Method and system for monitoring environmental air quality based on high-altitude unmanned aerial vehicle

Abstract

The present application relates to the technical field of intelligent sensing system, and more particularly to a method and system for stereoscopic monitoring of ambient air quality based on high-altitude unmanned aerial vehicle, which comprises: collecting initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each preset vertical height layer through the sensors carried by the high-altitude unmanned aerial vehicle; classifying the preset vertical height layers and determining the dry particulate matter concentration corresponding to the target dry height layer; decomposing the target state matrix corresponding to each humidity interference height layer; determining the dry particulate matter concentration corresponding to each humidity interference height layer according to the determined relative distribution coefficient sequence of dry particulate matter, and realizing stereoscopic monitoring of ambient air quality based on the dry particulate matter concentration corresponding to different height layers. The present application takes into account the humidity interference at different altitudes, and objectively quantifies the dry particulate matter concentration corresponding to different height layers, thereby improving the rationality of stereoscopic monitoring of ambient air quality.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensing system technology, specifically to a method and system for three-dimensional monitoring of ambient air quality based on high-altitude unmanned aerial vehicles (UAVs). Background Technology

[0002] Three-dimensional monitoring of ambient air quality is often a crucial operational method for analyzing the transport channels of atmospheric pollution. In conventional practice, monitoring personnel typically use high-altitude drones equipped with miniature optical sensors to acquire spatial distribution data of particulate matter through vertical flight. However, the atmospheric boundary layer often contains cloud bases or airflow convergence zones with extremely high relative humidity. When airborne optical sensors pass through such high-humidity air masses, aerosol particles often absorb moisture and expand in volume, resulting in an artificially inflated apparent concentration of water-containing particulate matter output by the optical sensors.

[0003] To address the aforementioned water vapor interference issue, ground-based fixed monitoring stations typically incorporate high-energy-consuming physical heating and dehumidification sample inlet tubes at the front end of the detection equipment. However, due to limitations in flight payload and battery power, lightweight UAVs often cannot accommodate such large-scale heating and dehumidification hardware. As a second-best option, using conventional static humidity empirical formulas for post-processing compensation often fails to match the actual water absorption properties of upper-air masses, which are not consistently present at different altitudes. This approach can easily lead to blind and excessive subtraction of sensor readings, potentially erasing the true increase in pollution while eliminating water vapor interference. This can result in poor accuracy in quantifying dehumidified dry particulate matter concentrations, and consequently, inaccurate results in comprehensive environmental air quality monitoring. Summary of the Invention

[0004] To address the technical problem of poor rationality in three-dimensional ambient air quality monitoring due to the inaccuracy of quantification of wet and dry particulate matter concentration, this invention proposes a method and system for three-dimensional ambient air quality monitoring based on high-altitude unmanned aerial vehicles (UAVs).

[0005] In a first aspect, the present invention provides a method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV), the method comprising: Using sensors mounted on a high-altitude drone, the initial particulate matter concentration, black carbon concentration, and relative humidity were collected for each preset vertical altitude layer. Based on the height of the preset vertical height layer and its corresponding relative humidity, the preset vertical height layer is classified to obtain the target drying height layer and the humidity interference height layer, and the dehumidification dry particulate matter concentration corresponding to each target drying height layer is determined. Based on the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each humidity interference height layer and the number of preset vertical height layers before it, construct the target state matrix corresponding to each humidity interference height layer; The target state matrix corresponding to each humidity interference height layer is decomposed to obtain the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer; Based on the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer, determine the dry particulate matter relative distribution coefficient sequence corresponding to each humidity interference height layer; Based on the relative distribution coefficient sequence of dry particulate matter, the concentration of dehumidified dry particulate matter corresponding to each humidity interference height layer is determined, and three-dimensional monitoring of ambient air quality is achieved based on the dehumidified dry particulate matter concentrations corresponding to different height layers.

[0006] In conjunction with the first aspect above, in one possible implementation, classifying the preset vertical height layers according to their height and corresponding relative humidity to obtain a target dry height layer and a humidity interference height layer includes: The maximum height of each preset vertical height layer is determined as the reference height corresponding to each preset vertical height layer; If the reference height corresponding to the preset vertical height layer is less than or equal to the preset low-altitude height threshold, then the preset vertical height layer will be determined as the target low-altitude height layer. If the reference height corresponding to the preset vertical height layer is greater than the preset low-altitude height threshold, then the preset vertical height layer will be determined as the target high-altitude height layer. If the average relative humidity of the target upper altitude layer and the preset number of preset vertical altitude layers before it is greater than the aerosol deliquescent humidity threshold, then the target upper altitude layer is determined as the humidity interference altitude layer, wherein the reference height corresponding to the preset number of preset vertical altitude layers is less than or equal to the preset low altitude threshold. If the average relative humidity of the target upper altitude layer and the preset number of preset vertical altitude layers before it is less than or equal to the aerosol deliquescence humidity threshold, then the target upper altitude layer is determined as the initial dry altitude layer. The target low-altitude layer and the initial dry altitude layer are collectively referred to as the target dry altitude layer.

[0007] In conjunction with the first aspect above, in one possible implementation, determining the dehumidified dry particulate matter concentration corresponding to each target drying height layer includes: The pre-obtained surface baseline dry particulate matter concentration is used to determine the dehumidified dry particulate matter concentration corresponding to each target low-altitude layer. The initial particulate matter concentration corresponding to each initial drying height layer is determined as the dehumidified dry particulate matter concentration corresponding to each initial drying height layer.

[0008] In conjunction with the first aspect above, in one possible implementation, the step of constructing a target state matrix corresponding to each humidity interference height layer based on the initial particulate matter concentration, black carbon concentration, and relative humidity corresponding to each humidity interference height layer and a preset number of preset vertical height layers preceding it includes: Any humidity interference height layer is designated as a marked interference height layer, and the marked interference height layer and a predetermined number of predetermined vertical height layers preceding it constitute a marked height layer sequence; The normalized values ​​of the initial particulate matter concentrations corresponding to all preset vertical height layers in the labeled height layer sequence are used to form a labeled particulate matter concentration vector. The normalized values ​​of the black carbon concentration corresponding to all preset vertical height layers in the marked height layer sequence are used to form a marked black carbon concentration vector. The normalized values ​​of relative humidity corresponding to all preset vertical height layers in the marked height layer sequence are used to form a marked relative humidity vector; Using the labeled particulate matter concentration vector, labeled black carbon concentration vector, and labeled relative humidity vector as column vectors, a target state matrix corresponding to the labeled interference height layer is constructed.

[0009] In conjunction with the first aspect above, in one possible implementation, the decomposition of the target state matrix corresponding to each humidity interference height layer to obtain the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer includes: Any humidity interference height layer is designated as the marked interference height layer, and the target state matrix corresponding to the marked interference height layer is designated as the marked state matrix; Based on the marked state matrix, the pre-acquired initial height activation weight matrix and initial feature load matrix are iteratively updated to obtain the target height activation weight matrix and target feature load matrix corresponding to the marked interference height layer. The elements in the initial height activation weight matrix and the initial feature load matrix are all random numbers within a preset value range.

[0010] In conjunction with the first aspect above, in one possible implementation, the step of iteratively updating the pre-acquired initial height activation weight matrix and initial feature load matrix based on the labeled state matrix to obtain the target height activation weight matrix and target feature load matrix corresponding to the labeled interference height layer includes: Based on the candidate high activation weight matrix, the candidate feature load matrix, and the labeled state matrix, the candidate high activation weight matrix and the candidate feature load matrix are updated to obtain the reference high activation weight matrix and the reference feature load matrix. The candidate high activation weight matrix is ​​the initial high activation weight matrix or the high activation weight matrix obtained in the previous update, and the candidate feature load matrix is ​​the initial feature load matrix or the feature load matrix obtained in the previous update. The error loss function value between the product of the reference height activation weight matrix and the reference feature load matrix and the labeled state matrix is ​​determined as the reference loss function value corresponding to this update. Similarly, the reference loss function value corresponding to the previous update is obtained. The absolute value of the difference between the reference loss function value corresponding to the current update and the previous update is determined as the target loss change corresponding to the current update. Similarly, the target loss change corresponding to the previous update is obtained. If the change in target loss corresponding to this update and the change in target loss corresponding to the previous update are both less than the preset convergence threshold, then the update is stopped, and the reference height activation weight matrix and the reference feature load matrix are respectively determined as the target height activation weight matrix and the target feature load matrix corresponding to the marked interference height layer. If the number of updates corresponding to this update reaches the preset maximum number of iterations, the update will stop, and the reference height activation weight matrix and reference feature load matrix will be determined as the target height activation weight matrix and target feature load matrix corresponding to the marked interference height layer, respectively.

[0011] In conjunction with the first aspect above, in one possible implementation, determining the dry particulate matter relative distribution coefficient sequence corresponding to each humidity interference height layer based on the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer includes: Any humidity interference height layer is designated as the marked interference height layer, and the target height activation weight matrix and target feature load matrix corresponding to the marked interference height layer are designated as the marked height activation weight matrix and the marked feature load matrix, respectively. The quotient between the second and third elements in each row of the labeled feature load matrix is ​​determined as the reference quotient for each row of the labeled feature load matrix; The row with the largest reference quotient value is selected from the labeled feature load matrix and taken as the dry particulate matter evolution component. The row number of the dry particulate matter evolution component in the labeled feature load matrix is ​​determined as the labeled row number. The column whose column number in the marker height activation weight matrix is ​​equal to the marker row number is determined as the dry particulate matter relative distribution coefficient sequence corresponding to the marker interference height layer.

[0012] In conjunction with the first aspect above, in one possible implementation, determining the dehumidified dry particulate matter concentration corresponding to each humidity interference height layer based on the dry particulate matter relative distribution coefficient sequence includes: Any humidity interference height layer is designated as the marked interference height layer, and a preset number of preset vertical height layers preceding the marked interference height layer are designated as temporary vertical height layers, thus obtaining a preset number of temporary vertical height layers. Based on the concentration of dehumidified dry particulate matter corresponding to a preset number of temporary vertical height layers, and the coefficient values ​​of the preset number of temporary vertical height layers in the sequence of relative distribution coefficients of dry particulate matter corresponding to the marked interference height layer, the absolute mass concentration index corresponding to the marked interference height layer is determined. The calibrated dry particulate matter concentration corresponding to the marked interference height layer is determined based on the absolute mass concentration index corresponding to the marked interference height layer and the coefficient value corresponding to the marked interference height layer in the dry particulate matter relative distribution coefficient sequence corresponding to the marked interference height layer. The dehumidified dry particulate matter concentration corresponding to the marked interference height layer is determined based on the calibrated dry particulate matter concentration corresponding to the marked interference height layer.

[0013] In conjunction with the first aspect above, in one possible implementation, determining the dehumidified dry particulate matter concentration corresponding to the marked interference height layer based on the calibrated dry particulate matter concentration corresponding to the marked interference height layer includes: If the calibrated dry particulate matter concentration corresponding to the marked interference height layer is greater than or equal to the preset theoretical lower limit threshold, then the calibrated dry particulate matter concentration corresponding to the marked interference height layer is determined as the dehumidified dry particulate matter concentration corresponding to the marked interference height layer. If the calibrated dry particulate matter concentration corresponding to the marked interference height layer is less than the preset theoretical lower limit threshold, then the dehumidified dry particulate matter concentration corresponding to the marked interference height layer is set to the preset theoretical lower limit threshold.

[0014] Secondly, the present invention provides a three-dimensional environmental air quality monitoring system based on a high-altitude unmanned aerial vehicle (UAV), comprising a processor and a memory, wherein the processor is used to process instructions stored in the memory to implement the method in the first aspect or any possible implementation thereof, specifically, the system includes: Multiple data acquisition modules are used to collect the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each preset vertical height layer through sensors on the airborne high-altitude UAV; The height layer classification and concentration determination module is used to classify the preset vertical height layer according to its height and corresponding relative humidity, to obtain the target dry height layer and humidity interference height layer, and to determine the dehumidified dry particulate matter concentration corresponding to each target dry height layer. The target state matrix construction module is used to construct the target state matrix corresponding to each humidity interference height layer based on the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each humidity interference height layer and the number of preset vertical height layers before it. The matrix decomposition module is used to decompose the target state matrix corresponding to each humidity interference height layer to obtain the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer. The relative distribution coefficient sequence determination module is used to determine the relative distribution coefficient sequence of dry particulate matter corresponding to each humidity interference height layer based on the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer. The concentration determination and quality detection module is used to determine the concentration of dehumidified dry particulate matter corresponding to each humidity interference height layer based on the relative distribution coefficient sequence of dry particulate matter, and to realize three-dimensional monitoring of ambient air quality based on the concentration of dehumidified dry particulate matter corresponding to different height layers.

[0015] The present invention has the following beneficial effects: The present invention provides a three-dimensional ambient air quality monitoring method based on high-altitude unmanned aerial vehicles (UAVs). This method considers humidity interference at different altitudes and relatively objectively quantifies the concentration of dehumidified dry particulate matter (DPM) at different altitude levels. It solves the technical problem of poor rationality in three-dimensional ambient air quality monitoring caused by poor quantification of DPM concentration, thus improving the rationality of DPM concentration quantification and consequently enhancing the overall rationality of three-dimensional ambient air quality monitoring. Specifically, the present invention classifies preset vertical altitude levels based on their height and corresponding relative humidity, obtaining relatively dry target dry altitude levels and humidity-affected altitude levels. By analyzing the initial particulate matter concentration, black carbon concentration, and relative humidity corresponding to each humidity-affected altitude level and a preset number of preceding preset vertical altitude levels, the DPM concentration corresponding to each humidity-affected altitude level is quantified. Finally, based on the DPM concentrations at different altitude levels, three-dimensional ambient air quality monitoring can be achieved, improving the rationality of DPM concentration quantification at humidity-affected altitude levels and thus enhancing the overall rationality of three-dimensional ambient air quality monitoring. Attached Figure Description

[0016] Figure 1 This is a flowchart of a three-dimensional monitoring method for ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to the present invention. Figure 2This is a schematic diagram of the composition structure of a three-dimensional ambient air quality monitoring system based on a high-altitude unmanned aerial vehicle (UAV) according to the present invention. Figure 3 This is a schematic diagram of the structure of a computer device according to the present invention. Detailed Implementation

[0017] refer to Figure 1 This paper illustrates the flowchart of a three-dimensional environmental air quality monitoring method based on a high-altitude unmanned aerial vehicle (UAV) according to the present invention. The method includes the following steps: Step S1: Using sensors onboard a high-altitude drone, collect the initial particulate matter concentration, black carbon concentration, and relative humidity corresponding to each preset vertical height layer.

[0018] High-altitude drones can be those flying at altitudes exceeding 5000 meters or significantly above the lower troposphere. They typically take off from the ground and ascend to the required altitude under their own power. Preset vertical altitude layers can be pre-defined altitude ranges based on actual conditions; for example, their corresponding length could be 10 meters. The maximum height of the highest preset vertical altitude layer can be the maximum altitude achievable by the high-altitude drone. In practice, preset vertical altitude layers can be defined as altitude ranges of 10 meters vertically from the ground. For example, the first preset vertical altitude layer could be [0 meters, 10 meters]; the second could be (10 meters, 20 meters). Initial particulate matter concentration can be the total concentration measurement value output by an optical particle counter. Black carbon concentration can be the carbon particle measurement value output by a miniature black carbon meter. Relative humidity can be the percentage of relative humidity (RH) output by a temperature and humidity probe. The optical particle counter, miniature black carbon meter, and temperature and humidity probe can be onboard sensors of the high-altitude drone.

[0019] As an example, taking a preset vertical height layer (100 meters, 110 meters) as an example, when the high-altitude drone reaches 110 meters, the initial particulate matter concentration corresponding to the preset vertical height layer can be collected by the optical particle counter on the high-altitude drone, the black carbon concentration corresponding to the preset vertical height layer can be collected by the micro black carbon meter on the high-altitude drone, and the relative humidity corresponding to the preset vertical height layer can be collected by the temperature and humidity probe on the high-altitude drone.

[0020] It should be noted that because the miniature optical particulate matter sensor onboard the drone is not equipped with a heated dehumidification sample inlet tube, its measurement output readings often exhibit varying degrees of volume expansion as the moisture content in the air increases. This often means that the initial reading measured by the airborne optical sensor, i.e., the initial particulate matter concentration, may contain a mixture of the mass of the aerosol solid particles themselves and the increased volume due to water vapor adsorption. To reduce the interference of moisture on air quality detection, the initial reading measured by the airborne optical sensor can be dehumidified to obtain the dehumidified dry particulate matter concentration. Pure carbon particles (such as black carbon from vehicle exhaust or biomass combustion) are essentially non-hydrophilic and often do not undergo significant hygroscopic expansion even in high humidity environments, making them a useful reference for eliminating moisture interference. Relative humidity characterizes the abundance of water vapor available for aerosol absorption.

[0021] Step S2: Based on the height of the preset vertical height layer and its corresponding relative humidity, classify the preset vertical height layer to obtain the target dry height layer and the humidity interference height layer, and determine the dehumidified dry particulate matter concentration corresponding to each target dry height layer.

[0022] As an example, this step may include the following steps: The first step is to determine the maximum height of each preset vertical height layer as the reference height corresponding to each preset vertical height layer.

[0023] For example, the reference height corresponding to the preset vertical height layer (100 meters, 110 meters) can be 110 meters.

[0024] The second step is to determine the preset vertical height layer as the target low-altitude layer if the reference height corresponding to the preset vertical height layer is less than or equal to the preset low-altitude height threshold.

[0025] The preset low-altitude threshold can be a pre-set upper limit of the low-altitude range based on the actual situation, such as 100 meters.

[0026] The third step is to determine the preset vertical height layer as the target high-altitude layer if the reference height corresponding to the preset vertical height layer is greater than the preset low-altitude height threshold.

[0027] Fourth step: If the average relative humidity of the target upper altitude layer and the preset number of preset vertical altitude layers before it is greater than the aerosol deliquescence humidity threshold, then the target upper altitude layer is determined as the humidity interference altitude layer.

[0028] The preset quantity can be a number pre-set based on actual conditions, and it can be less than or equal to the number of target low-altitude layers, such as 4. The reference height corresponding to the preset number of preset vertical height layers can be less than or equal to the preset low-altitude threshold, meaning that at least a preset number of target low-altitude layers exist. For example, if the preset quantity is 4, then the reference height corresponding to the 4th preset vertical height layer can be less than or equal to the preset low-altitude threshold, meaning that at least 4 target low-altitude layers exist in this embodiment of the invention. The aerosol deliquescence humidity threshold can be a threshold pre-set based on actual conditions, and it can represent the physical critical point at which environmental humidity causes aerosols to undergo nonlinear water absorption and expansion, such as 65%.

[0029] Fifth step: If the average relative humidity of the target high-altitude layer and the number of preset vertical height layers before it is less than or equal to the aerosol deliquescence humidity threshold, then the target high-altitude layer is determined as the initial dry height layer.

[0030] The sixth step is to refer to the target low-altitude layer and the initial dry altitude layer collectively as the target dry altitude layer.

[0031] The seventh step is to determine the pre-acquired surface baseline dry particulate matter concentration (PM) to the dehumidified dry particulate matter concentration corresponding to each target low-altitude layer.

[0032] Among them, the surface baseline dry particulate matter concentration can be particulate matter concentration data collected by a ground-based fixed monitoring station equipped with a physically heated and dehumidified sample inlet tube and located closest to the high-altitude UAV.

[0033] It should be noted that the surface baseline dry particulate matter concentration can characterize the dry particulate matter concentration after eliminating moisture interference, that is, it can represent the true dry matter mass concentration under the current surface environment. Since the target low-altitude layer is close to the surface environment, and the dry matter concentration in the atmospheric mixing layer is usually uniformly distributed within a range of 100 meters below the surface, the surface baseline dry particulate matter concentration can be directly used to characterize the dehumidified dry particulate matter concentration corresponding to the target low-altitude layer.

[0034] Step 8: Determine the initial particulate matter concentration corresponding to each initial drying height layer, and then determine the dehumidified dry particulate matter concentration corresponding to each initial drying height layer.

[0035] It should be noted that if the airspace traversed by the drone is extremely dry, the aerosol may not have absorbed any moisture or expanded at all. In this case, the measurement value output by the optical sensor is often essentially the true dry particulate matter concentration. Specifically, the initial dry altitude layer often has low moisture content and is relatively dry. Therefore, the initial particulate matter concentration output by the optical particle counter at the initial dry altitude layer can represent the dehumidified dry particulate matter concentration corresponding to the initial dry altitude layer.

[0036] Step S3: Based on the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each humidity interference height layer and the preset number of preset vertical height layers before it, construct the target state matrix corresponding to each humidity interference height layer.

[0037] As an example, this step may include the following steps: The first step is to designate any humidity interference height layer as the marked interference height layer, and to form a marked height layer sequence by combining the marked interference height layer and a predetermined number of preset vertical height layers preceding it.

[0038] Among them, the height layers in the height layer sequence can be sorted according to their height.

[0039] The second step is to construct a labeled particulate matter concentration vector by normalizing the initial particulate matter concentration values ​​corresponding to all preset vertical height layers in the above labeled height layer sequence.

[0040] It should be noted that the normalized values ​​in the embodiments of the present invention can be achieved by range normalization or normalization algorithms such as maximum-minimum normalization.

[0041] For example, one method for obtaining the normalized value of the initial particulate matter concentration is to determine the maximum value of the initial particulate matter concentration corresponding to all preset vertical height layers in the marked height layer sequence as the particulate matter concentration reference value, and to determine the initial particulate matter concentration corresponding to any preset vertical height layer in the marked height layer sequence as the temporary particulate matter concentration. The ratio between the temporary particulate matter concentration and the particulate matter concentration reference value can be determined as the normalized value of the temporary particulate matter concentration.

[0042] In reality, if the particulate matter concentration reference value is 0, it is theoretically considered to be in an absolutely clean state. However, in practice, it is almost impossible for it to be 0. If the measured particulate matter concentration reference value is indeed 0, then it can be considered to be in an extremely clean atmospheric environment, but this is almost impossible in the real Earth's atmosphere. Therefore, if the normally measured particulate matter concentration reference value is 0, then the marked interference height layer can be considered to be in an extremely clean atmospheric environment at high altitude. The dehumidified dry particulate matter concentration corresponding to the marked interference height layer can be directly set to 0. Thus, the relevant content of quantifying the dehumidified dry particulate matter concentration corresponding to the marked interference height layer recorded in steps S3 to S6 can be skipped, and there is no need to normalize the initial particulate matter concentration.

[0043] The third step is to construct a black carbon concentration vector by normalizing the black carbon concentration values ​​corresponding to all preset vertical height layers in the above-mentioned marked height layer sequence.

[0044] For example, one method for obtaining the normalized value of black carbon concentration is to determine the maximum value of black carbon concentration corresponding to all preset vertical height layers in the marked height layer sequence as the black carbon reference concentration value, and to determine the black carbon concentration corresponding to any preset vertical height layer in the marked height layer sequence as the temporary black carbon concentration. The ratio between the temporary black carbon concentration and the black carbon reference concentration value can be determined as the normalized value of the temporary black carbon concentration.

[0045] In practice, if the reference concentration value of black carbon is 0, it often means that the black carbon concentration is not a reference. In this case, the initial particulate matter concentration corresponding to the marked interference height layer can be directly recorded as the dewetting dry particulate matter concentration corresponding to the marked interference height layer. Thus, the relevant content of quantifying the dewetting dry particulate matter concentration corresponding to the marked interference height layer recorded in steps S3 to S6 can be skipped directly, and there is no need to normalize the black carbon concentration.

[0046] The fourth step is to construct a marker relative humidity vector by taking the normalized values ​​of the relative humidity corresponding to all preset vertical height layers in the above marker height layer sequence.

[0047] For example, one method for obtaining the normalized value of relative humidity is to determine the maximum value of relative humidity corresponding to all preset vertical height layers in the marked height layer sequence as the relative humidity reference value, and to determine the relative humidity corresponding to any preset vertical height layer in the marked height layer sequence as the temporary relative humidity. The ratio between the temporary relative humidity and the relative humidity reference value can be determined as the normalized value of the temporary relative humidity.

[0048] It should be noted that since the average relative humidity of the marked interference height layer and the preset number of preset vertical height layers before it is greater than the aerosol deliquescence humidity threshold, the relative humidity reference value is often not 0. Therefore, the relative humidity reference value can be used as the denominator.

[0049] The fifth step involves constructing the target state matrix corresponding to the marked interference height layer using the marked particulate matter concentration vector, the marked black carbon concentration vector, and the marked relative humidity vector as column vectors.

[0050] In this context, elements at the same index in the labeled particulate matter concentration vector, labeled black carbon concentration vector, and labeled relative humidity vector can correspond to the same preset vertical height layer. The target state matrix can be... The matrix, , where n is the preset quantity. A matrix, that is, an N x 3 matrix.

[0051] Step S4: Decompose the target state matrix corresponding to each humidity interference height layer to obtain the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer.

[0052] It should be noted that although the hygroscopic expansion process of aerosols exhibits an exponential nonlinear change on a macroscopic scale, within a series of consecutive altitude layers traversed by a drone (i.e., an extremely short vertical spatial slice only tens of meters thick), the co-evolution of optical readings caused by the increase in particulate matter dry matter concentration and changes in environmental humidity can often be approximated as a local linear superposition of two independent trend components. The target state matrix can be deconstructed into a set of mutually independent eigenvectors, specifically extracting two independent source components from the mixed readings: the dry matter increment trend and the moisture adsorption increment trend. The target altitude activation weight matrix is ​​equivalent to a composition table of these two factors, showing the proportions of black carbon, concentration measurements, and humidity in each factor. The target characteristic load matrix represents the degree of influence of these two factors, indicating whether each layer is more affected by the increase in dry matter or the increase in water vapor within a series of altitude layers.

[0053] As an example, this step may include the following steps: The first step is to determine any humidity interference height layer as the marked interference height layer, and to determine the target state matrix corresponding to the marked interference height layer as the marked state matrix.

[0054] The second step involves iteratively updating the pre-obtained initial height activation weight matrix and initial feature load matrix based on the aforementioned labeled state matrix, to obtain the target height activation weight matrix and target feature load matrix corresponding to the aforementioned labeled interference height layer.

[0055] In this matrix, the elements of both the initial high-level activation weight matrix and the initial feature load matrix can be random numbers within a preset numerical range. The preset numerical range can be a normalized range pre-set according to actual conditions, and it can be (0, 1). The initial high-level activation weight matrix can be... The matrix, , where n is a preset number. The initial feature loading matrix can be... The matrix. The matrix is ​​an N-row, 2-column matrix. A matrix, that is, a 2x3 matrix.

[0056] For example, based on the aforementioned labeled state matrix, iteratively updating the pre-acquired initial height activation weight matrix and initial feature load matrix to obtain the target height activation weight matrix and target feature load matrix corresponding to the aforementioned labeled interference height layer may include the following sub-steps: The first sub-step involves updating the candidate height activation weight matrix and the candidate feature load matrix based on the candidate height activation weight matrix, the candidate feature load matrix, and the aforementioned labeled state matrix. This update is denoted as the current update, resulting in the reference height activation weight matrix and the reference feature load matrix.

[0057] The candidate high activation weight matrix can be either the initial high activation weight matrix or the high activation weight matrix obtained in the previous update. The candidate feature payload matrix can also be either the initial feature payload matrix or the feature payload matrix obtained in the previous update.

[0058] It should be noted that during the first update, the candidate high activation weight matrix and the candidate feature loading matrix can be the initial high activation weight matrix and the initial feature loading matrix, respectively; during each subsequent update, the candidate high activation weight matrix and the candidate feature loading matrix can be the high activation weight matrix and the feature loading matrix obtained from the previous update, respectively.

[0059] For example, the formulas for determining the reference height activation weight matrix and the reference feature load matrix can be: ; ; Where M is the reference feature loading matrix, H is the reference high-level activation weight matrix, m is the candidate feature loading matrix, and h is the candidate high-level activation weight matrix. This represents the Hadamard product of matrix elements, i.e., multiplying the elements corresponding to each position. T represents the transpose of the matrix. v is the state marker matrix. It is an adjustment factor preset according to the actual situation, mainly used to prevent the denominator from being 0. For example, it can be... .

[0060] It should be noted that M can represent the feature load matrix in this update iteration. It can represent the ideal gradient direction matrix obtained by projecting the height weight distribution onto the labeled state matrix. It can represent the actual projection error basis reconstructed by multiplying the feature matrix and the weight matrix. The larger the value, the further the feature load matrix may deviate from the true data distribution, and the larger the update pace will be. It can represent the ideal projection gradient of the labeled state matrix onto the characteristic load space. It can represent the actual feature deviation after reconstruction. This can force the activation weights of the corresponding features on the labeled interference height layer to increase adaptively, thereby gradually approximating the local linear evolution law.

[0061] The second sub-step involves determining the error loss function value between the product of the reference height activation weight matrix and the reference feature load matrix and the aforementioned labeled state matrix as the reference loss function value for this update. Similarly, the reference loss function value for the previous update is obtained.

[0062] The method for obtaining the error loss function value between the product of the reference height activation weight matrix and the reference feature load matrix and the aforementioned labeled state matrix can be as follows: The product of the reference height activation weight matrix and the reference feature load matrix is ​​determined as the undetermined matrix, and the error loss function value between the undetermined matrix and the labeled state matrix is ​​denoted as the reference loss function value corresponding to this update. The error loss function can be constructed based on the Frobenius norm of Euclidean distance.

[0063] It should be noted that the method for obtaining the reference loss function value corresponding to the previous update can be the same as the method for obtaining the reference loss function value corresponding to this update, and will not be repeated here.

[0064] The third sub-step is to determine the absolute value of the difference between the reference loss function value corresponding to the current update and the previous update as the target loss change corresponding to the current update. Similarly, the target loss change corresponding to the previous update is obtained.

[0065] It should be noted that the method for obtaining the change in target loss corresponding to the previous update can be the same as the method for obtaining the change in target loss corresponding to this update, and will not be repeated here.

[0066] In the fourth sub-step, if the change in target loss corresponding to this update and the change in target loss corresponding to the previous update are both less than the preset convergence threshold, then the update is stopped, and the reference height activation weight matrix and the reference feature load matrix are respectively determined as the target height activation weight matrix and the target feature load matrix corresponding to the above-mentioned marked interference height layer.

[0067] The preset convergence threshold can be a threshold set in advance based on the actual situation, for example, it can be... .

[0068] In the fifth sub-step, if the number of updates corresponding to this update reaches the preset maximum number of iterations, the update is stopped, and the reference height activation weight matrix and reference feature load matrix are respectively determined as the target height activation weight matrix and target feature load matrix corresponding to the above-mentioned marked interference height layer.

[0069] The preset maximum number of iterations can be a maximum number of iterations that is set in advance based on the actual situation, such as 1000 iterations.

[0070] Step S5: Determine the relative distribution coefficient sequence of dry particulate matter corresponding to each humidity interference height layer based on the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer.

[0071] As an example, this step may include the following steps: The first step is to determine any humidity interference height layer as the marked interference height layer, and then determine the target height activation weight matrix and the target feature load matrix corresponding to the marked interference height layer as the marked height activation weight matrix and the marked feature load matrix, respectively.

[0072] The second step is to determine the quotient between the second element (black carbon load) and the third element (humidity load) in each row of the above-mentioned marked feature load matrix as the reference quotient for each row of the above-mentioned marked feature load matrix.

[0073] The third step is to select the row with the largest reference quotient from the above-mentioned marker feature load matrix as the dry particulate matter evolution component, and to determine the row number of the above-mentioned dry particulate matter evolution component in the above-mentioned marker feature load matrix as the marker row number.

[0074] It should be noted that since pure carbon particulate matter often does not expand under high humidity, the component representing the increase in dry matter tends to have a higher loading weight in the black carbon channel, while its loading weight in the relative humidity channel is often extremely low. Therefore, for each row in the labeled feature loading matrix, the quotient of the element value in the second column (black carbon loading) and the element value in the third column (humidity loading) of that row can be calculated. The row vector with the largest quotient can be extracted and identified as the dry particulate matter evolution component.

[0075] The fourth step is to determine the column number in the above-mentioned marker height activation weight matrix that is equal to the above-mentioned marker row number as the dry particulate matter relative distribution coefficient sequence corresponding to the above-mentioned marker interference height layer.

[0076] For example, if the row number is the first row, the sequence of relative distribution coefficients of dry particulate matter corresponding to the marked interference height layer can be the first column in the marked height activation weight matrix, which can be used as the baseline evolution trend for concentration conversion. It can be a set of dimensionless values ​​representing the evolution of relative strength.

[0077] Step S6: Based on the relative distribution coefficient sequence of dry particulate matter, determine the concentration of dehumidified dry particulate matter corresponding to each humidity interference height layer, and realize three-dimensional monitoring of ambient air quality based on the concentration of dehumidified dry particulate matter corresponding to different height layers.

[0078] As an example, this step may include the following steps: The first step is to designate any humidity interference height layer as the marked interference height layer, and to designate a preset number of preset vertical height layers before the marked interference height layer as temporary vertical height layers, thus obtaining a preset number of temporary vertical height layers.

[0079] The second step is to determine the absolute mass concentration index corresponding to the marked interference height layer based on the concentration of dehumidified dry particulate matter corresponding to a preset number of temporary vertical height layers, and the coefficient values ​​corresponding to the preset number of temporary vertical height layers in the sequence of relative distribution coefficients of dry particulate matter corresponding to the marked interference height layer.

[0080] In practice, there are at least a predetermined number of target low-altitude layers. Therefore, when calculating the dehumidified dry particulate matter concentration at the first humidity interference height layer, the predetermined number of target low-altitude layers with known dehumidified dry particulate matter concentrations can be referenced. This layer-by-layer calculation ensures that there are also predetermined number of height layers with known dehumidified dry particulate matter concentrations preceding other humidity interference height layers. Therefore, when calculating the dehumidified dry particulate matter concentration corresponding to the marked interference height layer, the dehumidified dry particulate matter concentrations corresponding to the predetermined number of temporary vertical height layers are known.

[0081] It should be noted that the highly activated weight matrix can be The matrix, , where n is a preset number. That is, one row of the high-activation weight matrix corresponds to one height layer. Therefore, the dry particulate matter relative distribution coefficient sequence is a column of the high-activation weight matrix. Thus, one coefficient value in the dry particulate matter relative distribution coefficient sequence corresponds to one height layer.

[0082] For example, the formula for determining the absolute mass concentration index corresponding to the marked interference height layer can be: ; Where A is the absolute mass concentration index corresponding to the marked interference height layer. n is the preset quantity. i is the sequence number of the temporary vertical height layer. It is the concentration of dehumidified dry particulate matter corresponding to the i-th temporary vertical height layer. It is the coefficient value corresponding to the i-th temporary vertical height layer in the sequence of relative distribution coefficients of dry particulate matter corresponding to the marked interference height layer. It is an adjustment factor that is preset according to the actual situation, mainly used to prevent the denominator from being 0. For example, it can be 0.000001.

[0083] It should be noted that A can represent the calculated mean of the sliding base, or it can represent the absolute mass concentration equivalent to one unit of the relative distribution coefficient of dry particulate matter within the window formed by the marked interference height layer and a preset number of temporary vertical height layers. The meaning can be that the historically determined concentration is divided by its relative proportion in the algorithm, thereby deducing the baseline absolute concentration when the proportion is 1.

[0084] The third step is to determine the calibrated dry particulate matter concentration corresponding to the above-mentioned marked interference height layer based on the absolute mass concentration index corresponding to the above-mentioned marked interference height layer and the coefficient value corresponding to the above-mentioned marked interference height layer in the dry particulate matter relative distribution coefficient sequence corresponding to the above-mentioned marked interference height layer.

[0085] For example, the formula for determining the calibration dry particulate matter concentration corresponding to the marked interference height layer can be: ; Where P is the calibrated dry particulate matter concentration corresponding to the labeled interference height layer. A is the absolute mass concentration index corresponding to the labeled interference height layer. t is the coefficient value corresponding to the labeled interference height layer in the dry particulate matter relative distribution coefficient sequence corresponding to the labeled interference height layer.

[0086] The fourth step, determining the dehumidified dry particulate matter concentration corresponding to the marked interference height layer based on the calibrated dry particulate matter concentration, may include the following sub-steps: In the first sub-step, if the calibrated dry particulate matter concentration corresponding to the above-mentioned marked interference height layer is greater than or equal to the preset theoretical lower limit threshold, then the calibrated dry particulate matter concentration corresponding to the above-mentioned marked interference height layer is determined as the dehumidified dry particulate matter concentration corresponding to the above-mentioned marked interference height layer.

[0087] The preset theoretical lower limit threshold can be a lower limit threshold set in advance based on actual conditions. The formula corresponding to the preset theoretical lower limit threshold can be: ; in, This is the preset theoretical lower limit threshold. BC represents the black carbon concentration corresponding to the marked interference height layer. This is an empirical coefficient for the proportion of black carbon mass, preset based on actual conditions; for example, it could be 0.05. In reality, if the local average black carbon concentration is... The average dry concentration of PM2.5 is ,but It can take the value 0.05.

[0088] It should be noted that occasional intense air mass convection at high altitudes can cause data asynchrony and misalignment in airborne multidimensional sensors within a very short period. This extreme sampling deviation, when fed into the non-negative matrix factorization algorithm, can lead to a significant underestimation or even near-zero value of the calculated calibrated dry particulate matter concentration. However, according to basic atmospheric physics, since independently measured black carbon concentrations (BC) within the same layer have significant values, and black carbon represents only a very small fraction of the total dry particulate matter (PM2.5 or PM10), the actual total dry particulate matter concentration should generally be greater than the concentration of this black carbon component. Therefore, based on the black carbon concentration, a lower limit for the dehumidified dry particulate matter concentration can be obtained, i.e., a preset theoretical lower limit threshold.

[0089] The second sub-step is to set the concentration of calibrated dry particulate matter corresponding to the above-mentioned marked interference height layer to the preset theoretical lower limit threshold if the concentration of dehumidified dry particulate matter corresponding to the above-mentioned marked interference height layer is less than the preset theoretical lower limit threshold.

[0090] The fifth step is to achieve three-dimensional monitoring of ambient air quality based on the concentration of dehumidified and dry particulate matter at different altitude levels.

[0091] For example, if the concentration of dehumidified dry particulate matter corresponding to a preset vertical height level is less than or equal to a preset compliance concentration threshold, the ambient air quality at that preset vertical height level can be determined to meet the standards; if the concentration of dehumidified dry particulate matter corresponding to a preset vertical height level is greater than the preset compliance concentration threshold, the ambient air quality at that preset vertical height level can be determined to fail to meet the standards. The preset compliance concentration threshold can be a threshold set in advance based on actual conditions; for example, it can be... .

[0092] refer to Figure 2 Based on the same inventive concept as the above-described method embodiments, this invention provides a three-dimensional environmental air quality monitoring system based on a high-altitude unmanned aerial vehicle (UAV). This system includes a processor and a memory. The processor processes instructions stored in the memory to implement the steps of a three-dimensional environmental air quality monitoring method based on a high-altitude UAV, specifically including: Multiple data acquisition modules 201 are used to collect the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each preset vertical height layer through the sensors on the airborne high-altitude UAV; The height layer classification and concentration determination module 202 is used to classify the preset vertical height layer according to the height of the preset vertical height layer and its corresponding relative humidity, to obtain the target dry height layer and the humidity interference height layer, and to determine the dehumidified dry particulate matter concentration corresponding to each target dry height layer. The target state matrix construction module 203 is used to construct the target state matrix corresponding to each humidity interference height layer based on the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each humidity interference height layer and the preset number of preset vertical height layers before it. The matrix decomposition module 204 is used to decompose the target state matrix corresponding to each humidity interference height layer to obtain the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer. The relative distribution coefficient sequence determination module 205 is used to determine the relative distribution coefficient sequence of dry particulate matter corresponding to each humidity interference height layer based on the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer. The concentration determination and quality detection module 206 is used to determine the concentration of dehumidified dry particulate matter corresponding to each humidity interference height layer based on the relative distribution coefficient sequence of dry particulate matter, and to realize three-dimensional monitoring of ambient air quality based on the concentration of dehumidified dry particulate matter corresponding to different height layers.

[0093] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. For example, as shown... Figure 3 As shown, the computer device 300 includes: a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302. When the processor 302 executes the computer program 303, the computer device can execute any of the aforementioned three-dimensional environmental air quality monitoring methods based on high-altitude UAVs.

[0094] In summary, this invention categorizes preset vertical height layers based on their height and corresponding relative humidity, resulting in a relatively dry target dry height layer and a humidity-affected height layer. By analyzing the initial particulate matter concentration, black carbon concentration, and relative humidity corresponding to each humidity-affected height layer and a preset number of preceding preset vertical height layers, the dehumidified dry particulate matter concentration corresponding to each humidity-affected height layer is quantified. Ultimately, based on the dehumidified dry particulate matter concentrations corresponding to different height layers, three-dimensional monitoring of ambient air quality can be achieved, improving the rationality of quantifying the dehumidified dry particulate matter concentrations corresponding to humidity-affected height layers, thereby enhancing the rationality of three-dimensional ambient air quality monitoring.

Claims

1. A method for three-dimensional monitoring of ambient air quality based on high-altitude unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Using sensors mounted on a high-altitude drone, the initial particulate matter concentration, black carbon concentration, and relative humidity were collected for each preset vertical height layer. Based on the height of the preset vertical height layer and its corresponding relative humidity, the preset vertical height layer is classified to obtain the target drying height layer and the humidity interference height layer, and the dehumidification dry particulate matter concentration corresponding to each target drying height layer is determined. Based on the initial particulate matter concentration, black carbon concentration and relative humidity corresponding to each humidity interference height layer and the number of preset vertical height layers before it, construct the target state matrix corresponding to each humidity interference height layer; The target state matrix corresponding to each humidity interference height layer is decomposed to obtain the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer; Based on the target height activation weight matrix and target feature load matrix corresponding to each humidity interference height layer, determine the dry particulate matter relative distribution coefficient sequence corresponding to each humidity interference height layer; Based on the relative distribution coefficient sequence of dry particulate matter, the concentration of dehumidified dry particulate matter corresponding to each humidity interference height layer is determined, and three-dimensional monitoring of ambient air quality is achieved based on the dehumidified dry particulate matter concentrations corresponding to different height layers.

2. The method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 1, characterized in that, The step of classifying the preset vertical height layers according to their height and corresponding relative humidity to obtain target dryness height layers and humidity interference height layers includes: The maximum height of each preset vertical height layer is determined as the reference height corresponding to each preset vertical height layer; If the reference height corresponding to the preset vertical height layer is less than or equal to the preset low-altitude height threshold, then the preset vertical height layer will be determined as the target low-altitude height layer. If the reference height corresponding to the preset vertical height layer is greater than the preset low-altitude height threshold, then the preset vertical height layer will be determined as the target high-altitude height layer. If the average relative humidity of the target upper altitude layer and the preset number of preset vertical altitude layers before it is greater than the aerosol deliquescent humidity threshold, then the target upper altitude layer is determined as the humidity interference altitude layer, wherein the reference height corresponding to the preset number of preset vertical altitude layers is less than or equal to the preset low altitude threshold. If the average relative humidity of the target upper altitude layer and the preset number of preset vertical altitude layers before it is less than or equal to the aerosol deliquescence humidity threshold, then the target upper altitude layer is determined as the initial dry altitude layer. The target low-altitude layer and the initial dry altitude layer are collectively referred to as the target dry altitude layer.

3. The method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 2, characterized in that, Determining the concentration of dehydrated dry particulate matter corresponding to each target drying height layer includes: The pre-obtained surface baseline dry particulate matter concentration is used to determine the dehumidified dry particulate matter concentration corresponding to each target low-altitude layer. The initial particulate matter concentration corresponding to each initial drying height layer is determined as the dehumidified dry particulate matter concentration corresponding to each initial drying height layer.

4. The method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 1, characterized in that, The step involves constructing a target state matrix corresponding to each humidity interference height layer based on the initial particulate matter concentration, black carbon concentration, and relative humidity corresponding to each humidity interference height layer and a preset number of preset vertical height layers prior to it, including: Any humidity interference height layer is designated as a marked interference height layer, and the marked interference height layer and a predetermined number of predetermined vertical height layers preceding it constitute a marked height layer sequence; The normalized values ​​of the initial particulate matter concentrations corresponding to all preset vertical height layers in the labeled height layer sequence are used to form a labeled particulate matter concentration vector. The normalized values ​​of the black carbon concentration corresponding to all preset vertical height layers in the marked height layer sequence are used to form a marked black carbon concentration vector. The normalized values ​​of relative humidity corresponding to all preset vertical height layers in the marked height layer sequence are used to form a marked relative humidity vector; Using the labeled particulate matter concentration vector, labeled black carbon concentration vector, and labeled relative humidity vector as column vectors, a target state matrix corresponding to the labeled interference height layer is constructed.

5. A method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 1, characterized in that, The decomposition of the target state matrix corresponding to each humidity interference height layer yields the target height activation weight matrix and target feature load matrix for each humidity interference height layer, including: Any humidity interference height layer is designated as the marked interference height layer, and the target state matrix corresponding to the marked interference height layer is designated as the marked state matrix; Based on the marked state matrix, the pre-acquired initial height activation weight matrix and initial feature load matrix are iteratively updated to obtain the target height activation weight matrix and target feature load matrix corresponding to the marked interference height layer. The elements in the initial height activation weight matrix and the initial feature load matrix are all random numbers within a preset value range.

6. A method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 5, characterized in that, The step of iteratively updating the pre-acquired initial height activation weight matrix and initial feature load matrix based on the labeled state matrix to obtain the target height activation weight matrix and target feature load matrix corresponding to the labeled interference height layer includes: Based on the candidate high activation weight matrix, the candidate feature load matrix, and the labeled state matrix, the candidate high activation weight matrix and the candidate feature load matrix are updated to obtain the reference high activation weight matrix and the reference feature load matrix. The candidate high activation weight matrix is ​​the initial high activation weight matrix or the high activation weight matrix obtained in the previous update, and the candidate feature load matrix is ​​the initial feature load matrix or the feature load matrix obtained in the previous update. The error loss function value between the product of the reference height activation weight matrix and the reference feature load matrix and the labeled state matrix is ​​determined as the reference loss function value corresponding to this update. Similarly, the reference loss function value corresponding to the previous update is obtained. The absolute value of the difference between the reference loss function value corresponding to the current update and the previous update is determined as the target loss change corresponding to the current update. Similarly, the target loss change corresponding to the previous update is obtained. If the change in target loss corresponding to this update and the change in target loss corresponding to the previous update are both less than the preset convergence threshold, then the update is stopped, and the reference height activation weight matrix and the reference feature load matrix are respectively determined as the target height activation weight matrix and the target feature load matrix corresponding to the marked interference height layer. If the number of updates corresponding to this update reaches the preset maximum number of iterations, the update will stop, and the reference height activation weight matrix and reference feature load matrix will be determined as the target height activation weight matrix and target feature load matrix corresponding to the marked interference height layer, respectively.

7. The method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 1, characterized in that, The step of determining the relative distribution coefficient sequence of dry particulate matter for each humidity interference height layer based on the target height activation weight matrix and target feature load matrix for each humidity interference height layer includes: Any humidity interference height layer is designated as the marked interference height layer, and the target height activation weight matrix and target feature load matrix corresponding to the marked interference height layer are designated as the marked height activation weight matrix and the marked feature load matrix, respectively. The quotient between the second and third elements in each row of the labeled feature load matrix is ​​determined as the reference quotient for each row of the labeled feature load matrix; The row with the largest reference quotient value is selected from the labeled feature load matrix and taken as the dry particulate matter evolution component. The row number of the dry particulate matter evolution component in the labeled feature load matrix is ​​determined as the labeled row number. The column whose column number in the marker height activation weight matrix is ​​equal to the marker row number is determined as the dry particulate matter relative distribution coefficient sequence corresponding to the marker interference height layer.

8. A method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 1, characterized in that, The step of determining the concentration of dehumidified dry particulate matter corresponding to each humidity interference height layer based on the relative distribution coefficient sequence of dry particulate matter includes: Any humidity interference height layer is designated as the marked interference height layer, and a preset number of preset vertical height layers preceding the marked interference height layer are designated as temporary vertical height layers, thus obtaining a preset number of temporary vertical height layers. Based on the concentration of dehumidified dry particulate matter corresponding to a preset number of temporary vertical height layers, and the coefficient values ​​of the preset number of temporary vertical height layers in the sequence of relative distribution coefficients of dry particulate matter corresponding to the marked interference height layer, the absolute mass concentration index corresponding to the marked interference height layer is determined. The calibrated dry particulate matter concentration corresponding to the marked interference height layer is determined based on the absolute mass concentration index corresponding to the marked interference height layer and the coefficient value corresponding to the marked interference height layer in the dry particulate matter relative distribution coefficient sequence corresponding to the marked interference height layer. The dehumidified dry particulate matter concentration corresponding to the marked interference height layer is determined based on the calibrated dry particulate matter concentration corresponding to the marked interference height layer.

9. A method for three-dimensional monitoring of ambient air quality based on a high-altitude unmanned aerial vehicle (UAV) according to claim 8, characterized in that, The step of determining the dehumidified dry particulate matter concentration corresponding to the marked interference height layer based on the calibrated dry particulate matter concentration corresponding to the marked interference height layer includes: If the calibrated dry particulate matter concentration corresponding to the marked interference height layer is greater than or equal to the preset theoretical lower limit threshold, then the calibrated dry particulate matter concentration corresponding to the marked interference height layer is determined as the dehumidified dry particulate matter concentration corresponding to the marked interference height layer. If the calibrated dry particulate matter concentration corresponding to the marked interference height layer is less than the preset theoretical lower limit threshold, then the dehumidified dry particulate matter concentration corresponding to the marked interference height layer is set to the preset theoretical lower limit threshold.

10. A three-dimensional ambient air quality monitoring system based on a high-altitude unmanned aerial vehicle (UAV), characterized in that, It includes a processor and a memory, the processor being used to process instructions stored in the memory to implement the three-dimensional environmental air quality monitoring method based on a high-altitude unmanned aerial vehicle according to any one of claims 1-9.

Images