Underground tunnel driving risk evaluation method based on dynamic Bayesian network

Abstract

The invention provides an underground tunnel driving risk evaluation method based on a dynamic Bayesian network. The method comprises the following steps: constructing an underground tunnel driving risk evaluation index system; determining Bayesian network nodes according to the underground tunnel driving risk evaluation index system, and constructing an underground tunnel driving risk dynamic Bayesian network model; obtaining a driving risk assessment result according to the dynamic Bayesian network model of the driving risk of the underground tunnel; and performing driving improvement according to the driving risk assessment result. The method can effectively identify and predict the driving risk probability of different road sections of the underground tunnel, provides risk early warning for drivers, and provides data support for a traffic management department to optimize an induction system and an accident prevention strategy, thereby improving the driving safety and traffic efficiency in the tunnel.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel driving risk assessment, and in particular to an underground tunnel driving risk assessment method based on a dynamic Bayesian network. Background Art

[0002] With the acceleration of urbanization and the increase in population density, underground space is being utilized more and more. The existence of urban underground tunnels not only facilitates transportation, but also provides the possibility for sustainable development of cities. Through underground tunnels, cities can manage space more effectively, reduce the pressure of ground transportation, and improve environmental quality.

[0003] Due to the small internal space, closed and dim environment, and poor visual conditions of urban underground tunnels, the traffic environment in the diversion section is complex. Vehicles in this section will change lanes many times, there are many conflict areas, and the risk of accidents is relatively high. Summary of the invention

[0004] Based on this, it is necessary to provide an underground tunnel driving risk assessment method based on a dynamic Bayesian network to address the above technical issues.

[0005] A method for evaluating driving risk in an underground tunnel based on a dynamic Bayesian network comprises the following steps:

[0006] Construct an indicator system for evaluating the risk of driving in underground tunnels;

[0007] According to the underground tunnel driving risk assessment index system, determining the Bayesian network nodes, and constructing a dynamic Bayesian network model for underground tunnel driving risk;

[0008] Obtaining a driving risk assessment result according to the underground tunnel driving risk dynamic Bayesian network model;

[0009] Driving improvements are performed based on the driving risk assessment results.

[0010] In one embodiment, constructing an underground tunnel driving risk assessment index system includes:

[0011] Obtain the driver's physiological indicators, vehicle operation characteristics and lane change characteristic indicators in underground tunnels;

[0012] Selecting the physiological index, the vehicle operation characteristic and the lane change characteristic index according to the index optimization principle to obtain an index for constructing an underground tunnel driving risk assessment index system;

[0013] An underground tunnel driving risk assessment index system is constructed according to the indicators for constructing an underground tunnel driving risk assessment index system.

[0014] In one embodiment, according to the underground tunnel driving risk evaluation index system, determining the Bayesian network nodes and constructing the underground tunnel driving risk dynamic Bayesian network model includes:

[0015] According to the underground tunnel driving risk assessment index system, determining the Bayesian network nodes and obtaining a dynamic Bayesian network structure;

[0016] Acquire real vehicle data, and calculate the root node prior probability of the dynamic Bayesian network structure according to the real vehicle data;

[0017] The conditional probability of each node is calculated according to the root node prior probability and the dynamic Bayesian network structure to obtain a dynamic Bayesian network model of underground tunnel driving risk.

[0018] In one embodiment, obtaining real vehicle data, and calculating the root node prior probability of the dynamic Bayesian network structure according to the real vehicle data includes:

[0019] Acquire real vehicle data, and obtain expert evaluation information based on the real vehicle data;

[0020] Determining a corresponding probability classification according to the evaluation information;

[0021] Get the triangular fuzzy number corresponding to the probability classification;

[0022] A root node prior probability is calculated according to the probability classification and the triangular fuzzy number.

[0023] In one embodiment, calculating the root node prior probability according to the probability classification and the triangular fuzzy number includes:

[0024] Calculate the arithmetic mean of the fuzzy probabilities of multiple expert judgements:

[0025]

[0026] in, represents the arithmetic mean of the fuzzy probabilities of multiple experts’ evaluations, l represents the total number of experts, represents the root node X given by the kth expert i The triangular fuzzy number corresponding to the linguistic variable of the probability of being in state j, Three parameters representing triangular fuzzy numbers; e i ' j 、m i ' j 、f i ' j Three parameters representing the arithmetic mean of the fuzzy probabilities;

[0027] The mean area method is used to fuzzy process the arithmetic mean of the fuzzy probability and convert the fuzzy probability into an exact probability:

[0028]

[0029] Among them, L′ ij represents the exact probability, e i ' j 、m i ' j 、f i ' j Three parameters representing the arithmetic mean of the fuzzy probabilities;

[0030] The precise probabilities of each state of the root node are normalized to obtain the prior probability of the root node:

[0031]

[0032] Among them, L ij represents the root node prior probability, L′ ij represents the exact probability, j represents being in state j, r i Indicates the number of different states of the root node.

[0033] In one embodiment, calculating the conditional probability of each node according to the root node prior probability and the dynamic Bayesian network structure includes:

[0034] Calculate the initial conditional probability of each node based on the Leaky Noisy-or Gate expansion model according to the root node prior probability;

[0035] According to the initial conditional probability, parameter learning is performed through the EM algorithm to obtain the conditional probability.

[0036] In one embodiment, calculating the initial conditional probability of each node based on the Leaky Noisy-or Gate expansion model according to the root node prior probability includes:

[0037] The connection probability is calculated according to the following formula:

[0038]

[0039] P(Y|X i )=P i +P all -P i P all

[0040]

[0041] Among them, P(Y|X i) indicates that the intermediate node Y is in the root node X i Connection probability under the existence condition, P i Represents the intermediate nodes and the root node X i The connection probability, P all Represents the intermediate nodes and the root node X i The connection probability of the root node outside is Represents the intermediate node Y and the root node X i Connection probability under the condition of the existence of external root nodes;

[0042] Calculate the initial conditional probability of the intermediate nodes based on the connection probability:

[0043]

[0044] Among them, P i (Y) represents the initial conditional probability of the intermediate node, P l represents the connection probability between the intermediate nodes and the unknown factors, P i Represents the intermediate nodes and the root node X i The connection probability, X p represents the set of root nodes connected to the intermediate node Y, i represents X p A root node index in .

[0045] In one embodiment, the driving risk assessment result obtained according to the underground tunnel driving risk dynamic Bayesian network model includes:

[0046] Obtaining a dynamic Bayesian network model of underground tunnel driving risk, and obtaining a dynamic change process of risk probability according to the dynamic Bayesian network model of underground tunnel driving risk;

[0047] Obtaining the posterior probability of each root node according to the underground tunnel driving risk dynamic Bayesian network model, comparing and analyzing the prior probability and the posterior probability of the root node, and determining the key factors leading to driving risk;

[0048] The driving risk dynamic Bayesian network model of the underground tunnel is used to calculate the driving risk occurrence probability of the root node event in different sections of the underground tunnel when the root node event does not occur or occurs 100%; and the sensitivity analysis result of the root node is obtained according to the driving risk occurrence probability;

[0049] The driving risk assessment result is obtained according to the dynamic change process of the risk probability, the key factors causing the driving risk and the sensitivity analysis result of the root node.

[0050] In one embodiment, obtaining the posterior probability of each root node according to the underground tunnel driving risk dynamic Bayesian network model includes:

[0051] The posterior probability is calculated by the following formula:

[0052]

[0053] Among them, p(Xi=Y|T=Y) represents the probability of event X occurring under the premise of event T occurring, that is, the posterior probability; p(Xi) represents the probability of event X occurring, that is, the prior probability; P(T=Y|Xi=Y) represents the probability of event T occurring under the premise of event X occurring; Y represents the state of the event, i represents the index of a sub-event or factor of event X, and n represents the total number of sub-events or factors related to event X.

[0054] A risk assessment system for driving in an underground tunnel based on a dynamic Bayesian network is used to implement the above-mentioned risk assessment method for driving in an underground tunnel based on a dynamic Bayesian network, comprising:

[0055] System construction module, used to construct the risk assessment index system for driving in underground tunnels;

[0056] A model building module, used to determine the Bayesian network nodes according to the underground tunnel driving risk evaluation index system, and to build a dynamic Bayesian network model for underground tunnel driving risk;

[0057] A result evaluation module, used to obtain a driving risk evaluation result according to the underground tunnel driving risk dynamic Bayesian network model;

[0058] The driving improvement module is used to perform driving improvement according to the driving risk assessment result.

[0059] Compared with the prior art, the advantages and beneficial effects of the present invention are that the present invention can effectively identify and predict the probability of driving risks in different sections of underground tunnels, provide risk warnings for drivers, and provide data support for traffic management departments to optimize induction systems and accident prevention strategies, thereby improving driving safety and traffic efficiency in tunnels. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of a process of an underground tunnel driving risk assessment method based on a dynamic Bayesian network in one embodiment;

[0061] Figure 2 A schematic diagram of an underground tunnel driving risk assessment index system in one embodiment;

[0062] Figure 3 A schematic diagram of a dynamic Bayesian network model of underground tunnel driving risk in an embodiment;

[0063] Figure 4 A comparison diagram of driving risk probabilities on different road sections in an embodiment;

[0064] Figure 5 A schematic diagram of a Bayesian reverse reasoning diagram of a diversion area section in an embodiment;

[0065] Figure 6 A comparison diagram of the prior probability and the posterior probability of a root node in one embodiment;

[0066] Figure 7 A schematic diagram of the probability of occurrence of driving risk at a root node in one embodiment;

[0067] Figure 8 The figure is a schematic diagram of the structure of an underground tunnel driving risk assessment system based on a dynamic Bayesian network in one embodiment. DETAILED DESCRIPTION

[0068] Before describing the specific embodiments of the present invention, the overall concept of the present invention is described as follows:

[0069] The present invention is mainly developed based on the risk process of driving in underground tunnels. Currently, there is no efficient method for risk assessment of underground tunnels.

[0070] Therefore, the present invention proposes a risk assessment method for underground tunnel driving based on a dynamic Bayesian network, which uses a dynamic Bayesian network to evaluate and predict the driving risks of different sections of underground tunnels (main line, diversion area, ramp), and provide risk warnings for drivers to better make path decisions and ensure the driving safety of drivers in tunnels.

[0071] After introducing the overall concept of the present invention, in order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below through specific implementation methods in conjunction with the accompanying drawings.

[0072] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of this specification should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of this specification do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0073] In one embodiment, Figure 1 As shown, a method for evaluating the risk of driving in an underground tunnel based on a dynamic Bayesian network is provided, comprising the following steps:

[0074] Step S101, constructing an underground tunnel driving risk assessment index system.

[0075] Specifically, based on the reference to relevant domestic and foreign literature, according to the principle of index optimization, the physiological indicators of drivers in underground tunnels, vehicle operation characteristics and lane change characteristics are screened to determine the underground tunnel driving risk assessment index system. The specific index optimization principles include: correlation principle, multi-angle principle, measurability principle, comparability principle, trend principle, etc. Indicators are selected according to one or more of the above principles.

[0076] On this basis, the risk assessment index system for driving in underground tunnels is constructed, including:

[0077] Obtain the driver's physiological indicators, vehicle operation characteristics and lane change characteristic indicators in underground tunnels;

[0078] Selecting the physiological index, the vehicle operation characteristic and the lane change characteristic index according to the index optimization principle to obtain an index for constructing an underground tunnel driving risk assessment index system;

[0079] An underground tunnel driving risk assessment index system is constructed according to the indicators for constructing an underground tunnel driving risk assessment index system.

[0080] Specifically, the risk assessment index system for driving in underground tunnels is as follows: Figure 2 The indicators of the underground tunnel driving risk assessment index system include: speed A1, acceleration A2, scanning amplitude B1, scanning speed B2, pupil area change rate B3, gaze duration B4, heart rate change rate B5, lane change speed change C1, lane change duration C2, etc.

[0081] Step S102, determining Bayesian network nodes according to the underground tunnel driving risk assessment index system, and constructing a dynamic Bayesian network model for underground tunnel driving risk.

[0082] Specifically, a dynamic Bayesian network model of driving risk in different sections of underground tunnels is first constructed, the prior probability of the root node of the dynamic Bayesian network model is obtained, and the conditional probability parameters of the dynamic Bayesian network in different sections of underground tunnels are learned to obtain the dynamic Bayesian network model of driving risk in underground tunnels.

[0083] On this basis, according to the underground tunnel driving risk evaluation index system, the Bayesian network nodes are determined, and the underground tunnel driving risk dynamic Bayesian network model is constructed, including:

[0084] According to the underground tunnel driving risk assessment index system, determining the Bayesian network nodes and obtaining a dynamic Bayesian network structure;

[0085] Acquire real vehicle data, and calculate the root node prior probability of the dynamic Bayesian network structure according to the real vehicle data;

[0086] The conditional probability of each node is calculated according to the root node prior probability and the dynamic Bayesian network structure to obtain a dynamic Bayesian network model of underground tunnel driving risk.

[0087] Specifically, based on the underground tunnel driving risk evaluation index system, the Bayesian network nodes are determined, and the dynamic Bayesian network structure of different sections of the underground tunnel is constructed. Among them, the decision index in the evaluation index system corresponds to the target node of the dynamic Bayesian network model, the first-level index in the evaluation index system corresponds to the intermediate node of the dynamic Bayesian network model, and the second-level index corresponds to the root node of the dynamic Bayesian network model.

[0088] The root node prior probability is obtained through real vehicle data, and the root node prior probability of the dynamic Bayesian network structure is calculated based on the real vehicle data.

[0089] The dynamic Bayesian network conditional probability parameter learning is carried out under different sections of underground tunnels to obtain the dynamic Bayesian network model of underground tunnel driving risk. Figure 3 As shown, based on Figure 2 A dynamic Bayesian network model of underground tunnel driving risk constructed by the underground tunnel driving risk assessment index system.

[0090] On this basis, obtaining real vehicle data, and calculating the root node prior probability of the dynamic Bayesian network structure according to the real vehicle data includes:

[0091] Acquire real vehicle data, and obtain expert evaluation information based on the real vehicle data;

[0092] Determining a corresponding probability classification according to the evaluation information;

[0093] Get the triangular fuzzy number corresponding to the probability classification;

[0094] A root node prior probability is calculated according to the probability classification and the triangular fuzzy number.

[0095] Specifically, the root node prior probability is obtained through real vehicle data. With the help of fuzzy set theory, experts in this field judge the occurrence of root node events, convert the natural language evaluation used by experts into fuzzy numbers, and finally conduct a comprehensive analysis to obtain a more accurate prior probability value of the root node event. Discretize the Bayesian network nodes and process all nodes into binomial nodes with values ​​of discrete data {0,1}, representing the two states of {low risk, high risk}, and observe the prior probability distribution of each observation node.

[0096] On this basis, calculating the root node prior probability according to the probability classification and the triangular fuzzy number includes:

[0097] Calculate the arithmetic mean of the fuzzy probabilities of multiple expert judgements:

[0098]

[0099] in, represents the arithmetic mean of the fuzzy probabilities of multiple experts’ evaluations, l represents the total number of experts, represents the root node X given by the kth expert i The triangular fuzzy number corresponding to the linguistic variable of the probability of being in state j, Three parameters representing triangular fuzzy numbers; e i ' j 、m i ' j 、f i ' j Three parameters representing the arithmetic mean of the fuzzy probabilities;

[0100] The mean area method is used to fuzzy process the arithmetic mean of the fuzzy probability and convert the fuzzy probability into an exact probability:

[0101]

[0102] Among them, L′ ij represents the exact probability, e i ' j 、m i ' j 、f i ' j Three parameters representing the arithmetic mean of the fuzzy probabilities;

[0103] The precise probabilities of each state of the root node are normalized to obtain the prior probability of the root node:

[0104]

[0105] Among them, L ij represents the root node prior probability, L′ ijrepresents the exact probability, j represents being in state j, r i Indicates the number of different states of the root node.

[0106] Specifically, when the fuzzy set of underground tunnel driving risk factors is parameterized, a piecewise linear function is obtained, and its membership function form is:

[0107]

[0108] Among them, μ(x) is the membership function, which means that for any x∈U, there is a number μ(x)∈[0,1] corresponding to it. Triangular fuzzy numbers are represented by three parameters e, m, and f, denoted as (e, m, f).

[0109] According to the relevant literature, the probability classification standard is formulated, and the probability classification is divided into: "very high", "high", "high", "medium", "low", "low" and "very low". The corresponding relationship between each linguistic variable and triangular fuzzy number is shown in Table 1.

[0110] Table 1. The triangular fuzzy numbers corresponding to the natural language variables of event probability

[0111] Serial number Natural language variables Triangular fuzzy numbers 1 Very high (0.9,1.0,1.0) 2 high (0.7,0.9,1.0) 3 High (0.5,0.7,0.9) 4 medium (0.3,0.5,0.7) 5 Low (0.1,0.3,0.5) 6 Low (0,0.1,0.3) 7 Very low (0,0,0.1)

[0112] The calculation process of the prior probability value of the root node event is as follows:

[0113] If the number of experts is l, the root node X given by the kth expert i The linguistic variables of the probability of being in state j can be converted into triangular fuzzy numbers according to Table 1

[0114] The arithmetic mean of the fuzzy probabilities of multiple expert judgements is calculated as:

[0115]

[0116] The mean area method is used to fuzzy process the arithmetic mean of the fuzzy probability and convert the fuzzy probability into an exact probability:

[0117]

[0118] The precise probabilities of each state of the root node are normalized to obtain the prior probability of the root node.

[0119]

[0120] On this basis, the conditional probability of each node is calculated according to the root node prior probability and the dynamic Bayesian network structure, including:

[0121] Calculate the initial conditional probability of each node based on the Leaky Noisy-or Gate expansion model according to the root node prior probability;

[0122] According to the initial conditional probability, parameter learning is performed through the EM algorithm to obtain the conditional probability.

[0123] Specifically, the Leaky Noisy-or Gate extension model is used to calculate the conditional probability of each node. The EM algorithm is used for parameter learning to iteratively correct the initialized conditional probability distribution.

[0124] On this basis, according to the root node prior probability based on the Leaky Noisy-or Gate expansion model, the initial condition probability of each node is calculated including:

[0125] The connection probability is calculated according to the following formula:

[0126]

[0127] P(Y|X i )=P i +P all -P i P all

[0128]

[0129] Among them, P(Y|X i ) indicates that the intermediate node Y is in the root node X i Connection probability under the existence condition, P i Represents the intermediate nodes and the root node X i The connection probability, P all Represents the intermediate nodes and removes the root node X i The connection probability of the root node outside is Represents the intermediate node Y and the root node X i Connection probability under the condition of the existence of external root nodes;

[0130] Calculate the initial conditional probability of the intermediate nodes based on the connection probability:

[0131]

[0132] Among them, P i (Y) represents the initial conditional probability of the intermediate node, P l represents the connection probability between the intermediate nodes and the unknown factors, P i Represents the intermediate nodes and the root node X i The connection probability, X p represents the set of root nodes connected to the intermediate node Y, i represents Xp A root node index in .

[0133] Specifically, suppose there is X at the intermediate node Y. i and X all (X i Other factors) 2 root nodes, unknown factor X l represents the set of all omitted factors, P i , P all and P l Represent the corresponding connection probabilities respectively, and we can get:

[0134] P(Y|X i )=P i +P all -P i P all

[0135]

[0136] The connection probability of each root node corresponding to the intermediate node Y:

[0137]

[0138] The conditional probability of the intermediate node Y is:

[0139]

[0140] After calculating the conditional probability of the intermediate nodes, the EM algorithm is used for parameter learning to iteratively correct the initialized conditional probability distribution. The calculation process is as follows:

[0141] First, initialize θ (0) Design accuracy ε level correction value To estimate

[0142]

[0143] Step E calculates missing data e * Statistics of e * The probability distribution of is:

[0144]

[0145] In the formula, Indicates in e * and The probability distribution of e when it is unknown; Indicates e * and The joint distribution of X i represents the i-th variable; N ijk Represents all Xi and π(X i )The number of possible joint distributions.

[0146] The M step is under the given conditions The maximum a posteriori probability and maximum likelihood estimates are calculated below.

[0147] The maximum a posteriori probability calculation formula is:

[0148]

[0149] The maximum likelihood estimation formula is:

[0150]

[0151] Step S103, obtaining a driving risk assessment result according to the underground tunnel driving risk dynamic Bayesian network model.

[0152] Specifically, GENIE 4.1 Bayesian network software was introduced, and the driving risk assessment results were obtained based on the dynamic Bayesian network model of underground tunnel driving risk.

[0153] On this basis, the driving risk assessment results obtained according to the underground tunnel driving risk dynamic Bayesian network model include:

[0154] Obtaining a dynamic Bayesian network model of underground tunnel driving risk, and obtaining a dynamic change process of risk probability according to the dynamic Bayesian network model of underground tunnel driving risk;

[0155] Obtaining the posterior probability of each root node according to the underground tunnel driving risk dynamic Bayesian network model, comparing and analyzing the prior probability and the posterior probability of the root node, and determining the key factors leading to driving risk;

[0156] The driving risk dynamic Bayesian network model of the underground tunnel is used to calculate the driving risk occurrence probability of the root node event in different sections of the underground tunnel when the root node event does not occur or occurs 100%; and the sensitivity analysis result of the root node is obtained according to the driving risk occurrence probability;

[0157] The driving risk assessment result is obtained according to the dynamic change process of the risk probability, the key factors causing the driving risk and the sensitivity analysis result of the root node.

[0158] Specifically, GENIE 4.1 Bayesian network software was introduced to infer and evaluate the dynamic Bayesian network model of different sections of the underground tunnel. Figure 4 The dynamic changes of driving risk probability in the main line section, diversion area section and ramp section are shown. Figure 4It can be seen that the driving risk probability of different sections gradually increases during the entire driving process. Among them, the driving risk probability of the main line section within 10 seconds increases from the initial 18% to 19.8%, and the driving risk probability of the diversion area section within 10 seconds increases from the initial 24.2% to 25.7%, which is the highest driving risk among the three sections. The driving risk probability of the ramp section within 10 seconds increases from the initial 11.1% to 12.7%.

[0159] Then, based on GENIE 4.1 Bayesian network software, dynamic Bayesian reverse diagnosis analysis is used to obtain the posterior probability of each root node, and the key factors leading to driving risks are determined by comparing the prior probability and posterior probability of the root node. Assuming that the driving risk of the target node underground tunnel occurs, its probability is set to 1, and the posterior probability of all root nodes is obtained through Bayesian reverse diagnosis reasoning. Taking the diversion area section as an example, the Bayesian reverse reasoning diagram of the diversion area section is shown in the figure below. Figure 5 As shown, the prior probability and posterior probability of the root node are compared. Figure 6 As shown. Figure 6 It can be seen that the basic events with more significant probability increases are A1 (speed), A2 (acceleration), C1 (lane change speed change), and C2 (lane change duration). Therefore, these root nodes are the most critical factors causing driving risks in underground tunnels.

[0160] Calculate the probability of driving risk under the condition that the root node event does not occur or 100% occurs in different sections of underground tunnels. Analyze the influence of each root node on the driving risk of underground tunnels and provide a clear path for the risk management of underground tunnels. Taking the diversion area section as an example, set the evidence of each root node in the diversion area section to the non-occurrence state State = 0 and the 100% occurrence state State = 1, and calculate the probability of underground tunnel driving risk under the two states. The sensitivity analysis results of the root node are shown in Figure 2. Figure 7 As shown. Figure 7 It can be seen that the root nodes A1 (speed), A2 (acceleration), C1 (lane change speed change), and C2 (lane change duration) are the key factors affecting the risk of driving in underground tunnels, which is consistent with the results of the Bayesian reverse reasoning analysis.

[0161] Step S104: Perform driving improvement according to the driving risk assessment result.

[0162] Specifically, according to the results of bidirectional reasoning and sensitivity analysis (driving risk assessment results), dynamic speed limit signs and variable information boards can be set on the main line section to adjust the speed limit according to real-time traffic flow, thereby reducing speed fluctuations and sudden braking of vehicles. In the diversion area section, an intelligent guidance system is set up to guide vehicles to choose the appropriate lane before entering the diversion area, reducing lane change conflicts and bottleneck effects. At the same time, clearer lane markings and lane change prompts are used to help drivers prepare in advance and reduce dangerous lane change operations. In the ramp section, traffic conflicts can be reduced by widening lanes and adding buffer zones. In addition, reasonable configuration of deceleration signs and smooth road surfaces can enhance drivers' safety driving awareness and driving stability.

[0163] The underground tunnel driving risk assessment method based on dynamic Bayesian network provided by the present invention has the following advantages:

[0164] (1) The present invention constructs a set of underground tunnel driving risk rating index system and proposes a driving risk assessment method for different sections of underground tunnels based on a dynamic Bayesian network. The evaluation index system is used as the node of the Bayesian network. According to real-time data (such as vehicle speed, acceleration, etc.), the driving risk in different sections of underground tunnels is dynamically assessed and the future risk change trend is predicted.

[0165] (2) The dynamic Bayesian network model proposed in the present invention can not only evaluate the current risk status, but also predict future status changes, and provide early warnings of dangerous driving behaviors or potential risks that may occur in different sections of underground tunnels, thereby providing drivers with more timely and accurate risk warnings, helping drivers adjust their driving behaviors, and effectively improving the driving safety level in underground tunnels.

[0166] (3) The dynamic Bayesian network model proposed in the present invention has a two-way reasoning function, which can not only predict or infer the future trend of underground tunnel driving risk based on known driving data (such as speed, acceleration, etc.), but also analyze the influence of each node on the underground tunnel driving risk, thereby providing a clear path for underground tunnel risk management.

[0167] It should be noted that the method of the embodiment of the present invention can be performed by a single device, such as a computer or a server. The method of this embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present invention, and the multiple devices will interact with each other to complete the described method.

[0168] It should be noted that some embodiments of the present invention are described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0169] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present invention also provides an underground tunnel driving risk assessment system based on a dynamic Bayesian network.

[0170] refer to Figure 8 , an underground tunnel driving risk assessment system based on dynamic Bayesian network, including:

[0171] The system construction module 801 is used to construct an underground tunnel driving risk assessment index system;

[0172] A model building module 802 is used to determine the Bayesian network nodes according to the underground tunnel driving risk evaluation index system, and to build a dynamic Bayesian network model for underground tunnel driving risk;

[0173] A result evaluation module 803, used to obtain a driving risk evaluation result according to the underground tunnel driving risk dynamic Bayesian network model;

[0174] The driving improvement module 804 is used to perform driving improvement according to the driving risk assessment result.

[0175] For the convenience of description, the above system is described as being divided into various modules according to their functions. Of course, when implementing the present invention, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0176] The system of the above embodiment is used to implement a corresponding underground tunnel driving risk assessment method based on a dynamic Bayesian network in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0177] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0178] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0179] Where specific details have been set forth to describe exemplary embodiments of the present invention, it will be apparent to those skilled in the art that embodiments of the present invention may be implemented without these specific details or with variations thereto. Therefore, these descriptions should be considered illustrative rather than restrictive. Although the present invention has been described in conjunction with specific embodiments of the present invention, many replacements, modifications, and variations of these embodiments will be apparent to those of ordinary skill in the art based on the foregoing description.

[0180] The embodiments of the present invention are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for evaluating driving risk in underground tunnels based on dynamic Bayesian networks, characterized in that: include: Construct an indicator system for evaluating the risk of driving in underground tunnels; According to the underground tunnel driving risk assessment index system, determining the Bayesian network nodes, and constructing a dynamic Bayesian network model for underground tunnel driving risk; Obtaining a driving risk assessment result according to the underground tunnel driving risk dynamic Bayesian network model; Driving improvements are performed based on the driving risk assessment results.

2. According to claim 1, a method for evaluating the risk of driving in an underground tunnel based on a dynamic Bayesian network is characterized in that: The construction of the underground tunnel driving risk assessment index system includes: Obtain the driver's physiological indicators, vehicle operation characteristics and lane change characteristic indicators in underground tunnels; Selecting the physiological index, the vehicle operation characteristic and the lane change characteristic index according to the index optimization principle to obtain an index for constructing an underground tunnel driving risk assessment index system; An underground tunnel driving risk assessment index system is constructed according to the indicators for constructing an underground tunnel driving risk assessment index system.

3. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 1 is characterized in that: Determining the Bayesian network nodes according to the underground tunnel driving risk assessment index system and constructing the underground tunnel driving risk dynamic Bayesian network model includes: According to the underground tunnel driving risk assessment index system, determining the Bayesian network nodes and obtaining a dynamic Bayesian network structure; Acquire real vehicle data, and calculate the root node prior probability of the dynamic Bayesian network structure according to the real vehicle data; The conditional probability of each node is calculated according to the root node prior probability and the dynamic Bayesian network structure to obtain a dynamic Bayesian network model of underground tunnel driving risk.

4. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 3 is characterized in that: The obtaining of real vehicle data and calculating the root node prior probability of the dynamic Bayesian network structure according to the real vehicle data comprises: Acquire real vehicle data, and obtain expert evaluation information based on the real vehicle data; Determining a corresponding probability classification according to the evaluation information; Get the triangular fuzzy number corresponding to the probability classification; A root node prior probability is calculated according to the probability classification and the triangular fuzzy number.

5. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 4 is characterized in that: The calculating of the root node prior probability according to the probability classification and the triangular fuzzy number comprises: Calculate the arithmetic mean of the fuzzy probabilities of multiple expert judgements: in, represents the arithmetic mean of the fuzzy probabilities of multiple experts’ evaluations, l represents the total number of experts, represents the root node X given by the kth expert i The triangular fuzzy number corresponding to the linguistic variable of the probability of being in state j, Three parameters representing triangular fuzzy numbers; e i ′ j 、m i ′ j 、f i ′ j Three parameters representing the arithmetic mean of the fuzzy probabilities; The mean area method is used to fuzzy process the arithmetic mean of the fuzzy probability and convert the fuzzy probability into an exact probability: Among them, L ′ ij represents the exact probability, e i ′ j 、m i ′ j 、f i ′ j Three parameters representing the arithmetic mean of the fuzzy probabilities; The precise probabilities of each state of the root node are normalized to obtain the prior probability of the root node: Among them, L ij represents the root node prior probability, L ′ ij represents the exact probability, j represents being in state j, r i Indicates the number of different states of the root node.

6. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 3 is characterized in that: The calculating the conditional probability of each node according to the root node prior probability and the dynamic Bayesian network structure includes: Calculate the initial conditional probability of each node based on the Leaky Noisy-or Gate expansion model according to the root node prior probability; According to the initial conditional probability, parameter learning is performed through the EM algorithm to obtain the conditional probability.

7. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 6 is characterized in that: The step of calculating the initial condition probability of each node based on the Leaky Noisy-or Gate expansion model according to the root node prior probability includes: The connection probability is calculated according to the following formula: P(Y∣X i )=P i +P all -P i P all Among them, P(Y|X i ) indicates that the intermediate node Y is in the root node X i The probability of connection under the condition of existence. P i Represents the intermediate nodes and the root node X i The connection probability, P all Represents the intermediate nodes and the root node X i The connection probability of the root node outside is Represents the intermediate node Y and the root node X i Connection probability under the condition of the existence of external root nodes; Calculate the initial conditional probability of the intermediate nodes based on the connection probability: Among them, P i (Y) represents the initial conditional probability of the intermediate node, P l represents the connection probability between the intermediate nodes and the unknown factors, P i Represents the intermediate nodes and the root node X i The connection probability, X p represents the set of root nodes connected to the intermediate node Y, i represents X p A root node index in .

8. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 1 is characterized in that: The driving risk assessment result obtained according to the underground tunnel driving risk dynamic Bayesian network model includes: Obtaining a dynamic Bayesian network model of underground tunnel driving risk, and obtaining a dynamic change process of risk probability according to the dynamic Bayesian network model of underground tunnel driving risk; Obtaining the posterior probability of each root node according to the underground tunnel driving risk dynamic Bayesian network model, comparing and analyzing the prior probability and the posterior probability of the root node, and determining the key factors leading to driving risk; The driving risk dynamic Bayesian network model of the underground tunnel is used to calculate the driving risk occurrence probability of the root node event in different sections of the underground tunnel when the root node event does not occur or occurs 100%; and the sensitivity analysis result of the root node is obtained according to the driving risk occurrence probability; The driving risk assessment result is obtained according to the dynamic change process of the risk probability, the key factors causing the driving risk and the sensitivity analysis result of the root node.

9. The underground tunnel driving risk assessment method based on dynamic Bayesian network according to claim 1 is characterized in that: The step of obtaining the posterior probability of each root node according to the underground tunnel driving risk dynamic Bayesian network model includes: The posterior probability is calculated by the following formula: Among them, p(Xi=Y|T=Y) represents the probability of event X occurring under the premise of event T occurring, that is, the posterior probability; p(Xi) represents the probability of event X occurring, that is, the prior probability; P(T=Y|Xi=Y) represents the probability of event T occurring under the premise of event X occurring; Y represents the state of the event, i represents the index of a sub-event or factor of event X, and n represents the total number of sub-events or factors related to event X.

10. An underground tunnel driving risk assessment system based on dynamic Bayesian network, characterized in that: A method for evaluating driving risk in an underground tunnel based on a dynamic Bayesian network as claimed in any one of claims 1 to 9, comprising: System construction module, used to construct the risk assessment index system for driving in underground tunnels; A model building module, used to determine the Bayesian network nodes according to the underground tunnel driving risk evaluation index system, and to build a dynamic Bayesian network model for underground tunnel driving risk; A result evaluation module, used to obtain a driving risk evaluation result according to the underground tunnel driving risk dynamic Bayesian network model; The driving improvement module is used to perform driving improvement according to the driving risk assessment result.

Images