressure and flows to the detection window of the particle detection system; the sheath fluid flows from all sides of the sheath fluid tube to the nozzle and surrounds the lunar soil sample particle before being ejected from the nozzle; the control module is used to sort the lunar soil sample particles into at least two collectors according to the control instructions of the particle identification system.

[0021] Optionally, the detection method of the particle detection system includes: a laser emits a focused beam, and a photodetector collects forward scattering and side scattering signals; if the sample contains fluorescently labeled or naturally photoluminescent particles, a fluorescence detector simultaneously collects one fluorescence signal or multiple fluorescence signals of different wavelengths, an image acquisition device captures two-dimensional morphology images of lunar soil sample particles, and a microelectrode measuring device measures the surface electrical properties of the particles.

[0022] Optionally, after the particle detection system completes data acquisition, it uses a timestamp system to synchronously bind the images, optical signals, and electrical data of soil sample particles for each month to construct a multimodal feature set.

[0023] The multimodal feature set is standardized.

[0024] A multimodal deep learning neural network model is constructed based on a structure of convolutional neural network + multilayer perceptron + splicing and fusion module.

[0025] Supervised learning training is performed on the multimodal deep learning neural network model to obtain a well-trained multimodal deep learning neural network model;

[0026] The standardized data is input into a trained multimodal deep learning neural network model to achieve particle feature recognition. After recognition is completed, real lunar soil particles are collected.

[0027] Repeat the above process for the simulated lunar soil, compare the identification data of the real lunar soil with that of the simulated lunar soil, and complete the screening process.

[0028] Compared with the prior art, this application has at least the following beneficial effects:

[0029] 1. This invention prepares a lunar soil sample solution from real lunar soil sample particles and sends it into the flow chamber by the liquid pressure of the microflow rate control system. After the lunar soil sample particles flow into the sample channel, they pass through the detection system in a single-particle arrangement. Each lunar soil sample particle is then detected and identified by a particle identification system. Simulated lunar soil sample particles are then prepared into a simulated lunar soil sample solution and identified and compared by the particle identification system. The particle identification system then feeds back the sorting information to the particle screening system, which places the simulated lunar soil sample into different collectors, achieving high-throughput particle screening and recovery. This solves the problem of large differences between the properties of existing simulated lunar soil samples and real lunar soil, and provides an efficient tool for in-situ utilization of lunar resources and research on extraterrestrial materials.

[0030] 2. The detection system of the present invention collects multimodal data such as forward / lateral scattering, fluorescence excitation and electrical acquisition of particles, which is conducive to realizing multi-dimensional characterization at the single particle level and can meet the analysis needs of lunar soil particles in the context of diverse characterization.

[0031] 3. Due to the preciousness and scarcity of lunar soil samples, and the limitations of sampling volume and experimental costs, this application can avoid pollution and waste caused by mixed use, and help improve the efficiency of sample utilization. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of the simulated lunar soil particle screening device based on real lunar soil in the embodiment.

[0034] Figure 2 This is a schematic diagram of the structure of the multi-channel detector in the embodiment;

[0035] Figure 3 This is a schematic diagram of the structure of the multimodal deep learning neural network model in the embodiment;

[0036] Figure 4 This is a flowchart of the model training process in the embodiment;

[0037] Figure 5 This is a flowchart illustrating the extraction of deep representations using a three-layer perceptron in the embodiment.

[0038] Reference numerals: 1-Microflow control system, 2-Flow chamber, 3-Particle detection system, 4-Particle identification system, 5-Particle screening system, 11-Compressed gas source, 12-Sheath fluid container, 13-Sample container, 14-Pressure valve, 21-Sample channel, 22-Sheath fluid tube, 31-Laser, 32-Detection system, 41-Computer, 42-Information storage, 51-Control module, 52-Collector. Detailed Implementation

[0039] 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. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0040] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other. It should also be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments; similar or identical parts between embodiments can be referred to interchangeably.

[0041] In the description of this invention, it should be noted that the terms "front," "rear," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of this invention is usually placed in during use, or the orientation or positional relationship that is commonly understood by those skilled in the art. They are only used to facilitate the description of this invention and to simplify the description, and are not intended to 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] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0043] This embodiment discloses a simulated lunar soil particle screening device based on real lunar soil, which is obtained by modifying a flow cytometer. For example... Figure 1 As shown, the simulated lunar soil particle screening device based on real lunar soil specifically includes a microflow rate control system 1, a flow chamber 2, a particle detection system 3, a particle identification system 4, and a particle screening system 5. Both the particle detection system 3 and the particle screening system 5 are connected to the particle identification system 4.

[0044] The microfluidic control system 1 includes a compressed air source 11, a sheath fluid container 12, and a sample container 13. The compressed air source 11 is connected to the sheath fluid container 12 and the sample container 13 via air vents, and a pressure valve 14 is installed on the air vents. The sample container 13 is used to store lunar soil sample solution.

[0045] The flow chamber 2 includes a sample channel 21 and a sheath fluid tube 22. The sheath fluid tube 22 is concentrically located outside the sample channel 21. The sample container 13 is connected to the rear end of the sample channel 21 through a pipe, and the sheath fluid container 12 is connected to the rear end of the sheath fluid tube 22 through a pipe. The sample channel 21 has a nozzle at the front end. The suspension of a single lunar soil sample particle is ejected from the nozzle at the front end of the sample channel 21 under the action of liquid flow pressure. The sheath fluid flows from all sides of the sheath fluid tube 22 to the nozzle, surrounds the outer periphery of the lunar soil sample particle, and is ejected from the nozzle.

[0046] The particle detection system 3 is used to detect particles in lunar soil samples.

[0047] Optionally, the particle detection system 3 includes a laser 31 for irradiating and exciting lunar soil sample particles and a detection system 32. The laser 31 is used to irradiate and excite lunar soil sample particles to generate one or more optical response signals, the type of which depends on the physical composition and type of the particles. The detection system 32 includes a photodetector for collecting the fluorescence generated by the lunar soil sample particles after photoexcitation, an image acquisition device for collecting image signals of the lunar soil sample particles, and a microelectrode measuring device for measuring the surface electrical properties of the lunar soil sample particles.

[0048] Because some particles possess natural fluorescence response capabilities due to their physical composition, in another embodiment, the particle detection system 3 also includes a fluorescence detector for acquiring fluorescence signals; the fluorescence detector is also a type of photodetector. It is worth noting that there can be one or more fluorescence detectors. Figure 2 As shown, when multiple fluorescence detectors are used, the combination structure consists of multiple detectors, a spectrometer, and filters, with each channel corresponding to a different center wavelength (e.g., 530nm, 605nm, 690nm). Some lunar soil sample particles are excited to emit multi-band fluorescence signals under laser irradiation, which are then passed through different filters and entered into the photodetectors of each channel. The light intensity of each particle at different wavelengths is recorded simultaneously for subsequent component analysis and particle screening. Furthermore, the autofluorescence properties of minerals are utilized to avoid contamination of lunar soil samples by chemical labeling.

[0049] For example, when the particles are silicate particles, laser 31 can simultaneously excite silicate particles to produce a fluorescence response, excite metal oxide particles to produce a Raman scattering signal, and excite all particles to produce elastic scattered light. A fluorescence detector can then receive these three types of signals in parallel, constructing a component fingerprint database to achieve non-destructive classification and identification.

[0050] Preferably, the image acquisition device can be a CMOS image acquisition device.

[0051] The particle recognition system 4 includes a computer 41 and an information storage device 42.

[0052] The particle screening system 5 includes a control module 51 and at least two collectors 52. The control module 51 is connected to a computer 41. The control module 51 includes deflectors, pore controls, or optical sorters, etc. Among them, the deflector can be an electromagnetic deflector.

[0053] The working principle of the simulated lunar soil particle sorting device based on real lunar soil is as follows: Lunar soil samples are mixed with deionized water to prepare a lunar soil solution. Compressed gas is introduced through a pressure valve to control the solution flow rate. The solution flows into the sample channel in the classification system and passes through the device's laser sorting system in a single-particle arrangement. Multimodal data such as forward / lateral scattering, fluorescence excitation, and electrical properties of the particles are collected. Intelligent algorithms are used to perform functions such as particle shape differentiation, component identification, and particle size sorting. The sorting information is fed back to the classification control module, and the data-collected particles are sorted and placed into different particle collection tubes to complete the classification and identification. This device solves the core problems of low efficiency, high destructiveness, and insufficient component identification accuracy in lunar soil sample particle sorting, and provides a flow cytometry sorting system suitable for large-sized, irregular mineral particles.

[0054] Optionally, in some embodiments, for lunar soil with a wide range of particle sizes, a large-aperture nozzle with an aperture of 50–500 μm is used to support the passage of large-sized particles, prevent clogging, adapt to the high hardness and multi-faceted characteristics of lunar soil, and combine with a high-viscosity sheath fluid formulation to achieve stable flow of 0.1–1000 μm particles, thus solving the clogging problem.

[0055] Optionally, the flow chamber can be a ceramic-coated flow chamber, which helps reduce lunar soil abrasion.

[0056] Optionally, in some embodiments, the laser 31 integrates dual laser beams (405nm+785nm). The laser 31 can excite characteristic fluorescence of minerals (such as silicate fluorescence peaks). Under laser irradiation, the particles simultaneously generate forward scattering (FSC) and side scattering (SSC) signals. These signals are received by a high-sensitivity photodetector array to accurately capture optical feature information related to physical parameters such as particle size and morphological roughness.

[0057] Optionally, in some embodiments, a silicon carbide ceramic coating is used with a sapphire detection window to reduce lunar soil wear and improve equipment lifespan.

[0058] Based on the simulated lunar soil particle screening device based on real lunar soil, this embodiment also discloses a method for using the simulated lunar soil particle screening device based on real lunar soil, including the following steps:

[0059] Step 1: Dissolve real lunar soil sample particles in deionized water and disperse the particles to obtain a real lunar soil sample solution. Inject the real lunar soil sample particle suspension into the sample channel using compressed gas, and simultaneously inject sheath fluid into the sheath fluid tube, so that the real lunar soil sample particles are positioned in the center of the flow channel under the sheath fluid and pass through in a sequential manner. When the real lunar soil sample particles pass through the detection window of the particle detection system, the particle detection system detects the real lunar soil sample particles and transmits the detection data to the particle recognition system, which performs memory processing and analysis. After detection, the real lunar soil sample particles are collected by the collector.

[0060] The core purpose of first introducing real lunar regolith particles is to provide a comparison standard for the subsequent screening of simulated lunar regolith. As the original sample of lunar material, real lunar regolith's physical, optical, electrical, and mineral composition characteristics are the target parameters that the simulated lunar regolith needs to match. By processing the real lunar regolith first, the particle detection system can collect data, which serves as a reference benchmark for judging whether the simulated lunar regolith meets the requirements, thereby achieving a high-precision match between the simulated and real lunar regolith across multiple dimensions.

[0061] The memory processing and analysis of the particle identification system mainly involves storing the laser excitation optical signal transmitted from the particle detection system, the fluorescence band and intensity collected by the fluorescence detector, the image signal captured by the image acquisition device, and the electrical signals of the particle surface obtained by the microelectrode measuring device. The acquired information is then processed and analyzed through data standardization, feature extraction, and other techniques.

[0062] Step 2, Sample Preparation

[0063] The simulated lunar soil sample particles were initially dried and foreign matter removed to ensure that the particles did not clump together. The sample was then added to deionized water, and the particles were dispersed using ultrasonic treatment and surfactants to prevent aggregation. The sample concentration was controlled so that only a single particle passed through the detection zone with each injection to prevent overlap. Finally, the prepared lunar soil sample solution was stored in sample container 13.

[0064] Step 3, Sample delivery

[0065] A suspension of lunar soil sample particles is injected into sample channel 21 using compressed gas, simultaneously with the injection of sheath fluid (deionized water or low-conductivity buffer solution). This ensures that the lunar soil sample particles are positioned at the center of the channel and pass through sequentially as individual particles, enveloped by the sheath fluid. Meanwhile, the microfluidic system 1 controls the flow rate within a typical range (1–100 μL / min), forming a stable and continuous flow of single particles. The system maintains flow stability through a precision pressure or pump control platform, preventing air bubbles or blockage.

[0066] Step 4, Sample Testing

[0067] When lunar soil sample particles pass through the detection window of particle detection system 3, laser 31 emits a focused beam. Forward scattering (FSC) and side scattering (SSC) signals are collected by a photodetector to reflect the particle's size, structure, and roughness. If the sample contains fluorescently labeled or naturally photoluminescent particles, the fluorescence channel simultaneously collects fluorescence intensities at different wavelengths. The multi-channel detector records the fluorescence information for each wavelength, reflecting the particle's compositional characteristics. Simultaneously, an image acquisition unit is activated to capture two-dimensional morphological images of the particles as they pass through the detection point, acquiring grayscale and RGB images to capture contours, textures, and edge features. A microelectrode measuring instrument measures the particle surface electrical properties, obtaining zeta potential data to identify the mineral type and dispersion stability of the particles. Finally, a timestamp system synchronously binds the image, optical signal, and electrical data of each particle to construct a multimodal feature set.

[0068] It is worth noting that the binding here refers to associating the feature data of the same particle collected from different sensor channels such as image, scattering, fluorescence, and electrical properties with the particle's ID number and saving it as a particle feature dataset.

[0069] Step 5: Data Processing and Storage

[0070] After the particle detection system detects particles in the simulated lunar soil sample, it transmits the detection data to the particle identification system. The particle identification system compares and contrasts the particles based on the detection data and controls the control module based on the comparison results.

[0071] The comparison, based on real lunar soil characteristic data, compares the characteristic data of simulated lunar soil particles, including physical and electrical characteristics:

[0072] Physical characteristics include: whether the particle size is within the actual lunar soil particle size distribution range and whether the morphology is similar; optical characteristics: whether the forward / side scattering signal intensity matches and whether the fluorescence band and intensity are consistent (reflecting mineral composition); electrical characteristics include whether the zeta potential is within the potential range of the actual lunar soil (reflecting surface electrical properties); comprehensive characteristics: the overall similarity between the combination of single-particle "optical-image-electrical" characteristics and the feature set of the actual lunar soil.

[0073] The comparison results show different levels of matching. This embodiment includes high matching, medium matching, and low matching.

[0074] High matching degree means that the similarity between the multi-dimensional features of the simulated particles and the features of the real lunar soil is greater than or equal to a preset threshold (e.g., 80%), and the key parameters match. The particle screening system controls the particles to be imported into the "high matching degree collector" to complete the screening.

[0075] Medium matching refers to a similarity between 50% and 80%, where some non-critical parameters differ but the core features match. This is imported into the "Medium Match Collector".

[0076] Low matching degree refers to similarity <50%, where key parameters (such as particle size out of range, large fluorescence band deviation) do not match, and is imported into the "exclusion collector".

[0077] In some embodiments, the particle identification system 4 processes the detection data and then identifies the particles in the lunar soil sample. The particle identification system 4 records all detection data, identification results, and physical sorting paths of each particle as structured entries and stores them in a database, supporting subsequent retrieval, tracing, and analysis. It also displays particle size distribution, particle type proportion, and characteristic parameter statistics on a visual interface, allowing for the selection of specific particle types for analysis or recovery based on task requirements.

[0078] The particle detection system collects and records all raw sensor data as each particle passes by, and associates this multi-channel data into a complete particle feature record using particle identification numbers. Following the particle detection system, the particle recognition system intelligently processes the synchronized multimodal features, performing tasks such as classification and identification.

[0079] Step 5: Particle sorting

[0080] Based on the identification results of the particle identification system 4, the particle identification system 4 sends a deflection command to the control module 51; and activates the electromagnetic deflection, airflow control or optical sorter to guide the target particles into different collectors 52 to achieve real-time physical separation.

[0081] Optionally, in some embodiments, particle velocity and particle density can be monitored in real time. In an exemplary embodiment, the velocity monitoring method is as follows: two laser beams are arranged on the flow channel, the distance between the two laser beams is known, and when a particle sequentially blocks the two laser beams, the time difference Δt is recorded, and the velocity v = distance / Δt.

[0082] In other embodiments, the flow velocity is monitored by continuously capturing images of particle flow using a high-speed camera, identifying changes in particle position, and calculating the displacement / time to obtain the flow velocity.

[0083] In an exemplary embodiment, the particle density monitoring method is as follows: since the device is equipped with a detection window, a record is triggered every time a particle passes through, and the number of triggers per unit time is counted to calculate the particle number density.

[0084] By monitoring flow rate and particle density, the control strategy is continuously optimized based on real-time feedback through a control optimization algorithm. This drives the microfluidic control system to adjust the sheath fluid pressure in real time, ensuring stable particle flow and single-particle passage through the detection window. This avoids blockage or particle overlap, thereby improving detection throughput and identification accuracy.

[0085] The control optimization algorithm here can be an adaptive control algorithm or a machine learning-based reinforcement learning control strategy, used to dynamically adjust the sheath fluid pressure based on the real-time feedback data of the collected flow rate and particle density.

[0086] Example 2

[0087] The particle recognition system 4 in this embodiment uses a deep learning model to achieve particle recognition. Specifically, the intelligent recognition method includes the following steps:

[0088] Step 1: Based on the CNN+MLP+concatenation and fusion module structure, construct a multimodal deep learning neural network model, such as... Figure 1 As shown.

[0089] Step 2, model training, such as Figure 2 As shown, it includes:

[0090] 2.1 Supervised learning training was conducted based on a set of labeled standard particle samples; the training data came from lunar soil images and spectral datasets with manual annotation.

[0091] 2.2 The loss function is selected according to the task, such as classification cross-entropy, regression MSE or clustering loss;

[0092] 2.3. Apply data augmentation techniques (image rotation, noise perturbation, multimodal missing data simulation) to improve the robustness of the model;

[0093] 2.4 Adjust hyperparameters such as learning rate, batch size, and network depth to obtain the optimal initial model.

[0094] Step 3: Model optimization and verification.

[0095] 3.1 Evaluate model performance on independent validation sets and analyze metrics such as accuracy, recall, and F1 score;

[0096] 3.2. Locate the main sources of error through the confusion matrix (such as two types of particles being easily confused), and optimize accordingly;

[0097] 3.3 The model undergoes pruning, distillation, quantization, and other operations to compress the model size and improve inference speed;

[0098] 3.4. Incremental learning is supported. When the system encounters new types of particles during operation, it can be fine-tuned online to continuously optimize the model and obtain a well-trained multimodal deep learning neural network model.

[0099] In some embodiments, the optimization algorithm uses the Adam optimizer, with an initial learning rate of 0.001, a batch size of 32, and 100 training epochs.

[0100] Step 4: Using the trained multimodal deep learning neural network model, perform intelligent particle recognition, including:

[0101] 4.1 Acquire raw data of lunar soil sample particles, including multi-channel data such as optical signals, image data, and surface electrical signals, and synchronously bind the data of the same particle in different channels;

[0102] In some embodiments, the optical signals include forward scattering signals, side scattering signals, and multi-channel fluorescence signals.

[0103] 4.2 Standardize the raw data, including image cropping and normalization, signal filtering and baseline correction, and convert the image into an image tensor + numerical vector model input format to ensure consistent data input format.

[0104] Cropping aims to extract particle regions, highlight the characteristics of the particles themselves, reduce irrelevant information input, and improve recognition accuracy. Image normalization is used to unify the range of pixel values, avoid numerical shifts caused by varying illumination intensity, and improve stability. Signal filtering removes noise from scattering and fluorescence signals, focusing on effective information. Signal baseline correction corrects fluctuations in sensor zero-point and background signals, ensuring comparability of data collected in different batches or at different times, and improving data consistency. Image-to-tensor + numerical vector conversion unifies data from different modalities into a standard format that can be processed by neural networks.

[0105] 4.3 Input the standardized data into the trained multimodal deep learning neural network model, and output the category, confidence score, and group assignment. Specifically:

[0106] Image data is input into a convolutional neural network (CNN) to extract shape features;

[0107] Numerical features such as particle size, scattering intensity, and fluorescence intensity are input into a multilayer perceptron (MLP) for feature expansion.

[0108] The feature fusion module integrates information from different modalities. The feature fusion module refers to processing, aligning, and combining feature vectors from different sources to generate a fused feature vector in a unified semantic space, which is used to jointly model feature vectors from different modalities. This is equivalent to the task of the feature fusion layer.

[0109] Depending on the task requirements, set a classification head (such as identifying categories like vitreous / pyroxene / olivine), a regression head (predicting parameters such as elastic modulus), or a clustering head (to achieve unsupervised particle grouping).

[0110] Perform tasks such as classification (e.g., particle type), clustering (e.g., particle similarity grouping), and regression (e.g., physical property parameter estimation); output results such as particle category, confidence level, and group affiliation.

[0111] The process includes: classification to determine the category of particles for efficient grouping and sorting applications; clustering to automatically group particles based on feature similarity when the category is unknown, discovering potential particle categories or structural patterns; regression to predict continuous physical properties of particles, such as elastic modulus and density; and outputting the category, confidence level, and group affiliation to provide the identification conclusion and confidence level for each particle, facilitating subsequent screening, retrieval, statistical analysis, and optimization.

[0112] In some embodiments, the input features are represented as:

[0113] X = {I,S,F,Z}

[0114] In the above formula, I represents the image tensor, which is the boundary, texture and morphological features of the particle captured by the single-particle RGB of the CMOS camera; S represents the scattering vector, which includes forward scattering (FSC) and side scattering (SSC) intensity values ​​to represent particle size and structural roughness; F represents the multi-channel fluorescence vector, which reflects the material composition of the particle; Z represents the electrical characteristics, which are inputs of the surface electrical properties of the particle and are used to identify the mineral type and dispersion characteristics of the particle.

[0115] In some embodiments, the classic convolutional network ResNet18 is used to extract deep features of the image, compressing the original image information into a low-dimensional representation, such as texture direction and edge sharpness, which is beneficial for subsequent classifier recognition.

[0116]

[0117] In the above formula, I represents the input particle image, and ResNet18() represents using a ResNet18 convolutional neural network to extract features from the image; feat The output image feature vector is represented as a 512-dimensional vector; This indicates that the vector is a 512-dimensional vector in the real number space.

[0118] In some embodiments, all non-image data is stitched together using the following formula:

[0119]

[0120] In the above formula, S represents the scattered light feature vector, including forward scattering (FSC) and side scattering (SSC); F represents the fluorescence intensity vector; z represents the electrical eigenvalue; and Concat represents concatenating the feature vectors of the above three modes along the feature dimension. raw The concatenated original numerical mode input vector The spliced ​​feature dimensions consist of 2 (scattering) + N (number of fluorescence channels) + 1 (electrical).

[0121] In some embodiments, such as Figure 3 As shown, a three-layer perceptron is used to extract deep representations. The first layer rapidly expands the information dimensionality, handles nonlinear relationships between features, and maps the original low-dimensional numerical features to a high-dimensional space, outputting a high-dimensional representation. The second layer extracts the interaction features between each feature dimension and stimulates high-order feature combinations through a nonlinear activation function. The third layer performs dimensionality reduction and compression on the feature representation, outputting a stable, low-dimensional, and compact numerical vector as a unified representation for fusion and alignment with image features.

[0122] S feat =W3·σ(W2·σ(W1·S) raw +b1)+b2)+b3

[0123] In the above formula, S feat Output feature vectors; W1, W2, and W3 are the weight matrices of the three fully connected layers, used for the linear transformations of the first, second, and third layers, respectively; σ is the non-linear activation function; b1, b2, and b3 are the bias vectors corresponding to each layer; S raw Original input feature vector (such as numerical modal features such as particle size, scattered light intensity, fluorescence intensity, etc.).

[0124] A multi-branch fusion neural network is employed. The image branch uses a pre-trained convolutional neural network to extract image features, outputting a 1×512 image feature vector. The signal branch uses scattering, fluorescence, and electrical features to concatenate a 1×(2+N+1) vector (where N is the number of fluorescence channels), and extracts signal features through a three-layer fully connected network (64, 128, and 256 nodes, respectively). The feature fusion layer concatenates the image features and signal features to form a 1×768 feature vector, which is then fed into the output layer after passing through a fusion layer with 256 nodes and ReLU activation.

[0125]

[0126] In the above formula, H is the concatenated fused feature vector, Concat() is the feature concatenation operation, which concatenates the two vectors along the feature dimension, and I feat S is the image feature vector. feat F is the numerical modal eigenvector. final The final feature vector output after fusion is σ(), where σ is the non-linear activation function and W is the final feature vector. f Let b be the weight matrix. f This is the bias vector.

[0127] The output layer is for classification tasks: For classification tasks, the output layer uses the Softmax activation function, and the number of output nodes equals the total number of particle categories. The model maps the fused features to a category probability distribution and uses the category corresponding to the maximum probability as the final prediction result; this maximum probability value can also be used as the recognition confidence output. For regression tasks, the output layer uses a linear activation function to predict one or more continuous physical properties. For clustering tasks, the model first uses t-SNE or UMAP methods to perform nonlinear dimensionality reduction on the intermediate high-dimensional feature representation, mapping the data to a low-dimensional visible space. Then, the DBSCAN density clustering algorithm is used to complete unsupervised grouping and output the cluster label to which each particle belongs.

[0128] In classification tasks, the output layer employs the Softmax activation function, and the dimension of the output vector equals the number of target particle categories. Each dimension corresponds to the predicted probability of a particle type, with the maximum value representing the final classification result predicted by the model. This maximum probability value also serves as a confidence index for particle identification. In regression tasks, the output layer uses a linear activation function, outputting one or more continuous numerical variables to predict the physical properties of particles, such as particle size estimates, roughness indices, and mean fluorescence intensity. In clustering tasks, the model performs nonlinear dimensionality reduction on intermediate feature representations using t-SNE or UMAP methods, mapping high-dimensional features to a two-dimensional or three-dimensional embedding space. The model then further performs density clustering analysis on the dimensionality reduction results using the DBSCAN algorithm. Finally, a clustering label for each particle is output, useful for unsupervised particle grouping or exploration of unknown particle types. This clustering result is an optional output, primarily used to assist in classification system validation, sorting strategy optimization, and visualization interpretation.

[0129] Preferably, the classification uses cross-entropy loss, outputs a probability distribution, determines which mineral category it belongs to, and optimizes using the cross-entropy loss function:

[0130] y class =Softmax(W c ·F final +b c )

[0131]

[0132] In the above formula, y class represents the probability distribution belonging to each class; Softmax() is the activation function used to convert the neural network output into probability form; W c F is the classification weight matrix; final b is the feature vector after multimodal fusion; c For classification bias vector, For the loss function value of the classification task, y kThis is the one-hot encoding of the real label. If the real category is the k-th category, then this position is 1, and the rest are 0. This determines the probability that the model belongs to the k-th class, where K is the total number of classes.

[0133] Preferably, the regression uses continuous variables for prediction, such as particle size and roughness value, and is optimized using mean square error.

[0134]

[0135] In the above formula, For continuous values ​​predicted by the model, such as particle size estimates, modulus, and other physical properties. The regression value, y true This represents the actual measurement results.

[0136] In some embodiments, weighted fusion can be used to introduce fusion weights for each modality. These weights can be: fixedly designed, such as empirical values ​​or hyperparameters; learned and automatically adjusted through neural networks or attention mechanisms; or dynamically generated, such as adaptively adjusted based on modality quality / missing conditions.

[0137] The model output results are linked with the control module in real time, mapping the classification results into digital control signals to drive the electromagnetic deflection or airflow sorting module to complete the precise guidance of target particles, realizing closed-loop intelligent identification and sorting.

[0138] This embodiment demonstrates the construction of a multi-parameter fusion model, jointly analyzing forward scattering (FSC, particle size), side scattering (SSC, surface roughness), and characteristic fluorescence intensity (FL, component identification), to construct a three-dimensional classification threshold. Constructing the three-dimensional classification threshold involves building a three-dimensional feature space based on three optical parameters, and classifying or discriminating particle categories within this space. In this three-dimensional space, different particle categories have different distributions on the FSC-SSC-FL coordinate axes. By setting multiple decision planes, it is determined which category a particle belongs to or which sorting channel it enters.

[0139] This embodiment uses a trained convolutional neural network (CNN) to identify scattered light signal patterns, distinguishing particle shapes (e.g., angular vs. spherical) and compositional types (e.g., olivine, pyroxene). During the sorting execution phase, the system evaluates the brittleness characteristics of the particles based on the identification results and automatically selects either a flexible sorting mode or a standard sorting mode. The flexible sorting mode, by reducing the electric field deflection voltage or decreasing the airflow impact force, is suitable for structurally fragile or highly important particles, ensuring particle integrity and recovery quality during the sorting process.

[0140] This embodiment solves the problem of detecting and sorting large, irregular particles through hardware-algorithm collaborative optimization, and realizes the identification of unlabeled components.

[0141] This invention integrates microfluidic transport technology and a multimodal sensor array to achieve real-time image acquisition and multi-parameter intelligent identification of particles ranging from micrometers to submicrometers. It has online classification and physical sorting capabilities and can be extended to lunar in-situ analysis platforms, deep space exploration missions, and planetary particulate matter scientific research, becoming one of the core analytical technologies for deep space exploration and extraterrestrial resource development.

[0142] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A simulated lunar soil particle screening device based on real lunar soil, characterized in that, include: It has a flow chamber comprising a sample channel, a nozzle disposed at the front end of the sample channel, and a sheath fluid tube concentrically disposed outside the sample channel; A microflow rate control system for delivering sheath fluid to the sheath fluid tube of the flow chamber and lunar soil sample solution to the sample channel of the flow chamber; A particle detection system used to detect particles in lunar soil samples and transmit the detection data to a particle identification system; A particle identification system used to identify particles in simulated lunar soil samples based on real lunar soil detection data and to send control commands to a particle screening system based on the identification results; and A particle screening system with a control module and a collector; A single lunar soil sample particle suspension is ejected from the nozzle at the front end of the sample channel under the action of liquid flow pressure and flows to the detection window of the particle detection system; the sheath fluid flows from all sides to the nozzle through the sheath fluid tube and surrounds the lunar soil sample particle before being ejected from the nozzle; the control module is used to sort the simulated lunar soil sample particles into at least two collectors according to the control instructions of the particle identification system.

2. The simulated lunar soil particle screening device based on real lunar soil according to claim 1, characterized in that, The particle detection system includes: A laser used to excite particles in lunar soil samples by irradiation; A photodetector used to collect optical signals generated by photoexcitation of lunar soil sample particles; An image acquisition device for acquiring particle image signals from lunar soil samples; and Microelectrode measuring instrument used to measure the surface electrical properties of lunar soil sample particles.

3. The simulated lunar soil particle screening device based on real lunar soil according to claim 2, characterized in that, The particle detection system also includes a fluorescence detector for collecting fluorescence signals.

4. The simulated lunar soil particle screening device based on real lunar soil according to claim 2, characterized in that, The particle detection system also includes a multi-channel fluorescence detector for synchronously acquiring fluorescence signals in different wavelength bands.

5. The simulated lunar soil particle screening device based on real lunar soil according to any one of claims 1-4, characterized in that, The particle identification system includes a computer and an information storage device.

6. The simulated lunar soil particle screening device based on real lunar soil according to any one of claims 1-4, characterized in that, The microfluidic control system includes a compressed gas source, a sheath fluid container, and a sample container. The compressed gas source is connected to the sheath fluid container and the sample container through a vent pipe. The sample container is used to store lunar soil sample solution, and the sheath fluid container is used to store sheath fluid.

7. The method of using the simulated lunar soil particle screening device based on real lunar soil as described in any one of claims 1-6, characterized in that, Includes the following steps: First, real lunar soil sample particles are dissolved in deionized water and dispersed to obtain a real lunar soil sample solution. The suspension is then injected into the sample channel using compressed gas, while sheath fluid is simultaneously injected into the sheath fluid tube. This ensures the real lunar soil sample particles are positioned at the center of the channel and pass through sequentially, one particle at a time, enveloped by the sheath fluid. When a real lunar soil sample particle passes through the detection window of the particle detection system, the system detects the particle and transmits the detection data to the particle recognition system, which processes and analyzes the data. After detection, the real lunar soil sample particles are collected by a collector. Then, the simulated lunar soil sample particles are dissolved in deionized water and dispersed to obtain a simulated lunar soil sample solution. The simulated lunar soil sample particle suspension is injected into the sample channel by compressed gas, and sheath fluid is injected into the sheath fluid tube simultaneously, so that the simulated lunar soil sample particles are located in the center of the flow channel under the sheath fluid and pass through in a single arrangement. When the simulated lunar soil sample particles pass through the detection window of the particle detection system, the particle detection system detects the simulated lunar soil sample particles and transmits the detection data to the particle recognition system. The particle recognition system compares and contrasts the particles according to the detection data and controls the control module according to the comparison results. The control module operates according to the control instructions of the particle identification system, sorting lunar soil sample particles into at least two collectors.

8. The method of using the simulated lunar soil particle screening device based on real lunar soil according to claim 7, characterized in that, The detection methods of the particle detection system include: The laser emits a focused beam, and a photodetector collects the forward and side-scattered signals. An image acquisition device captures two-dimensional morphological images of lunar soil sample particles; a microelectrode measuring device measures the surface electrical properties of the particles.

9. The method of using the simulated lunar soil particle screening device based on real lunar soil according to claim 8, characterized in that, The detection method of the particle detection system also includes: if the sample contains fluorescently labeled or naturally photoluminescent particles, the fluorescence detector simultaneously collects one fluorescence signal or multiple fluorescence signals of different wavelengths.

10. The method of using the simulated lunar soil particle screening device based on real lunar soil according to claim 8 or 9, characterized in that, After the particle detection system completes data acquisition, it synchronously binds the images, optical signals, and electrical data of soil sample particles for each month through a timestamp system to construct a multimodal feature set. The multimodal feature set is standardized. A multimodal deep learning neural network model is constructed based on a structure of convolutional neural network + multilayer perceptron + splicing and fusion module. Supervised learning training is performed on the multimodal deep learning neural network model to obtain a well-trained multimodal deep learning neural network model; The standardized data is input into a trained multimodal deep learning neural network model to achieve particle recognition.

Images