A Highway Traffic Congestion Determination Method Based on Two-Dimensional Fuzzy Cloud and Simultaneous Introduction of Upstream Congestion Status

Abstract

This invention discloses a method for highway traffic congestion identification based on two-dimensional fuzzy cloud and simultaneously incorporating upstream congestion status. The method includes: acquiring highway congestion identification indicators and congestion levels; filtering the congestion identification indicators to obtain congestion indicator pairs; fuzzifying the congestion indicator pairs; constructing a congestion identification rule base based on the fuzzification result and congestion level; extracting the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the antecedent of each rule and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the consequent of each rule in the congestion identification rule base using a cloud generator, and combining these with upstream congestion status influence factors to construct a two-dimensional fuzzy cloud model for highway traffic congestion identification; identifying real-time congestion indicators using the two-dimensional fuzzy cloud model to obtain real-time congestion level values, and obtaining the final congestion level based on the real-time congestion level values ​​for accurate traffic congestion status identification.

Description

Technical Field

[0001] This invention belongs to the field of traffic engineering technology, and in particular relates to a method for determining highway traffic congestion based on two-dimensional fuzzy cloud and simultaneously incorporating upstream congestion status. Background Technology

[0002] To fundamentally solve traffic congestion, the first challenge should be accurately identifying traffic congestion conditions. Currently, numerous studies utilize various models and methods based on congestion indicators to determine road traffic congestion, specifically analyzing the operational state and degree of congestion at a given moment. However, existing models and methods often lack consideration of upstream congestion conditions and a quantitative mechanism for the transmission of congestion pressure between upstream and downstream areas, leading to low accuracy. Therefore, accurately identifying traffic congestion conditions on highways based on dynamic highway data is of great significance, laying a solid foundation for efficient highway traffic congestion management. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a highway traffic congestion discrimination method based on two-dimensional fuzzy cloud and simultaneously incorporating upstream congestion status, thereby resolving the issues present in the existing technologies.

[0004] To achieve the above objectives, this invention provides a highway traffic congestion discrimination method based on two-dimensional fuzzy clouds while simultaneously incorporating upstream congestion status, comprising:

[0005] Obtain congestion identification indicators and congestion levels for highways; filter the congestion identification indicators to obtain congestion identification indicator pairs; perform fuzzification processing on the congestion identification indicator pairs;

[0006] Based on the fuzzification results and the degree of congestion, a congestion discrimination rule base is constructed;

[0007] Based on the congestion level corresponding to the congestion discrimination index, the cloud generator extracts the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the antecedent of each rule and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the consequent of each rule in the congestion discrimination rule base.

[0008] Based on the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of each rule's antecedent and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the rule's consequent, and combined with the upstream congestion status influence factor, a two-dimensional fuzzy cloud model for highway traffic congestion discrimination is constructed.

[0009] A two-dimensional fuzzy cloud model for highway traffic congestion is used to identify real-time congestion indicators, thereby obtaining a real-time congestion level value. The final congestion level is then determined based on this real-time congestion level value.

[0010] Optionally, the process of obtaining the congestion identification index includes:

[0011] The system acquires multi-source heterogeneous traffic data from highways, preprocesses the data to obtain statistical data, extracts feature data from the statistical data to obtain congestion discrimination indicators, which include spatial and temporal discriminant indicators. The spatial discriminant indicators include traffic volume, density, space occupancy rate, number of blocked vehicles, average travel time, and average travel speed. The temporal discriminant indicators include travel time ratio, congestion duration, cumulative vehicle delay, and average traffic delay.

[0012] Optionally, the process of obtaining the congestion level includes:

[0013] Statistical analysis of traffic density on different sections of highways is conducted to draw a cumulative probability distribution map. Different quantiles in the cumulative probability distribution map are used as the definition criteria to classify and evaluate traffic density according to the definition criteria, thereby obtaining the congestion level corresponding to different road sections.

[0014] Optionally, the process of filtering the congestion identification indicators includes:

[0015] Rough set theory is used to process congestion discrimination indicators into an indicator set. Road segment samples with different levels of congestion are integrated into a sample set. The indicator set and the sample set are integrated into a road segment congestion indicator data system. Based on the road segment congestion indicator data system, an identifiable matrix is ​​constructed. The identifiable matrix is ​​reduced to obtain a set of indicators that significantly affect congestion. The importance of the congestion discrimination indicators in the set of indicators that significantly affect congestion is calculated and ranked. The two most important congestion discrimination indicators are selected to obtain the congestion discrimination indicator pairs.

[0016] Optionally, the process of fuzzifying the congestion identification indicators includes:

[0017] Statistical analysis is performed on the congestion discrimination indicators in the congestion discrimination indicator pair, and cumulative probability distribution maps of the indicators are plotted respectively. Different quantiles in the cumulative probability distribution maps of the indicators are used as the indicator definition criteria. The congestion discrimination indicators are graded and evaluated according to the indicator definition criteria to obtain the fuzzy level of the congestion discrimination indicators.

[0018] Optionally, the process of constructing the congestion determination rule base includes:

[0019] In the fuzzification process, the fuzzy levels of different congestion discrimination indicators in the congestion discrimination indicator pair are combined in pairs to obtain different combinations. The congestion level with the highest frequency of occurrence under different combinations is used as the congestion level corresponding to the congestion discrimination indicator pair. Different rules are constructed based on the corresponding fuzzy results and the corresponding congestion levels of the congestion discrimination indicator pairs to obtain different rules to build a congestion discrimination rule library.

[0020] Optionally, the process of obtaining the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of each rule's antecedent and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of each rule's consequent includes:

[0021] The reverse cloud generator processes the sample data corresponding to each rule in the congestion discrimination rule base to obtain the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the rule antecedent, wherein the two-dimensional cloud parameters include the expected value, entropy value, and hyperentropy value of different congestion discrimination indicators; the reverse cloud generator also processes the sample data corresponding to the congestion degree of each rule in the congestion discrimination rule base to obtain the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the consequent of each rule, wherein the one-dimensional cloud parameters include the expected value, entropy value, and hyperentropy value of the consequent.

[0022] Optionally, the number and proportion of vehicles from the upstream road segment to the current road segment can be statistically analyzed to obtain the corresponding upstream congestion status impact factor.

[0023] Optionally, the real-time congestion identification indicators are obtained as follows:

[0024] Based on the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the rule antecedent, the real-time congestion discrimination index is processed to generate two-dimensional random values. The membership degree of the two-dimensional random values ​​is then calculated using the congestion impact factor.

[0025]

[0026] Among them, y i For membership, x1 and x2 are the real-time congestion discrimination indicators in the congestion discrimination indicator pair, Ex1 represents the expected value of the congestion discrimination parameter in the two-dimensional cloud parameters, Ex2 represents the expected value of another congestion discrimination parameter in the two-dimensional cloud parameters, and ΔEx i , ΔEn i En represents the adjustment amount by which upstream congestion affects the expected value and entropy of the current road segment for different congestion indices i. 1i and En 2i The different entropy values ​​of the congestion discrimination parameter in the two-dimensional cloud parameters represent the congestion impact factor, and N represents the congestion impact factor. j→i T represents the number of vehicles from upstream segment j to the current segment i. j→i λ represents the average travel time from upstream road segment j to current road segment i, and λ represents the time decay coefficient.

[0027] Extract the rule corresponding to the maximum membership degree from the membership degrees, randomly generate a one-dimensional normal random value based on the one-dimensional cloud parameters corresponding to the rule, and iteratively calculate the maximum membership degree based on the one-dimensional normal random value:

[0028]

[0029] Where y1 is the maximum membership degree, T represents the congestion level value, and Enti Represents a 3D normally distributed random value. This represents the expected value in the one-dimensional cloud parameters;

[0030] Repeat the conversion process between the maximum membership degree and the one-dimensional normal random value until the maximum number of iterations is reached to obtain the final congestion level value.

[0031] On the other hand, the present invention provides a highway traffic congestion discrimination system based on two-dimensional fuzzy cloud and simultaneously incorporating upstream congestion status, characterized in that it is used to execute the above-mentioned method.

[0032] Compared with the prior art, the present invention has the following advantages and technical effects:

[0033] 1. By using rough set theory to screen congestion discrimination index pairs, the most significant indexes affecting congestion discrimination are identified, thereby greatly improving the accuracy of the two-dimensional fuzzy cloud model for highway traffic congestion discrimination.

[0034] 2. By using a data-driven approach, a two-dimensional fuzzy cloud congestion discrimination model that takes into account both fuzziness and randomness is constructed. The final output value of the model is determined by two input variables, which solves the problem of complex uncertainty reasoning of road segment congestion status, realizes quantitative classification of congestion degree and qualitative state discrimination, and greatly improves the accuracy and applicability of highway segment congestion status discrimination.

[0035] 3. Introduce upstream congestion status to quantify the mechanism of congestion pressure transmission between upstream and downstream road sections, so as to improve the adaptability and recognition accuracy of the two-dimensional fuzzy cloud congestion discrimination model to actual traffic conditions. Attached Figure Description

[0036] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0037] Figure 1 This is a schematic diagram of the overall method flow according to an embodiment of the present invention;

[0038] Figure 2 This is a one-dimensional reverse cloud generator according to an embodiment of the present invention;

[0039] Figure 3 This is a schematic diagram of the distribution of the membership degree one-dimensional cloud on the universe of discourse in an embodiment of the present invention;

[0040] Figure 4 This is a two-dimensional reverse cloud generator according to an embodiment of the present invention;

[0041] Figure 5 This is a two-dimensional single rule generator according to an embodiment of the present invention;

[0042] Figure 6 This is a two-dimensional multi-rule generator according to an embodiment of the present invention. Detailed Implementation

[0043] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0044] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0045] This invention provides a highway traffic congestion discrimination method based on two-dimensional fuzzy clouds that incorporates upstream congestion status. The purpose of this invention is to solve the problem of accurate traffic congestion status discrimination. Existing identification methods mainly suffer from the main drawback of rarely incorporating upstream congestion status into their discrimination models or methods, lacking a quantitative mechanism for the transmission of congestion pressure between upstream and downstream, resulting in low discrimination accuracy. Therefore, this invention, based on dynamic highway data, is of great significance for accurately discerning the traffic congestion status of highway segments, laying a solid foundation for subsequent efficient management of highway traffic congestion. The process is as follows: 1. Based on the detection data of highway field electromechanical equipment such as ETC gantries, extract various congestion indicators related to highway segment congestion; 2. Use density indicators to quantify the degree of congestion and classify congestion levels; 3. Based on rough set theory, screen the congestion indicators, selecting the two congestion indicators with the highest importance to the segment congestion level as the selected congestion discrimination indicator pair; 4. Fuzzify the selected congestion discrimination indicator pairs; 5. Based on sample data and combined with expert experience, construct a congestion discrimination rule base with dual indicator input. 6. Extract the two-dimensional cloud parameters of the two-dimensional fuzzy cloud for each rule's antecedent and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud for the rule's consequent from historical congestion index data using the reverse cloud algorithm; 7. Introduce the upstream congestion state influence factor to correct the membership degree of the rule's antecedent two-dimensional fuzzy cloud; 8. Based on the antecedent two-dimensional cloud parameters, the corrected membership degree introduced by the upstream congestion state, and the consequent cloud parameters, construct a two-dimensional fuzzy cloud model for highway traffic congestion discrimination that incorporates the upstream congestion state; 9. Determine the congestion degree value, and obtain the final congestion level based on the congestion degree value.

[0046] The above technical solution is described in detail below:

[0047] The purpose of this invention is to address the problems in current traffic congestion management models or methods that fail to incorporate upstream congestion status, resulting in a lack of quantitative mechanisms for the transmission of congestion pressure between upstream and downstream sectors and consequently low accuracy. This invention proposes a highway traffic congestion discrimination method based on two-dimensional fuzzy clouds that incorporates upstream congestion status. The specific details of this solution are as follows:

[0048] Combination Figure 1 This invention provides a method for determining highway traffic congestion based on two-dimensional fuzzy clouds by introducing upstream congestion status. The specific method flow is as follows:

[0049] Step 1: Based on the detection data of highway field electromechanical equipment such as ETC gantries, extract various congestion indicators related to highway congestion.

[0050] Step 2: Use density indicators to quantify the degree of congestion and classify congestion levels.

[0051] Step 3: Based on rough set theory, the congestion indicators obtained in Step 1 are screened, and the two congestion indicators with the highest importance to the congestion level of the road segment are selected as the congestion discrimination indicator pair (X1, X2).

[0052] Step 4: The congestion identification indexes selected in Step 3 are fuzzed for X1 and X2 respectively.

[0053] Step 5: Based on the threshold values ​​for congestion discrimination indicators and congestion levels determined in Steps 2 and 4, construct a congestion discrimination rule base with dual input indicators based on sample data and expert experience.

[0054] Step 6: Based on the rule base obtained in Step 5, extract the two-dimensional cloud parameters (Ex1, En1, He1; Ex2, En2, He2) of the two-dimensional fuzzy cloud of the rule antecedent and the one-dimensional cloud parameters (Ey, En, He) of the one-dimensional fuzzy cloud of the rule consequent from the historical congestion index data using the reverse cloud algorithm.

[0055] Step 7: Introduce the upstream congestion state influence factor γ to correct the membership degree of the two-dimensional fuzzy cloud with the rule antecedent.

[0056] Step 8: Based on the antecedent two-dimensional cloud parameters obtained in Step 7, the corrected membership degree of the upstream congestion state, and the consequent cloud parameters, construct a two-dimensional fuzzy cloud model for highway traffic congestion discrimination that incorporates the upstream congestion state.

[0057] Step 9: Based on the two-dimensional fuzzy cloud model for highway traffic congestion discrimination obtained in Step 8, which incorporates upstream congestion status, determine the congestion level value, and obtain the final congestion level based on the congestion level value.

[0058] As one embodiment, in step 1, based on the detection data of highway field electromechanical equipment such as ETC gantries, various congestion indicators related to highway congestion are extracted; the specific process is as follows:

[0059] Based on multi-source heterogeneous data such as ETC gantry data, toll station data, toll road section data, toll plaza data, meteorological monitoring data, and holiday event logs, statistical data are obtained through data cleaning and spatiotemporal alignment. Traffic indicators related to traffic congestion during holidays and peak hours, such as traffic volume, congestion duration, maximum queue length, travel time ratio, average travel time, average travel speed, flow rate, speed, density, number of blocked vehicles, and space occupancy rate, are extracted as congestion discrimination indicators.

[0060] Among the existing data types, ETC gantry signage data will be used as the main source of historical data, and corresponding congestion discrimination indicators will be extracted from both spatial and temporal dimensions.

[0061] The main indicators for spatial dimensions include: traffic volume, density, space occupancy, number of blocked vehicles, average travel time, and average travel speed; the main indicators for temporal dimensions include: travel time ratio, congestion duration, cumulative vehicle delay, and average traffic delay.

[0062] In some embodiments, step 2 uses a density index to quantify the degree of congestion and classify congestion levels; the specific process is as follows:

[0063] First, the traffic density of each road segment was statistically analyzed and a cumulative probability distribution map was drawn. Then, using the four characteristic quantiles of 15%, 30%, 50%, and 85% as the definition criteria, the congestion level of the road segment was divided into a multi-level evaluation system, and finally a graded model with five gradients including {"smooth", "basically smooth", "mildly congested", "moderately congested" and "severely congested"} was formed.

[0064] As one embodiment, in step 3, the congestion indicators obtained in step 1 are screened based on rough set theory, and the two congestion indicators with the highest importance to the road segment congestion level are selected as the selected congestion discrimination indicator pair (X1, X2); the specific process is as follows:

[0065] Based on the congestion level classification model obtained in step 2, the congestion level of all road segment samples is determined. Using rough set theory, the congestion indicators obtained in step 1 are denoted as set X, and all road segment samples are denoted as U, resulting in a road segment congestion indicator data system S = (U, X). Subsequently, an identifiable matrix M(S) of S is constructed. The identifiable matrix is ​​reduced to obtain the selected set of congestion indicators X' that significantly influence the road segment congestion level. The importance of each congestion indicator in X' to the road segment congestion level is calculated and ranked. The two most important congestion indicators are selected as the congestion discrimination indicator pair (X1, X2). In this content:

[0066] Let S = (U,X) be a data system of road segment sample congestion discrimination index, and U = {u1,u2,…,u...} n}, u i Let X be the sample of the i-th road segment, and there are n road segment samples in total. X = {x1, x2, ..., x...} m}, x1 to x m Let D = {d(u1), d(u2), ..., d(u...}}, where m are the congestion indicators for the corresponding road segment samples. n )},d(u i Let be the accident rate of road segment unit i. Matrix M(S) = (c ij ) n×n For the congestion discrimination index data system S, the identifiable matrix is ​​defined as follows: identifiable matrix element c ij The possible values ​​are as follows:

[0067]

[0068] Where i and j represent the sample numbers of road segments, and n represents the total number of road segment samples.

[0069] Identifiable matrix element c ij It means that it can make the road segment sample u i With road segment sample u j The set of all congestion indicator variables that are distinguished, taking x1∨x2∨...∨x m The disjunctive normal form. Clearly, taking u... i with u i+1 ,u i+2 ,…,u n The variable to be distinguished should be the conjunction c. i(i+1) ∧c i(i+2) ∧...∧c in Then the conjunction of all identifiable matrix elements can distinguish all road segment units in pairs.

[0070] The Rosetta software was used to implement the process of selecting congestion discrimination index variables based on rough set theory.

[0071] The importance σ(x) of the selected congestion discrimination indicators were respectively determined. i The calculation is performed using the following formula:

[0072]

[0073] In the formula: card represents the cardinality of the corresponding set, and pos represents the positive domain information of the road segment unit. x (D) = X_(D), where X_(D) represents the positive domain of D. A positive domain is the union of objects that can be unambiguously classified into a single decision class (the classification of D) among all equivalence classes of the conditional attribute set X. The conditional attribute set X represents the set of congestion discrimination indicators, and the decision attribute D represents the mapped congestion level.

[0074] Based on the calculation results, the importance σ(x) of each congestion discrimination indicator is assigned. i The indicators are sorted, and the two most important indicators are selected as the congestion judgment indicator pair (X1, X2).

[0075] As one embodiment, in step 4, the congestion discrimination index pair selected in step 3 is fuzzified for X1 and X2 respectively; the specific process is as follows:

[0076] Based on the sample data, statistical analysis was performed on the congestion discrimination index pair (X1, X2) selected in step 3, and cumulative probability distribution maps of X1 and X2 were plotted. Then, using the four characteristic quantiles of 15%, 30%, 50%, and 85% as the definition criteria, the congestion discrimination index was divided into five levels: "very low", "low", "medium", "high", and "very high".

[0077] As one embodiment, in step 5, based on the congestion discrimination index classification threshold and congestion level classification threshold determined in steps 2 and 4, a congestion discrimination rule base with dual index input is constructed according to sample data and expert experience; the specific process is as follows:

[0078] Based on the congestion indicators and congestion level classification standards in steps 2 and 4, X1 and X2 are each divided into five levels. Combining each level of X1 and X2 in pairs yields 25 possible scenarios: X1 is very low and X2 is very low, X1 is very low and X2 is low, X1 is very low and X2 is medium, X1 is very low and X2 is high, X1 is very low and X2 is very high, X1 is low and X2 is very low, X1 is low and X2 is low, X1 is low and X2 is medium, X1 is low and X2 is medium, X1 is low and X2 is high. 2 High, X1 Low and X2 Very High, X1 Medium and X2 Very Low, X1 Medium and X2 Low, X1 Medium and X2 Medium, X1 Medium and X2 High, X1 Medium and X2 Very High, X1 High and X2 Very Low, X1 High and X2 Low, X1 High and X2 Medium, X1 High and X2 High, X1 High and X2 Very High, X1 Very High and X2 Very Low, X1 Very High and X2 Low, X1 Very High and X2 Medium, X1 Very High and X2 High, X1 Very High and X2 Very High.

[0079] Statistical analysis was performed on each situation to identify the most frequent congestion level in each sample. Combined with expert experience, congestion discrimination rules were constructed based on dual congestion index inputs, such as: IF "X1 Low" AND "X2 Low" THEN "Congestion level is smooth".

[0080] As one embodiment, in step 6, based on the rule base obtained in step 5, the two-dimensional cloud parameters (Ex1, En1, He1; Ex2, En2, He2) of the two-dimensional fuzzy cloud of the rule antecedent and the one-dimensional cloud parameters (Ey, En, He) of the one-dimensional fuzzy cloud of the rule consequent are extracted from the historical congestion index data using the reverse cloud algorithm; the specific process is as follows:

[0081] Based on the rule base obtained in step 5, the two-dimensional cloud parameters (Ex1, En1, He1; Ex2, En2, He2) of the two-dimensional fuzzy cloud of the rule antecedent are extracted from the sample data using the inverse cloud algorithm. For example, sample data conforming to the rule 'IF "X1 is very low" AND "X2 is low" THEN "Congestion level is smooth"' are extracted from the sample data. The two-dimensional cloud parameters (Ex1, En1, He1; Ex2, En2, He2) of the two-dimensional fuzzy cloud of the rule antecedent are extracted using the inverse cloud algorithm. Similarly, the one-dimensional cloud parameters (Ey, En, He) of the one-dimensional fuzzy cloud of the rule consequent are extracted using the inverse cloud algorithm. In this content:

[0082] The model input consists of congestion index pairs, X1 and X2, and the output is the membership degree of each index pair to different levels of congestion. The inverse cloud algorithm extracts the two-dimensional cloud parameters (ex1, En1, He1; Ex2, En2, He2) of the two-dimensional fuzzy cloud for the rule antecedents from the congestion index sample data. Ex1, En1, He1; Ex2, En2, He2 represent the expected value, entropy value, and hyperentropy value of the first antecedent, and the expected value, entropy value, and hyperentropy value of the second antecedent, respectively.

[0083] For each congestion level (t) 畅通 t 基本畅通 t 轻度拥堵 t 中度拥堵 t 严重拥堵 Based on the principle of inverse cloud, the one-dimensional cloud parameters (Ey, En, He) of the one-dimensional fuzzy cloud of the rule consequent are extracted from the sample data. Ey, En, and He represent the expected value, entropy value, and hyperentropy value of the consequent, respectively.

[0084] The principle of one-dimensional inverse cloud is as follows: assuming a set of cloud droplets drop(x) conforming to a certain normal distribution, ... i ,y i ) as a sample, where x i ,y i Let the index parameters and their corresponding membership degrees represent the values, respectively. This will generate a set of numbers (Ex, En, He). It should be noted that the superscript ^ indicates a calculated value, and the absence of a superscript indicates a theoretical value. The principle is as follows: Figure 2 As shown.

[0085] The formula for a one-dimensional fuzzy cloud is as follows:

[0086] Sample mean for:

[0087]

[0088] In the formula x i Let i be the i-th sample.

[0089] First-order sample absolute central distance - is:

[0090]

[0091] In the formula, N is the number of samples.

[0092] Sample variance S 2 for:

[0093]

[0094] One-dimensional cloud parameter expectation value for:

[0095]

[0096] One-dimensional cloud parameter entropy for:

[0097]

[0098] One-dimensional cloud parameter hyperentropy for:

[0099]

[0100] Once the rules for judgment and the cloud digital features of each rule are obtained, the one-dimensional cloud congestion discrimination model is completed.

[0101] One-dimensional fuzzy cloud membership degree y i The calculation formula is as follows:

[0102]

[0103] Where X represents the index that needs to be judged, and En represents the normally distributed random value corresponding to X.

[0104] The positive cloud generator is used to generate subordinate clouds for each level of each indicator. The principle of the positive cloud generator is as follows: Figure 3 As shown, the schematic diagram of the membership degree cloud is as follows. Figure 3 As shown.

[0105] The principle of a two-dimensional reverse cloud generator is as follows: Figure 4 As shown, specifically:

[0106] Implementation of a 2D Inverse Cloud Generator:

[0107] Input: Several 3D points (x 1i ,x 2i ,y i ), i = 1, 2, 3, ..., n#

[0108] Output: Two-dimensional cloud parameters (Ex1, En1, He1; Ex2, En2, He2)

[0109] Begin{

[0110] Ex1=(x 11 +x 12 +…+x 1n ) / n

[0111] Ex2=(x 21 +x 22 +…+x 2n ) / n

[0112] For(j=1;j<=n;j++)

[0113] {Ifx 1j =Ex1thenx 1j →Set A1;

[0114] Ifx 2j =Ex2thenx 2j →Set A2;

[0115] };

[0116] (En1,He1)=CGx(A1);

[0117] (En2,He2)=CGx(A2);

[0118] };

[0119] End.

[0120] Where CGx represents the reverse cloud generator algorithm, x 1i x 2i This indicates the index in the congestion discrimination index, y i The above indicators represent the degree of membership and are obtained by statistical analysis of historical data.

[0121] The membership degree calculation formula for a two-dimensional fuzzy cloud is:

[0122]

[0123] Where (u1,u2) represents a congestion index pair, i.e., a massive sample obtained from the historical data of the road segment; (E x1 E x2 ) represents the expected value of the index (u1, u2), calculated using the formula in the aforementioned formula "Implementation of Two-Dimensional Inverse Cloud Generator"; (E n1i E n2i ) represents the entropy of (u1,u2), generated by the reverse cloud algorithm in the aforementioned formula "Implementation of Two-Dimensional Reverse Cloud Generator", where i represents the entropy generated each time through the reverse cloud (E). n1i E n2i ) is a random number that satisfies the cloud parameters, and it needs to be generated thousands of times in multiple loops (E n1i E n2i Based on this, multiple membership degrees (quantitative predictions of congestion levels) are obtained, and the average value is finally taken as the output.

[0124] As one embodiment, step 7 introduces an upstream congestion state influence factor γ to correct the membership degree of the two-dimensional fuzzy cloud, which is a rule antecedent; the specific process is as follows:

[0125] Based on the traffic density threshold classification results in step 2, when the traffic density of the upstream road segment of the evaluation road segment is identified as moderate or severe congestion, a congestion impact factor γ is introduced based on the two-dimensional fuzzy cloud parameters obtained in step 6 to correct the membership degree of the downstream adjacent road segment (i.e. the evaluation road segment), thereby quantifying the pressure propagation effect of upstream congestion on the downstream.

[0126] When the upstream road segment is identified as moderately or severely congested, the proposed correction formula for the membership degree of the downstream adjacent road segment and the quantitative formula for γ are as follows:

[0127]

[0128] In the formula: γ represents the congestion impact factor, with a value range of [0,1], and ΔEx i , ΔEn i This represents the correction amount by which upstream congestion affects the expected value and entropy of the current road segment for different congestion indicators. It is obtained through training with historical data. N j→i T represents the number of vehicles from upstream segment j to the current segment i. j→i λ represents the average travel time from upstream road segment j to current road segment i, and λ represents the time decay coefficient.

[0129] In the above formula, λ controls the time sensitivity of the upstream's impact on the downstream, and λ is set to 0.1. Based on experience, the propagation time of congestion between adjacent highway segments is generally 5-15 minutes, so λ is set to 0.1 min. -1 This represents the decay factor after 10 minutes (T=10). This means that 37% of the upstream impact is retained, which is consistent with the observation results in actual scenarios where downstream road sections are significantly affected by upstream congestion within 10 minutes.

[0130] With ΔEx i For example, we obtain ΔEx for each congestion discrimination index. i The process is as follows: Determine the target output, which is the actual change ΔEx of the expected value of relevant indicators in the downstream road section. i Prepare the dataset. For the selected congestion indicators, select road segments with moderate or severe congestion upstream and no independent events (accidents, etc.) downstream as samples. The time window t is within the attenuation range of the influence of λ (in this patent, t ≤ 15 min).

[0131] First, calculate the expected value Ex of the downstream congestion index for each sample in the rule base when there is no upstream congestion. base Then, calculate the actual expected value Ex of the downstream indicator per minute during the upstream congestion period for each sample. obs (Calculate the congestion index data of the downstream road segment every minute within the time window from 0 to t minutes after the upstream section begins to experience moderate congestion); Define the objective function.

[0132] ΔEx i =Ex obs -Ex base

[0133] The multiple linear regression method was used, with the input feature being the congestion level T of the upstream road segment. up (0-1 congestion quantization value obtained through the basic two-dimensional fuzzy cloud model), congestion duration t sp Distance L between upstream and downstream road sections gap The regression analysis results are as follows:

[0134] ΔEx i =a·Tup +b·t sp +c·L gap

[0135] In the formula, a, b, and c represent coefficients determined by the regression analysis, and ΔEx i This represents the expected value correction amount for the selected congestion assessment indicators, such as the expected value correction amount for traffic flow as ΔEx. Q The expected value correction for velocity is ΔEx. S .

[0136] Get ΔEn i The computational training method and ΔEx i Consistent. Finally, obtain the ΔEx of all congestion indicators. i ΔEn i The formula is then used to re-determine congestion levels in downstream road segments by substituting the corrected membership values.

[0137] As one embodiment, in step 8, based on the antecedent two-dimensional cloud parameters obtained in step 7, the corrected membership degree introducing the upstream congestion state, and the consequent cloud parameters, a two-dimensional fuzzy cloud model for highway traffic congestion discrimination introducing the upstream congestion state is constructed; the specific process is as follows:

[0138] Based on the antecedent two-dimensional cloud parameters obtained in step 7, the corrected membership degree incorporating the upstream congestion state, and the consequent cloud parameters, a complex rule generator is constructed for each rule, such as... Figure 5 As shown.

[0139] Then, the single-rule generators of each rule are combined to work together to construct a multi-rule generator, such as... Figure 6 As shown.

[0140] Congestion identification is performed on newly input data based on a two-dimensional fuzzy cloud model. For each individual rule, [the following is a description of the method used in the original text, which is not directly related to the congestion identification process]. As expected, To generate a two-dimensional random value (En) that conforms to a two-dimensional normal distribution, given the variance. 1i En 2i ), Based on the newly input congestion index, (En) is applied to (x1, x2). 1i En 2i Based on the membership calculation formula that introduces the congestion impact factor γ, the activation intensity, i.e. the membership degree y, is obtained when the input (x1, x2) is given in all single rule generator antecedents. i The specific calculation formula can be found in step 7, namely:

[0141]

[0142] y represents the index value of congestion level, that is, the quantitative value of congestion level. i The single rule corresponding to the largest value y1 in the middle, and the one-dimensional cloud parameters of the consequent given by this rule. Randomly generated As expected, A one-dimensional normal random value with variance (En) ti ), N represents a normal distribution. The formula is used to calculate the value of En at y1. ti The formula for calculating the T-value (i.e., the degree of congestion) under the given conditions is as follows:

[0143]

[0144] Where t represents the parameter corresponding to the degree of congestion, E xt It represents the expected value of the congestion level, a fuzzy cloud parameter indicating the congestion level—the expected value. Parameters with subscript 't' refer to parameters related to the congestion level. E nti The congestion level corresponds to the entropy of a one-dimensional cloud, where i represents the E generated each time the cloud is reversed. nti It is a random number that satisfies the cloud parameters; multiple iterations require generating thousands of E values. nti .

[0145] T represents the congestion level value, and y represents the membership degree value of the congestion level. Each iteration yields a predicted membership degree y. i The system calculates the mean of the corresponding membership degree, i.e. the congestion level value T. After repeating this process a thousand times, the average value T of all membership degree results is taken as the final output congestion level prediction value.

[0146] After multiple iterations (more than 1000), the average value of all membership degrees is finally output as the prediction result.

[0147] Figure 6 The final output T is the output of the two-dimensional fuzzy cloud multi-rule generator congestion discrimination model for the congestion discrimination index pair (X1, X2), which is the congestion degree value (i.e., traffic density value).

[0148] At this point, the two-dimensional fuzzy cloud model for judging highway traffic congestion by incorporating upstream congestion conditions has been completed.

[0149] As one embodiment, in step 9, based on the two-dimensional fuzzy cloud model for highway traffic congestion discrimination obtained in step 8, which incorporates the upstream congestion status, a congestion level value is determined, and the final congestion level is determined based on the congestion level value; the specific process is as follows:

[0150] The identified road segment discrimination index pair (X1, X2) is input into the two-dimensional fuzzy cloud model of highway traffic congestion discrimination obtained in step 8, which introduces the upstream congestion state. Finally, the congestion level value T is output. The obtained congestion level value T is substituted into the threshold division boundary of the five congestion levels in step 2 to obtain the final congestion level.

[0151] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for determining highway traffic congestion based on two-dimensional fuzzy clouds while simultaneously incorporating upstream congestion status, characterized in that, include: Obtain congestion indicators and congestion levels for highways; The congestion identification indicators are filtered to obtain congestion identification indicator pairs; The congestion identification indicators are fuzzified; The process of screening the congestion identification indicators includes: Rough set theory is used to process congestion discrimination indicators into an indicator set, and road segment samples with different levels of congestion are integrated into a sample set. The indicator set and the sample set are integrated into a road segment congestion indicator data system. Based on the road segment congestion indicator data system, an identifiable matrix is ​​constructed. The identifiable matrix is ​​reduced to obtain a set of indicators that significantly affect congestion. The importance of the congestion discrimination indicators in the set of indicators that significantly affect congestion is calculated and ranked. The two most important congestion discrimination indicators are selected to obtain the congestion discrimination indicator pairs. Based on the fuzzification results and the degree of congestion, a congestion discrimination rule base is constructed; The process of building the congestion judgment rule base includes: In the fuzzification process, the fuzzy levels of different congestion discrimination indicators in the congestion discrimination indicator pair are combined in pairs to obtain different combinations. The congestion level with the highest frequency of occurrence in different combinations is used as the congestion level corresponding to the congestion discrimination indicator pair. Different rules are constructed based on the fuzzy results and congestion levels corresponding to the congestion discrimination indicator pairs to obtain different rules to construct a congestion discrimination rule library. Based on the congestion level corresponding to the congestion discrimination index, the cloud generator extracts the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the antecedent of each rule and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the consequent of each rule in the congestion discrimination rule base. Based on the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of each rule's antecedent and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the rule's consequent, and combined with the upstream congestion status influence factor, a two-dimensional fuzzy cloud model for highway traffic congestion discrimination is constructed. The real-time congestion discrimination index is judged by a two-dimensional fuzzy cloud model for highway traffic congestion discrimination, and the real-time congestion level value is obtained. The final congestion level is obtained based on the real-time congestion level value. The acquisition of real-time congestion identification indicators is as follows: Based on the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the rule antecedent, the real-time congestion discrimination index is processed to generate two-dimensional random values. The membership degree of the two-dimensional random values ​​is then calculated using the congestion impact factor. in, For membership degree, x 1. x 2 is the congestion identification indicator for real-time congestion identification. This represents the expected value of the congestion discrimination parameter in the two-dimensional cloud parameters. The expected value of another congestion discrimination parameter in the two-dimensional cloud parameters. ， This represents the adjustment amount by which upstream congestion affects the expected value and entropy of the current road segment for different congestion indices i. and This represents the different entropy values ​​of the congestion discrimination parameter in the two-dimensional cloud parameters. Indicates the factors affecting congestion. This represents the number of vehicles traveling from upstream segment j to the current segment i. This represents the average travel time from upstream road segment j to the current road segment i. Indicates the time decay coefficient; Extract the rule corresponding to the maximum membership degree from the membership degrees, randomly generate a one-dimensional normal random value based on the one-dimensional cloud parameters corresponding to the rule, and iteratively calculate the maximum membership degree based on the one-dimensional normal random value: in, The maximum membership degree is represented by T, which indicates the degree of congestion. Represents a 3D normally distributed random value. This represents the expected value in the one-dimensional cloud parameters; Repeat the conversion process between the maximum membership degree and the one-dimensional normal random value until the maximum number of iterations is reached to obtain the final congestion level value.

2. The method according to claim 1, characterized in that, The process of obtaining the congestion identification indicators includes: The system acquires multi-source heterogeneous traffic data from highways, preprocesses the data to obtain statistical data, extracts feature data from the statistical data to obtain congestion discrimination indicators, which include spatial and temporal discriminant indicators. The spatial discriminant indicators include traffic volume, density, space occupancy rate, number of blocked vehicles, average travel time, and average travel speed. The temporal discriminant indicators include travel time ratio, congestion duration, cumulative vehicle delay, and average traffic delay.

3. The method according to claim 1, characterized in that, The process of obtaining the congestion level includes: Statistical analysis of traffic density on different sections of highways is conducted to draw a cumulative probability distribution map. Different quantiles in the cumulative probability distribution map are used as the definition criteria to classify and evaluate traffic density according to the definition criteria, thereby obtaining the congestion level corresponding to different road sections.

4. The method according to claim 1, characterized in that, The process of fuzzifying the congestion identification indicators includes: Statistical analysis is performed on the congestion discrimination indicators in the congestion discrimination indicator pair, and cumulative probability distribution maps of the indicators are plotted respectively. Different quantiles in the cumulative probability distribution maps of the indicators are used as the indicator definition criteria. The congestion discrimination indicators are graded and evaluated according to the indicator definition criteria to obtain the fuzzy level of the congestion discrimination indicators.

5. The method according to claim 1, characterized in that, The process of obtaining the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of each rule's antecedent and the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of each rule's consequent includes: The reverse cloud generator processes the sample data corresponding to each rule in the congestion discrimination rule base to obtain the two-dimensional cloud parameters of the two-dimensional fuzzy cloud of the rule antecedent, wherein the two-dimensional cloud parameters include the expected value, entropy value, and hyperentropy value of different congestion discrimination indicators; the reverse cloud generator also processes the sample data corresponding to the congestion degree of each rule in the congestion discrimination rule base to obtain the one-dimensional cloud parameters of the one-dimensional fuzzy cloud of the consequent of each rule, wherein the one-dimensional cloud parameters include the expected value, entropy value, and hyperentropy value of the consequent.

6. The method according to claim 1, characterized in that, By statistically analyzing and calculating the proportion of vehicles traveling from the upstream road segment to the current road segment, the corresponding upstream congestion status impact factor is obtained.

7. A highway traffic congestion discrimination system based on two-dimensional fuzzy cloud and simultaneously incorporating upstream congestion status, characterized in that, Used to perform the method according to any one of claims 1-6.

Images