CN115220132A - Method for forecasting pollen concentration in atmosphere - Google Patents

Abstract

The invention provides a method for forecasting pollen concentration in atmosphere, which belongs to the field of analysis and forecasting and mainly realizes data analysis, model training, risk forecasting and data display by using data such as weather, pollen, geographic information and the like. Compared with other current methods for preventing meteorological sensitive diseases, the method can realize automatic identification of pollen types, forecast pollen content according to different geographic positions by combining longitude and latitude, and provide reasonable prevention suggestions.

Description

A method for forecasting pollen concentration in the atmosphere

technical field

The invention relates to the field of analysis and forecasting, in particular to a method for forecasting pollen concentration in the atmosphere, which uses meteorological, pollen, geographic information and other data to realize data analysis, model training, and risk forecasting, can effectively avoid areas and seasons with high pollen incidence, and advance Take precautions.

Background technique

Weather, climate and atmospheric environment are closely related to human health. Climate change can affect human health through multiple channels and stages. For example, heat waves and floods can directly cause human discomfort and casualties, leading to a significant increase in regional mortality or the prevalence of infectious diseases. May indirectly affect water supply, agricultural production, food safety and the spread of vector-borne diseases. In the short term, the adjustment of various physiological functions of the human body is related to meteorological elements. When the weather changes sharply beyond the body's tolerance, it will cause a variety of diseases, such as cardiovascular and cerebrovascular diseases, respiratory diseases, infectious diseases, heat stroke, mental diseases, etc. . The onset, aggravation or weakening of these diseases are susceptible to weather or climate change, and have specific seasonal and climatic characteristics, so they are called meteorologically sensitive diseases. In the long run, the frequency of extreme climate events caused by global warming is increasing year by year, aggravating the degree, scope and uncertainty of such diseases. Air pollutants such as PM2.5 and O3 can affect the occurrence of cardiovascular and cerebrovascular diseases, respiratory diseases and other diseases, which are easy to aggravate when high temperature or cold wave occurs. This meteorological-environment interaction has a significant role in promoting the occurrence and development of meteorological-sensitive diseases. Some studies have been carried out in this field at home and abroad, but there is still a lack of sufficient research results and quantitative analysis evidence. At the same time, such research results are significantly affected by geographical and population characteristics, and general conclusions cannot be drawn. Foreign or foreign models cannot simply be copied. It is necessary to combine local population health, meteorological and environmental data to establish a disease meteorological risk forecast model that conforms to local conditions and meets local needs, so as to improve the practicability of healthy meteorological forecasting, early warning and assessment services.

Pollen in the air is one of the main airborne allergens that induce respiratory diseases. With the increase of urban greening, the number of patients with hay fever caused by allergy to plant pollen is increasing year by year, and pollen allergy disease has become one of the seasonal and epidemic diseases in the world. For most patients, effectively avoiding areas and seasons with high pollen incidence and taking preventive measures in advance are the most effective prevention methods at present.

SUMMARY OF THE INVENTION

In order to solve the above technical problems, the present invention provides a method for forecasting pollen concentration in the atmosphere, which narrows the development scope of weather-sensitive diseases, prompts sensitive people to avoid high pollen occurrence periods and areas, and effectively reduces the cost of medical expenditures in the whole society.

The technical scheme of the present invention is:

A method for forecasting pollen concentration in the atmosphere, automatically collecting pollen samples to generate image information, correlating the image information with site information for storage, using the collected pollen data for model training, realizing pollen artificial intelligence identification, and combining historical data to form pollen concentration statistics. Regression equation model to realize the prediction of pollen index.

further,

The pollen is sampled and made into glass slides; the automatic collection of pollen images is realized by using the ARM chip to program the free movement of the slide glass and digital image generation of the CMOS video sensor.

By combining the pollen concentration data observed under the corresponding meteorological conditions in the history, the model is trained through the artificial intelligence platform to form a regression equation model related to the pollen concentration and the region, time, and meteorological factors to predict the pollen concentration.

The pollen concentration index is calculated by observing the calculated pollen concentration data or the predicted pollen concentration data, combined with the pollen type and the degree of influence of the pollen concentration on human health.

Going further,

Automatically collect pollen images from the microscope, and use the image recognition algorithm of the artificial intelligence platform to train the automatic pollen identification, classification, and counting model, so as to automatically obtain the category information and quantity of various pollen on the collected images, and calculate their concentrations on this basis. At the same time, the artificial intelligence meteorological prediction algorithm is used, combined with historical meteorological information, to train the regression equation model of pollen concentration prediction, and then combined with the meteorological forecast information to predict the pollen concentration and species.

The pollen sampling device is a volumetric pollen collector. The air flow is fixed every minute. The atmospheric pollen particles are adsorbed on the polyester film and replaced once a week. The time resolution is daily. The collected samples are made of glycerol glue to make glass slides. Put it under a microscope for identification, and count the types and numbers of pollen families.

After the collection of image information is completed, the generated pollen image is automatically stored in the system through the interface, and is associated with the meteorological observation site collected by the pollen slide; the basic pollen collection picture library is formed for subsequent automatic pollen identification, Automatic counting, automatic classification, automatic concentration calculation provide data basis.

On the collected new pollen pictures, outline the pollen contours in the pictures, and classify the types of pollen to form a standard set of artificial intelligence automatic pollen identification model training;

After the artificial intelligence automatically recognizes and automatically classifies the pollen pictures, the artificial intelligence error labeling is corrected. The types of corrections included three types of pollen not identified in the picture, pollen not present in the picture identified, and pollen identified wrongly classified.

The beneficial effects of the present invention are

Compared with common solutions, the present invention can analyze the induction mechanism of local related diseases caused by abnormal changes in weather and climate on the basis of meteorological big data, environmental big data and disease big data. The degree of influence of disease patients, scientific prevention and early warning of weather-sensitive diseases and early precise intervention for vulnerable groups such as the elderly and children. In response to the increasing number of people with hay fever, intelligent pollen observation is carried out, combined with meteorological conditions, forecast pollen concentration, and give early warning to susceptible people.

Description of drawings

FIG. 1 is a schematic diagram of the work flow of the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work are protected by the present invention. scope.

The invention provides a method for forecasting pollen concentration in the atmosphere, which mainly uses meteorological, pollen, geographic information and other data to realize data analysis, model training, risk forecasting and data display. Compared with other current weather-sensitive disease prevention methods, this method can automatically identify pollen types, combine longitude and latitude, predict pollen content according to different geographic locations, and give rationalized prevention suggestions.

1. Pollen information: The atmospheric pollen sampling device is usually a volumetric pollen collector. The airflow is fixed every minute. The atmospheric pollen particles are adsorbed on the polyester film and replaced once a week. The time resolution is daily. The collected samples use glycerol. Make glass slides with glue, put them under a high-power microscope for identification, and count the types and numbers of pollen families. The atmospheric pollen sampling equipment used is an international sampler. The sporopollen identification personnel identified to ensure that the data is true and reliable.

2. ARM+CMOS automatic acquisition: add a simple CNC robotic arm and a CMOS video sensor on the basis of the binocular biological optical microscope commonly used in the station, and use the ARM chip to program the free movement of the slide and the digital image of the CMOS video sensor Generated to realize automatic collection of pollen images.

3. Meteorological data: Through the data API interface provided by the "Tianqing" system to provide external services, the meteorological-related data of the whole city are collected. The data range is the conventional meteorological observation data of various national meteorological stations in the past 10 years, including temperature, dew point, air pressure, and wind direction. , wind speed, total cloud cover, low cloud cover, visibility, 24-hour pressure change, weather phenomenon, daily maximum temperature, daily minimum temperature, 24-hour precipitation, sunshine hours, evaporation, etc., as well as weather situation, circulation index and other information .

4. Association between observation point data and image data: After the collection of image information is completed, the generated pollen image is automatically stored in the system through the interface, and is associated with the meteorological observation site collected by the pollen slide. The basic pollen collection picture library is formed, which provides the data basis for the subsequent automatic identification, automatic counting, automatic classification and automatic concentration calculation of pollen.

5. Model training, pollen identification: In the system, pollen image annotation needs to be performed in two steps. Both steps require personnel with extensive palynological expertise. In the first step, professionals use the system to draw the outline of the pollen in the picture on the new pollen picture collected by the system, use the professional knowledge, use the mouse on the PC or use the stylus pen on the PAD, and classify the types of pollen. , to form a standard set for artificial intelligence automatic pollen identification model training. Another step is that after the artificial intelligence automatically recognizes and automatically classifies the pollen pictures, professionals check the recognition effect and correct the artificial intelligence error labeling. The types of corrections included three types of pollen not identified in the picture, pollen not present in the picture identified, and pollen identified wrongly classified.

6. Machine learning algorithm: Pollen concentration refers to the number of pollen grains per 1000mm ² in the air within 24 hours. At present, the forecast of pollen concentration mainly uses the method of statistical forecasting, because meteorological factors have no linear relationship with pollen concentration, and its influence on pollen is very complex. Therefore, we chose to use machine learning algorithms to analyze the time distribution of pollen concentration, the correlation between pollen concentration and meteorological factors, the peak period of pollen concentration, and the peak period of pollen concentration. The factors that significantly affect the pollen concentration are combined with the pollen concentration data observed under the corresponding meteorological conditions in the history, and the model is trained through the artificial intelligence platform to form a regression equation model that the pollen concentration is related to regional, time and meteorological factors.

7. Pollen forecast: The system calculates the pollen concentration index according to the pollen concentration data obtained by observation and calculation or the predicted pollen concentration data, and according to the meteorological industry standard, combined with the pollen type and the degree of influence of the pollen concentration on human health. The index can be calculated according to different pollen concentrations and different groups of people, and provided to other systems issued to residents through API, reminding residents of the impact of pollen types and concentrations on different groups of people with hay fever, so as to avoid areas with high pollen incidence and season.

The present invention adopts a three-layer architecture, which is divided into: 1. User Interface layer; 2. Business Logic Layer; 3. Data Access layer.

The invention adopts a deployment and development framework based on micro-services, which can support rapid version replacement, realize low-cost capacity expansion, elastic expansion, adaptation to cloud environment, and maximize system resource utilization. Support mainstream UNIX, Linux, WindowsServer series operating systems to meet the installation and deployment in different server environments. The construction system of the present invention adopts the principle of centralized deployment, and is uniformly deployed on the private cloud platform. The mobile APP application terminal supports mainstream system versions such as Android and IOS, and adapts to different hardware devices.

The above descriptions are only preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, but not to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (9)

1. A method for forecasting the pollen concentration in the atmosphere is characterized in that,

the method comprises the steps of automatically collecting pollen samples to generate image information, storing the image information and site information in a related mode, conducting model training by using collected pollen data to achieve artificial intelligent identification of pollen, forming a regression equation model of pollen concentration by combining historical data, and achieving prediction of pollen indexes.

2. The method of claim 1,

sampling pollen and making into a glass slide; the automatic collection of the pollen image is realized by utilizing the ARM chip to program and control the free movement of the glass slide and the generation of the digital image of the CMOS video sensor.

3. The method of claim 2,

by combining pollen concentration data observed under corresponding meteorological conditions in history and carrying out model training through an artificial intelligence platform, a regression equation model related to the pollen concentration, the area, the time and the meteorological factors is formed, and the pollen concentration is predicted.

4. The method of claim 3,

and calculating the pollen concentration index by observing the calculated pollen concentration data or the predicted pollen concentration data and combining the influence degree of the pollen type and the pollen concentration on the human health.

5. The method of claim 4,

the method comprises the steps of automatically acquiring pollen images from a microscope, training a pollen automatic identification, classification and counting model by using an image identification algorithm of an artificial intelligence platform, automatically acquiring category information and quantity of various pollens on the acquired images, calculating the concentration of the pollens on the basis, training a regression equation model for pollen concentration prediction by using an artificial intelligence weather prediction algorithm and combining weather historical information, and predicting the pollen concentration and type by combining weather forecast information.

6. The method of claim 5,

the pollen sampling device is a positive displacement pollen collector, the airflow is fixed every minute, the atmospheric pollen particles are adsorbed on the polyester film and replaced once a week, the time resolution is day by day, the collected samples are made into slides by glycerol glue, and the slides are placed under a microscope for identification, and the types and the number of the pollen families are counted.

7. The method of claim 6,

automatically storing the generated pollen image into a system in an interface mode after the image information is collected, and associating the pollen image with a meteorological observation station collected by a pollen glass slide; and a basic pollen collection picture library is formed, and a data basis is provided for subsequent pollen automatic identification, automatic counting, automatic classification and automatic concentration calculation.

8. The method of claim 7,

on the collected brand-new pollen picture, sketching the pollen outline in the picture, and classifying the pollen types to form a standard set for artificial intelligent automatic pollen recognition model training;

after the pollen picture is automatically identified and classified by artificial intelligence, the artificial intelligence error label is corrected.

9. The method of claim 8,

the corrected types comprise three types of pollen which is not recognized in the picture, pollen which does not exist in the picture and recognized pollen classification errors.

Images