An ozone pollution prediction method based on meteorological similarity degree

Abstract

The application relates to the field of ozone pollution prediction, in particular to an ozone pollution prediction method based on meteorological similarity, which comprises the following steps: S1, collecting historical meteorological data and forecast meteorological data to form a historical meteorological two-dimensional meteorological data matrix; S2, adopting Euclidean distance similarity, cosine similarity and a dynamic time warping algorithm (DTW) to construct an ozone pollution prediction model based on meteorological similarity, and comparing the two-dimensional matrix of the forecast meteorological data with the historical meteorological data one by one. The application solves the problems that the prediction accuracy of the traditional ozone pollution prediction method in the prior art is prone to have a large deviation, the prediction result is not ideal, the data calculation amount is large during training, the time consumption is relatively long, and the complexity is high, and the application depends on the strong correlation between ozone concentration and meteorological conditions, only uses historical meteorological data and forecast meteorological data, the input data is relatively easy to obtain, data preprocessing is simple, and short-term ozone pollution prediction of national and municipal control sites can be realized.

Description

Technical Field

[0001] This invention relates to the field of ozone pollution prediction, and more specifically, to a method for predicting ozone pollution based on meteorological similarity. Background Technology

[0002] Ozone is an important greenhouse gas and photochemical oxidant in the atmosphere. It appears blue at room temperature and has strong oxidizing properties and a fishy odor. Most ozone is found in the upper atmosphere or stratosphere, with only about 10% distributed in the troposphere, near the Earth's surface.

[0003] Domestic and international scholars have conducted extensive research on ozone concentration forecasting and prediction methods, which can be mainly divided into dynamic modeling methods and data-driven methods. Dynamic modeling methods require high computer computation, which often leads to significant deviations in the accuracy of data prediction. Data-driven methods cannot consider the dynamic spatiotemporal correlation of complex time series data, resulting in unsatisfactory prediction results. They also involve large amounts of data computation during training, are time-consuming, and highly complex. Therefore, we propose an ozone pollution prediction method based on meteorological similarity to improve upon these methods. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing traditional ozone pollution prediction methods, which often suffer from significant deviations in prediction accuracy, unsatisfactory prediction results, large amounts of data computation during training, long processing times, and high complexity.

[0005] To achieve the above-mentioned objectives, this invention provides an ozone pollution prediction method based on meteorological similarity to improve the aforementioned problems.

[0006] The application is as follows:

[0007] An ozone pollution prediction method based on meteorological similarity includes the following steps:

[0008] S1. Collect historical meteorological data and forecast meteorological data to form a two-dimensional historical meteorological data matrix;

[0009] S2. Using Euclidean distance similarity, cosine similarity and dynamic time warping (DTW) algorithm, an ozone pollution prediction model based on meteorological similarity is constructed. The two-dimensional matrix of forecast meteorological data is compared with historical meteorological data one by one, and the historical similar period with the highest similarity and its predicted ozone concentration are output to realize short-term real-time prediction of ozone pollution at national and municipal control stations.

[0010] S3. Euclidean Distance Similarity: Suppose x and y are two points in n-dimensional space, the Euclidean distance between them is:

[0011]

[0012] S4. Cosine Similarity: Cosine similarity involves flattening a matrix into vectors, calculating the product of the vectors, and dividing by the magnitude. The cosine similarity algorithm is used to calculate the similarity of document data.

[0013]

[0014] S5. Dynamic Time Warping (DTW): This method measures the similarity between two time series of different lengths based on the nearest proximity principle. It is a non-linear warping technique. The DTW algorithm matches the changing trends of two time series data. In terms of meteorological similarity, it calculates the similarity of each meteorological data factor separately using DTW. The meteorological data factors are assigned weights of 0.45, 0.3, 0.15, 0.1, 0.1, and 0.1 according to their degree of influence on ozone pollution. The similarity of each factor is then summed with its weights to determine the overall similarity.

[0015] As a preferred technical solution of this application, it also includes step S6: Finally, by acquiring historical meteorological data and meteorological forecast data in real time, constructing a prediction chart visualization component, and constructing a data analysis interaction component, the short-term prediction results of ozone pollution at national and municipal monitoring stations are visualized.

[0016] As a preferred technical solution of this application, in step S3, when n=2, the Euclidean distance is the distance between two points on a plane. When using Euclidean distance to represent similarity, the smaller the distance, the greater the similarity. Generally, the following formula is used for conversion:

[0017]

[0018] As a preferred technical solution of this application, the meteorological data factors in step S5 include temperature (TMP), cumulative rainfall (PRE), wind direction (WD), wind speed (WS), relative humidity (RH), and pressure (PRS).

[0019] As a preferred technical solution of this application, the calculation formula in step S5 is as follows:

[0020] Similarity = DTW(A factor 1, B factor 1) * factor 1 weight + ... + DTW(A factor n, B factor n) * factor n weight.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0022] In the scheme of this application:

[0023] 1. In order to address the problems that traditional ozone pollution prediction methods in the prior art often have large deviations in prediction accuracy, unsatisfactory prediction results, large amount of data calculation during training, long time consumption, and high complexity, this application relies on the strong correlation between ozone concentration and meteorological conditions, uses only historical meteorological data and forecast meteorological data, the input data is easy to obtain, the data preprocessing is simple, and it can realize short-term ozone pollution prediction for national and municipal monitoring stations.

[0024] 2. Based on the strong correlation between ozone pollution and meteorological conditions, this application constructs an ozone pollution prediction model based on meteorological similarity. Different weights are assigned according to the degree of influence of different meteorological factors on ozone pollution, and meteorological similarity is compared. Multiple similarity algorithms (Euclidean distance similarity, cosine similarity, and dynamic time warping similarity) are used to compare the similarity of the two-dimensional meteorological data matrix (meteorological factors * time). The similarity algorithm is selected based on the accuracy of ozone prediction. Simultaneously, only the months and time periods with the most severe ozone pollution (May-September, 12-19:00) are selected for model training and optimization to further improve the accuracy of ozone pollution prediction. Finally, through the real-time acquisition of historical meteorological data and meteorological forecast data, the construction of prediction chart visualization components, and the construction of interactive components such as data analysis, the short-term prediction results of ozone pollution at national and municipal monitoring stations are visualized. Attached Figure Description

[0025] Figure 1 A flowchart illustrating the principle of the ozone pollution prediction method based on meteorological similarity provided in this application;

[0026] Figure 2 Meteorological data matrix charts for the ozone pollution prediction method based on meteorological similarity provided in this application. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] As described in the background section, traditional ozone pollution prediction methods tend to have large deviations in prediction accuracy, resulting in unsatisfactory prediction results. They also require a large amount of data computation during training, are time-consuming, and highly complex.

[0029] To address this technical problem, this invention provides an ozone pollution prediction method based on meteorological similarity, which is applied to the field of ozone pollution prediction.

[0030] For details, please refer to Figure 1 and Figure 2 An ozone pollution prediction method based on meteorological similarity includes the following steps:

[0031] S1. Collect historical meteorological data and forecast meteorological data to form a two-dimensional historical meteorological data matrix;

[0032] S2. Using Euclidean distance similarity, cosine similarity and dynamic time warping (DTW) algorithm, an ozone pollution prediction model based on meteorological similarity is constructed. The two-dimensional matrix of forecast meteorological data is compared with historical meteorological data one by one, and the historical similar period with the highest similarity and its predicted ozone concentration are output to realize short-term real-time prediction of ozone pollution at national and municipal control stations.

[0033] S3. Euclidean Distance Similarity: Suppose x and y are two points in n-dimensional space, the Euclidean distance between them is:

[0034]

[0035] S4. Cosine Similarity: Cosine similarity involves flattening a matrix into vectors, calculating the product of the vectors, and dividing by the magnitude. The cosine similarity algorithm is used to calculate the similarity of document data.

[0036]

[0037] S5. Dynamic Time Warping (DTW): This method measures the similarity between two time series of different lengths based on the nearest proximity principle. It is a non-linear warping technique. The DTW algorithm matches the changing trends of two time series data. In terms of meteorological similarity, it calculates the similarity of each meteorological data factor separately using DTW. The meteorological data factors are assigned weights of 0.45, 0.3, 0.15, 0.1, 0.1, and 0.1 according to their degree of influence on ozone pollution. The similarity of each factor is then summed with its weights to determine the overall similarity.

[0038] This invention relies on the strong correlation between ozone concentration and meteorological conditions, using only historical and forecast meteorological data. The input data is relatively easy to obtain, and the data preprocessing is simple, enabling short-term ozone pollution prediction at national and municipal monitoring stations.

[0039] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0040] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.

[0041] Example 1, please refer to Figure 1 and Figure 2 An ozone pollution prediction method based on meteorological similarity includes the following steps:

[0042] S1. Collect historical meteorological data and forecast meteorological data to form a two-dimensional historical meteorological data matrix;

[0043] S2. Using Euclidean distance similarity, cosine similarity and dynamic time warping (DTW) algorithm, an ozone pollution prediction model based on meteorological similarity is constructed. The two-dimensional matrix of forecast meteorological data is compared with historical meteorological data one by one, and the historical similar period with the highest similarity and its predicted ozone concentration are output to realize short-term real-time prediction of ozone pollution at national and municipal control stations.

[0044] S3. Euclidean Distance Similarity: Suppose x and y are two points in n-dimensional space, the Euclidean distance between them is:

[0045]

[0046] S4. Cosine Similarity: Cosine similarity involves flattening a matrix into vectors, calculating the product of the vectors, and dividing by the magnitude. The cosine similarity algorithm is used to calculate the similarity of document data.

[0047]

[0048] S5. Dynamic Time Warping (DTW): This method measures the similarity between two time series of different lengths based on the nearest proximity principle. It is a non-linear warping technique. The DTW algorithm matches the changing trends of two time series data. In terms of meteorological similarity, it calculates the similarity of each meteorological data factor separately using DTW. The meteorological data factors are assigned weights of 0.45, 0.3, 0.15, 0.1, 0.1, and 0.1 according to their degree of influence on ozone pollution. The similarity of each factor is then summed with its weights to determine the overall similarity.

[0049] It also includes step S6: Finally, by acquiring historical meteorological data and meteorological forecast data in real time, constructing a prediction chart visualization component, and constructing a data analysis interaction component, the short-term prediction results of ozone pollution at national and municipal monitoring stations are visualized.

[0050] In step S3, when n=2, the Euclidean distance is the distance between two points on a plane. When using Euclidean distance to represent similarity, the smaller the distance, the greater the similarity. The following formula is generally used for conversion:

[0051]

[0052] The meteorological data factors in step S5 include temperature (TMP), cumulative rainfall (PRE), wind direction (WD), wind speed (WS), relative humidity (RH), and pressure (PRS).

[0053] The calculation formula in step S5 is as follows:

[0054] Similarity = DTW(A factor 1, B factor 1) * factor 1 weight + ... + DTW(A factor n, B factor n) * factor n weight.

[0055] Example 2 further optimizes the ozone pollution prediction method based on meteorological similarity provided in Example 1. Specifically, the accurate collection of meteorological data includes the following steps:

[0056] S1. Selecting an observation site: The ground meteorological observation site should be located in a place that reflects the characteristics of meteorological elements over a large area of ​​the locality, is representative, avoids the influence of local topography, and is surrounded by open and flat areas. Avoid setting up the site on steep slopes, in depressions, or near tall buildings, railways, highways, or heat sources.

[0057] S2. Install and maintain measuring instruments;

[0058] S3. Data recording and storage: Ensure that all measuring instruments are equipped with data loggers to automatically and periodically record and store the measured data.

[0059] Temperature (TMP): A thermometer equipped with a data logger is used for timed data collection. The thermometer is placed in a radiation shield to prevent the influence of direct solar radiation and ground-reflected radiation on the instrument, while ensuring proper ventilation for the instrument's sensing part to accurately sense changes in the outside air temperature.

[0060] Accumulated Rainfall (PRE): Rainfall is measured and recorded using precipitation sensors.

[0061] Wind direction (WD) and wind speed (WS): A wind vane and a wind speed sensor are used. The wind vane is used to measure the wind direction, while the wind speed sensor calculates the wind speed by measuring the distance the air travels per unit time.

[0062] Relative humidity (RH): A humidity sensor measures the humidity in the air and transmits the data to a data logger.

[0063] Pressure: Atmospheric pressure is measured using a barometer.

[0064] Accurate meteorological data collection also includes the following steps:

[0065] S4. Data Quality Control: Regularly calibrate and maintain measuring instruments to ensure data accuracy, check and record the instrument's operating status, and any factors that may affect measurement accuracy;

[0066] S5. Data transmission and processing: Connect the measuring instrument to the network and data transmission system to observe data in real time, and use data processing software to clean, analyze and visualize the collected data.

[0067] Obviously, the embodiments described above are merely some embodiments of the present invention, not all embodiments. The accompanying drawings show preferred embodiments of the present invention, but do not limit the patent scope of the present invention. The present invention can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the patent protection scope of this invention.

Claims

1. A method for predicting ozone pollution based on meteorological similarity, characterized in that, Includes the following steps: S1. Collect historical meteorological data and forecast meteorological data to form a two-dimensional historical meteorological data matrix; S2. Using Euclidean distance similarity, cosine similarity and dynamic time warping algorithms, an ozone pollution prediction model based on meteorological similarity is constructed. The two-dimensional matrix of forecast meteorological data is compared with historical meteorological data one by one, and the historical similar period with the highest similarity and its predicted ozone concentration are output to realize short-term real-time prediction of ozone pollution at national and municipal control stations. S3. Euclidean Distance Similarity: Given two points x and y in n-dimensional space, the Euclidean distance between them is: ； S4. Cosine Similarity: Cosine similarity involves flattening a matrix into vectors, calculating the product of the vectors, and dividing by the magnitude. The cosine similarity algorithm is used to calculate the similarity of document data. ； S5. Dynamic Time Warping Algorithm: Based on the principle of proximity, this method measures the similarity between two time series of different lengths. The Dynamic Time Warping algorithm matches the changing trends of two time series data. In terms of meteorological similarity, DTW is used to calculate the similarity of each meteorological data factor. The meteorological data factors are assigned weights of 0.45, 0.3, 0.15, 0.1, 0.1, and 0.1 according to their degree of influence on ozone pollution. The similarity of each factor is then summed with its weights to form the overall similarity. The meteorological data factors include temperature, cumulative rainfall, wind direction, wind speed, relative humidity, and pressure. Similarity = DTW(Factor A1, Factor B1) × Factor 1 weight + ... + DTW(Factor An, Factor Bn) × Factor n weight.

2. The ozone pollution prediction method based on meteorological similarity according to claim 1, characterized in that, It also includes step S6: Finally, by acquiring historical meteorological data and meteorological forecast data in real time, constructing a prediction chart visualization component, and constructing a data analysis interaction component, the short-term prediction results of ozone pollution at national and municipal monitoring stations are visualized.

3. The ozone pollution prediction method based on meteorological similarity according to claim 2, characterized in that, In step S3, when n=2, the Euclidean distance is the distance between two points on a plane. When using Euclidean distance to represent similarity, the smaller the distance, the greater the similarity. The following formula is used for conversion: 。

Images