A method and system for evaluating contribution rate of meteorological conditions to air quality

Abstract

The present invention relates to the field of meteorological analysis, and specifically refers to a method and system for evaluating the contribution rate of meteorological conditions of air quality, the method comprising: constructing a sequence set, mean filtering, establishing an optimal meteorological model through a stepwise selection regression algorithm, and calculating the contribution rate. In view of the current method for evaluating the contribution rate of meteorological conditions of air quality, the traditional automatic variable selection algorithm used is prone to problems such as overfitting and collinearity, which in turn affects the problem of contribution rate evaluation. The present invention uses a stepwise selection regression algorithm to perform a forward stepwise comparative evaluation on the meteorological model, and combines multiple regression model evaluation criteria, including mean square error, Akaike information criterion and variance inflation factor, to solve the overfitting, underfitting and multicollinearity problems of the automatic variable selection algorithm. In addition, the algorithm used in the present invention further performs a significance check with a meteorological model with an optimal evaluation standard to ensure that the meteorological model has statistical significance and better predictive ability.

Description

Technical Field

[0001] The present invention relates to the field of meteorological analysis, and in particular to a method and system for evaluating the contribution rate of meteorological conditions to air quality. Background Art

[0002] With the rapid development of industrialization and urbanization, air pollution has become increasingly serious, causing great impacts on human health and ecosystems. Therefore, developing a method and system that can accurately evaluate the contribution of meteorological conditions to air quality is crucial for formulating effective air quality management strategies and reducing pollutant emissions. However, the current air quality meteorological condition contribution rate assessment method is prone to overfitting and collinearity problems when using traditional automatic variable selection algorithms, making it difficult to find the optimal regression model, which in turn affects the accuracy and reliability of the contribution rate assessment. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a method and system for evaluating the contribution rate of meteorological conditions of air quality. In view of the current method for evaluating the contribution rate of meteorological conditions of air quality, when using the traditional automatic variable selection algorithm, it is easy to produce problems such as overfitting and collinearity, and it is difficult to find the optimal regression model, which in turn affects the accuracy and reliability of the contribution rate evaluation. The present invention adopts a stepwise selection regression algorithm to perform a forward stepwise comparative evaluation on the meteorological model, combining multiple regression model quality evaluation criteria rather than a single indicator, including mean square error, Akaike information criterion and variance inflation factor, to solve the overfitting, underfitting and multicollinearity problems of the automatic variable selection algorithm. In addition, unlike simple forward and backward selection, the algorithm used in the present invention further performs a significance check with a meteorological model with the optimal evaluation criteria to ensure that the meteorological model has statistical significance and better predictive ability, thereby improving the stability and reliability of the meteorological model.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a method for evaluating the contribution rate of meteorological conditions to air quality, the method comprising the following steps:

[0005] S1: Obtain data on air pollutants and meteorological elements, and construct pollutant series and meteorological series sets;

[0006] S2: Mean filtering of pollutant series and meteorological series sets;

[0007] S3: Establish the optimal meteorological model through the stepwise selection regression algorithm;

[0008] S4: Calculate the contribution rate of meteorological elements according to the optimal meteorological model.

[0009] In step S2, the mean filtering of the pollutant sequence and the meteorological sequence set is specifically as follows: subtract the step length of each sequence in the pollutant sequence and the meteorological sequence set from The sliding mean of .

[0010] In step S3, the specific steps of establishing the optimal meteorological model through the stepwise selection regression algorithm are as follows:

[0011] S31: Set initialization: Take each sequence in the meteorological sequence set as an independent variable and the pollutant sequence as the dependent variable, establish an empty model that does not contain any independent variables, and establish a meteorological model set Add the empty model to the meteorological model set and create a deletion set ,at this time is an empty set;

[0012] S32: Building a new weather model: Traversal Each weather model in , add a The independent variables not included are used to establish a linear regression equation to form a new meteorological model ,Will Add to ;

[0013] S33: Screening is performed according to mean square error, Akaike information criterion and variance inflation factor. The specific steps are as follows:

[0014] S331: Calculation The mean square error and The mean square error ;

[0015] S332: Calculation Akaike Information Criterion and Akaike Information Criterion ;

[0016] S333: Define variance inflation factor threshold ,calculate Variance inflation factor ;

[0017] S334: If and and , then it is believed that Better than and will Join , otherwise Join ;

[0018] S34: Repeat S32 until the traversal is completed, remove All of the The elements that produce the intersection;

[0019] S35: Finding the optimal mean square error meteorological model and the optimal Akaike information criterion meteorological model: Traversal , the meteorological model with the smallest mean square error is taken as the optimal mean square error meteorological model , the meteorological model with the smallest Akaike information criterion is taken as the optimal Akaike information criterion meteorological model ;

[0020] S36: Significance check: Setting the mean square error significance threshold and Akaike information criterion significance threshold , traverse ,for and , and use the t-test method to calculate each meteorological model Significant difference in mean square error The significant difference between the Akaike information criterion and , if with and of All less than and All less than , then keep , otherwise from Delete ;

[0021] S37: Obtain the optimal meteorological model: The mean square error and Akaike information criterion of each meteorological model in are forward processed, and then each meteorological model is scored using the superior and inferior solution distance method. The meteorological model with the highest score is the optimal meteorological model.

[0022] In step S4, the contribution rate of meteorological elements is calculated according to the optimal meteorological model by dividing the weight of each meteorological element in the optimal meteorological model by the sum of all weights of the optimal meteorological model. The specific formula is as follows:

[0023] ;

[0024] in Indicates the category The contribution rate of meteorological elements, Indicates the category The weight of meteorological elements in the optimal meteorological model, represents the number of independent variables in the optimal meteorological model, Indicates The weight of the independent variable.

[0025] The present invention also provides an air quality meteorological condition contribution rate assessment system, including a data acquisition module, a data processing module, a model building module, a contribution rate calculation module and a user interaction module:

[0026] The data acquisition module collects meteorological data, and divides the meteorological data into two types: air pollutants and meteorological elements, and stores them in the database;

[0027] The data processing module constructs pollutant sequences and meteorological sequence sets based on meteorological data and performs mean filtering operations;

[0028] The model building module uses a stepwise selection regression algorithm to establish the optimal meteorological model based on the pollutant sequence and meteorological sequence set;

[0029] The contribution rate calculation module calculates the contribution rate of meteorological elements according to the optimal meteorological model;

[0030] The user interaction module calls the contribution rate calculation module to calculate and present the contribution rate of the meteorological elements according to the air pollutants and meteorological elements selected by the user.

[0031] The beneficial results achieved by the present invention using the above scheme are as follows:

[0032] In view of the current air quality meteorological condition contribution rate assessment method, when using the traditional automatic variable selection algorithm, it is easy to produce overfitting, collinearity and other problems, and it is difficult to find the optimal regression model, which in turn affects the accuracy and reliability of the contribution rate assessment. The present invention adopts a stepwise selection regression algorithm to conduct a forward stepwise comparative assessment of the meteorological model, combining multiple regression model quality assessment criteria rather than a single indicator, including mean square error, Akaike information criterion and variance inflation factor, to solve the overfitting, underfitting and multicollinearity problems of the automatic variable selection algorithm. In addition, unlike simple forward and backward selection, the algorithm used in the present invention further performs a significance check with a meteorological model with the optimal evaluation criteria to ensure that the meteorological model has statistical significance and better predictive ability, thereby improving the stability and reliability of the meteorological model. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flow chart of a method for evaluating contribution rate of meteorological conditions to air quality proposed by the present invention;

[0034] Figure 2 This is a flowchart of step S3.

[0035] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0037] Example 1, see Figure 1 , Figure 1 This is a flow chart of a method for evaluating the contribution rate of meteorological conditions to air quality proposed by the present invention. In this embodiment, the method comprises the following steps:

[0038] S1: Obtain data on air pollutants and meteorological elements, and construct pollutant series and meteorological series sets;

[0039] S2: Mean filtering of pollutant series and meteorological series sets;

[0040] S3: Establish the optimal meteorological model through the stepwise selection regression algorithm;

[0041] S4: Calculate the contribution rate of meteorological elements according to the optimal meteorological model.

[0042] Example 2: Based on the above example, this example realizes the construction of pollutant sequence and meteorological sequence set. The specific contents are as follows:

[0043] In this embodiment, the data of air pollutants and meteorological elements are the data observed by the meteorological station in Hohhot, Inner Mongolia from 2018 to 2024, where air pollutants include PM 10 、PM 2.5 、NO 2 , O 3 and SO 2 The daily concentration of meteorological elements is the daily maximum temperature, daily minimum temperature, daily average temperature, wind force, maximum wind direction, average wind speed, average daily precipitation, average daily air pressure, average daily relative humidity, UV index and sunshine duration.

[0044] Constructing pollutant sequences: Based on PM 10 、PM 2.5 、NO 2 , O 3 and SO 2 The data are used to construct PM 10 Sequence, PM 2.5 Sequence, NO 2 Sequence, O 3 Sequence and SO2 sequence.

[0045] Construct meteorological sequence: Based on the data of daily maximum temperature, daily minimum temperature, daily average temperature, wind force, maximum wind direction, average wind speed, daily precipitation, daily average air pressure, daily average relative humidity, UV index and sunshine duration, construct the maximum temperature sequence, daily minimum temperature sequence, daily average temperature sequence, wind force sequence, maximum wind direction sequence, average wind speed sequence, daily precipitation sequence, daily average air pressure sequence, daily average relative humidity sequence, UV index sequence and sunshine duration sequence on a daily basis, and the above sequences constitute the meteorological sequence set.

[0046] Embodiment 3: Based on the above embodiment, this embodiment implements mean filtering, and the specific contents are as follows:

[0047] For each pollutant sequence and meteorological sequence, the impact of weather trends is reduced by using mean filtering. Specifically, a window with a step size of 7 is slid across the sequence, and the mean within the window is calculated, and the mean is subtracted from each position in the window.

[0048] Example 4, see Figure 2 , Figure 2 This is a flow chart of step S3. Based on the above embodiment, this embodiment realizes the establishment of the optimal meteorological model through a stepwise selection regression algorithm. The specific contents are as follows:

[0049] S31: Set initialization: Take each sequence in the meteorological sequence set as an independent variable and the pollutant sequence as the dependent variable, establish an empty model that does not contain any independent variables, and establish a meteorological model set Add the empty model to the meteorological model set and create a deletion set ,at this time is an empty set;

[0050] In this embodiment, PM 10 The sequence was used as the dependent variable, and the maximum temperature sequence, the daily minimum temperature sequence, the daily average temperature sequence, the wind speed sequence, the maximum wind direction sequence, the average wind speed sequence, the daily average precipitation sequence, the daily average air pressure sequence, the daily average relative humidity sequence, the ultraviolet index sequence and the sunshine duration sequence were used as independent variables to establish the relationship between meteorological elements and PM 10 Linear relationship meteorological model.

[0051] S32: Building a new weather model: Traversal Each weather model in , add a The independent variables not included are used to establish a linear regression equation to form a new meteorological model ,Will Add to ;

[0052] S33: Screening is performed according to mean square error, Akaike information criterion and variance inflation factor. The specific steps are as follows:

[0053] S331: Calculation The mean square error and The mean square error , the specific calculation formula of mean square error is as follows:

[0054] ;

[0055] in, represents the mean square error, represents the number of samples, express Time nodes, The weather model indicates The calculated value of the time node, express The actual value of the time node.

[0056] S332: Calculation Akaike Information Criterion and Akaike Information Criterion , the specific calculation formula of Akaike information criterion is as follows:

[0057] ;

[0058] in, represents the Akaike information criterion, represents the maximum likelihood function of the meteorological model, Indicates the number of meteorological model parameters.

[0059] S333: Define variance inflation factor threshold ,calculate Variance inflation factor In this embodiment Set to 5, the specific formula of the variance inflation factor is as follows:

[0060] ;

[0061] in, Indicates the added independent variable Negative correlation coefficients of regression analysis were performed on the remaining independent variables.

[0062] S334: and and Weather model Join ,Will or or Weather model Join ;

[0063] S34: Repeat S32 until the traversal is completed, remove All of the The elements that produce the intersection;

[0064] S35: Finding the optimal mean square error meteorological model and the optimal Akaike information criterion meteorological model: Traversal , the meteorological model with the smallest mean square error is taken as the optimal mean square error meteorological model , the meteorological model with the smallest Akaike information criterion is taken as the optimal Akaike information criterion meteorological model ;

[0065] S36: Significance check: Setting the mean square error significance threshold and Akaike information criterion significance threshold , traverse ,for and , and use the t-test method to calculate each meteorological model Significant difference in mean square error The significant difference between the Akaike information criterion and , retain and and of All less than and All less than The meteorological model, from Delete the and of Are greater than or equal to or Are greater than or equal to The meteorological model in this embodiment and All are set to 0.05;

[0066] S37: Obtain the optimal meteorological model: The mean square error and Akaike information criterion of each meteorological model in the forward processing are processed. The forward processing of the mean square error is specifically to subtract the mean square error of the current meteorological model from the maximum mean square error, and the forward processing of the Akaike information criterion is specifically to subtract the Akaike information criterion of the current meteorological model from the maximum Akaike information criterion to obtain the forward matrix ,in for The number of meteorological models in is the number of evaluation indicators. In this embodiment .

[0067] Then, the superior and inferior solution distance method is used to score each meteorological model. Specifically, the forward matrix is ​​normalized to obtain ,definition The maximum set of elements in each column , The minimum set of elements in each column , calculate the maximum distance Distance from minimum value , the specific formula is:

[0068] ;

[0069] ;

[0070] in, Indicates evaluation indicators, Indicates The maximum value of the evaluation index, Indicates The minimum value of the evaluation index, express Middle Weather model The value of an evaluation indicator.

[0071] Calculate the unnormalized score for each weather model , the specific formula is as follows:

[0072] ;

[0073] in, Indicates A weather model.

[0074] Calculate the normalized score for each weather model , the specific formula is as follows:

[0075] ;

[0076] The meteorological model with the highest normalized score is the optimal meteorological model.

[0077] Embodiment 5: Based on the above embodiment, this embodiment realizes the calculation of the contribution rate of meteorological elements. The specific contents are as follows:

[0078] The weight of each meteorological element in the optimal meteorological model is divided by the sum of all weights of the optimal meteorological model. The specific formula is as follows:

[0079] ;

[0080] in Indicates the category The contribution rate of meteorological elements, Indicates the category The weight of meteorological elements in the optimal meteorological model, represents the number of independent variables in the optimal meteorological model, Indicates The weight of the independent variable.

[0081] Embodiment 6: Based on the above embodiments, this embodiment implements an air quality meteorological condition contribution rate assessment system, which specifically includes: a data acquisition module, a data processing module, a model building module, a contribution rate calculation module and a user interaction module:

[0082] The data acquisition module collects meteorological data, and divides the meteorological data into two types: air pollutants and meteorological elements, and stores them in the database;

[0083] The data processing module constructs pollutant sequences and meteorological sequence sets based on meteorological data and performs mean filtering operations;

[0084] The model building module uses a stepwise selection regression algorithm to establish the optimal meteorological model based on the pollutant sequence and meteorological sequence set;

[0085] The contribution rate calculation module calculates the contribution rate of meteorological elements according to the optimal meteorological model;

[0086] The user interaction module calls the contribution rate calculation module to calculate and present the contribution rate of the meteorological elements according to the air pollutants and meteorological elements selected by the user.

Claims

1. A method for evaluating the contribution rate of meteorological conditions to air quality, characterized by: The following steps are involved: S1: Obtain data on air pollutants and meteorological elements, and construct pollutant series and meteorological series sets; S2: Mean filtering of pollutant series and meteorological series sets; S3: Establish the optimal meteorological model through the stepwise selection regression algorithm; S4: Calculate the contribution rate of meteorological elements according to the optimal meteorological model; The S3 specifically includes the following steps: S31: Set initialization: take each sequence in the meteorological sequence set as an independent variable and the pollutant sequence as the dependent variable, establish an empty model that does not contain any independent variables, establish a meteorological model set and add the empty model to the meteorological model set, establish a deletion set, and the deletion set is an empty set at this time; S32: construct a new meteorological model: traverse the meteorological model set, the traversed meteorological model is the current meteorological model, add an independent variable that is not included in the current meteorological model and establish a linear regression equation to form a new meteorological model, and add the new meteorological model to the meteorological model set; S33: Screening based on mean square error, Akaike information criterion and variance inflation factor; S34: Repeat S32 until the traversal is completed, and remove all elements in the meteorological model set that intersect with the deletion set; S35: Finding the optimal mean square error meteorological model and the optimal Akaike information criterion meteorological model: traversing the meteorological model set, taking the meteorological model with the smallest mean square error as the optimal mean square error meteorological model, and taking the meteorological model with the smallest Akaike information criterion as the optimal Akaike information criterion meteorological model; S36: significance check: set the mean square error significance threshold and the Akaike information criterion significance threshold, traverse the meteorological model set, and for the optimal mean square error meteorological model and the optimal Akaike information criterion meteorological model, use the t-check method to calculate the significance difference of the mean square error and the significance difference of the Akaike information criterion of the current meteorological model respectively. If the significance difference of the mean square error is less than the mean square error significance threshold and the significance difference of the Akaike information criterion is less than the Akaike information criterion significance threshold, then retain the current meteorological model, otherwise delete the current meteorological model from the meteorological model set; S37: Obtaining the optimal meteorological model: performing forward processing on the mean square error and Akaike information criterion of each meteorological model in the meteorological model set, and then using the superior and inferior solution distance method to score each meteorological model, and the meteorological model with the highest score is the optimal meteorological model; The S33 specifically includes the following steps: S331: Calculate the mean square error of the new meteorological model and the mean square error of the current meteorological model; S332: Calculate the Akaike information criterion of the new meteorological model and the Akaike information criterion of the current meteorological model; S333: defining a variance inflation factor threshold and calculating the variance inflation factor of the new meteorological model; S334: If the mean square error of the new meteorological model is smaller than the mean square error of the current meteorological model, and the Akaike information criterion of the new meteorological model is smaller than the Akaike information criterion of the current meteorological model, and the variance inflation factor of the new meteorological model is smaller than the variance inflation factor threshold, then the new meteorological model is considered to be superior to the current meteorological model and the current meteorological model is added to the deletion set, otherwise the new meteorological model is added to the deletion set.

2. An air quality meteorological condition contribution rate assessment system, used to implement an air quality meteorological condition contribution rate assessment method as claimed in claim 1, characterized in that: It includes data collection module, data processing module, model building module, contribution rate calculation module and user interaction module; The data acquisition module collects meteorological data, and divides the meteorological data into two types: air pollutants and meteorological elements, and stores them in a database; The data processing module constructs a pollutant sequence and a meteorological sequence set according to the meteorological data, and performs a mean filtering operation; The model building module uses a stepwise selection regression algorithm to establish an optimal meteorological model based on the pollutant sequence and the meteorological sequence set; The contribution rate calculation module calculates the contribution rate of meteorological elements according to the optimal meteorological model; The user interaction module calls the contribution rate calculation module to calculate and present the contribution rate of the meteorological elements according to the air pollutants and meteorological elements selected by the user.

Images