Online automatic pollen monitoring system and method

Abstract

The invention discloses an on-line automatic pollen monitoring system and method, and relates to the technical field of environment detection, and the method comprises the steps: separating charged particles according to an aerodynamic diameter through an inertial impact effect, and collecting pollen particles in a target particle size range to the surface of a carrier tape; applying an electrostatic field to the carrier band to enable the surface of the carrier band to have charges with polarity opposite to the known polarity, and purging the carrier band by utilizing directional airflow to remove impurity particles which are not adsorbed; carrying out electrostatic neutralization treatment on the purified pollen particles, blowing the neutralized pollen particles from the surface of the carrier band into a detection channel by adopting pulse gas, and collecting corresponding light resistance pulse signals and multi-wavelength light scattering signals generated when the pollen particles pass through; and processing the light resistance pulse signal, extracting a dynamic characteristic sequence of the light resistance pulse signal along with time change, and judging whether the pollen particles are effective pollen particles or not based on particle physical characteristics reflected by the dynamic characteristic sequence.

Description

Technical Field

[0001] This application relates to the field of environmental monitoring technology, and more specifically to the field of automatic pollen monitoring. In particular, it relates to an online automatic pollen monitoring system and method. Background Technology

[0002] Pollen, a common biological particulate matter in the atmosphere, has a significant impact on human health, agricultural production, and the ecological environment due to its concentration and distribution. From a human health perspective, airborne pollen can easily trigger allergic diseases such as allergic rhinitis and asthma, significantly affecting the quality of life for sensitive populations. From an agricultural production perspective, pollen dispersal efficiency directly affects crop pollination success rates, thus influencing crop yields. Therefore, online monitoring of pollen in ambient air can provide crucial data support for allergy prevention and control, agricultural production guidance, and ecological environment assessment, and has significant practical implications. Currently, some monitoring methods struggle to effectively distinguish pollen from non-target components such as fine particulate matter in the air, easily leading to misidentification and deviations between monitored data and actual pollen concentrations, failing to meet the practical needs of high-precision online monitoring.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides an online automatic pollen monitoring system and method to solve the above-mentioned technical problems.

[0005] This application provides an online automatic pollen monitoring system, comprising: a charging module for charging particulate matter in ambient air with a known polarity charge by passing it through a corona discharge region; a separation module for separating the charged particulate matter according to its aerodynamic diameter and collecting pollen particles within a target particle size range onto a carrier surface via inertial impact; and a purification module for applying an electrostatic field to the carrier to charge its surface with a polarity opposite to the known polarity, thereby adsorbing and fixing pollen particles onto the carrier surface, and using a directional airflow to purge the carrier to remove unadsorbed impurity particles, thus achieving the purification of the pollen. The system includes a purification module, a neutralization module for electrostatically neutralizing purified pollen grains and using pulsed gas to purge the neutralized pollen grains from the carrier surface into the detection channel; a signal acquisition module for allowing pollen grains in the detection channel to pass through an optical sensor under the constraint of laminar sheath flow, and for acquiring the corresponding photoresist pulse signal and multi-wavelength light scattering signal generated when the pollen grains pass through; and a discrimination module for processing the photoresist pulse signal, extracting its dynamic feature sequence over time, and determining whether the pollen grains are valid pollen grains based on the particle physical properties reflected by the dynamic feature sequence.

[0006] This application provides an online automatic pollen monitoring method, comprising: passing particulate matter in ambient air through a corona discharge region to acquire a known polarity charge; separating the charged particulate matter according to its aerodynamic diameter and collecting pollen particles within a target particle size range onto a carrier surface by inertial impact; applying an electrostatic field to the carrier to acquire a charge of opposite polarity to the known polarity on the carrier surface, so that the pollen particles are adsorbed and fixed on the carrier surface, and using directional airflow to purge the carrier to remove unadsorbed impurity particles, thereby achieving in-situ purification of pollen; performing electrostatic neutralization treatment on the purified pollen particles, and using pulsed gas to purge the neutralized pollen particles from the carrier surface into a detection channel; allowing the pollen particles in the detection channel to pass through an optical sensor under the constraint of laminar sheath flow, and collecting the corresponding photoresist pulse signal and multi-wavelength light scattering signal generated when the pollen particles pass through; processing the photoresist pulse signal, extracting its dynamic feature sequence changing over time, and determining whether the pollen particles are valid pollen particles based on the particle physical properties reflected by the dynamic feature sequence.

[0007] Based on the embodiments provided in this application, precise screening and in-situ purification of target pollen are achieved, effectively solving the defect in the prior art where monitoring data is interfered with by non-target factors: by using corona discharge to give particles a known polarity charge, combined with inertial impact, pollen particles within the target particle size range are separated and collected according to aerodynamic diameter, thus achieving particle size screening of pollen in the first step; then, by applying an electrostatic field with the opposite polarity to the known polarity to the carrier, the pollen particles are stably adsorbed and fixed on the surface of the carrier, while directional airflow is used to sweep the carrier to remove unadsorbed impurity particles, forming an in-situ purification treatment of pollen, reducing the interference of impurities such as fine dust and non-target particle size on subsequent monitoring from the source, and providing high-purity pollen samples for subsequent detection. To ensure efficient transport of pollen particles to the detection stage and improve the accuracy of effective pollen identification, thus meeting the reliability requirements of online automatic monitoring, the following measures are taken: First, electrostatic neutralization is used to eliminate the charge on pollen particles. Then, pulsed gas is used to purge the neutralized pollen particles from the carrier surface to the detection channel. This avoids transmission loss caused by charge adsorption, ensuring that pollen particles can effectively enter the detection stage and reducing concentration monitoring deviations caused by pollen loss. Second, pollen particles in the detection channel pass through the optical sensor under the constraint of laminar sheath flow, ensuring that pollen particles pass through the detection area stably and orderly, improving the stability of photoresist pulse signal acquisition. At the same time, by processing the photoresist pulse signal to extract its dynamic feature sequence changing over time, and based on the particle physical characteristics reflected by this sequence, effective pollen particles can be identified. This further eliminates impurity particles that are similar in size to the target pollen but have different physical characteristics, significantly improving the accuracy of effective pollen identification. Ultimately, this solves the problem of insufficient monitoring data accuracy in existing technologies and meets the reliability and accuracy requirements of online automatic monitoring of ambient air pollen. Attached Figure Description

[0008] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a structural diagram of an optional online automatic pollen monitoring system according to an embodiment of this application; Figure 2 This is a flowchart of an optional online automatic pollen monitoring method according to an embodiment of this application; Figure 3 This is a flowchart of another optional online automatic pollen monitoring method according to an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an optional electronic device according to an embodiment of this application.

[0009] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0011] According to one aspect of the embodiments of this application, an online automatic pollen monitoring system is also provided. For example... Figure 1 As shown, the system includes: The charged module 101 is used to cause particulate matter in ambient air to be charged with a known polarity by passing through a corona discharge region; Separation module 102 is used to separate charged particles according to their aerodynamic diameter and collect pollen particles within the target particle size range onto the carrier surface by means of inertial impact. Purification module 103 is used to apply an electrostatic field to the carrier belt to make the surface of the carrier belt have a charge of opposite polarity to the known polarity, so that pollen particles are adsorbed and fixed on the surface of the carrier belt, and to use directional airflow to sweep the carrier belt to remove unadsorbed impurity particles, so as to achieve in-situ purification of pollen. Neutralization module 104 is used to electrostatically neutralize the purified pollen particles and use pulsed gas to purge the neutralized pollen particles from the carrier surface into the detection channel. The signal acquisition module 105 is used to allow pollen grains in the detection channel to pass through the optical sensor under the constraint of laminar sheath flow, and to acquire the corresponding photoresist pulse signal and multi-wavelength light scattering signal generated when the pollen grains pass through. The discrimination module 106 is used to process the photoresist pulse signal, extract its dynamic feature sequence that changes over time, and determine whether the pollen grains are valid pollen grains based on the particle physical characteristics reflected by the dynamic feature sequence.

[0012] refer to Figure 1 The pollen online automatic monitoring system of this application relates to the field of automatic pollen monitoring.

[0013] According to another aspect of the embodiments of this application, such as Figure 2 As shown, this application provides a method for online automatic pollen monitoring, including: S201 causes particulate matter in the ambient air to acquire a known polarity charge as it passes through the corona discharge region. In a specific embodiment of the invention, it is preferable to make the particulate matter carry a negative charge. This is achieved by constructing a corona discharge region, which consists of a pointed discharge electrode with a negative high voltage applied and a grounded collecting electrode plate. When ambient air flows through this region, the particulate matter therein collides with and captures negative ions in the ion wind generated by the negative corona discharge, thereby uniformly acquiring a known negative charge. Negative corona is chosen because it produces a more stable and uniform charging effect in the air and is less likely to generate byproducts such as ozone. In this embodiment, to achieve stable and efficient particle charging, the discharge current is controlled within the range of 1 microamp to 10 microamps. Too low a current will result in insufficient charging, while too high a current will easily produce arc discharge, disrupting the stability of the process.

[0014] S202 separates charged particles according to their aerodynamic diameter and collects pollen particles within the target particle size range onto the carrier surface through inertial impact. The determination of the "target particle size range" directly serves the core objective of pollen monitoring. Taking the monitoring of common wormwood and pine pollen as examples, their aerodynamic diameters are mainly distributed between 15 and 45 micrometers. The method of this invention aims to accurately capture particles within this range. Therefore, in the device design, the parameters of subsequent steps such as inertial impact separation are optimized to efficiently collect pollen particles in this size range, ensuring that the monitoring results accurately reflect the concentration of major allergenic pollens in the environment.

[0015] S203, by applying an electrostatic field to the carrier belt to make the surface of the carrier belt have a charge of opposite polarity to the known polarity, so that pollen particles are adsorbed and fixed on the surface of the carrier belt, and the carrier belt is swept by a directional airflow to remove unadsorbed impurity particles, so as to achieve in-situ purification of pollen. S204, the purified pollen grains are electrostatically neutralized and the neutralized pollen grains are blown from the carrier surface into the detection channel using pulsed gas; In this embodiment, an inertial impactor based on Stokes number design is used to achieve precise collection of pollen of the target particle size. By precisely designing the size and shape of the impactor nozzle and controlling the airflow speed, pollen particles of the target size cannot deviate due to their inertia from the airflow streamline, thus impacting and depositing on the carrier surface; while smaller particles (such as most dust and bacteria) bypass the carrier with the airflow, achieving preliminary separation by size.

[0016] After charged pollen grains are deposited onto the carrier tape, an electrostatic field is created by applying a DC voltage with the opposite polarity to the grain charge (e.g., a positive voltage is applied to the carrier tape if the grains are negatively charged). The field strength is optimized to be between 1 kV and 3 kV per millimeter. This field strength generates a sufficiently strong Coulomb force to firmly adhere the pollen grains to the carrier tape surface to resist interference from subsequent purge airflow, while avoiding excessively high field strength that could cause air breakdown.

[0017] Simultaneously with electrostatic adsorption, a clean airflow is introduced, parallel to the carrier surface and in the opposite direction to the particle collection. The speed of this airflow is controlled between 1 and 3 meters per second. This speed has been experimentally verified: it is sufficient to blow away smaller, weakly charged submicron-sized impurity particles (such as dust and mineral debris) that are not captured by electrostatic adsorption from the carrier surface; however, the shear force generated is far less than the adsorption force of the electrostatic field on the target pollen particles, thus ensuring the in-situ retention of pollen particles and achieving "in-situ purification".

[0018] S205 allows pollen grains in the detection channel to pass through the optical sensor under the constraint of laminar sheath flow, and collects the corresponding photoresist pulse signal and multi-wavelength light scattering signal generated when the pollen grains pass through. S206 processes the photoresist pulse signal, extracts its dynamic feature sequence that changes over time, and determines whether the pollen grains are valid pollen grains based on the particle physical characteristics reflected by the dynamic feature sequence.

[0019] The particle is identified as pollen and counted to output pollen concentration data in real time; the carrier belt is circulated and steps S201 to S206 are performed continuously to achieve uninterrupted online monitoring.

[0020] Furthermore, the separation and collection of pollen grains within the target size range via inertial impaction is achieved through a two-stage series of inertial impactors, wherein... The first-stage inertial impactor is configured to trap particles with an aerodynamic diameter greater than 10 micrometers; The second-stage inertial impactor is configured specifically to collect particles with aerodynamic diameters ranging from 8 to 12 micrometers; It should be explained that, in this embodiment, the first-stage impactor is used to trap coarse particles larger than 10 micrometers (such as willow catkins and large fibers), acting as a pre-filter to prevent them from interfering with subsequent precision collection. The second-stage impactor is specifically optimized for the critical aerodynamic diameter range of 8 to 12 micrometers, which covers the particle size of most important tree and grass pollen (such as birch and grass), ensuring efficient collection of target pollen while excluding smaller respirable dust.

[0021] The carrier tape is a polyester film coated with an indium tin oxide conductive film; this material was chosen as the carrier tape because its overall performance meets the requirements of this application. Conductivity: The indium tin oxide film provides a uniform conductive layer on the carrier surface, enabling the application of a uniform electrostatic field to achieve reliable electrostatic adsorption.

[0022] Flexibility: The polyester film substrate has good flexibility, which allows the carrier tape to be wound on the roller, enabling automatic and continuous feeding and replacement, and meeting the needs of long-term online monitoring.

[0023] Optical transparency: The material has high transparency in the visible light range, a property that offers potential possibilities for subsequent integrated transmission optical observation or calibration via a carrier.

[0024] Applying an electrostatic field to the carrier tape includes applying a DC voltage of 500 to 1500 volts to the conductive layer of the carrier tape; this voltage range is a performance window determined through systematic experiments. When the voltage is below 500 volts, the generated electrostatic force is insufficient to firmly fix pollen particles under all operating conditions, potentially leading to their loss during the purging step; while when the voltage is above 1500 volts, localized air breakdown (electric sparks) can easily occur at the edges of the carrier tape surface or where particles have already accumulated, generating ozone and interfering with stable system operation. Within this preferred range, the optimal balance between adsorption efficiency and system safety is ensured. The directional airflow is a negative pressure airflow opposite to the direction of carrier tape movement. Electrostatic neutralization is achieved through alternating current corona discharge; the pulsed gas is a clean, dry nitrogen pulse with a duration between 5 and 20 milliseconds, and its injection action is controlled by a solenoid valve.

[0025] Nitrogen was chosen as the pulse gas primarily because its inert nature prevents oxidation reactions that might occur near high-voltage discharge areas due to sparks or other unforeseen circumstances, preserving the pollen's natural chemical state—crucial for accurate subsequent activity assessment. The pulse duration was set between 5 and 20 milliseconds, a setting validated through fluid dynamics simulations and experiments. This duration generates sufficient instantaneous momentum to effectively purge neutralized pollen particles from the carrier surface into the detection channel while ensuring it is a brief "pulse" that does not continuously impact the detection channel, thus maximizing the stability of the laminar sheath flow within the detection channel. This is a prerequisite for obtaining high-precision optical detection signals. The jetting action is controlled by a high-speed solenoid valve to ensure precise timing.

[0026] Further, in S206, the photoresist pulse signal is processed to extract its dynamic feature sequence that changes over time, including: For each photoresist pulse signal, perform the following operations: Determine the start and end times of the photoresist pulse signal; Within the time period between the start and end times, the time axis is evenly divided into multiple consecutive time windows, with more than 5 time windows. It's important to explain that dividing the time axis of the photoresist pulse signal evenly into multiple consecutive time windows aims to convert the continuous analog signal into a characteristic sequence that can characterize the dynamic changes in particle shape. Setting the number of time windows to more than five is based on the analysis of the physical process of a typical pollen grain passing through the optical sensor beam. The basic photoresist pulse signal generated by a single pollen grain typically exhibits a relatively smooth single-peak shape. If the number of windows is too small (e.g., less than five), the sequence is too coarse and cannot effectively depict the subtle dynamics of the waveform; while too many windows reduce the amount of data within each window, decreasing the ability to resist random noise. Experiments have shown that setting the number of windows between six and twelve effectively smooths the signal while most sensitively capturing the light intensity occlusion patterns caused by the specific shape of the pollen grain (e.g., spherical, ellipsoidal, or spiky), laying the foundation for accurate subsequent discrimination.

[0027] For each time window, perform the following operations: Calculate the minimum value of the voltage signal within the time window; divide the minimum value by the minimum voltage value of the entire process of the photoresist pulse signal to obtain the normalized amplitude value of the time window; Divide the duration of this time window by the total duration of the photoresist pulse signal to obtain the local pulse width value of this time window; Arrange the normalized amplitude values ​​of all time windows in chronological order to form an amplitude sequence; Arrange the local pulse width values ​​of all time windows in chronological order to form a pulse width sequence.

[0028] It should be noted that normalizing the amplitude and pulse width in this invention is a crucial data preprocessing step. Its primary purpose is to eliminate interfering factors unrelated to the physical properties of the particles themselves. Specifically: Normalized Amplitude: The original amplitude of the photoresist pulse signal is strongly dependent on the particle's charge and velocity. A particle with a high charge and slow velocity, even with the same shape, will have a much larger signal amplitude than a particle with a low charge and fast velocity. By dividing the local minimum within each time window by the total minimum of the entire pulse signal, we obtain a relative amplitude sequence. This sequence eliminates the influence of absolute amplitude, thus making the morphological changes in the sequence only related to the shape and orientation of the particles.

[0029] Normalized pulse width: Similarly, the particle's passing velocity directly affects the total pulse width and the relative duration of each part. By calculating the ratio of local duration to total duration, we obtain a relative pulse width sequence that excludes velocity variables and more purely reflects the time-proportional characteristics of the beam being blocked by different parts of the particle.

[0030] After the above normalization process, the final amplitude sequence and pulse width sequence are "dynamic fingerprints" that eliminate interference from charge and velocity and can more fundamentally reflect the shape and structural characteristics of particles.

[0031] Based on the embodiments provided in this application, instead of relying on static characteristics such as the overall amplitude or width of traditional photoresist signals, the local details of the signal changing over time are quantified to capture the transient behavior of particles as they pass through the optical sensor. The normalized amplitude value, by dividing it by the minimum value of the signal throughout the entire process, eliminates the influence of absolute voltage fluctuations, making the characteristics more robust; the local pulse width value, by dividing it by the total duration of the signal, reflects the velocity distribution of the particles within the detection area. This time-series analysis can more precisely characterize the physical properties of particles, such as shape, surface roughness, or orientation changes as they pass through the sensor.

[0032] In online pollen monitoring scenarios, pollen particles in ambient air are often accompanied by impurities such as dust and fungal spores, and pollen itself may exhibit non-spherical or complex structures depending on the species. By extracting dynamic feature sequences, the system can effectively distinguish pollen from impurities: for example, pollen particles typically have smooth surfaces and uniform shapes, and their photoresistivity signals often exhibit stable temporal evolution patterns, while impurity particles may cause drastic signal fluctuations due to irregular shapes or aggregation behavior. Amplitude sequences can reveal continuous patterns of particle obstruction of the light path, while pulse width sequences can reflect the acceleration or rotational behavior of particles as they pass through the detection zone. This fine-grained feature extraction enhances the system's ability to identify the physical properties of particles, providing richer and more stable input data for subsequent discrimination and reducing the risk of misjudgment due to signal noise or particle overlap. Furthermore, since the features are calculated based on time window normalization, this method is adaptable to changes in sensor sensitivity or airflow fluctuations, improving the reliability of the monitoring system during long-term operation.

[0033] Further, in S206, based on the particle physical characteristics reflected by the dynamic feature sequence, determining whether a pollen grain is a valid pollen grain includes: Calculate the correlation coefficient between the amplitude sequence and the pre-stored standard bell curve; The standard bell-shaped curve is a key benchmark for determining whether a pollen grain is a typical single pollen grain. The establishment of this curve is a statistical modeling process based on a large amount of experimental data. Specifically, under controlled experimental conditions, a large number of known single pollen grains (such as fir and dandelion pollen) were collected and confirmed by microscopic examination. When these grains passed through an optical sensor, the system acquired the photoresist pulse signals they generated and extracted the amplitude sequence of each signal. Subsequently, the amplitude sequences of all qualified signals were normalized and aligned, and fitted using a Gaussian function to obtain the fitting curve corresponding to each sequence. Finally, hundreds or thousands of such fitting curves were averaged to obtain a smooth and representative "standard bell-shaped curve." This curve essentially reflects the concentrated manifestation of the relatively rounded and symmetrical morphological characteristics common to typical pollen grains in the photoresist signal.

[0034] Calculate the standard deviation of the pulse width sequence; When the correlation coefficient is higher than the first preset threshold and the standard deviation is lower than the second preset threshold, the photoresist pulse signal is determined to correspond to a valid pollen grain, and the count is performed. In one embodiment, the correlation coefficient is used to quantify the morphological linear similarity between the amplitude sequence of the measured particles and an ideal pollen model (standard bell curve). This invention uses the classic Pearson product-moment correlation coefficient for calculation, but its variable is explicitly defined as the sequence data in this scenario.

[0035] Correlation coefficient The calculation formula is as follows:

[0036] in, This represents the normalized amplitude value of the i-th time window in the amplitude sequence of the measured particle; This represents the standard amplitude value corresponding to the i-th time window on the standard bell curve; This represents the average value of all normalized amplitude values ​​in the entire amplitude sequence of the measured particle; Σ represents the average of all standard amplitude values ​​of the standard bell curve; Σ represents the summation over all time windows (from i=1 to i=N, where N is the number of time windows).

[0037] The range of its value is [-1, 1]. In this application, The closer the value is to 1, the more synchronized and consistent the amplitude sequence of the measured particle is with the morphology of the standard bell curve, that is, the closer the shape of the particle is to a regular and typical pollen grain.

[0038] The first preset threshold is used to measure the similarity between the amplitude sequence of the tested particles and the morphology of the "standard bell curve". The higher the threshold is set, the stricter the discrimination standard is, and the higher the requirement for the "typicality" of the particle morphology. This helps to exclude non-pollen particles with irregular shapes, but may also misidentify some pollen with special shapes. The threshold is set slightly lower, and the system has a stronger inclusiveness, but may introduce more impurities and interference.

[0039] Specific value examples: Example 1 (High-strictness scenario): In urban summer environments with high pollen concentrations and complex impurity backgrounds, or when focusing on monitoring pine pollen with very regular morphology, the first preset threshold can be set to 0.85. This means that only particles whose amplitude sequences are highly similar to the ideal pollen morphology (correlation coefficient > 0.85) will be counted, thus ensuring high purity of the counting results. Example 2 (High-sensitivity scenario): When pollen concentrations are low, or when monitoring herbaceous plant pollen with high morphological diversity (such as some grass pollen), to avoid missed detections, the first preset threshold can be appropriately relaxed to 0.75. This allows for the capture of more real pollen particles while ensuring the effectiveness of the core discrimination logic.

[0040] In one embodiment, the standard deviation is used to measure the dispersion of the normalized pulse width sequence of the tested particles around its mean, thereby determining the uniformity of the particle shape.

[0041] in, This represents the local pulse width value (i.e., the normalized duration) of the i-th time window in the pulse width sequence of the measured particle. This represents the arithmetic mean of all local pulse width values ​​in the entire pulse width sequence of the particle. Σ represents the total number of time windows; Σ represents the summation over all time windows (from i=1 to i=N).

[0042] The standard deviation σ quantifies the stability of the "relative width" of the beam obstruction at different points as the particle passes through the detection beam. A uniformly shaped, regular pollen grain (such as a standard sphere) will have a very low σ value because the pulse width values ​​are very similar across different time intervals. Conversely, a long, thin fiber or a highly irregular dust particle will have significant differences in the time intervals during which different parts of the grain obstruct the beam, leading to drastic fluctuations in the pulse width sequence and resulting in a very high σ value.

[0043] The second preset threshold is used to determine the stability of the "width variation" of the particle as it passes through the light beam. An ideal, uniformly shaped spherical or ellipsoidal pollen grain blocks the light beam with a gradual change in "width," resulting in minimal fluctuation in its normalized pulse width sequence, i.e., a low standard deviation. Conversely, a long, thin fiber or irregular dust particle blocks the light beam for significantly different durations, leading to a higher standard deviation in the pulse width sequence.

[0044] Specific value examples: Example 1 (High-strictness scenario): To effectively filter out significantly irregular impurities such as fibers and dust particles, the second preset threshold can be set at a lower level, such as 0.08. Any signal with a pulse width sequence standard deviation exceeding 0.08 will be judged as shape-unstable rather than a single pollen grain. Example 2 (High-inclusivity scenario): When the target pollen itself has a certain degree of irregularity (such as protruding surface patterns), or to accommodate slight fluctuations in carrier speed, the second preset threshold can be appropriately relaxed to 0.12. This allows the grains to "wobble" or "tumble" slightly more when passing through the beam, without being easily discarded.

[0045] In practical implementation, the first preset threshold and the second preset threshold must be used in conjunction to form a two-dimensional discrimination interval. For example, in one embodiment, the first preset threshold = 0.80 and the second preset threshold = 0.10 can be used in combination. This combination has been experimentally verified to effectively exclude the vast majority of common airborne non-pollen particles while ensuring a high capture rate for typical tree and grass pollen, achieving a good balance between accuracy and recall. The specification should clearly state that these thresholds are configurable parameters, and their optimal values ​​can be ultimately determined through controlled experiments on known samples under specific conditions.

[0046] Specifically, when the waveform of a photoresist pulse signal exhibits multiple separate peaks, or when the symmetry index of its waveform is lower than a preset symmetry threshold, it is determined that the photoresist pulse signal is generated by the superposition of multiple particles or by irregular impurities, and the photoresist pulse signal is discarded and not processed. The symmetry index of the waveform is obtained by calculating the similarity between the rising and falling edges of the photoresist pulse signal.

[0047] The waveform symmetry index is obtained by calculating the similarity between the rising and falling edges of the photoresist pulse signal. In one embodiment, the present invention proposes a symmetry index calculation method based on area ratio, which directly quantifies the similarity between the rising and falling edges of the photoresist pulse signal.

[0048] Symmetry index The calculation formula is as follows:

[0049] in, Represents the rising edge area, which is calculated as the area enclosed by the signal curve and the baseline (zero voltage line) between the start time and the peak time of the photoresist pulse signal. This represents the area over the falling edge. It is calculated as the area enclosed by the signal curve and the baseline between the peak time point and the end time point of the photoresist pulse signal.

[0050] The formula defines the symmetry index SI by calculating the ratio of the smaller to the larger area of ​​the rising and falling edges. For an ideally symmetrical pulse signal, the areas of the rising and falling edges are equal, and the SI value is 1. For an asymmetrical signal, such as a slow rise and a sharp fall (or vice versa), the two areas differ significantly, and the SI value will approach 0.

[0051] This area comparison method, compared to simple peak point symmetry checks, is better able to capture the overall morphological differences in the signal waveform. It is extremely sensitive to determining whether particles are regular, whether they are composed of multiple superimposed particles, or whether their shapes are severely irregular (such as fibers or irregular dust particles). In the discrimination logic of this invention, when the SI value of a pulse signal is lower than a preset symmetry threshold, it is determined to be generated by atypical particles and discarded, thereby significantly improving the accuracy of single pollen particle counting.

[0052] The preset symmetry threshold is a criterion threshold applied to symmetry indicators, used to directly filter out regularly shaped particles at the signal level. It aims to distinguish symmetrical signals generated by a single, regular pollen particle from asymmetrical signals generated by the superposition of multiple particles or irregular impurities.

[0053] An ideal, uniformly shaped, and singular spherical or ellipsoidal pollen grain, when passing through a light beam, generates a photoresistive pulse signal with a rising edge area ( ). ) and the area of ​​the falling edge ( The values ​​are very close, and the calculated SI value approaches 1.

[0054] Conversely, when multiple particles pass close together (signal superposition), or when the particles are long fibers or irregular dust particles, the rising and falling edge patterns of their signals will differ greatly, resulting in one area being much larger than the other, thus causing the SI value to approach 0.

[0055] Therefore, the preset symmetry threshold is set between 0 and 1. The higher the threshold, the more stringent the system's requirements for the regularity and uniformity of particle shape.

[0056] Specific value examples: Example 1 (High stringency scenario): In health early warning networks requiring extremely high counting accuracy, to minimize interference from any non-single pollen grains, the preset symmetry threshold can be set to 0.82. This means that only when the SI value ≥ 0.82, i.e., the areas of the rising and falling edges are very close, is the signal judged as highly symmetrical, and the grain will proceed to the next stage. This high threshold effectively filters out most signals caused by grain overlap or fibrous impurities. Example 2 (High inclusiveness scenario): In plant diversity surveys, there are many types of target pollen, some of which are not perfectly spherical or ellipsoidal in shape (e.g., some pollen with special protrusions). To avoid misclassifying these "uniquely shaped but still pollen" grains as impurities, the preset symmetry threshold can be appropriately relaxed to 0.65. This allows the system to accommodate more diverse pollen morphologies. Although this may introduce a small number of irregular impurities, it ensures the integrity of biodiversity data, which can then be further identified using other features (such as scattering patterns).

[0057] Based on the embodiments provided in this application, statistical features are combined with morphological analysis. The standard bell curve represents the signal waveform of an ideal pollen grain (typically exhibiting a smooth bell shape due to the spherical symmetry and uniform velocity of the grain). The correlation coefficient quantifies the similarity between the actual signal and the ideal model. The standard deviation of the pulse width sequence assesses the temporal stability of the signal, avoiding distortion caused by sudden changes in grain velocity or vibration interference. Symmetry indices further identify waveform distortions, such as multiple peaks or asymmetrical rising / falling edges, which are typically characteristics of multiple superimposed grains or impurity grains.

[0058] In real-time pollen monitoring, particle streams may contain pollen aggregates, abiotic particles (such as mineral dust), or foreign matter, all of which can generate abnormal photoresistivity signals. Through multi-feature fusion discrimination, the system can efficiently screen out single, complete pollen particles: a high correlation coefficient indicates regular particle shape and stable passing behavior; a low standard deviation indicates uniform particle passage through the detection zone; and symmetry indicators can exclude multi-peak signals caused by particle collisions or impurities. This discrimination mechanism significantly reduces false positives (e.g., miscounting impurities as pollen) and false negatives (e.g., undercounting deformed pollen), improving counting accuracy. Furthermore, automated discrimination is achieved through preset thresholds, reducing reliance on manual calibration and making it suitable for unattended online monitoring environments. Using a bell curve as a physical model, the dynamic behavior of particles is matched with ideal features, thereby reliably extracting target signals in complex backgrounds.

[0059] Furthermore, the method also includes: For each photoresist pulse signal, perform the following operations: S207, the correlation coefficient between the amplitude sequence of the photoresist pulse signal and the standard bell curve, and the standard deviation of the pulse width sequence are combined into the first discriminant vector; It is important to clarify that each photoresist pulse signal and its resulting first discrimination vector strictly correspond to a synchronously acquired multi-wavelength light scattering signal and its scattering spectrum. This is achieved through hardware timing control: when a particle blocks the detection optical path at the start of the photoresist pulse signal generation, the system synchronously triggers the acquisition of multi-wavelength scattering signals. Therefore, each processed particle simultaneously possesses a first discrimination vector and a scattering spectrum.

[0060] The scattering spectrum matching degree is a quantified similarity score used to evaluate the degree of agreement between the current particle's scattering spectrum and a pre-stored standard pollen scattering spectrum. Its calculation can be based on the principles of image matching or feature vector comparison. In a preferred embodiment, the system normalizes the scattering spectrum into a standard-sized two-dimensional intensity distribution map, and then calculates its structural similarity index with the standard spectrum. This index comprehensively compares brightness, contrast, and structural information, outputting a value between 0 and 1; the higher the value, the better the matching degree.

[0061] S208: The scattering spectrum formed by the multi-wavelength light scattering signal is matched with the pre-stored standard pollen scattering spectrum to obtain the spectrum matching degree, which is used as the second discrimination vector; S209, based on the first and second discrimination vectors, controls the microfluidic sorting chip to sort particles into three physically isolated channels: S210, when the correlation coefficient is higher than the first preset threshold, the standard deviation is lower than the second preset threshold, and the spectral matching degree is higher than the third preset threshold, the pollen grains are sorted into the first channel for collecting and confirming pollen; the activity analysis of the pollen grains sorted into the first channel is performed using the resource manager. S211, when the correlation coefficient is lower than the fourth preset threshold, or the standard deviation is higher than the fifth preset threshold, or the spectral matching degree is lower than the sixth preset threshold, the pollen grains are sorted to the second channel for removing impurities. It should be understood that the setting of these thresholds needs to be determined based on the analysis of the distribution of the first discriminant vector and the spectral matching degree in two-dimensional space after testing a large number of known types of particles (such as pure pollen, mineral dust, carbon black particles, and plant fibers).

[0062] Specifically, typical pollen grains are concentrated in regions with "high correlation coefficient, low standard deviation, and high spectral matching degree".

[0063] Spherical non-pollen particles (such as fog droplets) may have high spectral matching, but their photoresist signals may exhibit different dynamic characteristics (correlation coefficients or standard deviations) because they are droplets.

[0064] Fibrous impurities typically result in extremely low spectral matching (due to the significant difference between their scattering patterns and those of globular pollen) and anomalous light-blocking signals.

[0065] The third preset threshold (high spectral matching threshold) and the sixth preset threshold (low spectral matching threshold) are used to define the scattering feature boundaries between "clearly pollen" and "clearly non-pollen". The fourth and fifth preset thresholds serve as supplementary boundaries for the first discrimination vector. By analyzing the clustering of these particle groups in the data space, the optimal combination of thresholds that can distinguish different types of particles can be found.

[0066] The following specific example illustrates the setting of these thresholds: The third preset threshold (high threshold for spectral matching) can be set to 0.80. This means that when the matching degree between the particle's scattering spectrum and the standard pollen spectrum is higher than 0.80, it is considered "very similar to pollen" in terms of scattering optical characteristics. The fourth preset threshold (low threshold for correlation coefficient) can be set to 0.70. This means that when the correlation coefficient between the particle's photoresistance pulse signal amplitude sequence and the standard bell curve is lower than 0.70, its dynamic waveform characteristics are considered "severely deviating" from typical pollen morphology. The fifth preset threshold (high threshold for standard deviation) can be set to 0.15. This means that when the standard deviation of the particle's pulse width sequence is higher than 0.15, its shape uniformity when passing through the beam is considered "extremely unstable." The sixth preset threshold (low threshold for spectral matching) can be set to 0.40. This means that when the spectral matching degree of the particle's scattering spectrum is lower than 0.40, it is considered "completely unlike pollen" in terms of scattering optical characteristics.

[0067] In this configuration, a particle is sorted into the first channel if it meets all the following criteria: correlation coefficient > 0.70 (in fact, due to the condition of weight 4, it must be > 0.85), standard deviation < 0.15 (again, due to weight 4, it must be < 0.08), and spectral matching degree > 0.80. This ensures that the particles in the first channel are high-purity pollen with highly consistent optical characteristics.

[0068] The condition for being sorted to the second channel is "complete mismatch of any characteristic": correlation coefficient < 0.70, or standard deviation > 0.15, or spectral matching degree < 0.40. This effectively removes clearly non-pollen impurities directly from the system.

[0069] All "questionable particles" that do not meet the above strict conditions but are not completely non-compliant will enter the third channel for re-inspection.

[0070] S212, For pollen grains that do not meet the above two conditions, sort them to the third channel for temporary storage of pollen grains with doubts; use the resource manager to schedule and perform activity analysis on the pollen grains sorted to the third channel. Particles sorted to the first channel are marked as high-confidence pollen and immediately enter an activity analysis queue; particles sorted to the third channel are marked as questionable particles and temporarily stored in a buffer queue. The resource manager schedules suspicious particles in the buffer queue for activity analysis detection only when the activity analysis queue is empty; the resource manager is used to prioritize scheduling particles in the activity analysis queue for activity analysis detection.

[0071] It should be noted that the system configures a resource manager, which is a scheduling logic unit integrated within the system controller. The specific implementation of the resource manager is as follows: Hardware carrier: The resource manager is a set of dedicated scheduling firmware or programs running in the core processing unit of the system (such as an ARM series processor or an FPGA chip).

[0072] Logical Flow: Its workflow can be described as a clear state machine: Listening and Input: Continuously listen to the sorting results. Once a particle is sorted, its information (such as particle ID, feature data) is pushed into the corresponding queue data structure (activity analysis queue or buffer queue) according to its channel source. Priority Arbitration: Always prioritize checking the activity analysis queue. As long as the queue is not empty, take a particle task from its head and immediately trigger the subsequent activity analysis detection process. Idle State Detection and Scheduling: The resource manager switches to the idle scheduling state and begins checking the buffer queue only when the activity analysis queue is empty. If the buffer queue is not empty, take a task from its head for processing.

[0073] Looping and Reset: After completing the detection of one particle, the system immediately returns to the "Priority Arbitration" step to continue checking the activity analysis queue, forming a loop. When the system resets or receives a clear command, it will clear both queues.

[0074] A microfluidic sorting chip is a device integrating micron-scale fluid channels and electrodes. Its sorting principle is based on the dielectrophoresis effect or electrostatic deflection. In a preferred embodiment of the invention, the chip has multiple pairs of electrodes at the channel bifurcation points. When particles flow through the bifurcation, the system applies a momentary high-frequency AC voltage or DC pulse to specific electrode pairs based on the sorting decision of the resource manager. This electric field generates a lateral force on the particles (since they are already charged) sufficient to deflect them from the main flow, thus directing them into a preset collection channel, waste channel, or buffer channel. The entire sorting process is completed within milliseconds, ensuring matching with the particle flow rate.

[0075] Based on the embodiments provided in this application, classification reliability is improved through multi-sensor data fusion (photoresistance and light scattering). Specifically, light scattering signals can provide optical properties such as the internal structure and refractive index of particles, compensating for the shortcomings of photoresistance signals in component identification. Microfluidic sorting achieves the physical separation of particles, while the resource manager optimizes the allocation of detection resources through queue management, ensuring that high-value samples are processed first.

[0076] In pollen monitoring scenarios, particle composition is complex (e.g., pollen, spores, pollutants), and single sensors are easily limited. Combining light obscuration and light scattering characteristics can cross-verify particle identity: for example, pollen typically has a specific light scattering pattern (due to cell wall structure), while impurities may exhibit abnormal patterns. After sorting into three channels, high-confidence pollen directly enters the activity analysis queue, avoiding resource waste on clearly identified impurities; questionable particles are temporarily stored in a buffer queue and re-examined only when the system is idle, balancing real-time performance and accuracy. This hierarchical processing strategy is particularly suitable for sudden pollen outbreaks, allowing the system to quickly respond to high-concentration samples while avoiding overload through a buffering mechanism. Prioritized scheduling by the resource manager ensures that critical samples (such as high-confidence pollen) receive timely activity analysis, improving overall monitoring efficiency.

[0077] Furthermore, in S210, the activity of pollen grains sorted to the first channel is analyzed using a resource manager, including: A priority value is calculated for each pollen grain entering the first channel. The calculation rules for the priority value include: the lower the symmetry index of the waveform of the photoresist pulse signal and / or the lower the spectral matching degree, the higher the priority value is assigned. This defines an initial, fixed computational rule. It can be understood as the system's "factory settings" or "baseline strategy." This counterintuitive rule reflects profound insight. It assumes that particles with imperfect or atypical optical characteristics may contain greater informational value. These could be: rare species whose standard characteristics have not been fully learned by the system; pollen with abnormal morphology, potentially indicating varying maturity, activity, or damage; or aggregates with non-pollen particles, requiring in-depth analysis for definitive confirmation. Therefore, the purpose of this rule is to drive the system to proactively explore the "unknown" and "abnormal," rather than repeatedly confirming the "known."

[0078] In other words, this strategy is based on information gain as its core decision-making principle. The system treats the limited and expensive fluorescence lifetime detection as an opportunity to "ask questions." Asking a particle with very typical optical characteristics will likely yield an answer (i.e., its activity) that is similar to that of most typical pollen, resulting in low information gain, similar to confirming a known answer.

[0079] Questioning a particle with anomalous optical characteristics (low symmetry, low matching degree) may reveal: 1) the presence of a rare pollen species; 2) a pollen with a special activity state (e.g., inactive, germinating); or 3) a novel interfering substance that has not been previously identified. Regardless of the outcome, this provides significant information gain to the system, greatly enhancing its depth and breadth of understanding of the monitored environment. Therefore, prioritizing the detection of anomalous particles is an intelligent strategy that maximizes monitoring value and learning capabilities under resource-constrained conditions.

[0080] Maintain a detection resource counter based on the number of pulsed laser cycles and the cycle budget; The current particle's priority value is compared with the dynamic priority threshold. Only when the priority value is higher than the dynamic priority threshold and the current value of the detection resource counter is greater than zero, the execution activity analysis detection is triggered, and the detection resource counter is deducted after execution. The dynamic priority threshold is adjusted in reverse based on the ratio of the current remaining value of the detection resource counter to the total budget value.

[0081] To implement the above strategy and manage resources, this invention proposes the following dynamic threshold adjustment model:

[0082] in, This is a dynamic priority threshold. It's a threshold that changes as resources are consumed; a particle's priority value must be higher than this threshold to be detected. This is the base priority threshold. This is the initial threshold for the system when resources are plentiful, and it can be configured according to different monitoring sensitivity requirements (e.g., set to 0.5). To detect the current remaining value of the resource counter; The total budget value for detecting resource counters; K is the adjustment intensity coefficient, a constant greater than 0 (e.g., it can be set to 0.5). The larger the K value, the faster the threshold rises when resources decrease.

[0083] When resources are plentiful, the ratio in the formula approaches 1, and its negative Kth power also approaches 1. ≈ The system has a low barrier to entry, allowing a large number of particles to be detected.

[0084] As resources are continuously consumed, the ratio gradually decreases, while its negative K-th power value gradually increases, leading to... The numbers continue to rise. This means the system will become increasingly "selective," with only the highest priority particles—those that are the "most anomalous"—being eligible to consume the last of the precious detection resources.

[0085] This model achieves a transition from "lenient" to "strict" resource allocation in a non-linear and smooth manner, ensuring optimal resource utilization throughout the budget cycle. Based on the embodiments provided in this application, particle features are dynamically correlated with the system resource status. Specifically, the priority value depends inversely on symmetry and matching degree (i.e., the worse the feature, the higher the priority), so that particles with abnormal signals or low matching degree (which may be rare species or degenerated pollen) are given priority detection. The dynamic threshold is automatically adjusted according to the resource surplus, and the threshold is increased when resources are scarce to focus on the most critical samples.

[0086] In long-term online monitoring, activity analysis (such as fluorescence detection) is typically energy-intensive and resource-constrained. Adaptive strategies optimize resource allocation: for example, when resource counters are sufficient, the system can relax thresholds to perform more detections; when resources are scarce, only the highest-priority particle detection is triggered to avoid resource depletion. This extends the system's autonomous runtime, making it particularly suitable for field or power-constrained scenarios. Through priority values, the system can proactively focus on "abnormal" particles (e.g., low symmetry may indicate deformed pollen, low matching degree may indicate a new species), enhancing the comprehensiveness of monitoring. Dynamic thresholds transform resource budgets into decision parameters, enabling the system to self-regulate, ensuring high-confidence pollen detection while not missing potentially important samples.

[0087] Furthermore, such as Figure 3 As shown, the activity analysis detection in S210 and S212 includes a re-examination using a pulsed ultraviolet laser emitting two sub-pulses and coordinating gating to acquire fluorescence signals. Specifically, this includes: S301 is equipped with a pulsed ultraviolet laser with a dual-pulse drive circuit. Within the time window when pollen grains pass through the detection area, the pulsed ultraviolet laser is controlled to emit two sub-pulses with equal energy and an interval of 0.1 to 5 microseconds. The ultraviolet sub-pulse interval, set to 0.1 to 5 microseconds, is designed specifically for the fluorescence lifetime (typically in the nanosecond range) of endogenous pollen fluorophores (such as sporophytin and flavonoids). This interval achieves the key measurement objective: the first sub-pulse is used to excite fluorescence and measure its initial decay behavior; the second sub-pulse is applied when the fluorescence of the first pulse has almost completely decayed, but before severe irreversible photochemical changes occur. At this point, if pollen activity is high and the internal environment is stable, the fluorophore can be effectively excited again, and the decay curves of the two excitations are similar; if activity is low, disturbances in the intracellular environment may cause energy transfer or minor photobleaching of the fluorophore after the first excitation, resulting in a difference in the fluorescence lifetime (τB) obtained from the second excitation compared to the first (τA). This time window is the optimal choice for observing this subtle dynamic process.

[0088] S302 controls the opening time of the gating gate to synchronize it with each sub-pulse in order to suppress ambient background fluorescence interference; The timing of the gating is calibrated through the following steps: First, a standard material with a known fluorescence lifetime (such as fluorescent microspheres) is used for testing. The system acquires its continuous fluorescence decay signal, and by analyzing this signal, the gating activation time (delay) is precisely set to the moment after the laser pulse ends and before the fluorescence signal reaches its peak. Simultaneously, the gating activation duration is set to 3 to 4 times the known fluorescence lifetime value of the standard material. This calibration ensures that most of the fluorescence decay process is captured to obtain an accurate lifetime value, while minimizing early background stray light and noise after the laser pulse ends.

[0089] S303, within the gating window, synchronously acquires the fluorescence signals excited by two sub-pulses through a photodetector, and obtains the first fluorescence decay curve and the second fluorescence decay curve respectively; S304, The fluorescence lifetime value τA of the first fluorescence decay curve and the fluorescence lifetime value τB of the second fluorescence decay curve are calculated using a fitting algorithm. In this embodiment, the fitting algorithm is used to extract the fluorescence lifetime value from the fluorescence decay curve. Specifically, the collected fluorescence decay data (i.e., the curve of fluorescence intensity changing with time) is input into a nonlinear least squares fitting algorithm. This algorithm searches for an optimal single exponential decay function, i.e. Multiply by exp(-t / τ).

[0090] Where t represents the time calculated from the start point of the fluorescence decay process (i.e., the instant the laser pulse ends). I(t) represents the fluorescence intensity measured at time t. The initial fluorescence intensity at time t=0 represents a parameter to be fitted. τ represents the fluorescence lifetime, i.e., the time from the initial fluorescence intensity down to the fluorescence lifetime. The time required to obtain 1 / e (approximately 36.8%) of the result. This is the core objective parameter we solve for through fitting.

[0091] The task of the fitting algorithm is to find an optimal set of... The goal is to find the optimal parameter τ such that the overall error (usually expressed as the sum of squared residuals) between the theoretical curve calculated from model I(t) and all data points actually collected in the experiment is minimized. This is a standard optimization problem that can be solved using mature numerical methods such as the Levenberg-Marquardt algorithm. The system independently performs the above fitting process on the first and second fluorescence decay curves to obtain two fluorescence lifetime values ​​τA and τB. This minimizes the overall difference between the function curve and the experimentally collected data points. Through this iterative optimization calculation, the algorithm finally outputs the fluorescence lifetime value τ that best represents the characteristics of the decay curve. Performing this operation on the first and second decay curves respectively yields τA and τB.

[0092] S305, calculate the photostability index of the corresponding detected event according to the formula |τA-τB| / ((τA+τB) / 2); The photostability index (|τA-τB| / ((τA+τB) / 2)) is an indicator reflecting the ability of pollen grains to maintain their fluorescent physical properties under microsecond-level laser perturbation. Biologically, highly active pollen has an intact cell membrane system and a stable internal redox environment, which helps the fluorophore to quickly recover to its ground state after stimulation, exhibiting high photostability (i.e., the difference between τA and τB is small, and the index value is low). Conversely, pollen with low activity or inactivation has damaged cell structure and a chaotic internal environment, making it more susceptible to energy transfer or photoinduced damage, resulting in a significant difference in lifetime values ​​between the two measurements (high index value). This correlation has been verified by trypan blue staining: experiments have shown that pollen grain populations stained blue with trypan blue (damaged cell membrane, inactivation) under a microscope have a significantly higher average photostability index than unstained populations (intact cell membrane, high activity).

[0093] S306, compare the photostability index with a preset activity threshold based on the pollen species. When the photostability index is lower than the preset activity threshold, determine that the pollen grains captured in this detection event are highly active pollen.

[0094] Among these, the activity threshold based on pollen type needs to be preset according to the biological characteristics of different pollens. For example: Pine pollen has a relatively low metabolic rate and stable cellular components, so its activity threshold can be set at 0.08. When the photostability index is below 0.08, it is considered highly active pine pollen.

[0095] Dandelion pollen has a more active metabolism and is more sensitive to external stimuli, so the activity threshold can be appropriately relaxed to 0.12. Only when the photostability index is below 0.12 can it be considered as highly active dandelion pollen.

[0096] Based on the embodiments provided in this application, a dual-pulse time-resolved fluorescence technique is used: two sub-pulses excite the same particle within an extremely short interval, and the collected fluorescence decay curves should be consistent (highly active pollen); the photostability index quantifies the consistency difference between the two measurements, with a low index indicating stable fluorescence lifetime and corresponding to high activity. This method reduces errors caused by particle movement, laser fluctuations, or environmental interference in a single measurement.

[0097] Pollen viability is a core indicator for monitoring, but traditional fluorescence methods are susceptible to background fluorescence or photobleaching. The dual-pulse design allows for two independent measurements the instant a particle passes through the detection zone, assessing viability by comparing differences in fluorescence lifetime: highly viable pollen typically possesses stable fluorophores (such as NADPH) with consistent lifetimes; while degraded or dead pollen may experience lifetime fluctuations due to internal molecular changes. Gating and synchronization with sub-pulses suppress environmental background fluorescence (such as other fluorescent particles in the air), improving the signal-to-noise ratio. The photostability index, as a normalized indicator, avoids the dependence of absolute lifetime values ​​on laser energy, enhancing the method's robustness. This viability detection method can distinguish between pollen in different physiological states, providing more accurate data for pollen dispersal research or allergy early warning.

[0098] Furthermore, the synchronization between the gating gate and the dual-pulse drive circuit is achieved in the following way: When the dual-pulse drive circuit transmits each sub-pulse, it simultaneously outputs a synchronous trigger signal to the gating control circuit. The gating circuit turns on after receiving the synchronous trigger signal and then delays for a specific time. The specific delay time is calibrated according to the response characteristics of the photodetector and the circuit transmission delay. The duration of gating is set based on the expected value of fluorescence lifetime to ensure that the fluorescence decay process can be fully captured, and its range is 1 to 5 times the expected value of fluorescence lifetime.

[0099] Based on the embodiments provided in this application, signal acquisition timing is optimized through hardware-level synchronization: delay calibration ensures that the gating window is aligned with the fluorescence emission peak, avoiding signal loss due to circuit delay; duration setting is based on fluorescence lifetime characteristics, ensuring that most fluorescence photons are captured while minimizing background noise.

[0100] In fluorescence detection, timing accuracy directly impacts signal quality. Through a synchronization mechanism, the gating system activates only after sub-pulse excitation, effectively shielding against excitation light scattering and continuous background light (such as sunlight or indoor illumination), thus improving the signal-to-noise ratio of the fluorescence signal. Delay calibration compensates for detector response time and signal transmission delay, avoiding signal distortion caused by mismatch between the acquisition window and fluorescence emission time. The duration is set based on fluorescence lifetime (e.g., pollen fluorescence lifetime is typically in the nanosecond range), ensuring the window is long enough to capture the decay process without being excessively long and introducing noise. This precise timing control enables the system to operate stably in high-background environments, improving the accuracy and repeatability of activity analysis. Directly linking fluorescence physical properties (lifetime) with hardware parameters (delay, duration) achieves adaptive optimization of the acquisition window.

[0101] Furthermore, the method also includes: Record the waveform symmetry index of all pollen grains that have undergone activity analysis and their photostability index obtained by double-pulse fluorescence measurement. Regularly analyze the statistical correlation between waveform symmetry index and optical stability index in all recorded data; When the statistical correlation deviates from the historical benchmark within a preset range, the calculation rules for the priority values ​​are adjusted according to the direction and degree of the deviation. Specifically, the weight coefficients of the waveform symmetry index and the spectrum matching degree in the calculation are adjusted.

[0102] For statistical correlation, this embodiment uses the Pearson product-moment correlation coefficient to calculate the degree of linear correlation between the waveform symmetry index and the photostability index. This coefficient calculates the trend of the co-change of these two variables in all detection data over a fixed period (e.g., 24 hours).

[0103] Deviation within the preset range refers to the absolute difference between the calculated current Pearson correlation coefficient value and a historical value used as a benchmark (such as the system's initial calibration value or the average value of the previous week). The system presets a deviation threshold, for example, 0.15. This means that when |current correlation coefficient - historical benchmark coefficient| > 0.15, it is considered a "significant deviation" and the weight adjustment mechanism needs to be triggered.

[0104] In this embodiment, a later, dynamic optimization process is defined. After the system has been running for a period of time, the importance of each factor in the initial strategy will be reassessed based on the actual detection results.

[0105] When the system determines that the correlation has deviated significantly, it will adjust the system according to the "direction and degree of deviation," as follows: Deviation in direction: If the current correlation coefficient is lower than the historical benchmark, it indicates that the predictive ability of "waveform symmetry" for "optical stability" has deteriorated. In this case, the system should reduce the weight of the waveform symmetry index in the priority calculation.

[0106] Degree of deviation: The magnitude of the adjustment is proportional to the degree of deviation. An example quantitative adjustment rule is: New weight = Old weight * [1 - (deviation amount / 2)].

[0107] For example, if the old weight was 0.6, and the current correlation coefficient is 0.2 lower than the historical benchmark (i.e., a deviation of 0.2), then the new weight = 0.6 * [1 - (0.2 / 2)] = 0.6 * 0.9 = 0.54. The system will then update the weight of the waveform symmetry index from 0.6 to 0.54. Correspondingly, the weight of the spectral matching degree will be adjusted from 0.4 to 1 - 0.54 = 0.46 to maintain a total weight of 1. In this way, the system achieves quantitative self-optimization based on data feedback.

[0108] This is the system's "learning" process. For example: Scenario A: The system might discover that in certain seasons, the "symmetry index" is an unreliable indicator for predicting pollen activity (i.e., symmetry is unrelated to the photostability index), while the "spectral matching degree" is highly reliable. In this case, the system will automatically reduce the weight of the "symmetry index" while increasing the weight of the "spectral matching degree." Scenario B: Conversely, in another environment, the "symmetry index" might become more important. Scenario C: The system might also discover that both indicators are highly unreliable, requiring the search for new features; we limit ourselves to optimizing the weights of these two known parameters.

[0109] Phase 1 (Rule Enforcement): The system allocates detection resources based on a creative heuristic rule (anomaly priority).

[0110] The second stage (efficacy verification): The system obtains the true information of the particles (such as activity) through the expensive gold standard method (fluorescence lifetime detection).

[0111] The third stage (rule calibration): The system analyzes the correlation between the "predictions made by the initial rules" and the "actual results", and then dynamically adjusts the rules themselves (i.e., adjusts the weights) to make future predictions and priority allocation more accurate.

[0112] Based on the embodiments provided in this application, instead of statically relying on preset discrimination rules, the system continuously records the original signal characteristics (waveform symmetry index) and final activity verification results (photostability index) of each pollen grain that has undergone activity analysis, and periodically analyzes the statistical correlation between the two. When the system detects a deviation in this correlation, it indicates that the rule initially used to predict the "value" or "abnormality" of the grain based on characteristics such as waveform symmetry may no longer be optimal in the current environment, and thus actively adjusts the weight coefficients in the priority numerical calculation rule.

[0113] In long-term field online monitoring, environmental conditions (such as temperature, humidity, and pollutant composition) and pollen populations themselves may change, directly affecting the optical signal characteristics of the particles. For example, under specific humidity levels, the water absorption of pollen may cause slight changes in its shape, resulting in a decrease in the waveform symmetry of even highly active pollen. If the system adheres to the initial rules, it may incorrectly assign high priority to these normal pollen types, wasting detection resources. When the correlation between waveform symmetry and photostability (the true measure of activity) weakens, the system automatically reduces the weight of waveform symmetry in priority calculation, relying more on other, more stable features (such as spectral matching). This allows the system's discrimination logic to evolve with environmental drift and pollen population changes, maintaining high accuracy and resource allocation efficiency throughout long-term operation. This "self-learning" and "self-calibration" capability significantly reduces the risk of systematic misjudgment due to environmental changes, reduces the need for maintenance and manual recalibration, and greatly improves the robustness and intelligence of the monitoring network in complex real-world environments.

[0114] It should be noted that the embodiments implemented on the pollen online automatic monitoring system side in this application can be referenced with the embodiments implemented on the pollen online automatic monitoring method side, and will not be described in detail here.

[0115] According to another aspect of the embodiments of this application, an electronic device for implementing the above-described online automatic pollen monitoring method is also provided. This electronic device may be... Figure 4 The terminal device or server shown. This embodiment uses this electronic device as an example of a server. Figure 4 As shown, the electronic device includes a memory 402, a processor 404, and a transmission device 406. The memory 402 stores a computer program, and the processor 404 is configured to execute the steps in any of the above method embodiments through the computer program.

[0116] Optionally, in this embodiment, the aforementioned electronic device may be located in at least one of a plurality of network devices in a computer network.

[0117] Optionally, the transmission device 406 is used to receive or send data via a network. Specific examples of the network described above may include wired and wireless networks. In one example, the transmission device 406 includes a Network Interface Controller (NIC), which can be connected to other network devices and a router via a network cable to communicate with the Internet or a local area network. In another example, the transmission device 406 is a Radio Frequency (RF) module used to communicate with the Internet wirelessly. Furthermore, the electronic device also includes a display 408 and a connection bus 410, which connects the various module components within the electronic device.

[0118] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An on-line automatic pollen monitoring system, characterized in that, Comprising: a charging module for charging particles in ambient air with a known polarity by passing them through a corona discharge zone; a separation module for separating the charged particles by inertial impact and collecting pollen particles in a target size range to a carrying surface according to aerodynamic diameter; a purification module for applying an electrostatic field to the carrying surface to charge it with a polarity opposite to the known polarity, so that pollen particles are adsorbed and fixed to the carrying surface, and using directional airflow to blow the carrying surface to remove unadsorbed impurity particles, thereby achieving in-situ purification of pollen; a neutralization module for electrostatically neutralizing the purified pollen particles and using pulsed gas to blow the neutralized pollen particles from the carrying surface into a detection channel; a signal acquisition module for making the pollen particles in the detection channel pass through an optical sensor under the constraint of a laminar flow sheath, and acquiring corresponding light-blocking pulse signals and multi-wavelength light scattering signals generated by the pollen particles as they pass through; a discrimination module for processing the light-blocking pulse signals, extracting a dynamic feature sequence thereof varying with time, and discriminating whether the pollen particles are valid pollen particles based on the physical characteristics of the particles reflected by the dynamic feature sequence.

2. A method for on-line automatic monitoring of pollen, characterized by Comprising: S201, charging particles in ambient air with a known polarity by passing them through a corona discharge zone; S202, separating the charged particles by inertial impact and collecting pollen particles in a target size range to a carrying surface according to aerodynamic diameter; S203, applying an electrostatic field to the carrying surface to charge it with a polarity opposite to the known polarity, so that pollen particles are adsorbed and fixed to the carrying surface, and using directional airflow to blow the carrying surface to remove unadsorbed impurity particles, thereby achieving in-situ purification of pollen; S204, electrostatically neutralizing the purified pollen particles and using pulsed gas to blow the neutralized pollen particles from the carrying surface into a detection channel; S205, making the pollen particles in the detection channel pass through an optical sensor under the constraint of a laminar flow sheath, and acquiring corresponding light-blocking pulse signals and multi-wavelength light scattering signals generated by the pollen particles as they pass through; S206, processing the light-blocking pulse signals, extracting a dynamic feature sequence thereof varying with time, and discriminating whether the pollen particles are valid pollen particles based on the physical characteristics of the particles reflected by the dynamic feature sequence.

3. The method for on-line automatic monitoring of pollen according to claim 2, characterized in that, In S206, the processing of the light-blocking pulse signals and the extraction of a dynamic feature sequence thereof varying with time comprise: For each light-blocking pulse signal, the following operations are performed: determining the starting time point and the ending time point of the light-blocking pulse signal; dividing the time axis uniformly into a plurality of consecutive time windows within the time period between the starting time point and the ending time point, the number of time windows being greater than 5; for each time window, the following operations are performed: calculating the minimum value of the voltage signal in the time window; dividing the minimum value by the overall voltage minimum value of the light-blocking pulse signal to obtain the normalized amplitude value of the time window; Dividing the duration of the time window by the total duration of the photoresist pulse signal to obtain a local pulse width value of the time window; Arranging the normalized amplitude values of all time windows in time sequence to form an amplitude sequence; Arranging the local pulse width values of all time windows in time sequence to form a pulse width sequence.

4. The method for on-line automatic monitoring of pollen according to claim 3, characterized in that, In S206, the particle physical characteristics reflected by the dynamic characteristic sequence are used to determine whether the pollen particle is a valid pollen particle, including: Calculating the correlation coefficient of the amplitude sequence and the pre-stored standard bell-shaped curve; Calculating the standard deviation of the pulse width sequence; When the correlation coefficient is higher than a first preset threshold and the standard deviation is lower than a second preset threshold, it is determined that the photoresist pulse signal corresponds to a valid pollen particle, and counting is performed; When the waveform of a certain photoresist pulse signal presents multiple separated peak values, or the symmetry index of the waveform is lower than a preset symmetry threshold, it is determined that the photoresist pulse signal is generated by multiple particles superimposed or by irregular impurities, and the photoresist pulse signal is discarded; wherein the symmetry index of the waveform is obtained by calculating the similarity of the rising edge and the falling edge of the photoresist pulse signal.

5. The method for on-line automatic monitoring of pollen according to claim 2, characterized in that, The method further includes: For each photoresist pulse signal, the following operations are performed: In S207, the correlation coefficient of the amplitude sequence of the photoresist pulse signal and the standard bell-shaped curve, and the standard deviation of the pulse width sequence, are combined into a first discriminant vector; In S208, the scattering spectrum formed by the multi-wavelength light scattering signal is matched with a pre-stored standard pollen scattering spectrum to obtain a spectrum matching degree as a second discriminant vector; In S209, according to the first discriminant vector and the second discriminant vector, the microfluidic sorting chip is controlled to sort the particles into three physically isolated channels: In S210, when the correlation coefficient is higher than the first preset threshold, the standard deviation is lower than the second preset threshold, and the spectrum matching degree is higher than a third preset threshold, the pollen particle is sorted into a first channel for collecting confirmed pollen; the resource manager is used to perform activity analysis detection on the pollen particles sorted into the first channel; In S211, when the correlation coefficient is lower than a fourth preset threshold, or the standard deviation is higher than a fifth preset threshold, or the spectrum matching degree is lower than a sixth preset threshold, the pollen particle is sorted into a second channel for discharging impurities; In S212, for pollen particles that do not meet the above two conditions, they are sorted into a third channel for temporarily storing suspected pollen particles; the resource manager is used to schedule activity analysis detection on the pollen particles sorted into the third channel; wherein the particles sorted into the first channel are marked as high-confidence pollen and immediately enter an activity analysis queue; the particles sorted into the third channel are marked as suspected particles and temporarily stored in a buffer queue; only when the activity analysis queue is empty, the resource manager schedules the suspected particles in the buffer queue for activity analysis detection; the resource manager is used to preferentially schedule the particles in the activity analysis queue for activity analysis detection.

6. The method for on-line automatic monitoring of pollen according to claim 5, characterized in that, In S210, the resource manager performs activity analysis detection on the pollen grains sorted into the first channel, including: calculating a priority value for each pollen grain entering the first channel, the calculation rule of the priority value including: the lower the symmetry index of the light blocking pulse signal waveform and / or the lower the matching degree of the spectrum, the higher the assigned priority value; maintaining a detection resource counter based on the number of operations of the pulsed laser and the periodic budget; comparing the priority value of the current grain with a dynamic priority threshold, and only when the priority value is higher than the dynamic priority threshold and the current value of the detection resource counter is greater than zero, triggering the activity analysis detection, and deducting the detection resource counter after execution; wherein the dynamic priority threshold is inversely adjusted according to the ratio of the current remaining value of the detection resource counter to the total budget value.

7. The method for on-line automatic monitoring of pollen according to claim 5, characterized in that, The activity analysis detection in S210 and S212 includes rechecking by emitting two sub-pulses with a pulsed ultraviolet laser and collecting fluorescence signals with a gate control, specifically including: configuring a pulsed ultraviolet laser with a double-pulse driving circuit, controlling the pulsed ultraviolet laser to emit two sub-pulses with equal energy and an interval of 0.1 to 5 microseconds within the time window when the pollen grain passes through the detection area; controlling the opening time of the gate control to be synchronized with each sub-pulse to suppress environmental background fluorescence interference; within the gate window, synchronously collecting fluorescence signals excited by the two sub-pulses through a photodetector to obtain a first fluorescence decay curve and a second fluorescence decay curve respectively; using a fitting algorithm to calculate the fluorescence lifetime value τA of the first fluorescence decay curve and the fluorescence lifetime value τB of the second fluorescence decay curve respectively; calculating the photostability index corresponding to the detection event; comparing the photostability index with the activity threshold preset based on the pollen species, and when the photostability index is lower than the preset activity threshold, determining that the pollen grain captured in this detection event is high-activity pollen.

8. The method for on-line automatic monitoring of pollen according to claim 7, characterized in that, The synchronization of the gate control and the double-pulse driving circuit is achieved by the following methods: the double-pulse driving circuit outputs a synchronization trigger signal to the gate control circuit simultaneously when emitting each sub-pulse; the gate control circuit opens after a certain delay after receiving the synchronization trigger signal, and the specific delay is calibrated according to the response characteristics of the photodetector and the circuit transmission delay; the opening duration of the gate control is set according to the expected value of the fluorescence lifetime, and the range is 1 to 5 times the expected value of the fluorescence lifetime.

9. The method for on-line automatic monitoring of pollen according to claim 7, characterized in that, The method further includes: recording the waveform symmetry index of all pollen grains subjected to activity analysis detection and the photostability index obtained by double-pulse fluorescence measurement; periodically analyzing the statistical correlation between the waveform symmetry index and the photostability index in all recorded data; when the statistical correlation deviates from the historical baseline within a preset range, adjusting the calculation rule of the priority value according to the direction and degree of deviation, specifically adjusting the weight coefficients of the waveform symmetry index and the spectrum matching degree in the calculation.

10. The pollen online automatic monitoring method according to claim 2, characterized in that, the separation and collection of pollen particles in the target particle size range by inertial impact is achieved by two-stage inertial impactors connected in series, wherein, the first-stage inertial impactor is configured to intercept particles with an aerodynamic diameter greater than 10 microns; the second-stage inertial impactor is configured to specifically collect particles with an aerodynamic diameter in the range of 8 microns to 12 microns; the carrier tape is a polyester film coated with an indium tin oxide conductive film on the surface; applying an electrostatic field to the carrier tape includes applying a direct current voltage of 500 volts to 1500 volts to the conductive layer of the carrier tape; the directional airflow is a negative pressure airflow opposite to the movement direction of the carrier tape; the electrostatic neutralization treatment is achieved by alternating current corona discharge; and the pulsed gas is a nitrogen pulse with a duration of 5 milliseconds to 20 milliseconds, and the ejection action is controlled by a solenoid valve.

Images