Human-derived aerosol on-line monitoring system based on machine learning

Abstract

The invention relates to the technical field of aerosol concentration monitoring, in particular to a human source aerosol online monitoring system based on machine learning. Obtaining a historical concentration reference according to the non-humidity abnormal moment and the concentration time sequence; according to the historical concentration reference, the concentration value at the current moment and the range of the preset particle size grade, obtaining the concentration abnormality; whether correction is needed or not is judged according to the concentration abnormity and the humidity abnormity moment; and for any concentration time sequence corresponding to the minimum preset particle size grade and the maximum preset particle size grade, the humidity influence degree is obtained according to the correlation characteristics of the humidity at the historical recent preset number of humidity abnormal moments at the current moment and the concentration value at the same moment, and the abnormal continuous characteristics at the historical recent preset number of humidity abnormal moments. The concentration value at the current moment is corrected according to the humidity influence degree and the humidity abnormal characteristics, the corrected concentration value is obtained, and the acquisition accuracy of the aerosol concentration is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of aerosol concentration monitoring, and in particular to an online monitoring system for human-derived aerosols based on machine learning. Background Art

[0002] Human aerosols primarily refer to complex gaseous dispersions generated by the human respiratory tract, containing biomarkers of molecules produced by human metabolism. Medically, data such as particle size distribution, concentration, and chemical composition in human aerosols can be used to quickly and accurately diagnose and monitor conditions like asthma. However, the collection and monitoring of human aerosols is susceptible to the effects of ambient humidity. Excessively high or low humidity can cause changes in the particle size of the aerosols, deviating concentrations across different particle size ranges from actual values and impacting the accuracy of human aerosol analysis. Summary of the Invention

[0003] In order to solve the technical problem that environmental humidity can cause changes in particle size in human-derived aerosols, thereby affecting the accuracy of human-derived aerosol analysis, the present invention aims to provide an online human-derived aerosol monitoring system based on machine learning. The technical solutions adopted are as follows:

[0004] A data acquisition module is used to obtain the time series of ambient humidity and the time series of concentrations of all preset particle size classes of human-derived aerosols;

[0005] A concentration analysis module is configured to obtain a humidity abnormality moment and a non-humidity abnormality moment based on the ambient humidity time series; obtain a historical concentration benchmark for a preset particle size level based on the non-humidity abnormality moment and the concentration time series; and obtain a concentration abnormality degree based on a difference between the historical concentration benchmark and the concentration value at the current moment and the range of the preset particle size level;

[0006] A humidity analysis module is configured to determine whether the concentration value at the current moment needs to be corrected based on the concentration abnormality and the humidity abnormality moment. If correction is required, for any concentration time series corresponding to the minimum preset particle size level and the maximum preset particle size level: obtain the humidity impact of the arbitrary concentration time series based on the correlation characteristics of the humidity at the current moment and the concentration value at the same moment in the arbitrary concentration time series for the most recent preset number of humidity abnormal moments in history, and the abnormal persistence characteristics of the most recent preset number of humidity abnormal moments in history;

[0007] The concentration correction module is used to correct the concentration value at the current moment according to the humidity influence and humidity anomaly characteristics at the current moment, and obtain the corrected concentration value of the human-derived aerosol at the latest moment in the arbitrary concentration time series.

[0008] Furthermore, the step of obtaining the humidity abnormality moment and the non-humidity abnormality moment according to the ambient humidity time sequence includes:

[0009] The moment when the humidity in the ambient humidity time series is within the preset standard humidity range is regarded as the non-humidity abnormal moment, and the moment when the humidity in the ambient humidity time series is not within the preset standard humidity range is regarded as the humidity abnormal moment.

[0010] Furthermore, the step of obtaining a historical concentration benchmark of a preset particle size level according to the non-humidity abnormality time and the concentration time series includes:

[0011] The average value of the concentration values at all non-humidity abnormal moments in the concentration time series of the preset particle size level is calculated to obtain the historical concentration benchmark of the preset particle size level.

[0012] Furthermore, the step of obtaining the concentration abnormality according to the difference characteristics between the historical concentration reference and the current concentration value and the range of the preset particle size level includes:

[0013] Calculate the inverse of the difference between the maximum and minimum values of any preset particle size grade to obtain the particle size range weight; calculate the absolute value of the difference between the concentration value of the arbitrary preset particle size grade at the current moment and the historical concentration benchmark to obtain the concentration difference value; calculate the sum of the products of the concentration difference values corresponding to all preset particle size grades and the particle size range weight and normalize them to obtain the concentration anomaly at the current moment.

[0014] Furthermore, the step of judging whether the concentration value at the current moment needs to be corrected based on the concentration abnormality degree and the humidity abnormality moment includes:

[0015] If the current moment is a humidity abnormal moment and the concentration abnormality exceeds a preset abnormality threshold, the concentration value at the current moment needs to be corrected.

[0016] Furthermore, the step of obtaining the humidity influence degree of the arbitrary concentration time series based on the correlation characteristics of the humidity at the most recent preset number of humidity abnormal moments at the current moment and the concentration value at the same moment in the arbitrary concentration time series, and the abnormal persistence characteristics of the most recent preset number of humidity abnormal moments in the history includes:

[0017] Calculate the absolute value of the Pearson correlation coefficient between the humidity at the most recent preset number of humidity anomaly moments in history and the concentration value at the same moment in the arbitrary concentration time series to obtain a correlation degree value; for any humidity anomaly moment among the most recent preset number of humidity anomaly moments in history, calculate the number of other humidity anomaly moments connected to the arbitrary humidity anomaly moment to obtain a continuous characteristic value of the arbitrary humidity anomaly moment; calculate the sum of the squares of the continuous characteristic values of the most recent preset number of humidity anomaly moments in history and normalize them to obtain a humidity anomaly coefficient; calculate the product of the correlation degree value and the humidity anomaly coefficient to obtain the humidity influence degree.

[0018] Furthermore, the step of correcting the concentration value at the current moment according to the humidity influence and humidity anomaly characteristics at the current moment to obtain the corrected concentration value of the human-origin aerosol at the latest moment in the arbitrary concentration time series includes:

[0019] Calculating the absolute value of the difference between the current humidity value and the median of the preset standard humidity range and normalizing the difference to obtain a humidity difference degree; calculating the product of the humidity difference degree, the humidity influence degree, and a preset constant to obtain a concentration adjustment reference; calculating the sum of constant 1 and the concentration adjustment reference to obtain a first concentration adjustment coefficient; calculating the difference between constant 1 and the concentration adjustment reference to obtain a second concentration adjustment coefficient;

[0020] If the humidity value at the current moment exceeds the preset standard humidity range, the product of the concentration value at the current moment corresponding to the maximum preset particle size level and the second concentration adjustment coefficient is calculated to obtain the corrected concentration; the product of the concentration value at the current moment corresponding to the minimum preset particle size level and the first concentration adjustment coefficient is calculated to obtain the corrected concentration;

[0021] If the humidity value at the current moment is lower than the preset standard humidity range, calculate the product of the concentration value at the current moment corresponding to the maximum preset particle size level and the first concentration adjustment coefficient to obtain the corrected concentration; calculate the product of the concentration value at the current moment corresponding to the minimum preset particle size level and the second concentration adjustment coefficient to obtain the corrected concentration.

[0022] The present invention has the following beneficial effects:

[0023] In the present invention, obtaining the times of humidity anomaly and non-humidity anomaly can determine the aerosol concentration characteristics when undisturbed by external factors. Obtaining the historical concentration baseline can determine the normal level of human-derived aerosols under normal historical conditions, thereby facilitating the determination of whether the concentration is abnormal. Obtaining the concentration anomaly can be used to determine whether the overall aerosol concentration characteristics at the current moment are abnormal, and then, combined with the environmental humidity characteristics, determine whether correction is necessary. Obtaining the humidity influence can characterize the degree to which the aerosol concentration is affected by environmental humidity based on the changing correlation characteristics between humidity and aerosol concentration, as well as the duration of the humidity characteristics, thereby improving the accuracy of the aerosol concentration correction based on the humidity influence. Finally, the current concentration value is corrected based on the humidity influence and humidity anomaly characteristics to obtain a corrected concentration. The corrected concentration can reduce the impact of environmental humidity on aerosol concentration, making the obtained aerosol concentration more accurate, thereby improving the accuracy of aerosol characteristic analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0025] Figure 1 A block diagram of a human-derived aerosol online monitoring system based on machine learning is provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION

[0026] To further illustrate the technical means and effectiveness of the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a machine learning-based online monitoring system for human-derived aerosols. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0027] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0028] The following describes in detail a specific scheme of a human-derived aerosol online monitoring system based on machine learning provided by the present invention with reference to the accompanying drawings.

[0029] See also Figure 1 , which shows a block diagram of a human-derived aerosol online monitoring system based on machine learning provided by one embodiment of the present invention. The system includes the following modules:

[0030] The data acquisition module S1 is used to obtain the time series of environmental humidity and the time series of concentrations of all preset particle size classes of human-derived aerosols.

[0031] In an embodiment of the present invention, the implementation scenario is to correct the concentration value of human-derived aerosol measured under abnormal humidity conditions to improve the accuracy of concentration measurement. First, obtain the ambient humidity time series and the concentration time series of all preset particle size levels of human-derived aerosol. The ambient humidity time series is collected by a humidity sensor, and the concentration of human-derived aerosol is collected by a laser particle size analyzer. In an embodiment of the present invention, 5 common preset particle size levels are set, the first particle size range is 6-100nm, the second particle size range is 100-300nm, the third particle size range is 300nm-1μm, the fourth particle size range is 1-5μm, and the fifth particle size range is 5-10μm. The implementer can determine it by himself according to the implementation scenario. The collection time of the ambient humidity time series and the concentration time series is the same.

[0032] The concentration analysis module S2 is used to obtain the humidity abnormality moment and the non-humidity abnormality moment according to the ambient humidity time series; obtain the historical concentration benchmark of the preset particle size level according to the non-humidity abnormality moment and the concentration time series; and obtain the concentration abnormality degree according to the difference characteristics between the historical concentration benchmark and the concentration value at the current moment and the range of the preset particle size level.

[0033] Under normal circumstances, the concentration of human aerosols collected from the subject is similar at different times. If the aerosol concentration changes, it may be due to abnormal humidity or the presence of a large number of pathogens in the aerosol. Therefore, it is necessary to analyze the human aerosol concentration of the subject collected under normal circumstances to determine whether correction is needed. First, the humidity anomaly and non-humidity anomaly moments are obtained based on the ambient humidity time series. Specifically, the moments when the humidity in the ambient humidity time series is within the preset standard humidity range are regarded as non-humidity anomaly moments, and the moments when the humidity in the ambient humidity time series is outside the preset standard humidity range are regarded as humidity anomaly moments. In this embodiment of the present invention, the preset standard humidity range is 48%-52%. Within this range, the aerosol particle size in the air is unlikely to change. The implementer can determine this based on the implementation scenario.

[0034] Furthermore, aerosol particle size during non-humidity anomaly times is less susceptible to external influences, resulting in a more normal aerosol concentration during these times. Therefore, a historical concentration benchmark for a preset particle size class can be obtained based on the non-humidity anomaly times and the concentration time series. Preferably, in this embodiment of the present invention, obtaining the historical concentration benchmark includes calculating the average concentration value for all non-humidity anomaly times in the concentration time series for the preset particle size class to obtain the historical concentration benchmark for the preset particle size class. The historical concentration benchmark represents the normal aerosol concentration level for that preset particle size class in the absence of external interference. Furthermore, whether the concentration is abnormal can be determined based on the difference between the current aerosol concentration and the corresponding historical concentration benchmark. Under normal conditions of calm breathing, exhaled aerosol particles are smaller in size. However, unusual conditions such as deep breathing or coughing produce larger aerosol particles. Therefore, under normal conditions, aerosol concentration data with smaller particle sizes is more representative. Therefore, when calculating concentration anomalies for all particle size classes, the difference between the concentration data with smaller particle sizes is more important. Therefore, the degree of concentration anomaly is determined based on the difference between the historical concentration benchmark and the current concentration value, as well as the range of the preset particle size class.

[0035] Preferably, in an embodiment of the present invention, the step of obtaining the concentration abnormality includes: calculating the inverse of the difference between the maximum and minimum values of any preset particle size level to obtain the particle size range weight; because the smaller the particle size, the more representative the aerosol concentration is, the smaller the range of the preset particle size level is, and the larger the particle size range weight is. Calculate the absolute value of the difference between the concentration value of any preset particle size level at the current moment and the historical concentration benchmark to obtain the concentration difference value; the larger the concentration difference value, the more abnormal the concentration value of the preset particle size level at the current moment. Calculate the sum of the product of the concentration difference values corresponding to all preset particle size levels and the particle size range weight and normalize them to obtain the concentration abnormality at the current moment; the larger the concentration abnormality, the more abnormal the concentration of the human aerosol collected at the current moment.

[0036] The humidity analysis module S3 is used to determine whether the concentration value at the current moment needs to be corrected based on the concentration abnormality and the humidity abnormal moment. If correction is required, for any concentration time series corresponding to the minimum preset particle size level and the maximum preset particle size level: the humidity impact of any concentration time series is obtained based on the correlation characteristics of the humidity at the current moment and the concentration value at the same moment in any concentration time series for the latest preset number of humidity abnormal moments in history, and the abnormal persistence characteristics of the latest preset number of humidity abnormal moments in history.

[0037] If the current concentration abnormality is significant and the humidity differs significantly from the preset standard humidity range, the aerosol concentration is more likely to be affected by ambient humidity rather than an increase in pathogens. Therefore, the concentration abnormality and the time of the humidity abnormality are used to determine whether the current concentration value needs to be corrected. Specifically, if the current moment is a humidity abnormality and the concentration abnormality exceeds a preset abnormality threshold, the current concentration value needs to be corrected. In this embodiment of the present invention, the preset abnormality threshold is 0.6, which can be determined by the implementer based on the implementation scenario.

[0038] Furthermore, when the ambient humidity increases, the aerosol particles will absorb moisture, causing the particle size to increase. The size of the aerosol particles in the smaller particle size range exhaled by the human body will increase several times or even dozens of times, resulting in a higher concentration of aerosols in the larger particle size range; for example, the aerosol particles in the first particle size range will increase to the second particle size range, the aerosol particles in the second particle size range will increase to the third particle size range, and the aerosol particles in the fourth particle size range will increase to the fifth particle size range; ultimately, the aerosol concentration in the first particle size range will decrease, while the aerosol concentration in the fifth particle size range will increase, and the concentration changes in other particle size ranges will not be obvious. When ambient humidity decreases, the water in aerosol particles evaporates, causing the particles to shrink in size, reducing the size of aerosol particles in larger size ranges. For example, aerosol particles in the fifth size range will shrink to the fourth size range, those in the fourth size range will shrink to the third size range, and those in the second size range will shrink to the first size range. Ultimately, the aerosol concentration in the first size range increases, while the concentration in the fifth size range decreases. The concentrations in other size ranges do not change significantly. Therefore, the more abnormal the ambient humidity, the more likely it is to affect the concentrations of aerosols in the first and fifth size ranges. Therefore, the correlation between humidity anomalies and aerosol concentration can be analyzed. The stronger the correlation between the two, the more pronounced the humidity influences concentration. Humidity is generally stable and continuous. The longer the humidity persists, the more likely it is to correlate with aerosol concentration. The shorter the humidity anomaly persists, the more likely it is that the concentration anomaly is caused by factors other than humidity anomalies, perhaps human factors that alter aerosol particle size.

[0039] Furthermore, if correction is required, for any concentration time series corresponding to the minimum preset particle size level and the maximum preset particle size level: the humidity influence of any concentration time series is obtained based on the correlation characteristics of the humidity at the most recent preset number of humidity anomaly moments at the current moment and the concentration value at the same moment in any concentration time series, and the abnormal persistence characteristics of the most recent preset number of humidity anomaly moments in history. Preferably, in an embodiment of the present invention, the step of obtaining the humidity influence includes: calculating the absolute value of the Pearson correlation coefficient between the humidity at the most recent preset number of humidity anomaly moments in history and the concentration value at the same moment in any concentration time series, and obtaining a correlation degree value; it should be noted that the Pearson correlation coefficient belongs to the prior art, and the more correlated the change characteristics of the two data series are, the closer the absolute value of the Pearson correlation coefficient is to 1, and vice versa, the closer it is to 0; in an embodiment of the present invention, the most recent preset number of humidity anomaly moments in history are the 40 humidity anomaly moments closest to the current moment, and the implementer can determine it by himself according to the implementation scenario. For any humidity anomaly moment within a preset number of recent humidity anomaly moments, calculate the number of other humidity anomaly moments connected to that moment to obtain a continuous characteristic value for that moment. The number of other humidity anomaly moments connected to that moment reflects the duration of the continuous humidity anomaly. A larger continuous characteristic value indicates a longer humidity anomaly period, thus indicating that the aerosol concentration is more susceptible to humidity anomalies. Calculate the sum of the squares of the continuous characteristic values for the preset number of recent humidity anomaly moments and normalize the result to obtain a humidity anomaly coefficient. A larger sum of the squares of the continuous characteristic values indicates a longer humidity anomaly period, and thus a greater likelihood of causing aerosol concentration anomalies. Calculate the product of the correlation value and the humidity anomaly coefficient to obtain a humidity influence. A larger humidity influence indicates that the aerosol concentration is more susceptible to ambient humidity.

[0040] The concentration correction module S4 is used to correct the concentration value at the current moment according to the humidity influence and humidity anomaly characteristics at the current moment, and obtain the corrected concentration value of human-derived aerosol at the latest moment in any concentration time series.

[0041] After obtaining the humidity impact at the current moment, the concentration value at the current moment can be corrected based on the humidity impact and humidity anomaly characteristics to obtain a corrected concentration value for human-source aerosols at the latest moment in any concentration time series. Preferably, in an embodiment of the present invention, obtaining the corrected concentration value includes calculating the absolute value of the difference between the humidity value at the current moment and the median of a preset standard humidity range and normalizing the difference to obtain a humidity difference degree. The humidity difference degree reflects the difference between the current moment's humidity and normal humidity; the greater the humidity difference degree, the greater the impact on the aerosol concentration. The concentration adjustment reference is calculated by multiplying the humidity difference degree, the humidity impact, and a preset constant. In this embodiment of the present invention, the preset constant is 0.1 to adjust the range of the concentration adjustment reference for more accurate correction results. Implementers can determine this value based on their own implementation scenarios. The larger the concentration adjustment reference, the greater the degree of correction for the aerosol concentration. The sum of the constant 1 and the concentration adjustment reference is calculated to obtain a first concentration adjustment coefficient; the difference between the constant 1 and the concentration adjustment reference is calculated to obtain a second concentration adjustment coefficient.

[0042] Furthermore, if the humidity value at the current moment exceeds the preset standard humidity range, the aerosol concentration of the maximum preset particle size class will increase, and its concentration value needs to be lowered during correction; the aerosol concentration of the minimum preset particle size class will decrease, and its concentration value needs to be increased during correction. Therefore, the product of the concentration value at the current moment corresponding to the maximum preset particle size class and the second concentration adjustment coefficient is calculated to obtain the corrected concentration; in this embodiment of the present invention, the maximum preset particle size class is the fifth particle size range, and the corrected concentration of this preset particle size class is less than the collected concentration value. The product of the concentration value at the current moment corresponding to the minimum preset particle size class and the first concentration adjustment coefficient is calculated to obtain the corrected concentration; in this embodiment of the present invention, the minimum preset particle size class is the first particle size range, and the corrected concentration of this preset particle size class is greater than the collected concentration value. If the humidity value at the current moment is lower than the preset standard humidity range, the aerosol concentration of the maximum preset particle size class will decrease, and its concentration value needs to be increased during correction; the aerosol concentration of the minimum preset particle size class will increase, and its concentration value needs to be reduced during correction. Therefore, the corrected concentration is calculated by multiplying the current concentration value corresponding to the maximum preset particle size class by the first concentration adjustment coefficient. The corrected concentration is calculated by multiplying the current concentration value corresponding to the minimum preset particle size class by the second concentration adjustment coefficient. The corrected concentration at the current moment is more accurate than the collected concentration value, reducing the impact of ambient humidity on aerosol concentrations at the minimum and maximum preset particle size classes, thereby increasing the accuracy of subsequent aerosol characteristic analysis.

[0043] In summary, the embodiments of the present invention provide an online human-source aerosol monitoring system based on machine learning. A historical concentration baseline is obtained based on non-humidity anomaly moments and concentration time series. A concentration anomaly degree is obtained based on the historical concentration baseline, the concentration value at the current moment, and the range of preset particle size levels. Whether correction is required is determined based on the concentration anomaly degree and the humidity anomaly moment. If correction is required, for any concentration time series corresponding to the minimum preset particle size level and the maximum preset particle size level: The humidity impact degree is obtained based on the correlation characteristics of the humidity at the current moment and the concentration value at the same moment for the most recent preset number of humidity anomaly moments in history, and the abnormal persistence characteristics for the most recent preset number of humidity anomaly moments in history. The present invention corrects the concentration value at the current moment based on the humidity impact degree and the humidity anomaly characteristics to obtain a corrected concentration value, thereby improving the accuracy of aerosol concentration acquisition.

[0044] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0045] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A human-derived aerosol online monitoring system based on machine learning, characterized in that: The system includes the following modules: A data acquisition module is used to obtain the time series of ambient humidity and the time series of concentrations of all preset particle size classes of human-derived aerosols; A concentration analysis module is configured to obtain a humidity abnormality moment and a non-humidity abnormality moment based on the ambient humidity time series; obtain a historical concentration benchmark for a preset particle size level based on the non-humidity abnormality moment and the concentration time series; and obtain a concentration abnormality degree based on a difference between the historical concentration benchmark and the concentration value at the current moment and the range of the preset particle size level; A humidity analysis module is configured to determine whether the concentration value at the current moment needs to be corrected based on the concentration abnormality and the humidity abnormality moment. If correction is required, for any concentration time series corresponding to the minimum preset particle size level and the maximum preset particle size level: obtain the humidity impact of the arbitrary concentration time series based on the correlation characteristics of the humidity at the current moment and the concentration value at the same moment in the arbitrary concentration time series for the most recent preset number of humidity abnormal moments in history, and the abnormal persistence characteristics of the most recent preset number of humidity abnormal moments in history; The concentration correction module is used to correct the concentration value at the current moment according to the humidity influence and humidity anomaly characteristics at the current moment, and obtain the corrected concentration value of the human-derived aerosol at the latest moment in the arbitrary concentration time series.

2. The human-derived aerosol online monitoring system based on machine learning according to claim 1 is characterized in that: The step of obtaining the humidity abnormality time and the non-humidity abnormality time according to the ambient humidity time sequence comprises: The moment when the humidity in the ambient humidity time series is within the preset standard humidity range is regarded as the non-humidity abnormal moment, and the moment when the humidity in the ambient humidity time series is not within the preset standard humidity range is regarded as the humidity abnormal moment.

3. The human-derived aerosol online monitoring system based on machine learning according to claim 1 is characterized in that: The step of obtaining a historical concentration benchmark of a preset particle size level according to the non-humidity abnormality time and the concentration time series includes: The average value of the concentration values at all non-humidity abnormal moments in the concentration time series of the preset particle size level is calculated to obtain the historical concentration benchmark of the preset particle size level.

4. The human-derived aerosol online monitoring system based on machine learning according to claim 1 is characterized in that: The step of obtaining the concentration abnormality according to the difference characteristics between the historical concentration reference and the current concentration value and the range of the preset particle size level includes: Calculate the inverse of the difference between the maximum and minimum values of any preset particle size grade to obtain the particle size range weight; calculate the absolute value of the difference between the concentration value of the arbitrary preset particle size grade at the current moment and the historical concentration benchmark to obtain the concentration difference value; calculate the sum of the products of the concentration difference values corresponding to all preset particle size grades and the particle size range weight and normalize them to obtain the concentration anomaly at the current moment.

5. The human-derived aerosol online monitoring system based on machine learning according to claim 1 is characterized in that: The step of judging whether the concentration value at the current moment needs to be corrected based on the concentration abnormality degree and the humidity abnormality moment includes: If the current moment is a humidity abnormal moment and the concentration abnormality exceeds a preset abnormality threshold, the concentration value at the current moment needs to be corrected.

6. The human-derived aerosol online monitoring system based on machine learning according to claim 1, characterized in that: The step of obtaining the humidity influence of the arbitrary concentration time series based on the correlation characteristics of the humidity at the latest preset number of humidity abnormal moments at the current moment and the concentration value at the same moment in the arbitrary concentration time series and the abnormal persistence characteristics of the latest preset number of humidity abnormal moments in the history comprises: Calculate the absolute value of the Pearson correlation coefficient between the humidity at the most recent preset number of humidity anomaly moments in history and the concentration value at the same moment in the arbitrary concentration time series to obtain a correlation degree value; for any humidity anomaly moment among the most recent preset number of humidity anomaly moments in history, calculate the number of other humidity anomaly moments connected to the arbitrary humidity anomaly moment to obtain a continuous characteristic value of the arbitrary humidity anomaly moment; calculate the sum of the squares of the continuous characteristic values of the most recent preset number of humidity anomaly moments in history and normalize them to obtain a humidity anomaly coefficient; calculate the product of the correlation degree value and the humidity anomaly coefficient to obtain the humidity influence degree.

7. The human-derived aerosol online monitoring system based on machine learning according to claim 2, characterized in that: The step of correcting the concentration value at the current moment according to the humidity influence and humidity anomaly characteristics at the current moment to obtain the corrected concentration value of the human-origin aerosol at the latest moment in the arbitrary concentration time series includes: Calculating the absolute value of the difference between the current humidity value and the median of the preset standard humidity range and normalizing the difference to obtain a humidity difference degree; calculating the product of the humidity difference degree, the humidity influence degree, and a preset constant to obtain a concentration adjustment reference; calculating the sum of constant 1 and the concentration adjustment reference to obtain a first concentration adjustment coefficient; calculating the difference between constant 1 and the concentration adjustment reference to obtain a second concentration adjustment coefficient; If the humidity value at the current moment exceeds the preset standard humidity range, the product of the concentration value at the current moment corresponding to the maximum preset particle size level and the second concentration adjustment coefficient is calculated to obtain the corrected concentration; the product of the concentration value at the current moment corresponding to the minimum preset particle size level and the first concentration adjustment coefficient is calculated to obtain the corrected concentration; If the humidity value at the current moment is lower than the preset standard humidity range, calculate the product of the concentration value at the current moment corresponding to the maximum preset particle size level and the first concentration adjustment coefficient to obtain the corrected concentration; calculate the product of the concentration value at the current moment corresponding to the minimum preset particle size level and the second concentration adjustment coefficient to obtain the corrected concentration.

Images