Highway traffic operation risk real-time identification method and identification system under ice and snow conditions based on dynamic multi-layer fuzzy logic

Abstract

The application relates to a real-time identification method and identification system for highway traffic operation risk under ice and snow conditions based on dynamic multi-layer fuzzy logic, and belongs to the field of traffic safety.The application aims to solve the problems that the existing method cannot process the relationship of various influence factors, cannot give a quantitative index of traffic operation risk, and cannot guarantee the robustness of the model under a new environment, leading to low accuracy and poor real-time performance of the identification of highway traffic operation risk under ice and snow conditions.The model prediction result has high reliability in the case of missing accident data and traffic conflict data.The application processes the quantitative problem of the operation risk index by constructing a multi-layer fuzzy logic structure.The intelligent rule generation module is added to improve the prediction accuracy and robustness of the system under different environments.

Description

TECHNICAL FIELD

[0001] The application relates to a real-time identification method and system for highway traffic operation risk under ice and snow conditions, and belongs to the field of traffic safety. BACKGROUND

[0002] Ice and snow weather as common adverse weather has a prominent influence on highway traffic safety. The highway under ice and snow conditions has complex traffic operation influencing factors, including meteorological conditions, road surface state conditions, linear conditions and traffic flow conditions and various conditions. In order to accurately predict the traffic operation risk under ice and snow conditions in the case of missing accident and conflict data, it is urgent to propose a real-time discrimination system for highway traffic operation risk under ice and snow conditions which comprehensively considers various influencing conditions. At present, the most prominent problem of the comprehensive risk prediction model is how to deal with the relationship between various influencing factors, give a quantitative index to the traffic operation risk and ensure the robustness of the model in the new environment, and the real-time identification method for highway traffic operation risk under ice and snow conditions based on dynamic multi-layer fuzzy logic can solve the above problems. SUMMARY

[0003] The purpose of the application is to solve the problems that the existing method cannot deal with the relationship between various influencing factors, cannot give a quantitative index to the traffic operation risk and cannot guarantee the robustness of the model in the new environment, resulting in low accuracy and poor real-time performance of the identification of highway traffic operation risk under ice and snow conditions, and to propose a real-time identification method and system for highway traffic operation risk under ice and snow conditions based on dynamic multi-layer fuzzy logic.

[0004] The specific process of the real-time identification method for highway traffic operation risk under ice and snow conditions based on dynamic multi-layer fuzzy logic is as follows:

[0005] Step one, laying detection equipment, obtaining data based on the laid detection equipment;

[0006] Step two, preprocessing the data obtained in step one to obtain preprocessed data;

[0007] Step three, determining the environmental input variable based on the data obtained in steps one and two, and determining the bridge variable based on the environmental input variable;

[0008] The environmental input variable includes: wet slip coefficient, snow cover thickness, ice layer thickness, downhill slope, horizontal curve radius, short-time flow rate, cross-section speed difference, snow cover thickness increase rate, wind speed, relative humidity and light intensity;

[0009] The bridge variable includes ice and snow road traffic operation stability index and visibility;

[0010] Step four, dividing the fuzzy set of the environmental input variable and the bridge variable and determining the membership function;

[0011] Step 5: After dividing the fuzzy sets and determining the membership functions in step 4, the environment input variables and bridge variables are used to formulate original state generation rules, and the formulated original state generation rules are screened to obtain a valid state generation rule base;

[0012] Step 6: Based on the bridging variables obtained by dividing the fuzzy sets and determining the membership functions in step 4 and the “traffic operation risk under icy and snowy conditions”, determine the risk discrimination rules;

[0013] Step 7

[0014] Obtain new environment input variables, and preprocess the new environment input variables according to step 2 to obtain preprocessed environment input variable data;

[0015] The fuzzy matching method based on the minimum membership principle is used to process the pre-processed new input environment input variable combination and the valid state generation rule screened out in step 5 to determine whether the input variable combination is covered by the valid state generation rule screened out in step 5;

[0016] If the new input environment input variable combination is covered by the valid state generation rule after screening in step 5, then the valid state generation rule obtained in step 5 is used to perform fuzzy reasoning on the new environment input variable data and output the predicted value of the bridge variable; the risk discrimination rule in step 6 is used to perform fuzzy reasoning and output the final traffic operation risk under ice and snow conditions;

[0017] If the new input environment input variable combination is not covered by the valid state generation rules screened in step five, a new state generation rule is generated through a data-driven method, and the new state generation rule is used to perform fuzzy reasoning on the new environment input variable data to output the bridge variable prediction value; the risk judgment rule in step six is ​​used to perform fuzzy reasoning to output the final traffic operation risk under ice and snow conditions.

[0018] The real-time identification system for highway traffic operation risks under ice and snow conditions based on dynamic multi-layer fuzzy logic is used to execute the real-time identification method for highway traffic operation risks under ice and snow conditions based on dynamic multi-layer fuzzy logic.

[0019] The beneficial effects of the present invention are:

[0020] The present invention incorporates as many influencing factors of the complex traffic environment under ice and snow conditions as possible into the prediction system, so that the model prediction results have a high degree of credibility in the absence of accident data and traffic conflict data. Specifically: the present invention takes into account as many factors as possible that may affect traffic safety in the complex traffic environment under ice and snow conditions, such as the slip coefficient, snow thickness, ice thickness, wind speed, relative humidity and light intensity. This comprehensive incorporation of multiple influencing factors enables the model to more accurately reflect the complexity of the real traffic environment. Since traffic accident data and conflict data are often difficult to collect comprehensively, the present invention supplements the lack of missing data through model structure design and data fusion, thereby ensuring that in the absence of complete accident or conflict data, the prediction results still have a high degree of credibility and reliability.

[0021] The present invention uses fuzzy logic to fuzzy process variables to solve the problem of the relationship between various influencing factors in the comprehensive risk prediction model. Specifically: traditional methods and other models usually need to find the coupling relationship or quantitative relationship between variables, and have limitations when dealing with complex relationships between multiple factors. The present invention uses the advantages of fuzzy logic to fuzzify the various influencing factors in the traffic environment and solve the problem of complex and nonlinear relationships between these influencing factors. Fuzzy logic can describe different variables in a more natural way through fuzzy sets and membership functions, thereby better simulating complex cause-and-effect relationships. In this way, the model can more flexibly handle the complex interactions between influencing factors and improve the accuracy of prediction results.

[0022] The present invention proposes to add "bridging variables" to the fuzzy logic structure, and solves the problem of quantifying operational risk indicators by constructing a multi-layer fuzzy logic structure. Specifically: in the process of traffic risk assessment, it is often difficult to directly quantify and calculate risk indicators, especially under complex environmental conditions; and using a single layer of fuzzy logic to directly predict risks from environmental input variables mainly relies on experience, which is also inaccurate. The present invention proposes to introduce the concept of "bridging variables" into the fuzzy logic structure, and connect and transform multiple interrelated risk factors by constructing a multi-level fuzzy logic structure. The bridging variables play a mediating and regulating role, so that the multi-layer fuzzy logic structure can effectively integrate variables at different levels, and ultimately give a more accurate risk quantification result. This design can not only reflect the relative importance of variables at different levels, but also dynamically adjust the risk assessment method under different conditions.

[0023] The existing traffic risk assessment system usually relies on a pre-defined rule base, which is difficult to adapt to the changing traffic environment and real-time input data. The present application adds an intelligent rule generation module to improve the prediction accuracy and robustness of the system in different environments. Specifically: a general fuzzy logic-based traffic risk assessment system is usually based on a pre-defined rule base, which is difficult to adapt to the changing traffic environment and real-time input data. By adding an intelligent rule generation module, the present application can dynamically update and generate new rules during system operation, making it better adapt to the current traffic environment. For example, using clustering analysis and machine learning methods, the rule base is updated in real time, enabling the system to respond quickly to changes in different weather conditions, traffic conditions, etc. The introduction of this intelligent module improves the prediction accuracy of the system in various environments and enhances its robustness, i.e. when facing data noise, outliers and sudden situations, the system can still maintain good performance and stability. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 Flowchart of the present application;

[0025] Figure 2 System module structure diagram of the present application;

[0026] Figure 3 Fuzzy logic structure diagram of the real-time fuzzy evaluation module;

[0027] Figure 4a Relative humidity membership function curve in the environmental input variable;

[0028] Figure 4b Illumination intensity membership function curve in the environmental input variable;

[0029] Figure 4c Snow cover thickness growth rate membership function curve in the environmental input variable;

[0030] Figure 4d Wind speed visibility membership function curve in the environmental input variable;

[0031] Figure 4e Visibility membership function curve;

[0032] Figure 5a Wet slip coefficient membership function curve in the environmental input variable;

[0033] Figure 5b Snow cover thickness membership function curve in the environmental input variable;

[0034] Figure 5c Ice layer thickness membership function curve in the environmental input variable;

[0035] Figure 5dThe membership function curve diagram of the downhill slope in the environmental input variable;

[0036] Figure 5e The membership function curve diagram of the curve radius in the environmental input variable;

[0037] Figure 5f The membership function curve diagram of the short-time flow rate in the environmental input variable;

[0038] Figure 5g The membership function curve diagram of the cross-section velocity difference in the environmental input variable;

[0039] Figure 5h The membership function curve diagram of the traffic operation stability index on the icy and snowy road surface;

[0040] Figure 6 The membership function curve diagram of the traffic operation risk under the icy and snowy condition;

[0041] Figure 7 The intelligent rule generation module diagram;

[0042] Figure 8 The cumulative frequency distribution curve diagram of the variable. DETAILED DESCRIPTION

[0043] Embodiment one: the specific process of the real-time identification method of the highway traffic operation risk under the icy and snowy condition based on the dynamic multi-layer fuzzy logic is as follows:

[0044] Step one, laying the detection equipment and obtaining the data based on the laid detection equipment;

[0045] Step two, pre-processing the data obtained in step one to obtain the pre-processed data;

[0046] Step three, determining the environmental input variable based on the data obtained in steps one and two and determining the bridge variable based on the environmental input variable;

[0047] The environmental input variable includes the wet and slippery coefficient, the snow cover thickness, the ice layer thickness, the downhill slope (which can be obtained by knowing the road alignment), the curve radius (which can be obtained by knowing the road alignment), the short-time flow rate, the cross-section velocity difference, the snow cover thickness increasing speed, the wind speed, the relative humidity, and the light intensity;

[0048] The bridge variable includes the traffic operation stability index on the icy and snowy road surface and the visibility;

[0049] Step four, dividing the fuzzy sets of the environmental input variable and the bridge variable and determining the membership functions;

[0050] Step five, the original state generation rule is formulated for the environment input variable and the bridge variable after the fuzzy set is divided and the membership function is determined in step four, and the effective state generation rule library is obtained by screening the original state generation rule;

[0051] Step six, the risk discrimination rule is determined based on the bridge variable and the traffic operation risk under ice and snow conditions after the fuzzy set is divided and the membership function is determined in step four;

[0052] Step seven,

[0053] The new environment input variable is obtained, and the new environment input variable is preprocessed according to step two to obtain the preprocessed environment input variable data;

[0054] The preprocessed new input environment input variable combination and the effective state generation rule screened in step five are processed by using the fuzzy matching method (existing mature method) of the minimum membership degree principle to determine whether the input variable combination is covered by the effective state generation rule screened in step five;

[0055] If the new input environment input variable combination is covered by the effective state generation rule screened in step five, the new environment input variable data is subjected to fuzzy reasoning (mamdani method) by using the screened effective state generation rule obtained in step five, and the bridge variable prediction value is output; the final traffic operation risk under ice and snow conditions is output by using the risk discrimination rule in step six for fuzzy reasoning (mamdani method);

[0056] If the new input environment input variable combination is not covered by the effective state generation rule screened in step five, a new state generation rule is generated by a data-driven method, the new environment input variable data is subjected to fuzzy reasoning (mamdani method) by using the new state generation rule, and the bridge variable prediction value is output; the final traffic operation risk under ice and snow conditions is output by using the risk discrimination rule in step six for fuzzy reasoning (mamdani method).

[0057] Specific implementation method two: the difference between the specific implementation method and the specific implementation method one is that the detection device is laid in step one, and data is obtained based on the laid detection device;

[0058] The specific process is as follows:

[0059] The detection device is laid in the highway bottleneck section (such as the bend-slope combined section) that needs to be monitored in real time under ice and snow conditions;

[0060] The detection device is respectively a traffic operation monitoring device, a small weather station and a road surface state detection device;

[0061] The types of data detected by traffic operation monitoring equipment include: traffic flow per minute, single vehicle speed;

[0062] The data types detected by the small weather station are: wind speed, relative humidity, and light intensity;

[0063] The data types detected by the road condition detection equipment are: slip coefficient, snow thickness, and ice thickness.

[0064] The data types that the device is required to be able to detect are shown in Table 6-1.

[0065] Traffic operation monitoring equipment: Equipment model MPD-N2

[0066] Two types of road condition detection equipment: Jinzhou Sunshine YGLM-Z2 and Shandong Tianhe TH-LM2;

[0067] There are two types of small weather stations: Tianhe Environmental TH-QC5 and Jinzhou Sunshine PC-8.

[0068] Other steps and parameters are the same as those in the first embodiment.

[0069] Specific embodiment three: This embodiment differs from specific embodiment one or two in that in step two, the data acquired in step one is preprocessed to obtain preprocessed data;

[0070] The specific process is:

[0071] The raw data is processed into a data type that can be input into the real-time fuzzy evaluation module, that is, the data is processed into the data type of the specified environmental input variable. The short-term flow rate counts the number of vehicles passing through a specific road section within a short period of time (taken as 15 minutes) and converts this number into hourly flow rate; the cross-sectional velocity difference is the speed difference between different vehicles at the same detection point; the snow thickness growth rate is the rate of change of snow thickness over a certain period of time;

[0072] Step 2.1: Calculate the short-term flow rate; the expression is:

[0073]

[0074] Where, Q is the short-term flow rate; Q 15 is the number of vehicles detected in each 15-minute observation window; T is the length of the observation window in minutes;

[0075] Step 22: Calculate the cross-sectional velocity difference; the expression is:

[0076]

[0077] Where, is the average speed of the detected vehicles; V iis the speed of the i-th vehicle detected; n is the total number of vehicles detected in the observation window; D i is the absolute value of the difference between the bicycle speed and the average speed; D is the cross-sectional velocity difference, in km / h;

[0078] Step 2 and 3: Calculate the snow thickness growth rate; the expression is:

[0079]

[0080] Where R is the snow thickness growth rate in cm / min; Δt is the time interval; S t is the snow thickness detected at time t; S t+Δt is the snow thickness detected at time t+Δt.

[0081] Other steps and parameters are the same as those in the first or second embodiment.

[0082] Specific embodiment 4: This embodiment differs from any one of specific embodiments 1 to 3 in that, in step 3, the environmental input variables are determined based on the data obtained in steps 1 and 2, and the bridge variables are determined based on the environmental input variables;

[0083] Environmental input variables include: slip coefficient, snow thickness, ice thickness, downhill slope (can be obtained by knowing the road alignment), flat curve radius (can be obtained by knowing the road alignment), short-term flow rate, cross-sectional velocity difference, snow thickness growth rate, wind speed, relative humidity, and light intensity;

[0084] The bridging variables include traffic operation stability index and visibility on icy and snowy roads;

[0085] To address the difficulty in quantifying the final prediction indicator, "Traffic Operation Risk under Snowy and Icy Conditions," a "bridging variable" was introduced as an intermediate-level indicator. Based on the data type obtained from processing actual historical data in the previous step, a first-level fuzzy logic was constructed, using environmental input variables to first predict the bridging variable. Because there is currently no single indicator that comprehensively reflects traffic operation status based on road surface conditions, road alignment, and traffic flow, a "Traffic Operation Stability Index on Snowy and Icy Roads" was defined to reflect traffic operation status under these combined conditions.

[0086] The specific process is:

[0087] Step 3.1. Calculate the traffic operation stability index on icy and snowy roads based on the slip coefficient, snow cover thickness, ice thickness, short-term flow rate, cross-sectional velocity difference, downhill slope, and flat curve radius; the expression is:

[0088] TSI-ISR=∑w i ×w SI

[0089] = 0.325 x w SC + 0.1621 x w ST + 0.1391 x w IT + 0.0742 x w STF + 0.0496 x w SD + 0.1 x w DS + 0.15 x w HCR

[0090] TSI-ISR is the traffic operation stability index on ice and snow road;

[0091] w i w is the weight (obtained by principal component analysis), and the values are 0.325, 0.1621, 0.1391, 0.0742, 0.0496, 0.1, and 0.15, respectively;

[0092] w SI w is the grading score, and the grading method is consistent with the fuzzy set division. According to the principle that the higher the stability, the higher the score, the grading score is given. For example, the wet slip coefficient is divided into 5 levels, and according to the wet slip coefficient from small to large (the larger the wet slip coefficient, the higher the stability), 0.6, 1.2, 1.8, 2.4, and 3 points are given, respectively. The downhill slope is divided into 3 levels, and according to the downhill slope from small to large (the higher the downhill slope, the lower the stability), 3, 2, and 1 points are given, respectively.

[0093] SI = SC, ST, IT, STF, SD, DS, HCR, respectively, are the wet slip coefficient, snow cover thickness, ice layer thickness, short-time flow rate, cross-section velocity difference, downhill slope, and horizontal curve radius;

[0094] Step three, visibility is obtained by recognizing the monitoring video taken by the traffic operation detection device through OpenCV (OpenCV is an open-source computer vision library that provides rich image processing and analysis functions. OpenCV can be used to process monitoring video frames, and the visibility can be estimated through image processing algorithms).

[0095] After the rules, the visibility is determined based on the snow cover thickness increase rate, wind speed, relative humidity, and light intensity.

[0096] The other steps and parameters are the same as one of the first to third embodiments.

[0097] Embodiment five: this embodiment is different from one of the first to fourth embodiments in that the step four divides the fuzzy set for the environmental input variables and the bridge variables and determines the membership function;

[0098] The specific process is as follows:

[0099] Step four, divide the fuzzy sets of environmental input variables and bridge variables; the specific process is:

[0100] The environmental input variables include: wet skid resistance coefficient, snow cover thickness, ice layer thickness, downhill slope, horizontal curve radius, short-term flow rate, cross-section velocity difference, snow cover thickness increasing rate, wind speed, relative humidity, and light intensity;

[0101] The bridge variables include: ice and snow road traffic operation stability index and visibility;

[0102] Draw the cumulative frequency curves of wet skid resistance coefficient, snow cover thickness, ice layer thickness, downhill slope, horizontal curve radius, short-term flow rate, cross-section velocity difference, snow cover thickness increasing rate, wind speed, relative humidity, light intensity, ice and snow road traffic operation stability index, and visibility, respectively.

[0103] For the variables divided into 5 fuzzy sets, take 15%, 30%, 50%, and 85% of the cumulative frequency curve as the reference critical value, and adjust and determine it in combination with experience, and divide the variables into 5 fuzzy sets;

[0104] For the variables divided into 3 fuzzy sets, take 15% and 85% of the cumulative frequency curve as the reference critical value, and adjust and determine it in combination with experience, and divide the variables into 3 fuzzy sets;

[0105] The variables divided into 5 fuzzy sets are: wet skid resistance coefficient, snow cover thickness, ice layer thickness, horizontal curve radius, cross-section velocity difference, snow cover thickness increasing rate, wind speed, relative humidity, light intensity, ice and snow road traffic operation stability index, and visibility;

[0106] The 5 fuzzy sets from small to large are S, NS, M, PB, and B in turn;

[0107] The variables divided into 3 fuzzy sets are: downhill slope and short-term flow rate;

[0108] The 3 fuzzy sets from small to large are S, M, and B in turn.

[0109] The cumulative frequency curve diagram is shown in Figure 7 .

[0110] The fuzzy set range is determined according to the five or three sets divided by the boundary value. The specific division of the fuzzy set is shown in Table 1.

[0111] The most important feature of the membership function of the variable is the division of the fuzzy set, which is one of the main factors affecting the prediction result.

[0112] The membership function level and function type of the environmental input variable are shown in Table 2 and Table 3, and the membership function level and function type of the bridge variable are shown in Table 4.

[0113] Table 1 fuzzy set partition of variables

[0114]

[0115]

[0116]

[0117] Step four two, determine membership function.

[0118] Other steps and parameters are the same as one of the first four embodiments.

[0119] Embodiment six: the difference between this embodiment and one of the first five embodiments is that the membership function is determined in step four two (determine the type of function); the specific process is:

[0120] The number of membership functions of each variable is the same as the number of fuzzy sets partitioned, such as Figure 8 There are five membership functions for the wet slip coefficient, and there are five membership functions in total because it is divided into five fuzzy sets;

[0121] The variable is an environmental input variable and a bridge variable;

[0122] If it is the first or last membership function of the variable, the membership function is trapmf or zmf;

[0123] The trapmf is a trapezoidal membership function, and the zmf is a Z-shaped membership function;

[0124] If the data boundary is clear, resulting in fewer transition parts or the fuzzy set needs to be explicitly upper and lower bounded, then select trapmf (trapezoidal membership function); if the data shows a gradual change trend, and needs a more delicate gradual change description, then select zmf (Z-shaped membership function);

[0125] If it is the middle membership function of the variable (the membership function between the first and last membership functions), the membership function is trimf or gaussmf, if the data boundary is clear, select trimf (triangular membership function); if a smooth transition is needed, select gaussmf (Gaussian membership function);

[0126] The trimf is a triangular membership function; and the gaussmf is a Gaussian membership function;

[0127] Table 2 membership function level and function type of environmental input variable

[0128]

[0129]

[0130] Table 3 Membership function levels and function types of environmental input variables

[0131]

[0132] Table 4 Membership function levels and function types of bridging variables

[0133]

[0134]

[0135] Other steps and parameters are the same as one of the first to fifth embodiments.

[0136] The seventh embodiment is different from one of the first to sixth embodiments in that the step five formulates original state generation rules for the environmental input variables and the bridging variables after the step four divides the fuzzy sets and determines the membership functions, and the step five screens the formulated original state generation rules to obtain an effective state generation rule library.

[0137] The specific process is as follows:

[0138] The combination of the environmental input variables and the bridging variables after the step four divides the fuzzy sets and determines the membership functions at each time (the variable combination generated by the data detected at the same time after the calculation and processing of the step two and the step three) is taken as an original state generation rule;

[0139] The weight of the stability index rule of the original state generation rule is calculated; for the original rules with the same environmental input variable combination but different output bridging variable levels, the original state generation rule with the largest weight in the stability index rule is retained as an effective state generation rule, and the rest of the original state generation rules are discarded;

[0140] The weight of the visibility rule of the original state generation rule is calculated; for the original rules with the same environmental input variable combination but different output bridging variable levels, the original state generation rule with the largest weight in the visibility rule is retained as an effective state generation rule, and the rest of the original state generation rules are discarded;

[0141] The weight expression of the stability index rule of the original state generation rule is as follows:

[0142]

[0143] In the formula, w is the weight of the stability index rule of the ith original state generation rule; ∧ is the minimum operator, which means selecting the minimum value. ​The maximum membership value of the slippery coefficient of the original state generation rule for the i-th original state (the membership value can be known by knowing the membership function type); The maximum membership value of the snow thickness of the rule generating the original state of the i-th item; The maximum membership value of the ice thickness for the original state generation rule of the i-th item; The maximum membership of the downhill slope of the i-th original state generation rule; The maximum membership value of the flat curve radius of the original state generation rule for the i-th state; The maximum membership value of the short-term flow rate of the i-th original state generation rule; The maximum membership value of the cross-section velocity difference of the i-th original state generation rule;

[0144] The weight expression of the visibility rule of the original state generation rule is:

[0145]

[0146] Where, is the weight of the visibility rule of the original state generation rule of the i-th state; ∧ is the minimum operator, which means selecting the minimum value; The maximum membership value of the snow thickness growth rate of the i-th original state generation rule; The maximum membership value of the wind speed of the i-th original state generation rule; The maximum membership value of the relative humidity of the i-th original state generation rule; The maximum membership degree of the illumination intensity of the i-th original state generation rule;

[0147] Using the data collected by the deployed traffic operation monitoring equipment, meteorological detection equipment and road condition detection equipment, combined with video surveillance data, the snowfall period data from November to March was screened out. After cleaning the abnormal data, all data were processed into the data type of the environmental input variable; all data were divided into corresponding levels according to the fuzzy set division method of the membership function in step 4. The fuzzy sets correspond to S, NS, M, PB, and B levels from small to large (if divided into 3 fuzzy sets, they correspond to S, M, and B levels from small to large). The level combination at each corresponding moment is an original Fuzzy rules such as "If the slip coefficient is large, the snow thickness is small, the ice thickness is medium, the downhill slope is medium, the flat curve radius is large, the short-term flow rate is medium, and the cross-sectional velocity difference is medium, then the traffic operation stability index of icy and snowy roads is large" and "If the snow thickness growth rate is large, the wind speed is high, the relative humidity is high, and the light intensity is low, then the visibility is low" were selected. For the original rules with the same input conditions but different output conditions, the weight of the rules was used as a reference and combined with experience to screen them. 244 state generation rules for predicting the traffic operation stability index of icy and snowy roads and 253 state generation rules for predicting visibility were obtained.

[0148] After obtaining all the original state generation rules, the rules are filtered by their weights. The filtering objects are rules with the same input variable conditions but different output variable levels. Among these rules, the one with the largest weight is retained and the rest are discarded.

[0149] The weight expression of the stability index rule of the original state generation rule is:

[0150]

[0151] Where, is the weight of the stability index rule of the i-th original state generation rule; ∧ is the minimum operator, which means selecting the minimum value; The maximum membership value of the slippery coefficient of the original state generation rule for the i-th original state (the membership value can be known by knowing the membership function type); The maximum membership value of the snow thickness of the rule generating the original state of the i-th item; The maximum membership value of the ice thickness for the original state generation rule of the i-th item; The maximum membership of the downhill slope of the i-th original state generation rule; The maximum membership value of the flat curve radius of the original state generation rule for the i-th state; The maximum membership value of the short-term flow rate of the i-th original state generation rule; The maximum membership value of the cross-section velocity difference of the i-th original state generation rule;

[0152] After screening, there are 244 valid fuzzy rules, as shown in Table 5.

[0153] The weight expression of the visibility rule of the original state generation rule is:

[0154]

[0155] Where, is the weight of the visibility rule of the original state generation rule of the i-th state; ∧ is the minimum operator, which means selecting the minimum value; The maximum membership value of the snow thickness growth rate of the i-th original state generation rule; The maximum membership value of the wind speed of the i-th original state generation rule; The maximum membership value of the relative humidity of the i-th original state generation rule; The maximum membership degree of the illumination intensity of the i-th original state generation rule;

[0156] After screening, there are 253 valid fuzzy rules, as shown in Table 6.

[0157] Table 5 Rules for generating the traffic operation stability index state on snowy and icy roads

[0158]

[0159]

[0160] Table 6 Visibility status generation rules

[0161]

[0162] The other steps and parameters are the same as those in the first to sixth embodiments.

[0163] Specific embodiment eight: This embodiment differs from any one of specific embodiments one to seven in that, in step six, the risk discrimination rule is determined based on the bridge variables and the “traffic operation risk under icy and snowy conditions” after dividing the fuzzy set and determining the membership function in step four;

[0164] The specific process is:

[0165] Step 6.1: Determine the "risk level of traffic operation under icy and snowy conditions"; the specific process is as follows:

[0166] The traffic operation risk level under ice and snow conditions is set to a real number ranging from 0 to 5, and the fuzzy set of traffic operation risk levels under ice and snow conditions is divided into 5 levels, from small to large, 0-1, 1-2, 2-3, 3-4, and 4-5;

[0167] The membership function type corresponding to 0-1 is trapmf, and the corresponding level is S;

[0168] The membership function type corresponding to 1-2 is trimf, and the corresponding level is NS;

[0169] The membership function type corresponding to 2-3 is trimf, and the corresponding level is M;

[0170] 3-4 corresponds to the membership function type of trimf and the corresponding level is PB;

[0171] The membership function type corresponding to 4-5 is trapmf, and the corresponding level is B;

[0172] Step 62: Using a matrix-based rule-making method, combine each level of the bridge variable after partitioning the fuzzy set and determining the membership function in Step 4 into a two-dimensional matrix (25 types). Map the two-dimensional matrix to the "traffic operation risk level under icy and snowy conditions" determined in Step 61 to obtain the risk discrimination rules. The specific process is as follows:

[0173] When the traffic operation stability index level on icy and snowy roads is S and the visibility level is S, the traffic operation risk level under icy and snowy conditions is B;

[0174] When the traffic operation stability index level on icy and snowy roads is S and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is B;

[0175] When the traffic operation stability index level on icy and snowy roads is S and the visibility level is M, the traffic operation risk level under icy and snowy conditions is PB;

[0176] When the traffic operation stability index level on icy and snowy roads is S and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is PB;

[0177] When the traffic operation stability index level on icy and snowy roads is S and the visibility level is B, the traffic operation risk level under icy and snowy conditions is M;

[0178] When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is S, the traffic operation risk level under icy and snowy conditions is B;

[0179] When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is B;

[0180] When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is M, the traffic operation risk level under icy and snowy conditions is PB;

[0181] When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is M;

[0182] When the traffic operation stability index grade of icy and snowy road surface is NS, and the visibility grade is B, the traffic operation risk grade under icy and snowy conditions is NS;

[0183] When the traffic operation stability index grade of icy and snowy road surface is M, and the visibility grade is S, the traffic operation risk grade under icy and snowy conditions is PB;

[0184] When the traffic operation stability index grade of icy and snowy road surface is M, and the visibility grade is NS, the traffic operation risk grade under icy and snowy conditions is PB;

[0185] When the traffic operation stability index grade of icy and snowy road surface is M, and the visibility grade is M, the traffic operation risk grade under icy and snowy conditions is M;

[0186] When the traffic operation stability index grade of icy and snowy road surface is M, and the visibility grade is PB, the traffic operation risk grade under icy and snowy conditions is NS;

[0187] When the traffic operation stability index grade of icy and snowy road surface is M, and the visibility grade is B, the traffic operation risk grade under icy and snowy conditions is S;

[0188] When the traffic operation stability index grade of icy and snowy road surface is PB, and the visibility grade is S, the traffic operation risk grade under icy and snowy conditions is PB;

[0189] When the traffic operation stability index grade of icy and snowy road surface is PB, and the visibility grade is NS, the traffic operation risk grade under icy and snowy conditions is M;

[0190] When the traffic operation stability index grade of icy and snowy road surface is PB, and the visibility grade is M, the traffic operation risk grade under icy and snowy conditions is NS;

[0191] When the traffic operation stability index grade of icy and snowy road surface is PB, and the visibility grade is PB, the traffic operation risk grade under icy and snowy conditions is S;

[0192] When the traffic operation stability index grade of icy and snowy road surface is PB, and the visibility grade is B, the traffic operation risk grade under icy and snowy conditions is S;

[0193] When the traffic operation stability index grade of icy and snowy road surface is B, and the visibility grade is S, the traffic operation risk grade under icy and snowy conditions is M;

[0194] When the traffic operation stability index grade of icy and snowy road surface is B, and the visibility grade is NS, the traffic operation risk grade under icy and snowy conditions is NS;

[0195] When the traffic operation stability index grade of icy and snowy road surface is B, and the visibility grade is M, the traffic operation risk grade under icy and snowy conditions is S;

[0196] When the traffic operation stability index level on icy and snowy roads is B and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is S;

[0197] When the traffic operation stability index level on icy and snowy roads is B and the visibility level is B, the traffic operation risk level under icy and snowy conditions is S;

[0198] The fuzzy set partitioning and membership functions for the bridge variables in this layer of fuzzy logic are the same as those in the first layer. Since the final output variable, "Risk of Traffic Operation under Snowy Conditions," cannot be quantified, its fuzzy set range is set to a real number from 0 to 5, with 0 to 5 representing increasing risk. The membership functions and parameters for the risk of traffic operation under snowy conditions are shown in Table 7. The risk discrimination rules at this layer are determined empirically, such as "If visibility is low and the traffic operation stability index on snowy roads is low, then the risk of traffic operation under snowy conditions is high." This layer has a total of 25 risk discrimination rules.

[0199] How to determine parameters and rules, how to get predicted values

[0200] ① Determine the membership functions of input and output variables

[0201] Membership function type:

[0202] The input variables are the traffic operation stability index and visibility on icy and snowy roads, which are the output variables of the previous layer of fuzzy logic, and their membership function types are the same as before.

[0203] The output variable, traffic operation risk, is a real number with a numerical range of 0–5. Its fuzzy set is divided into five values: 0–1, 1–2, 2–3, 3–4, and 4–5. The corresponding five membership function types are determined using the same method as in step 4. The final selected function types are trapmf, trimf, trimf, trimf, and trapmf.

[0204] Table 7 Membership functions and parameters of traffic operation risk under icy and snowy conditions

[0205]

[0206] ②Determine the rules

[0207] When predicting traffic operation risks, the fuzzy logic input variables are divided into five fuzzy sets, forming 25 input combinations. Subsequently, as shown in Table 8, a matrix-based rule-making approach is employed to intuitively map input variable combinations to risk levels using a two-dimensional matrix. The determination of each risk level is also informed by empirical evidence.

[0208] Table 8 Fuzzy rule mapping matrix for predicting traffic operation risks

[0209]

[0210] Table 9 Risk Identification Rules

[0211]

[0212]

[0213] ③Get the predicted value

[0214] The type and parameters of the membership function, along with the rules, were manually entered into the MATLAB software. The software then used the Mamdani method for fuzzy inference and the centroid method (which calculates the centroid of the area under the output membership function curve to determine the precise output value) to defuzzify the data and directly output the risk prediction value.

[0215] The above steps 1 to 6 are for establishing the main functional module of the system - "real-time fuzzy evaluation module based on bridge variables".

[0216] The other steps and parameters are the same as those in the first to seventh embodiments.

[0217] Specific embodiment 9: This embodiment differs from any one of specific embodiments 1 to 8 in that, if there is no overriding in step 7, a new state generation rule is generated by a data-driven method; the specific process is as follows:

[0218] 1) Merge the new environment input variable with the environment input variable data in the original state generation rule base in step 5 to obtain merged environment input variable data, and obtain a bridge variable data set based on the bridge variable of the original state generation rule in step 5 (the bridge variable data set of the merged environment input variable data is still the original bridge variable data set). The environment input variable and the bridge variable of the original state generation rule in step 5 still correspond to each other as a complete rule;

[0219] The merged environmental input variable data and the original state generation rule bridge variables are converted into numerical variables (the input variable combination and the corresponding bridge variable are a rule, and each rule is still a group after the conversion). Different levels are mapped to different values, as shown in Table 10;

[0220] The conversion of the merged environment input variable data and the original state generation rule bridge variable into a numerical variable is specifically as follows:

[0221] When the level of the fuzzy set corresponding to the merged environmental input variable is S, the merged environmental input variable is converted to a value of 0;

[0222] When the level of the fuzzy set corresponding to the merged environmental input variable is NS, the merged environmental input variable is converted to a value of 1;

[0223] When the level of the fuzzy set corresponding to the merged environmental input variable is M, the merged environmental input variable is converted to a value of 2;

[0224] When the level of the fuzzy set corresponding to the merged environmental input variable is PB, the merged environmental input variable is converted to a value of 3;

[0225] When the level of the fuzzy set corresponding to the merged environmental input variable is B, the merged environmental input variable is converted to a value of 4;

[0226] When the rank of the fuzzy set corresponding to the bridge variable is S, the bridge variable is converted to the value 0;

[0227] When the rank of the fuzzy set corresponding to the bridge variable is NS, the bridge variable is converted to the value 1;

[0228] When the level of the fuzzy set corresponding to the bridge variable is M, the bridge variable is converted to the value 2;

[0229] When the level of the fuzzy set corresponding to the bridge variable is PB, the bridge variable is converted to the value 3;

[0230] When the fuzzy set corresponding to the bridge variable has a level of B, the bridge variable is converted to a value of 4;

[0231] Table 10 Label encoding of rules converted into numerical variables

[0232]

[0233] 2) Apply K-means clustering to the environmental input variable combinations converted into numerical values, and divide the data points into K clusters, each cluster corresponds to a centroid (each centroid is a combination of environmental input variables);

[0234] 3) Generate output conditions based on the centroids of each cluster containing the new input environment variable combination (each centroid is an input variable combination), and use the output conditions as the output conditions of the new fuzzy rule corresponding to the new input environment variable combination; the specific process is:

[0235] 31) Calculate the individual support Support(A); the expression is:

[0236]

[0237] Calculate the support of joint occurrence Support(A∪B); the expression is:

[0238]

[0239] Wherein, A represents the environmental input variable combination represented by the centroid of the cluster containing the new input environmental variable combination, B represents the output condition corresponding to A in the original state generation rule of step five; the total number of transactions represents the number of environmental input variable combinations converted into numerical values (i.e. the number of original unfiltered state generation rules);

[0240] 32), calculate the confidence degree Confidence(A→B); the expression is:

[0241]

[0242] The confidence degree represents the probability that condition B appears when condition A appears; the confidence degree of deriving a certain output condition B based on a specific input condition combination A; if the confidence degree is high, it means that the input combination is closely related to the output condition, thereby determining the output condition under the combination;

[0243] 33), retain the output condition B with the highest confidence degree as the output condition of the new fuzzy rule corresponding to the new input environmental variable combination.

[0244] The other steps and parameters are the same as one of the first to eighth embodiments.

[0245] The tenth embodiment is based on the dynamic multi-layer fuzzy logic-based real-time identification system for highway traffic operation risk under ice and snow conditions for executing the dynamic multi-layer fuzzy logic-based real-time identification method for highway traffic operation risk under ice and snow conditions.

[0246] The present application can also have other various embodiments, and those skilled in the art can make various corresponding changes and modifications according to the present application without departing from the spirit and essence of the present application, but these corresponding changes and modifications should all belong to the protection scope of the claims attached to the present application.

Claims

1. A real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic, characterized by: The specific process of the method is: Step 1: deploy testing equipment and obtain data based on the deployed testing equipment; Step 2: preprocess the data obtained in step 1 to obtain preprocessed data; Step 3: determining environmental input variables based on the data obtained in Step 1 and Step 2, and determining bridge variables based on the environmental input variables; Environmental input variables include: slip coefficient, snow thickness, ice thickness, downhill slope, flat curve radius, short-term flow rate, cross-sectional velocity difference, snow thickness growth rate, wind speed, relative humidity, and light intensity; The bridging variables include traffic operation stability index and visibility on icy and snowy roads; Step 4: Divide the environmental input variables and bridge variables into fuzzy sets and determine the membership functions; Step 5: After dividing the fuzzy sets and determining the membership functions in step 4, the environment input variables and bridge variables are used to formulate original state generation rules, and the formulated original state generation rules are screened to obtain a valid state generation rule base; Step 6: Based on the bridging variables obtained by dividing the fuzzy sets and determining the membership functions in step 4 and the "traffic operation risk under icy and snowy conditions", determine the risk discrimination rules; Step 7 Obtain new environment input variables, and preprocess the new environment input variables according to step 2 to obtain preprocessed environment input variable data; The fuzzy matching method based on the minimum membership principle is used to process the pre-processed new input environment input variable combination and the valid state generation rule screened out in step 5 to determine whether the input variable combination is covered by the valid state generation rule screened out in step 5; If the new input environment input variable combination is covered by the valid state generation rule after screening in step 5, then the valid state generation rule obtained in step 5 is used to perform fuzzy reasoning on the new environment input variable data and output the predicted value of the bridge variable; the risk discrimination rule in step 6 is used to perform fuzzy reasoning and output the final traffic operation risk under ice and snow conditions; If the new input environment input variable combination is not covered by the valid state generation rules screened in step five, a new state generation rule is generated through a data-driven method, and the new state generation rule is used to perform fuzzy reasoning on the new environment input variable data to output the bridge variable prediction value; the risk judgment rule in step six is ​​used to perform fuzzy reasoning to output the final traffic operation risk under ice and snow conditions.

2. The real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 1 is characterized by: In step 1, detection equipment is deployed, and data is acquired based on the deployed detection equipment. The specific process is as follows: Deploy detection equipment on highways where real-time monitoring of traffic operation risks under icy and snowy conditions is required; The testing equipment includes traffic operation monitoring equipment, weather stations and road condition testing equipment; The types of data detected by traffic operation monitoring equipment include: traffic flow per minute, single vehicle speed; The data types detected by the weather station are: wind speed, relative humidity, and light intensity; The data types detected by the road condition detection equipment are: slip coefficient, snow thickness, and ice thickness.

3. The real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 2 is characterized by: In the step 2, the data obtained in the step 1 is preprocessed to obtain preprocessed data; the specific process is: Step 2.1: Calculate the short-term flow rate; the expression is: Where, Q is the short-term flow rate; Q 15 is the number of vehicles detected in each 15-minute observation window; T is the length of the observation window in minutes; Step 22: Calculate the cross-sectional velocity difference; the expression is: Where, is the average speed of the detected vehicles; V i is the speed of the i-th vehicle detected; n is the total number of vehicles detected in the observation window; D i is the absolute value of the difference between the bicycle speed and the average speed; D is the cross-sectional velocity difference, in km / h; Step 2 and 3: Calculate the snow thickness growth rate; the expression is: Where R is the snow thickness growth rate in cm / min; Δt is the time interval; S t is the snow thickness detected at time t; S t+Δt is the snow thickness detected at time t+Δt.

4. The real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 3 is characterized by: In the step 3, the environmental input variables are determined based on the data obtained in the steps 1 and 2, and the bridging variables are determined based on the environmental input variables; Environmental input variables include: slip coefficient, snow thickness, ice thickness, downhill slope, flat curve radius, short-term flow rate, cross-sectional velocity difference, snow thickness growth rate, wind speed, relative humidity, and light intensity; The bridging variables include traffic operation stability index and visibility on icy and snowy roads; The specific process is: Step 3.

1. Calculate the traffic operation stability index on icy and snowy roads based on the slip coefficient, snow cover thickness, ice thickness, short-term flow rate, cross-sectional velocity difference, downhill slope, and flat curve radius; the expression is: TSI-ISR=∑w i ×w SI =0.325×w SC +0.1621×w ST +0.1391×w IT +0.0742×w STF +0.0496×w SD +0.1×w DS +0.15×w HCR Among them, TSI-ISR is the traffic operation stability index on icy and snowy roads; w i is the weight, and its values ​​are 0.325, 0.1621, 0.1391, 0.0742, 0.0496, 0.1, and 0.15 respectively; w SI to score the grading; SI = SC, ST, IT, STF, SD, DS, HCR, respectively, are the slip coefficient, snow thickness, ice thickness, short-term flow rate, cross-sectional velocity difference, downhill slope, and flat curve radius; Step 3.2: Visibility: Use OpenCV to identify the surveillance video captured by the traffic operation detection equipment to obtain visibility data.

5. The real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 4 is characterized by: In step 4, the environmental input variables and the bridge variables are divided into fuzzy sets and the membership functions are determined; the specific process is: Step 4.1: Divide the environmental input variables and bridge variables into fuzzy sets; the specific process is as follows: Environmental input variables include: slip coefficient, snow thickness, ice thickness, downhill slope, flat curve radius, short-term flow rate, cross-sectional velocity difference, snow thickness growth rate, wind speed, relative humidity, and light intensity; The bridging variables include: traffic operation stability index and visibility on icy and snowy roads; Draw the slip coefficient, snow thickness, ice thickness, downhill slope, flat curve radius, short-term flow rate, cross-sectional velocity difference, snow thickness growth rate, wind speed, relative humidity, light intensity, icy road traffic operation stability index, and visibility cumulative frequency curve respectively; For variables divided into 5 fuzzy sets, the 15%, 30%, 50%, and 85% positions on the cumulative frequency curve are used as reference critical values ​​to divide the variables into 5 fuzzy sets; For variables divided into three fuzzy sets, the 15% and 85% positions on the cumulative frequency curve are used as reference critical values ​​to divide the variables into three fuzzy sets; The variables divided into five fuzzy sets are: slip coefficient, snow thickness, ice thickness, flat curve radius, cross-sectional speed difference, snow thickness growth rate, wind speed, relative humidity, light intensity, icy road traffic operation stability index and visibility; The five fuzzy sets are S, NS, M, PB, and B from smallest to largest; The variables divided into three fuzzy sets are: downhill slope and short-term flow rate; The three fuzzy sets are S, M, and B from smallest to largest; Step 42: Determine the membership function.

6. The real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 5 is characterized by: The membership function is determined in step 42; the specific process is: The number of membership functions of each variable is the same as the number of fuzzy sets divided by the variable; The variables are environmental input variables and bridge variables; If it is the first or last membership function of a variable, the membership function is trapmf or zmf; The trapmf is a trapezoidal membership function, and zmf is a Z-type membership function; If it is an intermediate membership function of a variable, the membership function is trimf or gaussmf; The trimf is a triangle membership function; gaussmf is a Gaussian membership function.

7. The method for real-time identification of highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 6 is characterized by: In the step 5, the environment input variables and bridge variables after the fuzzy sets are divided and the membership functions are determined in the step 4 are formulated into original state generation rules, and the formulated original state generation rules are screened to obtain a valid state generation rule library; The specific process is: The combination of the environmental input variables and bridge variables at each moment after the fuzzy sets are divided and the membership functions are determined in step 4 is used as an original state generation rule; Calculate the weight of the stability index rule of the original state generation rule; for the original rules with exactly the same combination of environmental input variables but different levels of output bridge variables, retain the original state generation rule with the largest weight among the stability index rules as the valid state generation rule, and discard the remaining original state generation rules; Calculate the weight of the visibility rules of the original state generation rules; for original rules with exactly the same combination of environmental input variables but different output bridge variable levels, retain the original state generation rule with the largest weight among the visibility rules as the effective state generation rule, and discard the remaining original state generation rules; The weight expression of the stability index rule of the original state generation rule is: Where, is the weight of the stability index rule of the i-th original state generation rule; ∧ is the minimum operator; The maximum membership value of the slip coefficient of the original state generation rule for the i-th item; The maximum membership value of the snow thickness of the rule generating the original state of the i-th item; The maximum membership value of the ice thickness for the original state generation rule of the i-th item; The maximum membership of the downhill slope of the i-th original state generation rule; The maximum membership value of the flat curve radius of the original state generation rule for the i-th state; The maximum membership value of the short-term flow rate of the i-th original state generation rule; The maximum membership value of the cross-section velocity difference of the i-th original state generation rule; The weight expression of the visibility rule of the original state generation rule is: Where, is the weight of the visibility rule of the original state generation rule of the i-th state; ∧ is the minimum operator; The maximum membership value of the snow thickness growth rate of the i-th original state generation rule; The maximum membership value of the wind speed of the i-th original state generation rule; The maximum membership value of the relative humidity of the i-th original state generation rule; The maximum membership degree of the illumination intensity of the i-th original state generation rule.

8. The method for real-time identification of highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 7 is characterized by: In step 6, the risk discrimination rule is determined based on the bridge variables and "traffic operation risk under ice and snow conditions" after dividing the fuzzy set and determining the membership function in step 4; The specific process is: Step 6.1: Determine the "Risk Level of Traffic Operation under Snow and Ice Conditions"; the specific process is as follows: The traffic operation risk level under ice and snow conditions is set to a real number ranging from 0 to 5, and the fuzzy set of traffic operation risk levels under ice and snow conditions is divided into 5 levels, from small to large, 0-1, 1-2, 2-3, 3-4, and 4-5; The membership function type corresponding to 0-1 is trapmf, and the corresponding level is S; The membership function type corresponding to 1-2 is trimf, and the corresponding level is NS; The membership function type corresponding to 2-3 is trimf, and the corresponding level is M; 3-4 corresponds to the membership function type of trimf and the corresponding level is PB; The membership function type corresponding to 4-5 is trapmf, and the corresponding level is B; Step 62: Using a matrix-based rule-making method, combine each level of the bridge variable after partitioning the fuzzy set and determining the membership function in Step 4 into a two-dimensional matrix. Map the two-dimensional matrix to the "traffic operation risk level under icy and snowy conditions" determined in Step 61 to obtain the risk discrimination rule. The specific process is as follows: When the traffic operation stability index level on icy and snowy roads is S and the visibility level is S, the traffic operation risk level under icy and snowy conditions is B; When the traffic operation stability index level on icy and snowy roads is S and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is B; When the traffic operation stability index level on icy and snowy roads is S and the visibility level is M, the traffic operation risk level under icy and snowy conditions is PB; When the traffic operation stability index level on icy and snowy roads is S and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is PB; When the traffic operation stability index level on icy and snowy roads is S and the visibility level is B, the traffic operation risk level under icy and snowy conditions is M; When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is S, the traffic operation risk level under icy and snowy conditions is B; When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is B; When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is M, the traffic operation risk level under icy and snowy conditions is PB; When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is M; When the traffic operation stability index level on icy and snowy roads is NS and the visibility level is B, the traffic operation risk level under icy and snowy conditions is NS; When the traffic operation stability index level on icy and snowy roads is M and the visibility level is S, the traffic operation risk level under icy and snowy conditions is PB; When the traffic operation stability index level on icy and snowy roads is M and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is PB; When the traffic operation stability index level on icy and snowy roads is M and the visibility level is M, the traffic operation risk level under icy and snowy conditions is M; When the traffic operation stability index level on icy and snowy roads is M and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is NS; When the traffic operation stability index level on icy and snowy roads is M and the visibility level is B, the traffic operation risk level under icy and snowy conditions is S; When the traffic operation stability index level on icy and snowy roads is PB and the visibility level is S, the traffic operation risk level under icy and snowy conditions is PB; When the traffic operation stability index level on icy and snowy roads is PB and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is M; When the traffic operation stability index level on icy and snowy roads is PB and the visibility level is M, the traffic operation risk level under icy and snowy conditions is NS; When the traffic operation stability index level on icy and snowy roads is PB and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is S; When the traffic operation stability index level on icy and snowy roads is PB and the visibility level is B, the traffic operation risk level under icy and snowy conditions is S; When the traffic operation stability index level on icy and snowy roads is B and the visibility level is S, the traffic operation risk level under icy and snowy conditions is M; When the traffic operation stability index level on icy and snowy roads is B and the visibility level is NS, the traffic operation risk level under icy and snowy conditions is NS; When the traffic operation stability index level on icy and snowy roads is B and the visibility level is M, the traffic operation risk level under icy and snowy conditions is S; When the traffic operation stability index level on icy and snowy roads is B and the visibility level is PB, the traffic operation risk level under icy and snowy conditions is S; When the traffic operation stability index level on icy and snowy roads is B and the visibility level is B, the traffic operation risk level under icy and snowy conditions is S.

9. The real-time identification method for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic according to claim 8 is characterized by: If there is no overwriting in step 7, a new state generation rule is generated by a data-driven method; the specific process is: 1) Merging the new environment input variable with the environment input variable data in the original state generation rule base in step 5 to obtain merged environment input variable data, and obtaining a data set of bridge variables based on the bridge variables of the original state generation rule in step 5; Convert the merged environment input variable data and the original state generation rule bridge variables into numerical variables; The conversion of the merged environment input variable data and the original state generation rule bridge variable into a numerical variable is specifically as follows: When the level of the fuzzy set corresponding to the merged environmental input variable is S, the merged environmental input variable is converted to a value of 0; When the level of the fuzzy set corresponding to the merged environmental input variable is NS, the merged environmental input variable is converted to a value of 1; When the level of the fuzzy set corresponding to the merged environmental input variable is M, the merged environmental input variable is converted to a value of 2; When the level of the fuzzy set corresponding to the merged environmental input variable is PB, the merged environmental input variable is converted to a value of 3; When the level of the fuzzy set corresponding to the merged environmental input variable is B, the merged environmental input variable is converted to a value of 4; When the rank of the fuzzy set corresponding to the bridge variable is S, the bridge variable is converted to the value 0; When the rank of the fuzzy set corresponding to the bridge variable is NS, the bridge variable is converted to the value 1; When the level of the fuzzy set corresponding to the bridge variable is M, the bridge variable is converted to the value 2; When the level of the fuzzy set corresponding to the bridge variable is PB, the bridge variable is converted to the value 3; When the fuzzy set corresponding to the bridge variable has a level of B, the bridge variable is converted to a value of 4; 2) Apply K-means clustering to the combination of environmental input variables converted into numerical values ​​to divide the data points into K clusters, with each cluster corresponding to a centroid; 3) Generate output conditions based on the centroids of each cluster containing the new input environment variable combination, and use the output conditions as the output conditions of the new fuzzy rule corresponding to the new input environment variable combination; the specific process is: 31) Calculate the individual support Support(A); the expression is: Calculate the support of joint occurrence Support(A∪B); the expression is: Where A represents the environment input variable combination represented by the centroid of the cluster containing the new input environment variable combination, B represents the output condition corresponding to A in the original state generation rule of step 5; the total number of transactions represents the number of environment input variable combinations converted to numerical values; 32) Calculate the confidence level Confidence (A→B); the expression is: 33) Keep the output condition B with the highest confidence as the output condition of the new fuzzy rule corresponding to the new input environment variable combination.

10. A real-time identification system for highway traffic operation risks under icy and snowy conditions based on dynamic multi-layer fuzzy logic, characterized by: The system is used to execute the real-time identification method of highway traffic operation risks under ice and snow conditions based on dynamic multi-layer fuzzy logic as described in any one of claims 1 to 9.

Images