Air quality evaluation method and system facing knowledge clustering and evidence theory

Abstract

The application provides an air quality evaluation method and system facing knowledge clustering and evidence theory, and realizes comprehensive evaluation of air quality. The method first removes isolated points as outliers through knowledge clustering; then calculates six air quality sub-index vector means according to elements contained in each category; finally, the DS evidence theory is applied to fuse the air quality sub-index vector means, so that the comprehensive evaluation result of air quality of each category is obtained. The evaluation method can effectively analyze air quality and obtain accurate evaluation results, and plays a reference and support role for air quality evaluation and decision-making.

Description

Technical Field

[0001] This invention relates to the field of air quality management technology, particularly to the field of real-time air quality monitoring, and more specifically to an air quality evaluation method and system based on knowledge clustering and evidence theory. Background Technology

[0002] In existing technologies, air quality management involves constructing varying numbers of air quality monitoring stations distributed across different regions to assess urban air quality. The national standard evaluation method used is the Air Quality Index (AQI). However, existing AQI-based evaluation methods do not consider the impact of abnormal data, and the AQI value is the highest value among the six pollutant sub-indices, rather than a comprehensive evaluation method based on multiple pollutants. Urban air quality assessment methods based on the AQI standard select the highest sub-index among the six pollutants as the AQI, without conducting a comprehensive evaluation of multiple pollutants.

[0003] Comprehensive evaluation of urban air quality based on urban air quality monitoring stations falls under the category of urban air quality monitoring data fusion. In the field of ecological and environmental monitoring, scholars have conducted extensive research on data fusion methods. This paper proposes a DS evidence decision-making method for water quality monitoring by fusing data collected from water quality monitoring sensors and making decisions based on the DS evidence structure. To address the low accuracy of multi-sensor soil monitoring data fusion, an improved batch estimation adaptive weighted fusion algorithm is proposed. Multi-sensor data fusion technology is used to study the early warning mechanism for environmental monitoring at construction sites. Addressing the high accuracy requirements of sensor data in water quality monitoring, a data fusion algorithm based on the support degree function is proposed. The paper reviews the main techniques and methods of multi-source data fusion, including weighted average method, Kalman filtering, DS evidence theory, and rough set theory. It also categorizes and summarizes the research status and application prospects of multi-source data fusion technologies such as ground data fusion, ground-air data fusion, and multi-source remote sensing data fusion in the field of water quality monitoring and evaluation, and proposes relevant countermeasures and suggestions. To improve the accuracy of VOCs concentration prediction in the region, a three-level data fusion model for VOCs concentration monitoring information is constructed using data fusion theory. Research on air quality assessment methods mainly includes principal component analysis, fuzzy comprehensive evaluation, and DS evidence theory. However, these methods themselves have significant errors in comparing the assessment results with actual air quality. Furthermore, when multiple pollutants are present, these methods cannot provide accurate air quality assessments. Therefore, to more scientifically evaluate air quality, it is necessary to research comprehensive assessment methods that consider multiple pollutants. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides an air quality assessment method that uses real-time monitoring data from air quality monitoring stations as the data source, takes a comprehensive evaluation of urban air quality as the fusion objective, and is based on knowledge clustering and DS evidence theory for data fusion.

[0005] The first aspect of this invention discloses an air quality assessment method based on knowledge clustering and evidence theory, comprising:

[0006] Step S1: Input object , Let i be the air quality index vector monitored at the i-th monitoring point, where i is an integer greater than or equal to 1;

[0007] Step S2: Construct the similarity matrix between the input objects mentioned in step S1. The air quality index of the pollutants and The greater the similarity between them, The larger, and ;

[0008] Euclidean distance similarity was chosen: for ,as follows:

[0009] (1)

[0010] In the formula: function f is the Euclidean distance calculation formula;

[0011] Step S3: Calculate the air quality sub-index for the i-th monitoring point. Calculate threshold Threshold The calculation formula is as follows:

[0012] (2)

[0013] In the formula: ;

[0014] Step S4: Based on the threshold obtained in step S3 For the air quality index vector of the i-th monitoring point Calculate the initial equivalence relation ;based on The initial clustering pattern is obtained, denoted as CS;

[0015] Step S5: Based on the air quality sub-index of each pollutant initial equivalence relation , build The equivalence relation matrix R, where the th The row represents the i-th monitoring point, and the first column is a set. The second column is a set. ;

[0016] Step S6: Calculate the inter-class indistinguishability matrix After obtaining the initial clustering pattern CS, let... Define the indistinguishability between classes as follows:

[0017] (3)

[0018] In the formula: and Classes and class The number of objects contained in;

[0019] Step S7: If class and The indistinguishability between them is greater than or equal to the threshold. ,Right now If ≥θ, then merge the classes. and Otherwise, return to step S6;

[0020] Step S8: For the clustering results obtained in step S7, calculate the mean concentration of the six pollutants in each category, and then calculate the membership degree of each pollutant according to the following membership function;

[0021] when hour,

[0022] (4)

[0023] when hour,

[0024] (5)

[0025] when hour,

[0026] (6)

[0027] In the formula: Let be the air quality index vector monitored at the i-th monitoring point. For the first The first pollutant Level standard, For the first The first pollutant Membership degree of the level standard;

[0028] Step S9: Normalize the membership degrees to obtain the basic probability assignment of the DS evidence theory. ;

[0029] ;

[0030] In the formula: N is an integer greater than or equal to 1; Indicates the first Evidence allocated to Trust level; This is the xth proposition based on the DS theory;

[0031] Step S10: According to the synthesis rules, the basic probability assignment m of the DS evidence theory obtained in step S9 is subjected to multiple evidence fusion processing to obtain the probability set;

[0032] Step S11: Perform a comprehensive air quality evaluation using the probability set values ​​obtained in step S10. The air quality level corresponding to the highest probability value is the comprehensive evaluation result.

[0033] According to the method of the first aspect of the present invention, step S3 specifically involves: calculating the air quality sub-index of the i-th monitoring point using the difference method. threshold Each object threshold The calculation method is as follows: [The similarity matrix is ​​then used in the calculation.] Sort the elements in each row from largest to smallest to obtain the sorted matrix. Then No. Find the two similarity values ​​with the largest difference in the row, and the larger similarity value is the air quality sub-index vector of the i-th monitoring point. Corresponding threshold .

[0034] According to the method of the first aspect of the present invention, in step S4, the threshold obtained in step S3 is... For the air quality index vector of the i-th monitoring point Calculate the initial equivalence relation ,based on The initial clustering pattern, denoted as CS, is obtained as the matrix obtained in step S3. In the middle, when The value is greater than When, the air quality sub-index of the i-th monitoring point is... and the air quality sub-index of the j-th monitoring point It is indistinguishable; define the air quality index vector of the i-th pollutant. initial equivalence relation for ,in and for: , for The supplement to .

[0035] According to the method of the first aspect of the present invention, based on the air quality index vector of the i-th monitoring point initial equivalence relation , build The equivalence relation matrix R, where the th The row represents the air quality index vector of the i-th monitoring point. The first column is the set P = The second column is the set Q= The specific process of obtaining the initial clustering pattern CS is as follows:

[0036] Step S41: Input the initial equivalence relation ;

[0037] Step S42: i=i+1,

[0038] Step S43: When i <= n, P = {P ∩ {P i}}∪{Q∩{P i}};Q={P∩{UP i}}∪{Q∩{UP i Return to step S42; when i>n, output CS=P∪Q and exit.

[0039] According to the method of the first aspect of the present invention, in step S8, the pollutant level standard is: {excellent, good, lightly polluted, moderately polluted, heavily polluted}.

[0040] According to the method of the first aspect of the present invention, step S10 specifically comprises:

[0041] Choose the classic Dempster crafting rules, which are as follows:

[0042] ;

[0043] In the formula: ,in This is the xth proposition based on the DS theory.

[0044] According to the method of the first aspect of the present invention, in step S7: the merging condition is set to when and When the DS evidence theory evaluation results are consistent, then they are merged. and .

[0045] A second aspect of the present invention provides a system for an air quality assessment method based on knowledge clustering and evidence theory, the system being used to implement the air quality assessment method based on knowledge clustering and evidence theory of the first aspect.

[0046] The third aspect of the present invention discloses an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the program in the memory to implement the air quality assessment method based on knowledge clustering and evidence theory of the first aspect.

[0047] The fourth aspect of the present invention discloses a computer-readable storage medium storing a computer program implementing the air quality assessment method based on knowledge clustering and evidence theory of the first aspect.

[0048] In summary, the proposed solution of this invention has the following technical effects: This invention provides an air quality assessment method and system based on knowledge clustering and evidence theory. By combining knowledge-based clustering methods with a comprehensive evaluation approach based on evidence theory, it effectively analyzes the air data acquired by monitoring stations to obtain the current air quality assessment results. These results are more accurate and detailed than those of traditional assessment methods, providing a reference for air quality assessment and related decision-making. Attached Figure Description

[0049] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0050] Figure 1 This is the calculation process for obtaining the initial clustering pattern in this invention;

[0051] Figure 2 This is a flowchart of the comprehensive air quality evaluation process of this invention;

[0052] Figure 3 This is a comparative chart of the comprehensive air quality evaluation of the present invention;

[0053] Figure 4 This is the overall processing flow of the air quality assessment method provided by the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0055] Figure 4 A system flowchart for an air quality assessment method based on knowledge clustering and evidence theory, provided by this invention, includes:

[0056] Step S1: Input object , For the first The air quality index vector of each monitoring point, where i is an integer greater than or equal to 1;

[0057] Step S2: Construct a similarity matrix between objects. Air quality index vector of pollutants and The greater the similarity, The larger, and ;

[0058] Euclidean distance similarity was chosen: for ,as follows:

[0059] (1)

[0060] In the formula: function f is the Euclidean distance calculation formula, and the air quality index vector of pollutants. and The greater the similarity, The larger, and ;

[0061] Step S3: For the first Air quality index vector at each monitoring point Calculate threshold Threshold The calculation formula is as follows:

[0062] (2)

[0063] In the formula: ;

[0064] Step S4: For the first Air quality index vector at each monitoring point Calculate the initial equivalence relation ;based on The initial clustering pattern is obtained, denoted as CS;

[0065] Step S5: Based on the initial equivalence relation of each object , build The equivalence relation matrix R, where the th The line represents the first Air quality index vector at each monitoring point The first column is a set. The second column is a set. ;

[0066] Step S6: Calculate the inter-class indistinguishability matrix After obtaining the initial clustering pattern CS, let... Define the indistinguishability between classes as follows:

[0067] (3)

[0068] In the formula: and Classes and class The number of objects contained therein;

[0069] Step S7: If class and The indistinguishability between them is greater than or equal to the threshold. ,Right now Then merge class and Otherwise, return to step S6;

[0070] Step S8: For the clustering results obtained in step S7, calculate the mean concentration of the six pollutants in each category, and then calculate the membership degree of each pollutant according to the following membership function;

[0071] when hour,

[0072] (4)

[0073] when hour,

[0074] (5)

[0075] when hour,

[0076] (6)

[0077] In the formula: Let be the air quality index vector monitored at the i-th monitoring point. For the first The first pollutant Level standard, For the first The pollutant Membership degree of the level standard;

[0078] Step S9: Normalize the membership degrees to obtain the basic probability assignment of the DS evidence theory. ;

[0079] ;

[0080] In the formula: N is an integer greater than or equal to 1; Indicates the first Evidence allocated to Trust level; This is a proposition based on DS theory;

[0081] Step S10: Perform fusion processing of multiple pieces of evidence according to the synthesis rules;

[0082] Step S11: Conduct a comprehensive air quality evaluation based on the obtained probability values. The air quality level corresponding to the highest probability value is the comprehensive evaluation result.

[0083] Clustering algorithms are generally classified into hierarchical clustering, partitioning clustering, density-based and grid-based methods, etc. Existing technologies have proposed knowledge-oriented clustering methods. The knowledge representation system S is represented as a quadruple. . It is a finite set of objects, called the universe of discourse; It is a finite set of all attributes; It is the value range set of the attribute. It is an attribute The range of values; It is an information function. .

[0084] The following knowledge-oriented clustering algorithm is adopted:

[0085] 1) Input object .

[0086] 2) The similarity matrix between the constructed objects is: The most widely used similarity measurement methods are Euclidean distance similarity and cosine similarity. Euclidean distance similarity is chosen: for... The similarity is defined as shown in equation (1).

[0087] (1)

[0088] In the formula: object and The greater the similarity, The larger, and .

[0089] 3) For the first Air quality index vector at each monitoring point Calculate threshold The selection interpolation method is used to calculate the value of each object. threshold This method first uses the similarity matrix Sort the elements in each row from largest to smallest to obtain the sorted matrix. Then No. Find the two similarity scores with the largest difference in the row; the one with the larger similarity score is the object. threshold Threshold The calculation formula is shown in equation (2).

[0090] (2)

[0091] In the formula: .

[0092] 4) For the first Air quality index vector at each monitoring point Calculate the initial equivalence relation In the matrix In the middle, when The value is greater than At that time, the object is considered and object It is indistinguishable; defining an object initial equivalence relation for ,in and for: , for The supplement to .

[0093] 5) Based on The initial clustering pattern, denoted as CS, is obtained. This is based on the initial equivalence relations of each object. , build The equivalence relation matrix R, where the th Row Representation Object The first column is a set. The second column is a set. The algorithm flow for designing and calculating the initial clustering pattern CS is as follows: Figure 1 As shown.

[0094] 6) Calculate the inter-class indistinguishability matrix After obtaining the initial clustering pattern CS, the definition of inter-cluster indistinguishability is as follows: Define the indistinguishability between classes As shown in equation (3).

[0095] (3)

[0096] In the formula: and Classes and class The number of objects contained therein.

[0097] If class and The indistinguishability between them is greater than or equal to the threshold. ,Right now Then merge class and If the clustering is indistinguishable between clusters, return to step (6); otherwise, output the clustering results. Constructed inter-class indistinguishability matrix and the pre-set threshold Merge classes with inter-class similarity greater than or equal to a threshold, and recalculate the inter-class indistinguishability matrix until all inter-class indistinguishability is less than the threshold. until.

[0098] In step (7) of the knowledge-oriented clustering algorithm, when the class and The indistinguishability between them is greater than or equal to the threshold. ,Right now Then merge class and Due to the threshold The value needs to be preset manually and is greatly affected by human factors. Therefore, the condition for merging classes can be modified to: when the class and When the evaluation results of the DS evidence theory are consistent, then the merged category and .

[0099] The algorithm's performance is tested using a set of artificial datasets. The test dataset is shown in Table 1.

[0100] Table 1 Test Dataset

[0101] ;

[0102] The calculated similarity matrix is ​​shown in Table 2, where the underlined data represents the threshold. .

[0103] Table 2 Similarity Matrix

[0104] ;

[0105] The initial equivalence relations are shown in Table 3.

[0106] Table 3 Initial Equivalence Relations

[0107] ;

[0108] Initial clustering pattern The inter-class similarity matrix is ​​as follows:

[0109] ;

[0110] If a threshold is set If the inter-class similarity is less than 1, then the inter-class similarity is determined to be less than 1. Therefore, the initial clustering pattern CS of the test dataset is the final clustering result.

[0111] DS evidence theory is an effective method for handling fuzzy problems. Air quality assessment is an uncertain problem, and scholars have already conducted research on comprehensive air quality assessment based on DS evidence theory and its improved algorithms. The key to DS evidence theory is the identification frame, basic probability assignment (BPA), and synthesis formula, which mainly consists of the following five steps:

[0112] 1) Determine the identification framework. Based on the average concentrations of six pollutants obtained from knowledge-based clustering, this invention establishes an identification framework as: {Excellent, Good, Lightly Polluted, Moderately Polluted, and Heavily Polluted}, and establishes the ambient air quality levels and concentration limits for each pollutant as shown in Table 4.

[0113] Table 4 Concentration limits for various pollutants

[0114] ;

[0115] In Table 4: , , , , The unit of concentration is , The unit of concentration is .

[0116] 2) Calculate the membership function. The membership functions of the pollution factors are calculated as shown in equations (4), (5), and (6).

[0117] when hour,

[0118] (4)

[0119] when hour,

[0120] (5)

[0121] when hour,

[0122] (6)

[0123] In the formula: Let be the air quality index vector monitored at the i-th monitoring point. For the first The first pollutant Level standard, For the first The pollutant Membership degree of the level standard.

[0124] 3) Normalize the membership function to obtain the basic probability assignment of the DS evidence theory. .

[0125] ;

[0126] In the formula: , Indicates the first Evidence assigned to proposition Trust level.

[0127] 4) Multiple pieces of evidence are fused according to the synthesis rules. This invention selects the classic Dempster synthesis rules. Specifically, based on the data characteristics of this invention, the synthesis rules are as follows:

[0128] ;

[0129] In the formula: ;

[0130] 5) The air quality is comprehensively evaluated based on the obtained probability values, and the air quality level corresponding to the highest probability value is the comprehensive evaluation result.

[0131] The algorithm is tested using the dataset (6, 47, 0.5, 60, 120, 66) as an example to obtain the basic probability assignment matrix. as follows:

[0132] ;

[0133] The probability values ​​of the recognition frames obtained after Dempster synthesis are: The highest probability value corresponds to the level of "light pollution", which is consistent with the AQI evaluation results.

[0134] Using monitoring data from multiple air quality monitoring stations within an urban area at the same time as the data source, and based on a knowledge-oriented clustering algorithm, combined with DS evidence theory or fuzzy comprehensive evaluation method, a comprehensive air quality evaluation process is designed as follows: Figure 2 As shown. In Figure 2 The comprehensive evaluation process mainly consists of four steps: First, calculating the sub-indices of six pollutants according to the "Technical Regulations for Ambient Air Quality Index (AQI)"; second, performing a knowledge-oriented clustering algorithm based on the sub-indices, removing isolated points from the clustering results as they are considered outliers; third, calculating the average concentration of air quality monitoring data based on the clustering results; and fourth, applying the DS evidence theory to conduct a comprehensive air quality evaluation based on the average concentration of each cluster, and comparing and analyzing the results with the fuzzy comprehensive evaluation results.

[0135] An air quality comprehensive evaluation and analysis was conducted using hourly monitoring data from 35 monitoring stations in a certain city in 2022 as a sample, and inter-class similarity thresholds were set. A comprehensive air quality assessment was conducted using data from a randomly selected hour, and the results are shown in Table 5.

[0136] Table 5. Comprehensive Air Quality Assessment Results

[0137] ;

[0138] Table 5 shows that the air quality monitoring data are clustered into two categories: one containing 4 monitoring stations and the other containing 31 monitoring stations. Furthermore, the DS evidence theory... , , The four methods yielded consistent evaluation results for each category. The category with four monitoring stations received a comprehensive evaluation of "moderate pollution," while the category with 31 monitoring stations received a comprehensive evaluation of "good," and the AQI method also resulted in a "good" evaluation. This demonstrates that the method presented in this paper can automatically distinguish between high and low value areas, and its evaluation results are more refined and reasonable than those of the AQI method.

[0139] 264 data points were randomly selected from the hourly data throughout the year. The air quality levels were calculated using the AQI method and arranged from lowest to highest. The overall air quality assessment results are plotted as follows: Figure 3 As shown.

[0140] Depend on Figure 3 It can be seen that the changing trends of the DS evidence theory and the AQI method evaluation results are quite consistent; , The air quality comprehensive evaluation results of the two methods are completely consistent, and the trend of change is largely consistent with that of the AQI method evaluation results; The evaluation results of the method fluctuate greatly, with a significant amount of data deviating from the evaluation results of the AQI method.

[0141] Assuming the evaluation result of the AQI method is the actual value, and the comprehensive evaluation result of the method in this paper is the predicted value, we calculate Precision, Recall, and F1-score as a multi-classification problem. The calculation results are shown in Table 6.

[0142] Table 6. Precision, Recall, and F1-score values ​​for different methods.

[0143] ;

[0144] As shown in Table 6, the precision, recall, and F1-Score of the DS evidence theory are higher than those of the three fuzzy evaluation models.

[0145] The present invention also provides a system for an air quality assessment method based on knowledge clustering and evidence theory, the system being used to implement the first aspect of the air quality assessment method based on knowledge clustering and evidence theory.

[0146] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the program in the memory to implement the first aspect of the air quality assessment method based on knowledge clustering and evidence theory.

[0147] The present invention also provides a computer-readable storage medium storing a computer program implementing the air quality assessment method based on knowledge clustering and evidence theory of the first aspect.

[0148] In summary, the technical solution proposed in this invention has the following technical effects: The method provided by this invention can be used to construct a comprehensive environmental air quality assessment system. The system input has two parameters: one is the environmental air quality level and its corresponding concentration limits for various pollutants; the other is the inter-class similarity threshold. When the user sets the inter-class similarity threshold, inter-class merging is performed according to the threshold; when the user does not set the inter-class similarity threshold, inter-class merging is performed according to the evaluation results based on evidence theory. The comprehensive environmental air quality assessment results output by the system can serve as a supplement to the AQI assessment method, providing a certain reference and support for air quality assessment and decision-making.

[0149] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An air quality assessment method based on knowledge clustering and evidence theory, characterized in that, include: Step S1: Input object , Let i be the air quality index vector monitored at the i-th monitoring point, where i is an integer greater than or equal to 1; Step S2: Construct the similarity matrix between the input objects mentioned in step S1. The air quality sub-index of the monitoring point and The greater the similarity between them, The larger, and ; Euclidean distance similarity was chosen: for ,as follows: (1) In the formula: function f is the Euclidean distance calculation formula; Step S3: Calculate the air quality index vector for the i-th monitoring point. Calculate threshold Threshold The calculation formula is as follows: (2) In the formula: ; Step S4: Based on the threshold obtained in step S3 Air quality sub-index for the i-th monitoring point Calculate the initial equivalence relation ;based on The initial clustering pattern is obtained, denoted as CS; Step S5: Based on the air quality sub-index of each monitoring point initial equivalence relation , build The equivalence relation matrix R, where the th The row represents the i-th monitoring point, and the first column is a set. The second column is a set. ; Step S6: Calculate the inter-class indistinguishability matrix ; After obtaining the initial clustering pattern CS, let Define the indistinguishability between classes as follows: (3) In the formula: and Classes and class The number of objects contained in; Step S7: If class and The indistinguishability between them is greater than or equal to the threshold. ,Right now If ≥θ, then merge the classes. and Otherwise, return to step S6; Step S8: For the clustering results obtained in step S7, calculate the mean concentration of the six pollutants in each category, and then calculate the membership degree of each pollutant according to the following membership function; when hour, (4) when hour, (5) when hour, (6) In the formula: Let be the air quality index vector monitored at the i-th monitoring point. For the first The first pollutant Level standard, For the first The pollutant Membership degree of the level standard; Step S9: Normalize the membership degrees to obtain the basic probability assignment of the DS evidence theory. ; ； In the formula: N is an integer greater than or equal to 1; Indicates the first Evidence allocated to Trust level; This is the xth proposition based on the DS theory; Step S10: According to the synthesis rules, the basic probability assignment m of the DS evidence theory obtained in step S9 is subjected to multiple evidence fusion processing to obtain the probability set; Step S11: Perform a comprehensive air quality evaluation using the probability set values ​​obtained in step S10. The air quality level corresponding to the highest probability value is the comprehensive evaluation result.

2. The air quality assessment method based on knowledge clustering and evidence theory according to claim 1, characterized in that, Step S3 specifically involves: using the difference method to calculate the air quality sub-index for the i-th monitoring point. threshold The similarity matrix Sort the elements in each row from largest to smallest to obtain the sorted matrix. Then No. Find the two similarity scores with the largest differences, and the larger similarity score is the air quality sub-index of the i-th monitoring point. threshold .

3. The air quality assessment method based on knowledge clustering and evidence theory according to claim 2, characterized in that, In step S4, the threshold obtained in step S3 is used as a basis. Air quality sub-index for the i-th monitoring point Calculate the initial equivalence relation ,based on The initial clustering pattern, denoted as CS, is obtained as the matrix obtained in step S3. In the middle, when The value is greater than When, the air quality sub-index vector of the i-th monitoring point is... and the air quality index vector of the j-th monitoring point It is indistinguishable, and the air quality index vector of the i-th monitoring point is defined. initial equivalence relation for ,in and for: , for The supplement to .

4. The air quality assessment method based on knowledge clustering and evidence theory according to claim 3, characterized in that, Based on the air quality sub-index vector of the i-th monitoring point initial equivalence relation , build The equivalence relation matrix R, where the th The row represents the air quality index vector of the i-th monitoring point. The first column is the set P = The second column is the set Q= The specific process of obtaining the initial clustering pattern CS is as follows: Step S41: Input the initial equivalence relation ; Step S42: i=i+1, Step S43: When i <= n, P = {P ∩ {P i }}∪{Q∩{P i }};Q={P∩{UP i }}∪{Q∩{UP i Return to step S42; when i>n, output CS=P∪Q and exit.

5. The air quality assessment method based on knowledge clustering and evidence theory according to claim 1, characterized in that, In step S8, the pollutant level standards are: {Excellent, Good, Lightly Polluted, Moderately Polluted, and Heavily Polluted}.

6. The air quality assessment method based on knowledge clustering and evidence theory according to claim 5, characterized in that, Step S10 is as follows: Choose the classic Dempster crafting rules, which are as follows: ； In the formula: ,in This is the xth proposition based on the DS theory.

7. The air quality assessment method based on knowledge clustering and evidence theory according to claim 1, characterized in that, In step S7: the merging condition is set to when and When the DS evidence theory evaluation results are consistent, then they are merged. and .

8. A system for air quality assessment based on knowledge clustering and evidence theory, characterized in that, The system is used to implement the air quality assessment method based on knowledge clustering and evidence theory as described in any one of claims 1-7.

9. An electronic device, characterized in that: The system includes a memory and a processor. The memory stores a computer program, and the processor executes the program in the memory to implement the air quality assessment method based on knowledge clustering and evidence theory as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that implements the air quality assessment method based on knowledge clustering and evidence theory as described in any one of claims 1-7.

Images