CN107067727A - A kind of road traffic service level evaluation method based on fuzzy KNN characteristic matchings - Google Patents

Abstract

A road traffic service level evaluation method based on fuzzy KNN feature matching, which establishes a road traffic feature reference sequence and obtains road traffic benchmark data in different modes; extracts road traffic training data, and obtains road traffic difference data based on the benchmark data. Threshold processing to obtain the characteristics of the training data; fuzzy processing of the road traffic service level, combined with the characteristics of the training data, to complete the construction of the knowledge base; extract the road traffic test data; through the KNN feature matching algorithm, select the feature distance between the knowledge base and the test data The nearest k features, and calculate the corresponding fuzzy service level; add up the k groups of fuzzy service levels to get the service level corresponding to the maximum probability, which is the service level to which the current feature belongs, and complete the road traffic operation status Evaluation. The invention is simple to realize, does not need to perform a large amount of complex data calculations, and can effectively improve the processing speed.

Description

A Road Traffic Service Level Evaluation Method Based on Fuzzy KNN Feature Matching

technical field

The invention belongs to the field of road traffic service level evaluation, relates to the analysis and evaluation of road traffic data, and is a method for road traffic service level evaluation.

Background technique

With the development of society and economy, the number of cars keeps increasing, and road traffic problems become more serious. It is necessary to correctly evaluate the level of road traffic service in order to deal with traffic problems before they occur, and avoid traffic jams and accidents. The occurrence of such events provides a basis for the traffic management department to make decisions.

At present, most researches on the evaluation of road traffic service level focus on traffic congestion. The more mature evaluation indicators of traffic service level include connectivity, travel time reliability and road network capacity reliability. Other studies on traffic service level evaluation include traffic congestion evaluation weighted model, road network overall adaptability analysis method, space saturation index, road network dynamic traffic flow state estimation, etc. The above studies are mainly carried out at the road network level, without considering the traffic service level of a single road section, and the implementation process is relatively complicated.

Contents of the invention

In order to overcome the shortcomings of existing road traffic service level evaluation methods, the present invention provides a road traffic service level evaluation method based on a simplified algorithm and fuzzy KNN feature matching.

The technical solution adopted by the present invention to solve its technical problems is:

A road traffic service level evaluation method based on fuzzy KNN feature matching, comprising the following steps:

1) Establish a road traffic feature reference sequence to obtain road traffic benchmark data in different modes;

2) Extract road traffic training data, obtain road traffic difference data based on road traffic reference data in the same mode, and obtain the characteristics of the training data through thresholding processing;

3) Fuzzify the existing road traffic service level, and combine the characteristics of the training data to complete the construction of the knowledge base;

4) Extract road traffic test data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the test data through thresholding processing;

5) Through the KNN feature matching algorithm, select the k features closest to the test data features in the knowledge base, and obtain the corresponding fuzzy service level;

6) Add up the fuzzified service levels of k groups to obtain the service level corresponding to the maximum probability, which is the service level to which the current feature belongs, and complete the road traffic service level evaluation.

Further, in said step 1), a reference sequence of road traffic characteristics is established to obtain road traffic reference data in different modes, including the following steps:

1.1) Design the structure of the road traffic feature reference sequence

Assuming that the collection cycle of road traffic status data is Δt, the time format of the road traffic information template is shown in Figure 1.

Table 1 and Table 2 show the table format of the road traffic feature reference sequence.

Table 1. Road traffic feature reference sequence information table

Table 2. Road traffic feature reference sequence description table

1.2) Through relevant data preprocessing, establish a reference sequence of road traffic operation characteristics

The division of the road traffic operation mode is divided into two levels: the road network layer and the road section layer. The division logo of the traffic operation mode of the road network layer is set to divide the traffic operation mode of the road into the g seed mode, and the road section layer The division mark of the traffic operation mode divides the traffic operation mode of the road into h sub-modes, then the traffic operation mode of the road is divided into g×h types in total, recorded as the set _Mode = {M ₁₁ , M ₁₂ , ... , M _gh }, where the values of g and h are determined according to the division identification of the selected traffic operation mode;

Obtain representative road traffic state data of the target road section, perform data preprocessing, and input the road traffic state data after data preprocessing into the road traffic feature reference sequence, thereby establishing the road traffic feature reference sequence.

Further, in the step 2), the road traffic training data is extracted, based on the road traffic reference data under the same mode, the road traffic difference data is obtained, and through thresholding, the characteristics of the training data are obtained, and its general expression is as follows :

Among them, M _gh is the mode; Δt is the collection cycle of road traffic state data; (m*Δt) is the mth road traffic state data collection cycle, 0≤m≤N, and N represents the number of traffic information collected every day ; Indicates the training data at (m*Δt) time; Indicates the benchmark data at (m*Δt) time; S(m*Δt) represents the difference data between training data and benchmark data at (m*Δt) time; Represents the difference data between the training data and the benchmark data during the period from Δt to (m*Δt) under the mode M _gh ; Indicates the threshold; Represents the difference data after threshold processing in the period from Δt to (m*Δt) under the mode M _gh , which is the feature of the training data; Indicates the mapping rule between traffic state data and features.

Further, in said step 3), through the fuzzy division of existing road traffic data and road traffic service level, in combination with the characteristics of the training data, the construction of the knowledge base is completed, and its general expression is as follows:

Among them, φ represents the mapping relationship between the current road traffic state and the fuzzy traffic service level, and Los _m represents the fuzzy road traffic service level.

Combining (4) and (5), the relationship between the traffic operation state and the characteristics can be obtained:

Los _m ＝ω(he _m ) (6)

Among them, ω represents the mapping rule between the fuzzy service level and the traffic operation characteristics, so as to complete the construction of the knowledge base.

Further, in the step 4), the road traffic test data is extracted, based on the road traffic benchmark data under the same mode, the road traffic difference data is obtained, and the characteristics of the test data are obtained by thresholding, and its general expression is as follows :

in Indicates the road traffic test data at (m*Δt) time; MS(m*Δt) represents the difference data between test data and benchmark data at (m*Δt) time; Represents the difference data between the test data and the benchmark data in the time period from Δt to (m*Δt); Represents the difference data after threshold processing, which is the feature of the test data.

Further, in said step 5), by KNN feature matching algorithm, select the k features closest to the test data feature distance in the knowledge base, and obtain the corresponding fuzzy service level, the process is as follows:

5.1) Calculate the distance between the features of the training data and the features of the test data

dist(m)＝||The _Mgh (m)-he _Mgh (m)|| (10)

D _N (m)＝[dist ₁ (m)dist ₂ (m)...dist _N (m)] (11)

Among them, |||| means to obtain the Euclidean distance, dist(m) means the distance between the features of the test data and the features in the knowledge base during the period from Δt to (m*Δt), N means the number of groups of training data, D _N (m) Represents the set of feature distances from the features of N sets of test data to N sets of training data within the period from Δt to (m*Δt).

5.2) Find the features corresponding to the k closest distances, let the k features be s ₁ , s ₂ , ... s _k , according to the mapping relationship between the features and the fuzzy service level, then there is

(L ₁ , L ₂ , ... L _k ) = ω(s ₁ , s ₂ , ... s _k ) (12)

Among them, L ₁ , L ₂ , ... L _k represent the fuzzy service levels corresponding to s ₁ , s ₂ , ... s _k respectively. in, Respectively represent the probability of unimpeded, general congestion, and severe congestion under the mode M _gh , from Δt to (m*Δt).

In said step 6), add the service levels of k groups of fuzzification to obtain the service level corresponding to the maximum probability, which is the service level to which the current feature belongs, and complete the evaluation of the road traffic service level. The process is as follows:

Add the fuzzy service level probabilities corresponding to k features, then we have

Among them, Su _AK , Su _AK , and Su _AK respectively represent the probability of unimpeded traffic, general congestion, and severe congestion after integrating k features. Su _AK , Su _AK , the service level corresponding to the largest probability value among Su _AK is the service level to which the current feature belongs.

The technical idea of the present invention is: a road traffic service level evaluation method based on fuzzy KNN feature matching is proposed. The road traffic data of the same time period is extracted and divided into benchmark data, training data and test data. Perform thresholding processing on the difference data between the training data and the benchmark data to obtain the characteristics of the training data. The road traffic service level is fuzzified, combined with the current traffic operation status, and a knowledge base composed of training data characteristics and road traffic service level is constructed. Perform thresholding processing on the difference data between the test data and the benchmark data, obtain the characteristics of the test data, and calculate the Euclidean distance between the characteristics of the test data and the characteristics of the training data. Based on the KNN algorithm, select the fuzzy service level corresponding to the nearest k closest distances. Add up the fuzzy service levels corresponding to the k closest distances to obtain the service level corresponding to the maximum probability, which is the current service level, and realize the evaluation of road traffic service level.

This method only performs subtraction between data, obtains traffic operation characteristics, builds a knowledge base, and realizes road traffic service level evaluation through feature matching combined with fuzzy KNN algorithm. The method is simple to implement, does not need to perform a large number of complex data calculations, and can effectively improve the processing speed.

The beneficial effect of the present invention is mainly manifested in: by thresholding the difference data of the road traffic training data and the reference data of the same mode _Mgh , the characteristics of the training data are obtained, and the training is realized in combination with the fuzzy road traffic service level. The construction of the data knowledge base; obtain the characteristics of the test data in the same mode, and calculate the distance between the test data characteristics and the training data characteristics through the KNN algorithm. Select the fuzzy service levels corresponding to the smallest k group distances and add them up, and select the service level with the highest probability as the current road traffic service level to complete the evaluation of the road traffic service level.

Description of drawings

Fig. 1 is a schematic diagram of a time format of a road traffic information template.

Figure 2 is a schematic diagram of road traffic operation mode division.

Figure 3 is a flowchart of road traffic service level evaluation based on fuzzy KNN feature matching.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings.

Referring to Figures 1 to 3, a road traffic service level evaluation method based on fuzzy KNN feature matching includes the following steps:

1) Establish a reference sequence of road traffic characteristics to obtain road traffic reference data in different modes, including the following steps:

1.1) Design the structure of the road traffic feature reference sequence

Assuming that the collection cycle of road traffic status data is Δt, the time format of the road traffic information template is shown in Figure 1.

The table format of the road traffic feature reference sequence is shown in Table 1 and Table 2. Table 1 is the road traffic feature reference sequence information table, and Table 2 is the road traffic feature reference sequence description table.

Table 1

Table 2

1.2) Through relevant data preprocessing, establish a reference sequence of road traffic operation characteristics

The division of the road traffic operation mode is divided into two levels: the road network layer and the road section layer. The division logo of the traffic operation mode of the road network layer is set to divide the traffic operation mode of the road into the g seed mode, and the road section layer The division mark of the traffic operation mode divides the traffic operation mode of the road into h sub-modes, then the traffic operation mode of the road is divided into g×h types in total, recorded as the set _Mode = {M ₁₁ , M ₁₂ , ... , M _gh }, where the values of g and h are determined according to the division identification of the selected traffic operation mode;

Obtain representative road traffic state data of the target road section, perform data preprocessing, and input the road traffic state data after data preprocessing into the road traffic feature reference sequence, thereby establishing the road traffic feature reference sequence.

2) Extract road traffic training data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the training data after thresholding processing. The general expression is as follows:

Among them, M _gh is the mode; Δt is the collection cycle of road traffic state data; (m*Δt) is the mth road traffic state data collection cycle, 0≤m≤N, and N represents the number of traffic information collected every day ; Indicates the training data at (m*Δt) time and m*Δt time; Indicates the benchmark data at (m*Δt) time; S(m*Δt) represents the difference data between training data and benchmark data at (m*Δt) time; Represents the difference data between the training data and the benchmark data during the period from Δt to (m*Δt) under the mode M _gh ; Indicates the threshold; Represents the difference data after threshold processing in the period from Δt to (m*Δt) under the mode M _gh , which is the feature of the training data; Indicates the mapping rule between traffic state data and features.

3) Through the fuzzy division of the existing road traffic data and road traffic service level, combined with the characteristics of the training data, the construction of the knowledge base is completed, and its general expression is as follows:

Among them, φ represents the mapping relationship between the current road traffic state and the fuzzy traffic service level, and Los _m represents the fuzzy road traffic service level.

Combining (4) and (5), the relationship between the traffic operation state and the characteristics can be obtained:

Los _m ＝ω(he _m ) (6)

Among them, ω represents the mapping rule between the fuzzy service level and the traffic operation characteristics, so as to complete the construction of the knowledge base.

4) Extract road traffic test data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the test data through thresholding processing. The general expression is as follows:

in Indicates the road traffic test data at (m*Δt) time; MS(m*Δt) represents the difference data between test data and benchmark data at (m*Δt) time; Represents the difference data between the test data and the benchmark data in the time period from Δt to (m*Δt); Represents the difference data after threshold processing, which is the feature of the test data.

5) Through the KNN feature matching algorithm, select the k features closest to the test data features in the knowledge base, and obtain the corresponding fuzzy service level. The process is as follows:

5.1) Calculate the distance between the features of the training data and the features of the test data

dist(m)＝||The _Mgh (m)-he _Mgh (m)|| (10)

D _N (m)＝[dist ₁ (m)dist ₂ (m)...dist _N (m)] (11)

Among them, |||| means to obtain the Euclidean distance, dist(m) means the distance between the features of the test data and the features in the knowledge base during the period from Δt to (m*Δt), N means the number of groups of training data, D _N (m) Represents the set of feature distances from the features of N sets of test data to N sets of training data within the period from Δt to (m*Δt).

5.2) Find the features corresponding to the k closest distances, let the k features be s ₁ , s ₂ , ... s _k , according to the mapping relationship between the features and the fuzzy service level, then there is

(L ₁ , L ₂ , ... L _k ) = ω(s ₁ , s ₂ , ... s _k ) (12)

Among them, L ₁ , L ₂ , ... L _k represent the fuzzy service levels corresponding to s ₁ , s ₂ , ... s _k respectively. in, Respectively represent the probability of unimpeded, general congestion, and severe congestion under the mode M _gh , from Δt to (m*Δt).

6) Add up the fuzzy service levels of k groups to obtain the service level corresponding to the maximum probability, which is the service level to which the current feature belongs, and complete the road traffic service level evaluation. The process is as follows:

Add the fuzzy service level probabilities corresponding to k features, then we have

Among them, Su _AK , Su _AK , and Su _AK respectively represent the probability of unimpeded traffic, general congestion, and severe congestion after integrating k features. Su _AK , Su _AK , the service level corresponding to the largest probability value among Su _AK is the service level to which the current feature belongs.

Example: a road traffic service level evaluation method based on fuzzy KNN feature matching, comprising the following steps:

1) Establish a road traffic feature reference sequence to obtain road traffic benchmark data in different modes

Due to the similarity of road traffic flow at the same road section and corresponding time, the road traffic speed data of five expressways in Beijing on 7 working days in June 2011 (15, 16, 21, 22, 23, 24, 28) were selected Establish a road traffic feature reference sequence. The acquisition interval Δt of road traffic state data is 2min. The specific information of the road section is shown in Table 3.

Section ID section name HI2075a Central Conservatory of Music - Xibianmen Bridge HI3002b Deshengmen Bridge - Jishuitan Bridge HI7008a Baiqiao Street——Guangqumen Bridge HI7051a Moon Temple North Bridge - Moon Temple South Bridge HI7060b North Entrance of Dongzhimen Bridge—East of Xiaojie Bridge

table 3

The road traffic speed data on the 15th is used as the benchmark data set; the road traffic speed data of the same mode on the 16th, 21st, 22nd, 23rd, and 24th are used as training data to establish a knowledge base. The road traffic speed data on the 28th was used as an experimental data set to verify the algorithm.

2) Extract road traffic training data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the training data after thresholding processing. The general expression is as follows:

Among them, M _gh is the mode; Δt is the collection cycle of road traffic state data; (m*Δt) is the mth road traffic state data collection cycle, 0≤m≤N, and N represents the number of traffic information collected every day ; Indicates the training data at (m*Δt) time and m*Δt time; Indicates the benchmark data at (m*Δt) time; S(m*Δt) represents the difference data between training data and benchmark data at (m*Δt) time; Represents the difference data between the training data and the benchmark data during the period from Δt to (m*Δt) under the mode M _gh ; Indicates the threshold; Represents the difference data after threshold processing in the period from Δt to (m*Δt) under the mode M _gh , which is the feature of the training data; Indicates the mapping rule between traffic state data and features.

3) Through the fuzzy division of the existing road traffic data and road traffic service level, combined with the characteristics of the training data, the construction of the knowledge base is completed, and its general expression is as follows:

Among them, φ represents the mapping relationship between the current road traffic state and the fuzzy service level, and Los _m represents the fuzzy road traffic service level.

Combining (4) and (5), the relationship between the traffic operation state and the characteristics can be obtained:

Los _m ＝ω(he _m ) (6)

Among them, ω represents the mapping rule between the fuzzy service level and the traffic operation characteristics, so as to complete the construction of the knowledge base.

4) Extract road traffic test data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the test data through thresholding processing. The general expression is as follows:

in Indicates the road traffic test data at (m*Δt) time; MS(m*Δt) represents the difference data between test data and benchmark data at (m*Δt) time; Represents the difference data between the test data and the benchmark data in the time period from Δt to (m*Δt); Represents the difference data after threshold processing, which is the feature of the test data.

5) Through the KNN feature matching algorithm, select the k features closest to the test data features in the knowledge base, and obtain the corresponding fuzzy service level. The process is as follows:

5.1) Calculate the distance between the features of the training data and the features of the test data

dist(m)＝||The _Mgh (m)-he _Mgh (m)|| (10)

D _N (m)＝[dist ₁ (m)dist ₂ (m)...dist _N (m)] (11)

Among them, |||| means to obtain the Euclidean distance, dist(m) means the distance between the features of the test data and the features in the knowledge base during the period from Δt to (m*Δt), N means the number of groups of training data, D _N (m) Represents the set of feature distances from the features of N sets of test data to N sets of training data within the period from Δt to (m*Δt).

5.2) Find the features corresponding to the k closest distances, let the k features be s ₁ , s ₂ , ... s _k , according to the mapping relationship between features and service levels, then there is

(L ₁ , L ₂ , ... L _k ) = ω(s ₁ , s ₂ , ... s _k ) (12)

Among them, L ₁ , L ₂ , ... L _k represent the fuzzy service levels corresponding to features s ₁ , s ₂ , ... s _k respectively. in, Respectively represent the probability of unimpeded, general congestion, and severe congestion under the mode M _gh , from Δt to (m*Δt).

6) Add up the fuzzy service levels of k groups to obtain the service level corresponding to the maximum probability, which is the service level to which the current feature belongs, and complete the evaluation of road traffic operation status. The process is as follows:

Add the fuzzy service level probabilities corresponding to k features, then we have

Among them, Su _AK , Su _AK , and Su _AK respectively represent the probability of unimpeded traffic, general congestion, and severe congestion after integrating k features.

7) Parameter determination of road traffic service level evaluation method based on fuzzy KNN feature matching;

In the evaluation method of road traffic service level based on fuzzy KNN feature matching algorithm, the following parameters are designed:

can be made by with Obtain, can be made by It is decided that the value of k is between 3-10 and can be obtained through training. The parameter setting here is only an analysis of the approximate impact on the evaluation of road traffic operation status based on the KNN algorithm.

Since these parameters have different influences on the accuracy of the algorithm, analyzing the influence of each parameter on the accuracy of the algorithm alone cannot ensure the optimality of the algorithm. Therefore, the influence of all parameters on the evaluation of the road traffic operation status should be considered at the same time when analyzing the algorithm.

The deviation rate is introduced as the evaluation index of road traffic operation status, and its calculation formula is as follows:

Among them, PE represents the deviation rate, N _sr≠sm represents the number of traffic service levels that are different from the real ones, and N _sr represents the number of traffic service levels that are actually tested.

i.e. for different A corresponding NMAE exists. So there is the following equation:

which is There is a certain distribution relationship f with NMAE, looking for the corresponding when NMAE is the smallest That is the optimal parameter setting process. Therefore, the following model can be obtained:

finally The value of can be determined through the training of road traffic benchmark data and training data.

7) Experimental results

Obtain optimal parameters based on road traffic benchmark data and training data of the same modality The results of this experiment mainly evaluate the road traffic service level based on the vehicle speed value of the road section. The road traffic test data is extracted, and the evaluation of the test status is realized based on the KNN algorithm.

The deviation rate is selected as the evaluation index of road traffic service level, and its calculation is shown in formula (13). The deviation statistical analysis of the speed value of the experimental road section on June 28, 2011 is shown in Table 4 below.

Section ID time Deviation rate HI2075a 2011-6-28 11.53% HI3002b 2011-6-28 5.83% HI7008a 2011-6-28 10.00% HI7051a 2011-6-28 5.69% HI7060b 2011-6-28 12.22%

Table 4.

Claims (7)

1. a road traffic service level evaluation method based on fuzzy KNN feature matching, is characterized in that: described evaluation method comprises the following steps: 1) Establish a road traffic feature reference sequence to obtain road traffic benchmark data in different modes; 2) Extract road traffic training data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the training data through thresholding processing; 3) Fuzzify the existing road traffic service level, and combine the characteristics of the training data to complete the construction of the knowledge base; 4) Extract road traffic test data, obtain road traffic difference data based on the road traffic reference data in the same mode, and obtain the characteristics of the test data through thresholding processing; 5) Through the KNN feature matching algorithm, select the k features closest to the test data features in the knowledge base, and obtain the corresponding fuzzy service level; 6) Add up the fuzzified service levels of k groups to obtain the service level corresponding to the maximum probability, which is the service level to which the current feature belongs, and complete the road traffic service level evaluation. 2. a kind of road traffic service level evaluation method based on fuzzy KNN feature matching as claimed in claim 1, is characterized in that: in described step 1), set up road traffic characteristic reference sequence, obtain the road traffic under different modes Benchmark data, including the following steps: 1.1) Design the structure of the road traffic feature reference sequence Set the collection cycle of road traffic status data as Δt, the table format of the road traffic feature reference sequence is shown in Table 1 and Table 2, Table 1 is the road traffic feature reference sequence information table, and Table 2 is the road traffic feature reference sequence description table ; Table 1 Table 2 1.2) Through data preprocessing, establish a reference sequence of road traffic operation characteristics The division of the road traffic operation mode is divided into two levels: the road network layer and the road section layer. The division logo of the traffic operation mode of the road network layer is set to divide the traffic operation mode of the road into the g seed mode, and the road section layer The division mark of the traffic operation mode divides the traffic operation mode of the road into h sub-modes, then the traffic operation mode of the road is divided into g×h types in total, recorded as the set Mode = { _M11 , M12, ..., M _gh }, where the values of g and h are determined according to the division identification of the selected traffic operation mode; Obtain representative road traffic state data of the target road section, perform data preprocessing, and input the road traffic state data after data preprocessing into the road traffic feature reference sequence, thereby establishing the road traffic feature reference sequence. 3. a kind of road traffic service level evaluation method based on fuzzy KNN feature matching as claimed in claim 1 or 2, is characterized in that: in described step 2), extract road traffic training data, based on the road under the same modality Traffic reference data, obtain the road traffic difference data, after thresholding processing, obtain the characteristics of the training data, the general expression is as follows: <mrow> <mi>S</mi> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> <mo>=</mo> <msub> <mi>ST</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>SB</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> 1 <mrow> <msub> <mi>e</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mrow> <mi>S</mi> <mrow> <mo>(</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <mi>S</mi> <mrow> <mo>(</mo> <mn>2</mn> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mrow> <mi>S</mi> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </mrow> <mrow> <msub> <mi>he</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <msub> <mi>e</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>&lt;</mo> <msub> <mi>E</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>e</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <msub> <mi>e</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>&gt;</mo> <msub> <mi>E</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow> Among them, M _gh is the mode; Δt is the collection cycle of road traffic state data; (m*Δt) is the mth road traffic state data collection cycle, 0≤m≤N, and N represents the number of traffic information collected every day ; Indicates the training data at (m*Δt) time; Indicates the benchmark data at (m*Δt) time; S(m*Δt) represents the difference data between training data and benchmark data at (m*Δt) time; Represents the difference data between the training data and the benchmark data during the period from Δt to (m*Δt) under the mode M _gh ; Indicates the threshold; Represents the difference data after threshold processing in the period from Δt to (m*Δt) under the mode M _gh , which is the feature of the training data; Indicates the mapping rule between traffic state data and features. 4. a kind of road traffic service level evaluation method based on fuzzy KNN feature matching as claimed in claim 3, is characterized in that: in described step 3), by existing road traffic data and road traffic service level fuzzy division, Combined with the characteristics of the training data, the construction of the knowledge base is completed, and its general expression is as follows: <mrow> <msub> <mi>Los</mi> <mi>m</mi> </msub> <mo>=</mo> <mi>&amp;phi;</mi> <mrow> <mo>(</mo> <msub> <mi>ST</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mo>(</mo> <mi>m</mi> <mo>)</mo> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </mrow> Among them, φ represents the mapping relationship between the current road traffic state and the fuzzy service level, and Los _m represents the fuzzy road traffic service level. Combining (4) and (5), the relationship between traffic operation status and characteristics is obtained: Los _m ＝ω(he _m ) (6) Among them, ω represents the mapping rule between the fuzzy service level and the traffic operation characteristics, so as to complete the construction of the knowledge base. 5. a kind of road traffic service level evaluation method based on fuzzy KNN feature matching as claimed in claim 4, is characterized in that: in described step 4), extract road traffic real-time data, based on the road traffic benchmark under the same modality Data, to obtain road traffic difference data, through thresholding processing, to obtain the characteristics of real-time data, the general expression is as follows: <mrow> <mi>M</mi> <mi>S</mi> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> <mo>=</mo> <msub> <mi>SM</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>SB</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>7</mn> <mo>)</mo> </mrow> </mrow> <mrow> <msub> <mi>err</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mrow> <mi>M</mi> <mi>S</mi> <mrow> <mo>(</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <mi>M</mi> <mi>S</mi> <mrow> <mo>(</mo> <mn>2</mn> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mo>...</mo> </mtd> <mtd> <mrow> <mi>M</mi> <mi>S</mi> <mrow> <mo>(</mo> <mi>m</mi> <mo>*</mo> <mi>&amp;Delta;</mi> <mi>t</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>8</mn> <mo>)</mo> </mrow> </mrow> <mrow> <msub> <mi>The</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <msub> <mi>err</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>&lt;</mo> <msub> <mi>E</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>err</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <msub> <mi>err</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>&gt;</mo> <msub> <mi>E</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>9</mn> <mo>)</mo> </mrow> </mrow> in Indicates the real-time road traffic data at (m*Δt) time; MS(m*Δt) indicates the difference data between real-time data and benchmark data at (m*Δt) time; Indicates the difference data between the real-time data and the reference data in the time period from Δt to (m*Δt); Represents the difference data after threshold processing, which is the feature of real-time data. 6. a kind of road traffic service level evaluation method based on fuzzy KNN feature matching as claimed in claim 5, it is characterized in that: in described step 5), by KNN feature matching algorithm, select knowledge base and real-time data feature distance The nearest k features, and calculate the corresponding fuzzy service level, the process is as follows: 5.1) Calculate the distance between the features of the training data and the features of the real-time data dist(m)＝||The _Mgh (m)-he _Mgh (m)|| (10) D _N (m)＝[dist ₁ (m)dist ₂ (m)...dist _N (m)] (11) Among them, |||| means to obtain the Euclidean distance, dist(m) means the distance between the features of the real-time data and the features in the knowledge base during the period from Δt to (m*Δt), N means the number of groups of training data, D _N (m) Represents the set of feature distances between the features of N groups of real-time data and N groups of training data within the period from Δt to (m*Δt); 5.2) Find the features corresponding to the k closest distances, let the k features be s ₁ , s ₂ , ... s _k , according to the mapping relationship between features and service levels, then there is (L ₁ , L ₂ , ... L _k ) = ω(s ₁ , s ₂ , ... s _k ) (12) Among them, L ₁ , L ₂ , ... L _k represent the fuzzy service levels corresponding to s ₁ , s ₂ , ... s _k respectively. in, Respectively represent the probability of unimpeded, general congestion, and severe congestion under the mode M _gh , from Δt to (m*Δt). 7. a kind of road traffic operating state evaluation method based on fuzzy KNN feature matching as claimed in claim 6, is characterized in that: in described step 6), the service level of k group fuzzification is added up, draws maximum probability The corresponding service level is the service level to which the current feature belongs. To complete the road traffic service level evaluation, the process is as follows: Add the fuzzy service level probabilities corresponding to k features, then we have <mrow> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mrow> <msub> <mi>Su</mi> <mrow> <mi>A</mi> <mi>k</mi> </mrow> </msub> </mrow> </mtd> <mtd> <mrow> <msub> <mi>Su</mi> <mrow> <mi>B</mi> <mi>k</mi> </mrow> </msub> </mrow> </mtd> <mtd> <mrow> <msub> <mi>Su</mi> <mrow> <mi>C</mi> <mi>k</mi> </mrow> </msub> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mrow> <munderover> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>k</mi> </munderover> <msub> <mi>u</mi> <mrow> <mi>A</mi> <mi>i</mi> </mrow> </msub> <mrow> <mo>(</mo> <mrow> <msub> <mi>ST</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <munderover> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>k</mi> </munderover> <msub> <mi>u</mi> <mrow> <mi>B</mi> <mi>i</mi> </mrow> </msub> <mrow> <mo>(</mo> <mrow> <msub> <mi>ST</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> <mo>)</mo> </mrow> </mrow> </mtd> <mtd> <mrow> <munderover> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>k</mi> </munderover> <msub> <mi>u</mi> <mrow> <mi>C</mi> <mi>i</mi> </mrow> </msub> <mrow> <mo>(</mo> <mrow> <msub> <mi>ST</mi> <msub> <mi>M</mi> <mrow> <mi>g</mi> <mi>h</mi> </mrow> </msub> </msub> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> </mrow> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>13</mn> <mo>)</mo> </mrow> </mrow> Among them, Su _AK , Su _AK , and Su _AK respectively represent the probability of unimpeded traffic, general congestion, and severe congestion after integrating k features. Su _AK , Su _AK , the service level corresponding to the largest probability value among Su _AK is the service level to which the current feature belongs.