Traffic event detection method and device, communication equipment and medium

Abstract

The invention discloses a traffic event detection method and device, communication equipment and a medium, and relates to the field of wireless communication. According to the specific implementation scheme, driving data of a target vehicle on multiple road sections are input into a probability topic model to obtain first probability distribution of the target vehicle on the multiple road sections; according to the first probability distribution of the plurality of road sections, determining divergence values of driving tracks of the target vehicle on the plurality of road sections; and according to the divergence values, determining whether the target vehicle has traffic events in the plurality of road sections. Therefore, whether the vehicle has the traffic event or not can be detected according to the driving data collected by the related sensor (such as a GPS) on the vehicle, the detection mode does not need to deploy a large number of cameras and high-performance computing resources, the application cost can be reduced, the detection mode is not prone to being interfered by the external environment of the vehicle, and the detection accuracy is improved. The accuracy of a detection result can be improved.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a method, device, communication equipment and medium for detecting a traffic incident. Background Art

[0002] With the rapid development of urbanization and the increasing number of vehicles, traffic congestion and traffic accidents have become common problems in traffic scenes. These problems have brought huge challenges to traffic management and emergency response departments. The above departments need to detect traffic accidents in a timely and accurate manner to reduce the probability of traffic congestion and improve the safety of traffic travel.

[0003] In order to solve this problem, real-time traffic incident detection has become an extremely important research direction. At present, traffic incidents can be detected based on image processing technology, that is, traffic incidents can be detected by analyzing surveillance cameras or video data.

[0004] However, the above-mentioned traffic incident detection method, on the one hand, not only requires the deployment of a large number of cameras and high-performance computing resources, but also has high requirements for the layout and viewing angle of the cameras; on the other hand, it is easily affected by complex environment, weather, lighting and other factors, resulting in low accuracy of the detection results. Summary of the invention

[0005] The present application provides a method, apparatus, communication device and medium for detecting traffic incidents.

[0006] According to one aspect of the present application, a method for detecting traffic incidents is provided, comprising: obtaining driving data of a target vehicle on multiple road sections; inputting the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; according to the first probability distribution of the multiple road sections, determining a divergence value of the driving trajectory of the target vehicle on the multiple road sections; and according to the divergence value, determining whether a traffic incident occurs to the target vehicle on the multiple road sections.

[0007] As a possible implementation method, the number of reference traffic states is K, and the driving data of the target vehicle on multiple road sections is input into the probability topic model to obtain the first probability distribution of the target vehicle on the multiple road sections, including: for any road section among the multiple road sections, obtaining the distribution parameters and mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probability topic model; wherein k is a positive integer not greater than K; obtaining the second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; and determining the first probability distribution of the target vehicle on any road section according to the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

[0008] As a possible implementation manner, determining the divergence values ​​of the driving trajectories of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections includes:

[0009] For any road section among multiple road sections, a sub-divergence value of a target vehicle on any road section is determined according to a first probability distribution of any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; according to the sub-divergence values ​​of multiple road sections, a divergence value of a driving trajectory of the target vehicle on the multiple road sections is determined; wherein the sub-divergence value is used to indicate a difference between an actual traffic state of the target vehicle on any road section and multiple reference traffic states.

[0010] As a possible implementation method, it is determined whether a traffic incident occurs with a target vehicle in multiple road sections based on the divergence value, including: judging whether the divergence value is greater than a set threshold; when the divergence value is greater than the set threshold, determining that a traffic incident occurs with the target vehicle; when the divergence value is less than or equal to the set threshold, determining that no traffic incident occurs with the target vehicle in multiple road sections.

[0011] As a possible implementation method, determining whether a target vehicle has a traffic incident on multiple road sections based on a divergence value includes: obtaining a value range corresponding to at least one event type; comparing the divergence value with the value range corresponding to at least one event type; when the divergence value is within the value range corresponding to the target event type in at least one event type, determining that a traffic incident corresponding to the target event type has occurred on multiple road sections for the target vehicle; when none of the divergence values ​​are within the value range corresponding to at least one event type, determining that no traffic incident has occurred on the target vehicle on multiple road sections.

[0012] As a possible implementation method, the model parameters in the probabilistic topic model are determined by the following steps: obtaining an observation data set; wherein the observation data set includes observation data subsets of multiple reference sections; the observation data subset includes observation data obtained by multiple observations of the reference section, and the observation data is used to indicate the driving data of the floating vehicle on the reference section; obtaining the second probability distribution of each observation data in the observation data subset of any reference section on any reference section; wherein the second probability distribution is used to indicate the third posterior probability of the observation data of the corresponding reference section under multiple reference traffic conditions, and the second probability distribution contains unknown model parameters; determining the likelihood function of the observation data set based on the second probability distribution of all the observation data in the observation data set; and solving the likelihood function using the expectation maximization algorithm to obtain the model parameters.

[0013] As a possible implementation method, the likelihood function is solved by an expectation maximization algorithm to obtain model parameters, including: the likelihood function is solved by an expectation maximization algorithm to obtain initial model parameters; performing at least one round of iteration according to the likelihood function to update the initial model parameters; and using the initial model parameters updated by the last round of iteration as the model parameters.

[0014] As a possible implementation, the number of reference sections is S, and the first round of iteration process includes: for the nth observation data x in the observation data subset of the sth reference section, sn , determine the kth reference traffic state in the first round of iteration to meet x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration; where s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the sth reference road section; Determine the kth reference traffic state and x in the first round of iteration sn The fifth posterior probability under the condition of using the initial model parameters; according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iteration, determine the expected likelihood function of the model parameters in the first round of iteration; based on the Lagrange multiplier method, solve the expected likelihood function of the first round of iteration to obtain the unknown model parameters of the first round of iteration, and use them as the initial model parameters updated in the first round of iteration.

[0015] As a possible implementation method, the non-first round iteration process includes: obtaining the kth reference traffic state that satisfies x snand the sixth posterior probability under the conditions of the initial model parameters updated in the previous round of iteration process; according to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration process are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration process; according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process, the likelihood function is updated; when the updated likelihood function meets the set conditions, the iteration process is terminated.

[0016] As a possible implementation method, the non-first round of iteration process also includes: when the updated likelihood function does not meet the set conditions, the expected likelihood function of the previous round of iteration process is updated according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process to obtain the expected likelihood function updated in this round of iteration process; based on the Lagrange multiplier method, the expected likelihood function of the current round of iteration process is solved to obtain the initial model parameters updated in this round of iteration process.

[0017] As a possible implementation manner, the setting condition includes at least one of the following: the updated likelihood function converges; the logarithm of the updated likelihood function converges.

[0018] According to another aspect of the present application, a communication device is provided, including:

[0019] Memory, transceiver, processor:

[0020] A memory for storing computer programs; a transceiver for transmitting and receiving data under the control of a processor; a processor for reading the computer program in the memory and performing the following operations: obtaining driving data of a target vehicle on multiple road sections; inputting the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; determining a divergence value of the driving trajectory of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections; and determining whether a traffic incident occurs to the target vehicle on the multiple road sections according to the divergence value.

[0021] As a possible implementation method, the number of reference traffic states is K, and the processor executes inputting the driving data of the target vehicle on multiple road sections into the probabilistic topic model to obtain a first probability distribution of the target vehicle on multiple road sections, specifically: for any road section among the multiple road sections, obtain the distribution parameters and mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probabilistic topic model; wherein k is a positive integer not greater than K; obtain the second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; and determine the first probability distribution of the target vehicle on any road section based on the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

[0022] As a possible implementation method, the processor executes to determine the divergence value of the driving trajectory of the target vehicle in multiple road sections according to the first probability distribution of multiple road sections, specifically: for any road section among the multiple road sections, according to the first probability distribution of any road section, determine the sub-divergence value of the target vehicle in any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; according to the sub-divergence values ​​of the multiple road sections, determine the divergence value of the driving trajectory of the target vehicle in the multiple road sections; wherein the sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle in any road section and multiple reference traffic states.

[0023] As a possible implementation method, the processor executes to determine whether a traffic incident occurs on a target vehicle in multiple road sections based on the divergence value, specifically: judging whether the divergence value is greater than a set threshold; when the divergence value is greater than the set threshold, determining that a traffic incident occurs on the target vehicle; when the divergence value is less than or equal to the set threshold, determining that no traffic incident occurs on the target vehicle in multiple road sections.

[0024] As a possible implementation method, the processor executes to determine whether a traffic incident occurs to the target vehicle in multiple road sections based on the divergence value, specifically: obtaining a value range corresponding to at least one event type; comparing the divergence value with the value range corresponding to at least one event type; when the divergence value is within the value range corresponding to the target event type in at least one event type, determining that a traffic incident corresponding to the target event type occurs to the target vehicle in multiple road sections; when none of the divergence values ​​are within the value range corresponding to at least one event type, determining that no traffic incident occurs to the target vehicle in multiple road sections.

[0025] As a possible implementation method, the processor performs the following operations to determine the model parameters in the probabilistic topic model: obtaining an observation data set; wherein the observation data set includes observation data subsets of multiple reference sections; the observation data subset includes observation data obtained by multiple observations of the reference section, and the observation data is used to indicate the driving data of the floating vehicle on the reference section; obtaining the second probability distribution of each observation data in the observation data subset of any reference section on any reference section; wherein the second probability distribution is used to indicate the third posterior probability of the observation data of the corresponding reference section under multiple reference traffic conditions, and the second probability distribution contains unknown model parameters; determining the likelihood function of the observation data set based on the second probability distribution of all the observation data in the observation data set; and solving the likelihood function using an expectation maximization algorithm to obtain the model parameters.

[0026] As a possible implementation method, the processor executes the likelihood function and uses the expectation-maximization algorithm to solve the model parameters, specifically: the likelihood function is solved by the expectation-maximization algorithm to obtain initial model parameters; at least one round of iteration process is performed according to the likelihood function to update the initial model parameters; the initial model parameters updated by the last round of iteration process are used as the model parameters.

[0027] As a possible implementation, the number of reference sections is S, and the first round of iteration process includes: for the nth observation data x in the observation data subset of the sth reference section, sn , determine the kth reference traffic state in the first round of iteration to meet x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration; where s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the sth reference road section; Determine the kth reference traffic state and x in the first round of iteration sn The fifth posterior probability under the condition of using the initial model parameters; according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iteration, determine the expected likelihood function of the model parameters in the first round of iteration; based on the Lagrange multiplier method, solve the expected likelihood function of the first round of iteration to obtain the unknown model parameters of the first round of iteration, and use them as the initial model parameters updated in the first round of iteration.

[0028] As a possible implementation method, the non-first round iteration process includes: obtaining the kth reference traffic state that satisfies x snand the sixth posterior probability under the conditions of the initial model parameters updated in the previous round of iteration process; according to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration process are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration process; according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process, the likelihood function is updated; when the updated likelihood function meets the set conditions, the iteration process is terminated.

[0029] As a possible implementation method, the non-first round of iteration process also includes: when the updated likelihood function does not meet the set conditions, the expected likelihood function of the previous round of iteration process is updated according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process to obtain the expected likelihood function updated in this round of iteration process; based on the Lagrange multiplier method, the expected likelihood function of the current round of iteration process is solved to obtain the initial model parameters updated in this round of iteration process.

[0030] As a possible implementation manner, the setting condition includes at least one of the following: the updated likelihood function converges; the logarithm of the updated likelihood function converges.

[0031] According to another aspect of the present application, a traffic event detection device is provided, including: a first acquisition unit, used to acquire driving data of a target vehicle on multiple road sections; an input unit, used to input the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; a first determination unit, used to determine a divergence value of the driving trajectory of the target vehicle on multiple road sections according to the first probability distribution of the multiple road sections; and a second determination unit, used to determine whether a traffic event occurs to the target vehicle on the multiple road sections according to the divergence value.

[0032] As a possible implementation method, the number of reference traffic states is K, and the input unit is specifically used to: for any road section among the multiple road sections, obtain the distribution parameters and mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probabilistic topic model; wherein k is a positive integer not greater than K; obtain the second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; and determine the first probability distribution of the target vehicle in any road section according to the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

[0033] As a possible implementation method, the first determination unit is specifically used to: for any road section among multiple road sections, determine the sub-divergence value of the target vehicle in any road section according to the first probability distribution of any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; determine the divergence value of the driving trajectory of the target vehicle in the multiple road sections according to the sub-divergence values ​​of the multiple road sections; wherein the sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle in any road section and multiple reference traffic states.

[0034] As a possible implementation method, the second determination unit is specifically used to: determine whether the divergence value is greater than a set threshold; when the divergence value is greater than the set threshold, determine that a traffic incident has occurred with the target vehicle; when the divergence value is less than or equal to the set threshold, determine that no traffic incident has occurred with the target vehicle on multiple road sections.

[0035] As a possible implementation method, the second determination unit is specifically used to: obtain a value range corresponding to at least one event type; compare the divergence value with the value range corresponding to at least one event type; when the divergence value is within the value range corresponding to the target event type in at least one event type, determine that a traffic incident corresponding to the target event type occurs to the target vehicle in multiple road sections; when none of the divergence values ​​are within the value range corresponding to at least one event type, determine that no traffic incident occurs to the target vehicle in multiple road sections.

[0036] As a possible implementation method, the model parameters in the probabilistic topic model are determined using the following units:

[0037] A second acquisition unit is used to acquire an observation data set; wherein the observation data set includes observation data subsets of multiple reference road sections; the observation data subsets include observation data obtained by multiple observations of the reference road sections, and the observation data are used to indicate driving data of the floating vehicle on the reference road sections;

[0038] A third acquisition unit is used to obtain a second probability distribution of each observation data in the observation data subset of any reference road section in any reference road section; wherein the second probability distribution is used to indicate a third posterior probability of the observation data of the corresponding reference road section under multiple reference traffic conditions, and the second probability distribution contains unknown model parameters;

[0039] A third determining unit, configured to determine a likelihood function of the observation data set according to a second probability distribution of all observation data in the observation data set;

[0040] The computing unit is used to solve the likelihood function using the expectation maximization algorithm to obtain the model parameters.

[0041] As a possible implementation method, the computing unit is specifically used to: solve the likelihood function using an expectation maximization algorithm to obtain initial model parameters; perform at least one round of iteration according to the likelihood function to update the initial model parameters; and use the initial model parameters updated by the last round of iteration as the model parameters.

[0042] As a possible implementation, the number of reference sections is S, and the first round of iteration process includes: for the nth observation data x in the observation data subset of the sth reference section, sn , determine the kth reference traffic state in the first round of iteration to meet x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration; where s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the sth reference road section; Determine the kth reference traffic state and x in the first round of iteration sn The fifth posterior probability under the condition of using the initial model parameters; according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iteration, determine the expected likelihood function of the model parameters in the first round of iteration; based on the Lagrange multiplier method, solve the expected likelihood function of the first round of iteration to obtain the unknown model parameters of the first round of iteration, and use them as the initial model parameters updated in the first round of iteration.

[0043] As a possible implementation method, the non-first round iteration process includes: obtaining the kth reference traffic state that satisfies x sn and the sixth posterior probability under the conditions of the initial model parameters updated in the previous round of iteration process; according to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration process are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration process; according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process, the likelihood function is updated; when the updated likelihood function meets the set conditions, the iteration process is terminated.

[0044] As a possible implementation method, the non-first round of iteration process also includes: when the updated likelihood function does not meet the set conditions, the expected likelihood function of the previous round of iteration process is updated according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process to obtain the expected likelihood function updated in this round of iteration process; based on the Lagrange multiplier method, the expected likelihood function of the current round of iteration process is solved to obtain the initial model parameters updated in this round of iteration process.

[0045] As a possible implementation manner, the setting condition includes at least one of the following: the updated likelihood function converges; the logarithm of the updated likelihood function converges.

[0046] According to another aspect of the present application, a processor-readable storage medium is provided, characterized in that the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the traffic event detection method of the above embodiment.

[0047] According to another aspect of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the method for detecting traffic events of the above-mentioned embodiment of the present application is implemented.

[0048] The traffic event detection method, device, access network equipment and medium provided in the embodiments of the present application can detect whether a traffic event occurs in the vehicle based on the driving data collected by relevant sensors on the vehicle (such as GPS). This detection method does not require the deployment of a large number of cameras and high-performance computing resources, which can reduce application costs. In addition, this detection method is not easily affected by the external environment of the vehicle, which can improve the accuracy of the detection results.

[0049] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present application.

[0051] Figure 1 A flow chart of a method for detecting a traffic incident provided in an embodiment of the present application;

[0052] Figure 2 A flow chart of another method for detecting traffic incidents provided in an embodiment of the present application;

[0053] Figure 3 A flow chart of another method for detecting traffic incidents provided in an embodiment of the present application;

[0054] Figure 4 A flow chart of another method for detecting traffic incidents provided in an embodiment of the present application;

[0055] Figure 5 A flow chart of another method for detecting traffic incidents provided in an embodiment of the present application;

[0056] Figure 6A flow chart of another method for detecting traffic incidents provided in an embodiment of the present application;

[0057] Figure 7 A flow chart of another method for detecting traffic incidents provided in an embodiment of the present application;

[0058] Figure 8 This is a schematic diagram of the implementation principle of the embodiment of the present application;

[0059] Fig. 9 A schematic diagram of the structure of a communication device provided in an embodiment of the present application;

[0060] Fig.10 A schematic diagram of the structure of a traffic event detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] In the embodiments of the present application, the term "and / or" describes the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0062] In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.

[0063] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0064] In order to improve traffic congestion and traffic safety, real-time traffic incident detection requires accurate monitoring and identification of traffic incidents. In the related art, the following two methods are mainly used to detect traffic incidents: detection based on image processing and detection based on sensor data.

[0065] The first is the detection method based on image processing: traffic incidents are detected by analyzing surveillance camera or video data. This method uses image processing techniques, such as vehicle detection, target tracking, and behavior analysis, to extract key information from images or videos and classify traffic incidents based on this key information.

[0066] Although this method can identify traffic events such as traffic accidents, congestion and pedestrian violations, it requires the deployment of a large number of cameras and high-performance computing resources, and has certain limitations on lighting conditions and field of view.

[0067] The second method is based on sensor data: it mainly uses vehicle sensors or other traffic sensors to collect traffic-related sensor data, where the above sensors may include the vehicle's GPS (Global Positioning System), accelerometer, etc. By analyzing the sensor data, information related to traffic events can be extracted, such as the vehicle's speed, position change, and acceleration.

[0068] Although this method can monitor traffic congestion, detect traffic accidents and assess road conditions in real time and is applicable to a larger road network, it may be limited by sensor deployment and requires effective data collection and processing methods.

[0069] In summary, the detection method based on image processing is usually applied to video surveillance systems in traffic event detection scenarios. This method has at least the following disadvantages: on the one hand, this method has high requirements on the layout and viewing angle of the camera, and is easily affected by complex environment, weather, lighting and other factors, thereby reducing the accuracy of the detection results; on the other hand, the processing and analysis of large-scale video data requires high computing resources, resulting in poor real-time performance of the algorithm; on the other hand, the image processing method has limited detection effect on non-visual events (such as traffic congestion, abnormal behavior before an accident).

[0070] Detection methods based on sensor data mainly rely on data collected by vehicle sensors or roadside sensors. However, the data collected by sensors may be affected by errors, noise and incomplete data, resulting in reduced accuracy in traffic event detection. In addition, the deployment and maintenance costs of sensors are high, further limiting their feasibility in large-scale applications. At the same time, the collection and transmission of sensor data takes time, making the response relatively slow.

[0071] In practical applications, it is common to combine the above two detection methods to obtain more accurate and comprehensive traffic incident detection results, that is, by comprehensively utilizing image data and sensor data to detect traffic incidents, the accuracy and reliability of the detection results can be improved, and traffic management and emergency response capabilities can be further improved. In addition, with the continuous development of the Internet of Things and artificial intelligence technologies, the application of other emerging technologies, such as Internet of Vehicles, unmanned driving and deep learning, can also be explored to further improve the detection effect and performance of traffic incidents.

[0072] Therefore, in order to solve at least one of the above-mentioned problems, the embodiments of the present application provide a method, device, access network equipment and medium for detecting traffic events, wherein the method and the device are based on the same application concept. Since the principles of solving the problems by the method and the device are similar, the implementation of the device and the method can refer to each other, and the repeated parts will not be repeated.

[0073] The following is a description of the traffic incident detection method, apparatus, communication equipment and medium provided in the present application with reference to the accompanying drawings.

[0074] Figure 1 A flow chart of a method for detecting a traffic incident provided in an embodiment of the present application.

[0075] The traffic incident detection method of the embodiment of the present application can be executed by a communication device, wherein the communication device can be an access network device, a server, a communication device or a computer device of a traffic management and emergency response department, etc.

[0076] Herein, the access network device is taken as an example as a base station. The base station may include multiple cells that provide services for the terminal device. Depending on the specific application scenario, the base station may also be called an access point, or may be a device in the access network that communicates with the wireless terminal device through one or more sectors on the air interface, or other names. The base station may be used to interchange received air frames with Internet Protocol (IP) packets, and serve as a router between the wireless terminal device and the rest of the access network, wherein the rest of the access network may include an Internet Protocol (IP) communication network. The base station may also coordinate the attribute management of the air interface. For example, the base station involved in the embodiments of the present application may be a base station (Base Transceiver Station, BTS) in the Global System for Mobile communications (GSM) or Code Division Multiple Access (CDMA), or a base station (NodeB) in Wide-band Code Division Multiple Access (WCDMA), or an evolutionary Node B (eNB or e-NodeB) in the long term evolution (LTE) system, or a 5G base station (gNB) in the 5G network architecture (next generation system), or a home evolved Node B (HeNB), a relay node, a home base station (femto), a pico base station, etc., which is not limited in the embodiments of the present application. In some network structures, the base station may include a centralized unit (CU) node and a distributed unit (DU) node, and the centralized unit and the distributed unit may also be arranged geographically separately.

[0077] The terminal device may be a device that provides voice and / or data connectivity to a user, a handheld device with a wireless connection function, or other processing devices connected to a wireless modem. In different systems, the names of the terminal devices may also be different. For example, in a 5G system, the terminal device may be called a user equipment (UE). The wireless terminal device may communicate with one or more core networks (CN) via a radio access network (RAN). The wireless terminal device may be a mobile terminal device, such as a mobile phone (or "cellular" phone) and a computer with a mobile terminal device. For example, it may be a portable, pocket-sized, handheld, computer-built-in or vehicle-mounted mobile device that exchanges language and / or data with a wireless access network. For example, personal communication service (PCS) phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDA) and other devices. The wireless terminal device may also be referred to as a system, a subscriber unit, a subscriber station, a mobile station, a mobile station, a remote station, an access point, a remote terminal device, an access terminal device, a user terminal device, a user agent, and a user device, but is not limited in the embodiments of the present application.

[0078] like Figure 1 As shown, the traffic incident detection method may include the following steps:

[0079] Step S101, obtaining driving data of a target vehicle on multiple road sections.

[0080] In the embodiment of the present application, the target vehicle may be any vehicle, wherein there is no restriction on the vehicle type of the target vehicle, for example, the target vehicle may be a bus, a car, a taxi, etc.

[0081] There is no restriction on the number of target vehicles. For example, the number of target vehicles may be one or more.

[0082] In an embodiment of the present application, the driving data may be collected by relevant sensors (such as GPS) on the target vehicle, wherein the driving data may include but is not limited to: time data, location data represented by longitude and latitude, driving speed, acceleration, etc.

[0083] In an embodiment of the present application, driving data of a target vehicle on multiple road sections may be obtained.

[0084] Step S102, inputting the driving data of the target vehicle on multiple road sections into the probability topic model to obtain a first probability distribution of the target vehicle on multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions.

[0085] Among them, the reference traffic status, which can also be called the normal traffic status, includes but is not limited to the following traffic status: smooth, congested (dense vehicles), blocked (traffic is completely stagnant), accident (traffic is chaotic and the road is partially closed), etc.

[0086] In an embodiment of the present application, the driving data of the target vehicle on multiple road sections can be input into the probabilistic topic model to obtain the first probability distribution of the target vehicle on multiple road sections output by the probabilistic topic model, wherein the first probability distribution of each road section is used to indicate the posterior probability (referred to as the first posterior probability in this application) of the driving data of the target vehicle on the corresponding road section under multiple reference traffic conditions.

[0087] Step S103, determining the divergence values ​​of the driving trajectories of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections.

[0088] The divergence value is used to indicate the difference between the actual traffic state of the target vehicle on multiple road sections and the reference traffic state.

[0089] In an embodiment of the present application, the divergence value of the driving trajectory of the target vehicle in the multiple road sections can be determined according to the first probability distribution of the target vehicle in the multiple road sections.

[0090] Step S104: determining whether a traffic incident occurs with the target vehicle on multiple road sections based on the divergence value.

[0091] Among them, traffic incidents are to prevent the occurrence of traffic accidents, among which the types of traffic accidents include but are not limited to: minor accidents, general accidents, major accidents, and extremely serious accidents.

[0092] Among them, the event types of traffic incidents may include but are not limited to the following types: driving behavior or traffic safety (such as running a red light, turning right on the line, speeding, etc.), mixed traffic (such as pedestrians and animals mixing, ghosts poking their heads out, etc.), and traffic events (such as non-motor vehicles entering motor vehicle lanes, etc.).

[0093] In the embodiment of the present application, it is possible to determine whether a traffic incident occurs to the target vehicle in multiple road sections according to the divergence value of the driving trajectory of the target vehicle in multiple road sections. For example, when the divergence value is relatively large, it can be determined that a traffic incident occurs to the target vehicle, and when the divergence value is relatively small, it can be determined that no traffic incident occurs to the target vehicle.

[0094] The method for detecting traffic incidents in the embodiment of the present application is to input the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on multiple road sections; wherein the first probability distribution is used to indicate the first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; according to the first probability distribution of the multiple road sections, the divergence value of the driving trajectory of the target vehicle on the multiple road sections is determined; according to the divergence value, it is determined whether a traffic incident occurs to the target vehicle on the multiple road sections. Thus, it is possible to detect whether a traffic incident occurs to the vehicle based on the driving data collected by the relevant sensors (such as GPS) on the vehicle. This detection method does not need to deploy a large number of cameras and high-performance computing resources, which can reduce the application cost, and this detection method is not easily affected by the external environment of the vehicle, which can improve the accuracy of the detection results.

[0095] In order to clearly explain how to obtain the first probability distribution of the target vehicle on multiple road sections in any embodiment of the present application, the present application also proposes a method for detecting traffic events.

[0096] Figure 2 A schematic flow chart of another method for detecting traffic incidents provided in an embodiment of the present application.

[0097] like Figure 2 As shown, the traffic incident detection method may include the following steps:

[0098] Step S201, obtaining driving data of a target vehicle on multiple road sections.

[0099] For the explanation of step S201, please refer to the relevant description in any embodiment of the present application, which will not be repeated here.

[0100] Step S202 : for any road section among the multiple road sections, obtaining distribution parameters and a mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probabilistic topic model.

[0101] The number of reference traffic states is K, and k is a positive integer not greater than K.

[0102] In the embodiment of the present application, for any one of the multiple road sections, the distribution parameter θ corresponding to the kth reference traffic state in the model parameters in the probability topic model can be obtained. k and the mixing coefficient vector π sk .

[0103] Step S203: obtaining a second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied.

[0104] In the embodiment of the present application, for any one of the multiple road sections, the driving data of the target vehicle on the road section can be obtained when the distribution parameter θ of the kth reference traffic state is satisfied. k The posterior probability under the condition (referred to as the second posterior probability in this application). For example, if the driving data of the target vehicle on this road section is x, then the second posterior probability is p(x|θ k ).

[0105] Step S204, determining a first probability distribution of the target vehicle on any road section according to the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

[0106] The first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under a plurality of reference traffic conditions.

[0107] In the embodiment of the present application, the mixing coefficient vector π corresponding to the kth reference traffic state can be used. sk The product of the second posterior probability determines the first probability distribution of the target vehicle on any road section.

[0108] For example, the driving data of the target vehicle on the sth road section is marked as x, and the first probability distribution of the target vehicle on the sth road section is p(x|s), then:

[0109]

[0110] Step S205 , determining the divergence values ​​of the driving trajectories of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections.

[0111] Step S206, determining whether a traffic incident occurs with the target vehicle on multiple road sections according to the divergence value.

[0112] The explanation of steps S205 to S206 can be found in the relevant description of any disclosed embodiment, and will not be repeated here.

[0113] The traffic incident detection method of the embodiment of the present application can effectively determine the first probability distribution of the target vehicle in each road section according to the model parameters in the probabilistic topic model, so as to effectively detect whether a traffic incident occurs to the target vehicle according to the first probability distribution of the target vehicle in each road section, so as to improve the effectiveness and accuracy of traffic incident detection.

[0114] In order to clearly explain how to determine the divergence value of the driving trajectory of the target vehicle on multiple road sections in any embodiment of the present application, the present application also proposes a method for detecting traffic events.

[0115] Figure 3 A schematic flow chart of another method for detecting traffic incidents provided in an embodiment of the present application.

[0116] like Figure 3 As shown, the traffic incident detection method may include the following steps:

[0117] Step S301, obtaining driving data of a target vehicle on multiple road sections.

[0118] Step S302: input the driving data of the target vehicle on multiple road sections into the probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections.

[0119] The first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under a plurality of reference traffic conditions.

[0120] The explanation of steps S301 to S302 can be found in the relevant description in any embodiment of the present application and will not be repeated here.

[0121] Step S303 : for any road section among the multiple road sections, determine the sub-divergence value of the target vehicle in any road section according to the first probability distribution of any road section.

[0122] The sub-divergence value is negatively correlated with the first probability distribution, and the sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle on any road section and multiple reference traffic states.

[0123] In the embodiment of the present application, for any one of the multiple road sections, the sub-divergence value of the target vehicle in the road section can be determined according to the first probability distribution of the target vehicle in the road section.

[0124] As an example, the driving data of the target vehicle in the sth section is marked as x, the first probability distribution of the target vehicle in the sth section is p(x|s), and the sub-divergence value of the target vehicle in the sth section is d(s,x), then:

[0125]

[0126] Step S304: determining the divergence values ​​of the driving trajectory of the target vehicle on the multiple road sections according to the sub-divergence values ​​of the multiple road sections.

[0127] In the embodiment of the present application, the divergence value of the driving trajectory of the target vehicle in the multiple road sections can be determined according to the sub-divergence values ​​of the multiple road sections.

[0128] As an example, the mean, weighted sum, and cumulative value of the sub-divergence values ​​of the target vehicle in multiple road sections may be used as the divergence value of the driving trajectory of the target vehicle in the multiple road sections.

[0129] As an example, the number of marked sections is N, and the divergence value of the target vehicle's driving trajectory on multiple sections is D, then:

[0130]

[0131] Among them, i0 represents the starting index of the first road segment among multiple road segments, d i Represents the sub-divergence value of the i-th road segment.

[0132] Step S305: determining whether a traffic incident occurs with the target vehicle on multiple road sections according to the divergence value.

[0133] For the explanation of step S305, please refer to the relevant description in any embodiment of the present application, which will not be repeated here.

[0134] The traffic incident detection method of the embodiment of the present application can effectively calculate the sub-divergence value of the target vehicle in each road section according to the first probability distribution of the target vehicle in each road section, so as to effectively calculate the divergence value of the driving trajectory of the target vehicle in multiple road sections according to the sub-divergence value of the target vehicle in each road section, so as to improve the effectiveness and rationality of the divergence value calculation.

[0135] In order to clearly explain how any of the above embodiments determines whether a traffic incident occurs to a target vehicle in multiple road sections based on the divergence value, the present application also proposes a method for detecting a traffic incident.

[0136] Figure 4 A schematic flow chart of another method for detecting traffic incidents provided in an embodiment of the present application.

[0137] like Figure 4 As shown, the traffic incident detection method may include the following steps:

[0138] Step S401, obtaining driving data of a target vehicle on multiple road sections.

[0139] Step S402: input the driving data of the target vehicle on multiple road sections into the probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections.

[0140] The first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under a plurality of reference traffic conditions.

[0141] Step S403: determining the divergence values ​​of the driving trajectories of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections.

[0142] The explanation of steps S401 to S403 can be found in the relevant description in any embodiment of the present application and will not be repeated here.

[0143] Step S404, determine whether the divergence value is greater than a set threshold, if so, execute step S405, if not, execute step S406.

[0144] The threshold value may be set based on historical data or experience.

[0145] It should be noted that step S405 and step S406 are two parallel implementation methods. In actual application, only one needs to be executed.

[0146] Step S405, determining that a traffic incident occurs on multiple road sections for the target vehicle.

[0147] In an embodiment of the present application, when the divergence value of the driving trajectory of the target vehicle in multiple road sections is greater than a set threshold, it indicates that the difference between the actual traffic state and the reference traffic state of the target vehicle in multiple road sections is relatively large. At this time, it can be determined that traffic incidents have occurred in multiple road sections for the target vehicle.

[0148] Step S406, determining that no traffic incidents occurred on the target vehicle in multiple road sections.

[0149] In an embodiment of the present application, when the divergence value of the driving trajectory of the target vehicle in multiple road sections is less than or equal to a set threshold, it indicates that the difference between the actual traffic state and the reference traffic state of the target vehicle in multiple road sections is relatively small. At this time, it can be determined that no traffic incidents have occurred with the target vehicle in multiple road sections.

[0150] The traffic incident detection method of the embodiment of the present application can realize the detection of traffic incidents based on quantitative divergence values, so that the actual traffic status of the target vehicle can be monitored in real time, and abnormal or sudden traffic incidents can be quickly discovered to improve the efficiency of traffic management and emergency response.

[0151] In order to clearly explain how any of the above embodiments determines whether a traffic incident occurs to a target vehicle in multiple road sections based on the divergence value, the present application also proposes a method for detecting a traffic incident.

[0152] Figure 5 A schematic flow chart of another method for detecting traffic incidents provided in an embodiment of the present application.

[0153] like Figure 5 As shown, the traffic incident detection method may include the following steps:

[0154] Step S501, obtaining driving data of a target vehicle on multiple road sections.

[0155] Step S502: input the driving data of the target vehicle on multiple road sections into the probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections.

[0156] The first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under a plurality of reference traffic conditions.

[0157] Step S503: determining the divergence values ​​of the driving trajectories of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections.

[0158] The explanation of steps S501 to S503 can be found in the relevant description in any embodiment of the present application and will not be repeated here.

[0159] Step S504: Obtain a value range corresponding to at least one event type.

[0160] In an embodiment of the present application, the number of event types of traffic events may be at least one, wherein the value range corresponding to at least one event type is pre-set. For example, the value range corresponding to at least one event type may be set based on historical data or experience.

[0161] Step S505: compare the divergence value with a value range corresponding to at least one event type.

[0162] In an embodiment of the present application, the divergence value of the target vehicle's driving trajectory on multiple road sections can be compared with the value range of each event type to determine whether the divergence value is within the value range of each event type. If so, execute step S506, if not, execute step S507.

[0163] It should be noted that steps S506 and S507 are two parallel implementation methods. In actual application, only one needs to be executed.

[0164] Step S506: when the divergence value is within the value range corresponding to the target event type in at least one event type, it is determined that a traffic event corresponding to the target event type occurs on multiple road sections for the target vehicle.

[0165] In an embodiment of the present application, if the divergence value of the driving trajectory of the target vehicle in multiple road sections is within the value range corresponding to a certain event type (referred to as the target event type in this application), it can be determined that a traffic incident occurs with the target vehicle in multiple road sections, and the event type of the traffic incident is the target event type.

[0166] Step S507: When none of the divergence values ​​is within the value range corresponding to at least one event type, it is determined that no traffic incident occurs to the target vehicle on the multiple road sections.

[0167] In an embodiment of the present application, if the divergence values ​​of the driving trajectories of the target vehicle on multiple road sections are not within the value ranges corresponding to all event types, it can be determined that no traffic incidents occurred with the target vehicle on the multiple road sections.

[0168] The traffic event detection method of the embodiment of the present application can not only detect traffic events based on quantitative divergence values, but also detect the event type of traffic events to meet actual traffic management needs in different scenarios.

[0169] In order to clearly illustrate any of the above embodiments, the present application also proposes a method for detecting a traffic incident.

[0170] Figure 6 A schematic flow chart of another method for detecting traffic incidents provided in an embodiment of the present application.

[0171] like Figure 6 As shown, based on any of the above embodiments, the model parameters in the probabilistic topic model can be determined by the following steps:

[0172] Step S601, obtaining an observation data set; wherein the observation data set includes observation data subsets of multiple reference sections; the observation data subset includes observation data obtained by multiple observations of the reference sections, and the observation data is used to indicate the driving data of the floating vehicle on the reference sections.

[0173] The observation data may be collected by relevant sensors (such as GPS) in the floating vehicle, and the observation data may include time data, location data represented by longitude and latitude, driving speed, acceleration, and the like.

[0174] The length of the reference section is a set length (eg, 50 meters).

[0175] In the embodiment of the present application, each reference road section may be observed to obtain an observation data subset of each reference road section, and an observation data set may be generated based on the observation data subsets of each reference road section.

[0176] For example, if the number of reference sections is S, then the observation data subset of the sth reference section can be Where s = 1, 2, ..., S, N s represents the number of observations of the sth road segment, that is, N s For X s The number of observations contained in x sn represents the observation data obtained by the nth observation of the sth road segment, that is, x sn Represents X s The nth value in , where n=1,2,…,N s .

[0177] Step S602, obtaining a second probability distribution of each observation data in a subset of observation data of any reference road section in any reference road section; wherein the second probability distribution is used to indicate a third posterior probability of the observation data of the corresponding reference road section under multiple reference traffic conditions, and the second probability distribution contains unknown model parameters.

[0178] In an embodiment of the present application, for any reference road section, the second probability distribution of each observation data in the observation data subset of the reference road section can be obtained, wherein the second probability distribution of any observation data is used to indicate the posterior probability of the observation data of the reference road section under multiple reference traffic conditions (referred to as the third posterior probability in this application), and the second probability distribution contains unknown model parameters.

[0179] As an example, the observation data x in the observation data subset of the sth reference road segment sn , the second probability distribution of the s-th reference road segment is:

[0180]

[0181] Among them, θ k is the distribution parameter corresponding to the kth reference traffic state in the unknown model parameters, π sk is the mixing coefficient vector corresponding to the kth reference traffic state in the unknown model parameters.

[0182] Step S603: Determine the likelihood function of the observation data set according to the second probability distribution of all the observation data in the observation data set.

[0183] In the embodiment of the present application, the likelihood function of the observation data set may be determined according to the second probability distribution of all the observation data in the observation data set.

[0184] As an example, the labeled observation data set is X, and the likelihood function of the observation data set X is L(X), then:

[0185]

[0186] Step S604: using the expectation maximization algorithm to solve the likelihood function to obtain model parameters.

[0187] In the embodiment of the present application, the likelihood function can be solved using an expectation maximization algorithm to obtain optimal model parameters.

[0188] The traffic incident detection method of the embodiment of the present application can be implemented based on the expectation maximization algorithm to solve the optimal model parameters, and then based on the optimal model parameters, real-time monitoring of traffic incidents can be performed to improve the accuracy and reliability of the detection results.

[0189] In order to clearly illustrate any of the above embodiments, the present application also proposes a method for detecting a traffic incident.

[0190] Figure 7 A schematic flow chart of another method for detecting traffic incidents provided in an embodiment of the present application.

[0191] like Figure 7 As shown in the above Figure 6 Based on the illustrated embodiment, step S604 may include the following steps:

[0192] Step S701: Using the expectation maximization algorithm to solve the likelihood function, obtain initial model parameters.

[0193] In the embodiment of the present application, the likelihood function may be solved using an EM (Expectation Maximization) algorithm to obtain initial model parameters.

[0194] Step S702: performing at least one round of iteration according to the likelihood function to update the initial model parameters.

[0195] In the embodiment of the present application, at least one round of iteration process may be performed on the likelihood function to update the initial model parameters.

[0196] In any embodiment of the present application, the first round of iteration process may include:

[0197] 1. The nth observation data x in the observation data subset for the sth reference section sn , determine the kth reference traffic state in the first round of iteration to meet x snand the posterior probability under the condition of the unknown model parameters of the first round of iteration process (referred to as the fourth posterior probability in this application).

[0198] Where s is a positive integer not greater than S, S is the number of reference sections, and n is a positive integer not greater than N. s A positive integer, N s is the number of observation data contained in the observation data subset of the sth reference road segment.

[0199] For example, marking unknown model parameters as In the first round of iteration, the kth reference traffic state satisfies x sn and the unknown model parameters of the first iteration process The fourth posterior probability under the condition of is

[0200] 2. Determine the kth reference traffic state and x in the first round of iteration sn The posterior probability under the condition of using the initial model parameters (referred to as the fifth posterior probability in this application).

[0201] For example, if the initial model parameter is marked as Λ, then in the first round of iteration, the kth reference traffic state and x sn The fifth posterior probability under the condition of using the initial model parameter Λ is: p(k,x sn |Λ).

[0202] 3. According to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iteration, the expected likelihood function of the model parameters in the first round of iteration is determined.

[0203] In an embodiment of the present application, the expected likelihood function of the model parameters in the first round of iterations can be determined based on the fourth posterior probability and the fifth posterior probability corresponding to all observation data in the observation data set in the first round of iterations.

[0204] As an example, the expected likelihood function of the labeling model parameters is the Q function (the complementary cumulative distribution function of the standard normal distribution (or the right tail function of the standard normal distribution)), then:

[0205]

[0206] Among them, the Q function is an update equation for the model parameters, which can be understood as: the model parameters estimated by the previous EM iteration are known, while the model parameters of the next iteration are unknown, and need to be updated by iterative calculation. To implement the update of the model parameters, we can use the method of maximizing the likelihood function to obtain the update direction of the model parameters, and then repeat the update process.

[0207] 4. Based on the Lagrange multiplier method, the expected likelihood function of the first round of iteration is solved to obtain the unknown model parameters of the first round of iteration, and used as the initial model parameters updated in the first round of iteration.

[0208] In the embodiment of the present application, the expected likelihood function (Q function) of the first round of iteration can be solved based on the Lagrange multiplier method to obtain the unknown model parameters of the first round of iteration and use them as the initial model parameters updated in the first round of iteration. For example, the Lagrange multiplier can be introduced into the expected likelihood function (Q function) of the first round of iteration and the partial derivatives can be set to zero to solve the maximum value of the model parameters. And it is used as the initial model parameters updated in the first round of iteration process.

[0209] In any embodiment of the present application, the non-first round iteration process may include:

[0210] 1. Get the kth reference traffic state that satisfies x sn The posterior probability under the condition of the initial model parameters updated by the previous round of iterative process (referred to as the sixth posterior probability in this application).

[0211] For example, the initial model parameters updated in the previous iteration are marked as x sn The corresponding sixth posterior probability is:

[0212] 2. According to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration.

[0213] In an embodiment of the present application, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration process can be updated according to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, so as to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration process.

[0214] As an example, the mixing coefficient vector π of the kth reference traffic state updated in this round of iteration is sk The update formula can be, for example:

[0215]

[0216] As an example, the distribution parameter θ of the kth reference traffic state updated in this round of iteration isk The update formula can be, for example:

[0217]

[0218] in, That is, the partial derivative is set to zero.

[0219] 3. The mixing coefficient vector π corresponding to the kth reference traffic state updated according to this round of iteration process sk and distribution parameter θ k , and update the likelihood function.

[0220] In an embodiment of the present application, the likelihood function may be updated based on the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iterative process.

[0221] For example, the mixing coefficient vector π corresponding to the kth reference traffic state updated in this round of iteration can be sk and distribution parameter θ k , substitute into the likelihood function L(X) formula in step S603 (ie, formula (5)) to obtain the updated likelihood function.

[0222] 4. Determine whether the updated likelihood function meets the set conditions. If so, execute the subsequent step 5; if not, execute the subsequent step 6.

[0223] The setting condition is a pre-set condition. For example, the setting condition may include: the likelihood function after update converges, and / or the setting condition may include: the logarithm value (logL(X)) of the likelihood function after update converges.

[0224] In the embodiment of the present application, it can be determined whether the value of the updated likelihood function meets the set conditions. If so, the subsequent step 5 is executed; if not, the subsequent step 6 is executed.

[0225] It should be noted that step 5 and step 6 are two parallel implementation methods. In actual application, only one needs to be executed.

[0226] 5. When the updated likelihood function meets the set conditions, the iteration process ends.

[0227] 6. When the updated likelihood function does not meet the set conditions, the expected likelihood function of the previous round of iteration is updated according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration to obtain the expected likelihood function updated in this round of iteration.

[0228] In the embodiment of the present application, if the updated likelihood function does not meet the set conditions, the mixing coefficient vector π corresponding to the k-th reference traffic state obtained by the current iterative process can be further updated. sk and distribution parameter θ k , the expected likelihood function (Q function) of the previous round of iteration is updated to obtain the expected likelihood function updated in this round of iteration.

[0229] For example, the mixing coefficient vector π corresponding to the kth reference traffic state updated in this round of iteration process can be sk and distribution parameter θ k , the initial model parameters updated in the previous round of iterations Update the initial model parameters, and use the updated initial model parameters as Λ in formula (6) to obtain the updated Q function (wherein the updated Q function is an unknown quantity) and is used as the expected likelihood function updated in this round of iteration.

[0230] As an example, assuming that the posterior probability p follows a Poisson distribution, we have:

[0231]

[0232] Among them, the probability p follows the Poisson distribution, the mean and variance of the Poisson distribution are equal, λ k are the mean and variance of p.

[0233] By using the distribution parameter θ of the kth reference traffic state k Update and derive λ k Iteration, that is:

[0234]

[0235] Therefore, in the present application, formula (6) can be updated according to formulas (7)-(10) to obtain the updated Q function in this round of iterative process.

[0236] 7. Based on the Lagrange multiplier method, the expected likelihood function of this round of iteration is solved to obtain the initial model parameters updated in this round of iteration.

[0237] In the embodiment of the present application, the expected likelihood function (Q function) of this round of iterative process can be solved based on the Lagrange multiplier method to obtain the expected likelihood function (Q function) of this round of iterative process. And used as the initial model parameters updated in this round of iteration process. For example, the expected likelihood function (Q function) of this round of iteration process can be introduced with Lagrange multipliers and the partial derivatives can be set to zero to solve, so as to obtain the maximum value of the model parameters, and used as the initial model parameters updated in this round of iteration process.

[0238] Step S703: The initial model parameters updated in the last round of iteration are used as model parameters.

[0239] In an embodiment of the present application, the initial model parameters obtained by updating the last round of iteration process may be used as model parameters in the probabilistic topic model.

[0240] The traffic incident detection method of the embodiment of the present application iteratively updates the model parameters in the probabilistic topic model through multiple iterations, thereby improving the detection accuracy of traffic incidents, that is, improving the accuracy and reliability of the detection results.

[0241] This application aims to use probabilistic topic models to improve the detection method of real-time traffic events. By improving the accuracy and real-time performance of traffic event detection, it can effectively help traffic management and emergency response departments better deal with traffic congestion and traffic accidents, thereby optimizing the safety and traffic efficiency of modern transportation systems.

[0242] This application can more accurately identify potential traffic incidents and take timely measures to respond to and alleviate traffic problems, thereby providing citizens with a safer and smoother travel experience. By utilizing probabilistic topic models and machine learning classifiers, this application can parse features and patterns related to traffic incidents from social media and online text data, which allows traffic managers to have a more comprehensive understanding of current traffic conditions in order to adjust traffic flow and the allocation of emergency resources in a timely manner. In other words, this application can provide a powerful tool for traffic management and emergency response departments to better deal with traffic congestion and traffic accident problems, and through the detection of real-time traffic incidents, a safer and more efficient urban transportation system can be created to provide citizens with a better travel environment.

[0243] Specifically, this application proposes a real-time traffic incident detection method based on a probabilistic topic model. The method consists of model establishment, EM (Expectation Maximization) algorithm parameter estimation, and a detection model application unit. The implementation principle is as follows: Figure 8 As shown, the main implementation process includes the following steps:

[0244] 1. Obtain FCD (Floating Car Data, referred to as observation data in this application) through GPS, where FCD mainly includes time data and location data represented by longitude and latitude. Afterwards, a road of a set length (e.g., 50 meters) in a specific direction (or specified direction) on a specific or specified highway route can be defined as a road segment (or segment), all observation results are aggregated for each road segment, and the observation data subset of the sth road segment is obtained. And based on the collected observation data subsets of S road sections, an observation data set X is generated as the input data of the model.

[0245] Where s = 1, 2, ..., S, N s represents the number of observations of the sth road segment, that is, X s The number of observations contained in x sn represents the observation data obtained by the nth observation of the sth road segment, that is, x sn Represents X s The nth value in , where n=1,2,…,N s .

[0246] 2. Use the probabilistic topic model to associate traffic states (referred to as reference traffic states in this application, also known as conventional traffic states, such as smooth traffic, congestion (dense vehicles), jam (complete traffic stagnation), accidents (chaotic traffic and partial road closures), etc.) with probability distributions to establish a model. Through this model, the reference traffic states are represented as implicit topics, and their associations with FCDs are learned.

[0247] The number of marked reference traffic states is K, and the distribution parameter of the kth (where k = 1, 2, ..., K) traffic state is θ k , we can use the local mixing coefficient vector π s To compare and characterize the road segments, the probability distribution p(x|s) of a given s-th road segment is described by a mixture of K distributions as follows:

[0248]

[0249] Among them, π sk represents the mixing coefficient vector of the kth traffic state, and the π sk The following conditions must be met:

[0250] 0≤π sk ≤1

[0251] 3. On the basis of establishing the above model, in order to describe the traffic status of the road, the maximum likelihood estimation of the EM algorithm can be used to optimize the model parameters.

[0252] Given the set Λ of all model parameters in the model, under the mixed distribution model, the likelihood function of the entire observation data set X is:

[0253]

[0254] 4. After establishing the likelihood function of X, find the optimal model parameters by maximizing the likelihood function Among them, maximizing the likelihood function means finding the value of the model parameter that maximizes the probability of the data occurring under given observation data.

[0255] Define the Q function (complementary cumulative distribution function of the standard normal distribution (or the right tail function of the standard normal distribution), which is generally used to describe the probability that the value of a random variable is greater than a certain value), where the Q function is the expected likelihood function of the model parameters given the current model parameters.

[0256] The Q function is obtained by weighted summing of the posterior probabilities of the reference traffic state. By maximizing the Q function, the estimated values ​​of the model parameters can be updated and improved step by step, and the value of the likelihood function in formula (5) can be improved. For example, the Q function is defined as:

[0257]

[0258] It should be noted that the Q function is an update equation for model parameters, which can be understood as: the model parameters estimated by the previous EM iteration are known, while the model parameters of the next iteration are unknown and need to be updated by iterative calculation. To implement the update of model parameters, we can use the method of maximizing the likelihood function to obtain the update direction of the model parameters, and then repeat the update process.

[0259] Among them, iterative update is the update process for the model parameter Λ, p(k,x sn |Λ) indicates that under the condition of using the model parameter Λ, the kth reference traffic state and the observation data x sn The posterior probability of .

[0260] in, represents the posterior probability of the kth reference traffic state and satisfies the given observation data x sn and the model parameters estimated by the previous EM iteration conditions.

[0261] By introducing Lagrange multipliers into formula (6) and setting the partial derivatives to zero, the maximum value of the model parameters is obtained, thereby maximizing the Q function.

[0262] Furthermore, in order to gradually improve the model's fitting ability and make the likelihood function reach a local maximum, π sk The update equation is:

[0263]

[0264] Among them, the mixing coefficient π sk Depends on the posterior probability of the kth reference traffic state.

[0265] 5. Distribution parameter θ for the kth reference traffic statek Perform update iterations, and the update formula is:

[0266]

[0267] in, That is, the partial derivative is set to zero.

[0268] Based on the parameter updating method, the updated parameter values ​​are used to describe the probability distribution properties. Assuming that the vehicle speed is a non-negative integer in km / h (kilometers per hour), and the probability p conforms to the Poisson distribution, then:

[0269]

[0270] Among them, the probability p follows the Poisson distribution, the mean and variance of the Poisson distribution are equal, λ k are the mean and variance of p.

[0271] By using the distribution parameter θ of the kth reference traffic state k Update and derive λ k Iteration, that is:

[0272]

[0273] Repeat the above iterative process of each parameter until the log-likelihood function logL(X) converges, and the optimization of the model parameters is completed.

[0274] 6. Use the divergence function to quantitatively evaluate the difference between the current traffic state (CTS) and the reference traffic state (UTS). Specifically, define a probability distribution to represent the reference traffic state, and then use another probability distribution to represent the current traffic state. Then, by calculating the divergence value between the two probability distributions, the difference between the current traffic state and the reference traffic state can be quantitatively evaluated. The larger the divergence value, the greater the difference between the current traffic state and the reference traffic state.

[0275] The divergence function d(s,x) represents the difference between CTS and UTS when the observation data x is observed at the sth section. Therefore, the divergence of the observation data obtained by continuous observation can be quantified as the sum of the divergences of the observation data obtained by each observation. In order to meet the additivity requirement, d(s,x) is the negative logarithmic probability of the observation, that is:

[0276]

[0277] Among them, when the divergence function d(s,x) value is large, it means that the observed data is abnormal, otherwise, it means that the observed data is normal.

[0278] 7. After obtaining the quantitative divergence values, these divergence values ​​can be used for real-time detection of traffic events. In order to find the driving trajectory of the floating car with abnormal behavior, the divergence d(s,x) of each road segment traversed is measured and the entire driving trajectory of the floating car is averaged to avoid deviation.

[0279] The divergence D of the entire driving trajectory is the sum of the divergences of N consecutive road segments and is defined as:

[0280]

[0281] Where d is the measured sequence of d(s,x), d i is the i-th value in d, and i0 is the starting index.

[0282] The degree of abnormality of each floating vehicle is quantified by D, and a threshold is set. When the divergence value exceeds the threshold, it is determined that a traffic incident has occurred. The above threshold can be set based on historical data or experience to ensure high-accuracy traffic incident detection. By detecting traffic incidents based on quantitative divergence values, relevant personnel can monitor traffic conditions in real time and quickly discover abnormalities or emergencies, thereby improving the efficiency of traffic management and emergency response.

[0283] In summary, this application combines traffic status with probability distribution by introducing a probabilistic topic model, and analyzes the potential topic structure from traffic data. Therefore, the introduction of this model enables the model to automatically discover topics in traffic status and infer the distribution and association patterns of topics from traffic data. By applying the probabilistic topic model to real-time traffic data and monitoring changes in topic distribution, abnormal patterns related to traffic incidents can be accurately detected. When the topic distribution deviates significantly from the expected pattern, a rapid warning of possible traffic incidents can be issued.

[0284] That is, the real-time traffic incident detection method based on the probabilistic topic model provided by the present application can quickly detect and identify traffic incidents on real-time traffic data, and has high real-time performance. Through the application of the probabilistic topic model, the method can effectively deal with the noise and uncertainty in the traffic data, and extract useful information from a large amount of data, thereby improving the accuracy of traffic incident detection. In addition, the technology also has the advantages of adaptability and scalability, and can automatically adjust the model parameters to adapt to different traffic environments and event types. With the continuous growth and increasing complexity of traffic data, the method can maintain good detection performance and respond flexibly.

[0285] Compared with the prior art, the present invention has at least the following advantages:

[0286] 1. Real-time: It can detect traffic incidents in real time, respond to changes in traffic conditions in a timely manner, help traffic management departments take quick measures to deal with traffic incidents, and improve overall traffic efficiency.

[0287] 2. Automation: It can automate the detection process of traffic incidents without human intervention, reduce the impact of subjective judgment and human errors, and improve the accuracy and consistency of detection results.

[0288] 3. Diversity: It can discover multiple themes and patterns in traffic data and identify different types of traffic events, including traffic jams, accidents, construction, etc., providing more comprehensive and granular information.

[0289] 4. Scalability: It can be applied to various real-time traffic data sources, including traffic sensors, traffic cameras, etc., and the model can be flexibly adjusted and expanded as needed to meet the needs of different cities and traffic environments.

[0290] The technical solution provided in the embodiment of the present application can be applicable to a variety of systems, especially 5G systems. For example, the applicable system can be a global system of mobile communication (Global System of Mobile communication, referred to as GSM) system, code division multiple access (Code Division Multiple Access, referred to as CDMA) system, wideband code division multiple access (Wideband Code Division Multiple Access, referred to as WCDMA) general packet radio service (General Packet Radio Service, referred to as GPRS) system, long term evolution (long term evolution, referred to as LTE) system, LTE frequency division duplex (Frequency Division Duplex, referred to as FDD) system, LTE time division duplex (time division duplex, referred to as TDD) system, advanced long term evolution (Long Term Evolution Advanced, referred to as LTE-A) system, Universal Mobile Telecommunication System (Universal Mobile Telecommunication System, referred to as UMTS), Worldwide interoperability for Microwave Access (Worldwide interoperability for Microwave Access, referred to as WiMAX) system, 5G new air interface (NewRadio, referred to as NR) system, etc. These various systems include terminals and network equipment. The system may also include a core network part, such as an evolved packet system (EPS), a 5G system (5GS), etc.

[0291] In order to implement the above embodiments, the present application also proposes a communication device.

[0292] Fig. 9 A schematic diagram of the structure of a communication device provided in an embodiment of the present application.

[0293] like Fig. 9 As shown, the communication device may include: a transceiver 900 , a processor 910 , and a memory 920 .

[0294] Among them, the memory 920 is used to store computer programs; the transceiver 900 is used to send and receive data under the control of the processor 910; the processor 910 is used to read the computer program in the memory 920 and perform the following operations: obtain the driving data of the target vehicle on multiple road sections; input the driving data of the target vehicle on multiple road sections into the probability topic model to obtain the first probability distribution of the target vehicle on multiple road sections; wherein the first probability distribution is used to indicate the first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; according to the first probability distribution of multiple road sections, determine the divergence value of the driving trajectory of the target vehicle on multiple road sections; according to the divergence value, determine whether a traffic incident occurs to the target vehicle on multiple road sections.

[0295] The transceiver 900 is used to receive and send data under the control of the processor 910 .

[0296] Among them, Fig. 9 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 910 and various circuits of memory represented by memory 920 are linked together. The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 900 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which transmission medium may include a wireless channel, a wired channel, an optical cable, and other transmission media. The processor 910 is responsible for managing the bus architecture and general processing, and the memory 920 may store data used by the processor 910 when performing operations.

[0297] The processor 910 may be a CPU, an ASIC, an FPGA or a CPLD, and the processor 910 may also adopt a multi-core architecture.

[0298] As a possible implementation, the number of reference traffic states is K, and the processor 910 executes inputting the driving data of the target vehicle on multiple road sections into the probability topic model to obtain a first probability distribution of the target vehicle on multiple road sections, specifically: for any road section among the multiple road sections, obtaining the distribution parameters and mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probability topic model; wherein k is a positive integer not greater than K; obtaining the second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; and determining the first probability distribution of the target vehicle on any road section according to the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

[0299] As a possible implementation method, the processor 910 executes to determine the divergence value of the driving trajectory of the target vehicle in multiple road sections according to the first probability distribution of multiple road sections, specifically: for any road section among the multiple road sections, according to the first probability distribution of any road section, determine the sub-divergence value of the target vehicle in any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; according to the sub-divergence values ​​of the multiple road sections, determine the divergence value of the driving trajectory of the target vehicle in the multiple road sections; wherein the sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle in any road section and multiple reference traffic states.

[0300] As a possible implementation method, the processor 910 executes to determine whether a traffic incident occurs on the target vehicle in multiple road sections based on the divergence value, specifically: judging whether the divergence value is greater than a set threshold; when the divergence value is greater than the set threshold, determining that a traffic incident occurs on the target vehicle; when the divergence value is less than or equal to the set threshold, determining that no traffic incident occurs on the target vehicle in multiple road sections.

[0301] As a possible implementation method, the processor 910 executes to determine whether a traffic incident occurs to the target vehicle in multiple road sections based on the divergence value, specifically: obtaining a value range corresponding to at least one event type; comparing the divergence value with the value range corresponding to at least one event type; when the divergence value is within the value range corresponding to the target event type in at least one event type, determining that a traffic incident corresponding to the target event type occurs to the target vehicle in multiple road sections; when none of the divergence values ​​are within the value range corresponding to at least one event type, determining that no traffic incident occurs to the target vehicle in multiple road sections.

[0302] As a possible implementation, the processor 910 performs the following operations to determine the model parameters in the probabilistic topic model: obtaining an observation data set; wherein the observation data set includes observation data subsets of multiple reference sections; the observation data subset includes observation data obtained by multiple observations of the reference section, and the observation data is used to indicate the driving data of the floating vehicle on the reference section; obtaining the second probability distribution of each observation data in the observation data subset of any reference section on any reference section; wherein the second probability distribution is used to indicate the third posterior probability of the observation data of the corresponding reference section under multiple reference traffic conditions, and the second probability distribution contains unknown model parameters; determining the likelihood function of the observation data set based on the second probability distribution of all the observation data in the observation data set; and solving the likelihood function using an expectation maximization algorithm to obtain the model parameters.

[0303] As a possible implementation method, the processor 910 executes the likelihood function and uses the expectation maximization algorithm to solve it to obtain model parameters, specifically: the likelihood function is solved by the expectation maximization algorithm to obtain initial model parameters; at least one round of iteration process is performed according to the likelihood function to update the initial model parameters; the initial model parameters updated by the last round of iteration process are used as the model parameters.

[0304] As a possible implementation, the number of reference sections is S, and the first round of iteration process includes: for the nth observation data x in the observation data subset of the sth reference section, sn , determine the kth reference traffic state in the first round of iteration to meet x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration; where s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the sth reference road section; Determine the kth reference traffic state and x in the first round of iteration sn The fifth posterior probability under the condition of using the initial model parameters; according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iteration, determine the expected likelihood function of the model parameters in the first round of iteration; based on the Lagrange multiplier method, solve the expected likelihood function of the first round of iteration to obtain the unknown model parameters of the first round of iteration, and use them as the initial model parameters updated in the first round of iteration.

[0305] As a possible implementation method, the non-first round iteration process includes: obtaining the kth reference traffic state that satisfies x snand the sixth posterior probability under the conditions of the initial model parameters updated in the previous round of iteration process; according to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration process are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration process; according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process, the likelihood function is updated; when the updated likelihood function meets the set conditions, the iteration process is terminated.

[0306] As a possible implementation method, the non-first round of iteration process also includes: when the updated likelihood function does not meet the set conditions, the expected likelihood function of the previous round of iteration process is updated according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process to obtain the expected likelihood function updated in this round of iteration process; based on the Lagrange multiplier method, the expected likelihood function of the current round of iteration process is solved to obtain the initial model parameters updated in this round of iteration process.

[0307] As a possible implementation manner, the setting condition includes at least one of the following: the updated likelihood function converges; the logarithm of the updated likelihood function converges.

[0308] It should be noted that the communication device provided in the embodiment of the present application can achieve the above Figures 1 to 7 All the method steps implemented in the method embodiment can achieve the same technical effect, and the parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.

[0309] With the above Figures 1 to 7 Corresponding to the detection method of traffic incidents provided in the embodiment, the present application also provides a detection device for traffic incidents. Figures 1 to 7 The traffic event detection method provided in the embodiment corresponds to the traffic event detection method, so the implementation method of the traffic event detection method is also applicable to the traffic event detection device provided in the embodiment of the present application, and will not be described in detail in the embodiment of the present application.

[0310] Fig.10 A schematic diagram of the structure of a traffic event detection device provided in an embodiment of the present application.

[0311] like Fig.10 As shown, the traffic event detection device 1000 may include: a first acquisition unit 1010 , an input unit 1020 , a first determination unit 1030 , and a second determination unit 1040 .

[0312] The first acquisition unit 1010 is used to acquire the driving data of the target vehicle on multiple road sections.

[0313] Input unit 1020 is used to input the driving data of the target vehicle on multiple road sections into the probability topic model to obtain a first probability distribution of the target vehicle on multiple road sections; wherein the first probability distribution is used to indicate the first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions.

[0314] The first determining unit 1030 is used to determine the divergence value of the driving trajectory of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections.

[0315] The second determination unit 1040 is used to determine whether a traffic incident occurs to the target vehicle in multiple road sections according to the divergence value.

[0316] As a possible implementation method, the number of reference traffic states is K, and the input unit 1020 is specifically used to: for any road section among the multiple road sections, obtain the distribution parameters and mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probabilistic topic model; wherein k is a positive integer not greater than K; obtain the second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; and determine the first probability distribution of the target vehicle on any road section based on the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

[0317] As a possible implementation method, the first determination unit 1030 is specifically used to: for any road section among multiple road sections, determine the sub-divergence value of the target vehicle in any road section according to the first probability distribution of any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; determine the divergence value of the driving trajectory of the target vehicle in the multiple road sections according to the sub-divergence values ​​of the multiple road sections; wherein the sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle in any road section and multiple reference traffic states.

[0318] As a possible implementation method, the second determination unit 1040 is specifically used to: determine whether the divergence value is greater than a set threshold; when the divergence value is greater than the set threshold, determine that a traffic incident has occurred with the target vehicle; when the divergence value is less than or equal to the set threshold, determine that no traffic incident has occurred with the target vehicle on multiple road sections.

[0319] As a possible implementation method, the second determination unit 1040 is specifically used to: obtain a value range corresponding to at least one event type; compare the divergence value with the value range corresponding to at least one event type; when the divergence value is within the value range corresponding to the target event type in at least one event type, determine that a traffic incident corresponding to the target event type occurs to the target vehicle in multiple road sections; when none of the divergence values ​​are within the value range corresponding to at least one event type, determine that no traffic incident occurs to the target vehicle in multiple road sections.

[0320] As a possible implementation method, the model parameters in the probabilistic topic model can be determined using the following units:

[0321] A second acquisition unit is used to acquire an observation data set; wherein the observation data set includes observation data subsets of multiple reference road sections; the observation data subsets include observation data obtained by multiple observations of the reference road sections, and the observation data are used to indicate driving data of the floating vehicle on the reference road sections;

[0322] A third acquisition unit is used to obtain a second probability distribution of each observation data in the observation data subset of any reference road section in any reference road section; wherein the second probability distribution is used to indicate a third posterior probability of the observation data of the corresponding reference road section under multiple reference traffic conditions, and the second probability distribution contains unknown model parameters;

[0323] A third determining unit, configured to determine a likelihood function of the observation data set according to a second probability distribution of all observation data in the observation data set;

[0324] The computing unit is used to solve the likelihood function using the expectation maximization algorithm to obtain the model parameters.

[0325] As a possible implementation method, the computing unit is specifically used to: solve the likelihood function using an expectation maximization algorithm to obtain initial model parameters; perform at least one round of iteration according to the likelihood function to update the initial model parameters; and use the initial model parameters updated by the last round of iteration as the model parameters.

[0326] As a possible implementation, the number of reference sections is S, and the first round of iteration process includes: for the nth observation data x in the observation data subset of the sth reference section, sn , determine the kth reference traffic state in the first round of iteration to meet x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration; where s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the sth reference road section; Determine the kth reference traffic state and x in the first round of iteration snThe fifth posterior probability under the condition of using the initial model parameters; according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iteration, determine the expected likelihood function of the model parameters in the first round of iteration; based on the Lagrange multiplier method, solve the expected likelihood function of the first round of iteration to obtain the unknown model parameters of the first round of iteration, and use them as the initial model parameters updated in the first round of iteration.

[0327] As a possible implementation method, the non-first round iteration process includes: obtaining the kth reference traffic state that satisfies x sn and the sixth posterior probability under the conditions of the initial model parameters updated in the previous round of iteration process; according to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iteration process are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iteration process; according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process, the likelihood function is updated; when the updated likelihood function meets the set conditions, the iteration process is terminated.

[0328] As a possible implementation method, the non-first round of iteration process also includes: when the updated likelihood function does not meet the set conditions, the expected likelihood function of the previous round of iteration process is updated according to the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state updated in this round of iteration process to obtain the expected likelihood function updated in this round of iteration process; based on the Lagrange multiplier method, the expected likelihood function of the current round of iteration process is solved to obtain the initial model parameters updated in this round of iteration process.

[0329] As a possible implementation manner, the setting condition includes at least one of the following: the updated likelihood function converges; the logarithm of the updated likelihood function converges.

[0330] It should be noted that the above-mentioned traffic event detection device provided in the embodiment of the present application can achieve the above-mentioned Figures 1 to 7 All the method steps implemented in the method embodiment can achieve the same technical effect, and the parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.

[0331] It should be noted that each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0332] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program code.

[0333] In order to implement the above embodiments, the present application also proposes a processor-readable storage medium.

[0334] The processor-readable storage medium stores a computer program for causing the processor to execute the present application. Figures 1 to 7 The method for detecting a traffic incident of any embodiment.

[0335] Among them, the processor-readable storage medium can be any available medium or data storage device that can be accessed by the processor, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state drive (SSD)), etc.

[0336] In order to implement the above embodiments, the present application also proposes a computer program product.

[0337] The computer program product includes a computer program which, when executed by a processor, implements the present application. Figures 1 to 7 The method for detecting a traffic incident of any embodiment.

[0338] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0339] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer executable instructions. These computer executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0340] These processor executable instructions may also be stored in a processor readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the processor readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0341] These processor-executable instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0342] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for detecting a traffic incident, characterized in that: include: Obtain the driving data of the target vehicle on multiple road sections; Inputting the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; Determining, according to the first probability distribution of the plurality of road sections, a divergence value of the driving trajectory of the target vehicle on the plurality of road sections; It is determined whether a traffic incident occurs to the target vehicle in the plurality of road sections according to the divergence value.

2. The method according to claim 1, characterized in that The number of reference traffic states is K, and the driving data of the target vehicle on multiple road sections is input into a probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections, including: For any road section among the multiple road sections, obtaining a distribution parameter and a mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probabilistic topic model; wherein k is a positive integer not greater than K; Obtaining a second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; A first probability distribution of the target vehicle on any road section is determined according to the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

3. The method according to claim 1, characterized in that The step of determining the divergence value of the driving trajectory of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections comprises: For any road section among the multiple road sections, determine a sub-divergence value of the target vehicle in the any road section according to a first probability distribution of the any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; Determining, according to the sub-divergence values ​​of the multiple road sections, divergence values ​​of the driving trajectory of the target vehicle on the multiple road sections; The sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle on any road section and the multiple reference traffic states.

4. The method according to claim 1, characterized in that: Determining whether a traffic incident occurs to the target vehicle in the plurality of road sections according to the divergence value includes: Determine whether the divergence value is greater than a set threshold; When the divergence value is greater than the set threshold, determining that a traffic incident occurs to the target vehicle; When the divergence value is less than or equal to the set threshold, it is determined that the target vehicle has not had the traffic incident on the multiple road sections.

5. The method according to claim 1, characterized in that Determining whether a traffic incident occurs to the target vehicle in the plurality of road sections according to the divergence value includes: Get the value range corresponding to at least one event type; Comparing the divergence value with a value range corresponding to the at least one event type; When the divergence value is within a value range corresponding to a target event type in the at least one event type, determining that a traffic event corresponding to the target event type occurs to the target vehicle in the plurality of road sections; When none of the divergence values ​​are within the value range corresponding to the at least one event type, it is determined that no traffic event occurs to the target vehicle in the plurality of road sections.

6. The method according to any one of claims 1 to 5, characterized in that The model parameters in the probabilistic topic model are determined by the following steps: Acquire an observation data set; wherein the observation data set includes observation data subsets of multiple reference road sections; the observation data subsets include observation data obtained by multiple observations of the reference road section, and the observation data is used to indicate the driving data of the floating vehicle on the reference road section; Obtaining a second probability distribution of each observation data in the observation data subset of any reference road section in the any reference road section; wherein the second probability distribution is used to indicate a third posterior probability of the observation data of the corresponding reference road section under the multiple reference traffic conditions, and the second probability distribution contains the unknown model parameters; Determining a likelihood function of the observation data set according to a second probability distribution of all observation data in the observation data set; The likelihood function is solved by an expectation maximization algorithm to obtain the model parameters.

7. The method according to claim 6, characterized in that The likelihood function is solved by using an expectation maximization algorithm to obtain the model parameters, including: The likelihood function is solved by using an expectation maximization algorithm to obtain initial model parameters; Performing at least one round of iteration according to the likelihood function to update the initial model parameters; The initial model parameters updated in the last round of the iterative process are used as the model parameters.

8. The method according to claim 7, characterized in that The number of reference sections is S, and the first round of iteration process includes: The nth observation data x in the observation data subset for the sth reference road segment sn , determine the kth reference traffic state in the first round of iteration to meet the x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration process; wherein s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the s-th reference section; Determine the kth reference traffic state and the x in the first iteration process sn a fifth posterior probability conditional on using the initial model parameters; Determine the expected likelihood function of the model parameters in the first round of iterations according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iterations; Based on the Lagrange multiplier method, the expected likelihood function of the first round of iteration is solved to obtain the unknown model parameters of the first round of iteration, and the unknown model parameters are used as the initial model parameters updated in the first round of iteration.

9. The method according to claim 8, characterized in that The non-first round iteration process includes: Get the kth reference traffic state that satisfies the x sn The sixth posterior probability under the condition of the initial model parameters updated by the previous round of iteration process; According to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iterations are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iterations; The likelihood function is updated according to the mixing coefficient vector and distribution parameter corresponding to the kth reference traffic state updated in the current iterative process; When the likelihood function after update satisfies the set conditions, the iterative process ends.

10. The method according to claim 9, characterized in that The non-first round iteration process further includes: When the likelihood function after the update does not satisfy the set condition, the expected likelihood function of the previous round of iteration is updated according to the mixing coefficient vector and distribution parameter corresponding to the k-th reference traffic state updated in the current round of iteration to obtain the expected likelihood function updated in the current round of iteration; Based on the Lagrange multiplier method, the expected likelihood function of the current iterative process is solved to obtain the initial model parameters updated by the current iterative process.

11. The method according to claim 9, characterized in that The setting condition includes at least one of the following: The updated likelihood function converges; The updated logarithmic value of the likelihood function converges.

12. A communication device, characterized in that: Including memory, transceiver, processor; A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations: Obtain the driving data of the target vehicle on multiple road sections; Inputting the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; Determining, according to the first probability distribution of the plurality of road sections, a divergence value of the driving trajectory of the target vehicle on the plurality of road sections; It is determined whether a traffic incident occurs to the target vehicle in the plurality of road sections according to the divergence value.

13. The communication device according to claim 12, characterized in that The number of reference traffic states is K, and the processor executes inputting the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections, specifically: For any road section among the multiple road sections, obtaining a distribution parameter and a mixing coefficient vector corresponding to the kth reference traffic state in the model parameters in the probabilistic topic model; wherein k is a positive integer not greater than K; Obtaining a second posterior probability of the driving data of any road section under the condition that the distribution parameters of the kth reference traffic state are satisfied; A first probability distribution of the target vehicle on any road section is determined according to the product of the mixing coefficient vector corresponding to the kth reference traffic state and the second posterior probability.

14. The communication device according to claim 12, characterized in that The processor determines the divergence value of the driving trajectory of the target vehicle on the multiple road sections according to the first probability distribution of the multiple road sections, specifically: For any road section among the multiple road sections, determine a sub-divergence value of the target vehicle in the any road section according to a first probability distribution of the any road section; wherein the sub-divergence value is negatively correlated with the first probability distribution; Determining, according to the sub-divergence values ​​of the multiple road sections, divergence values ​​of the driving trajectory of the target vehicle on the multiple road sections; The sub-divergence value is used to indicate the difference between the actual traffic state of the target vehicle on any road section and the multiple reference traffic states.

15. The communication device according to claim 12, characterized in that The processor determines, according to the divergence value, whether a traffic incident occurs to the target vehicle in the plurality of road sections, specifically: Determine whether the divergence value is greater than a set threshold; When the divergence value is greater than the set threshold, determining that a traffic incident occurs to the target vehicle; When the divergence value is less than or equal to the set threshold, it is determined that the target vehicle has not had the traffic incident on the multiple road sections.

16. The communication device according to claim 12, characterized in that The processor determines, according to the divergence value, whether a traffic incident occurs to the target vehicle in the plurality of road sections, specifically: Get the value range corresponding to at least one event type; Comparing the divergence value with a value range corresponding to the at least one event type; When the divergence value is within a value range corresponding to a target event type in the at least one event type, determining that a traffic event corresponding to the target event type occurs to the target vehicle in the plurality of road sections; When none of the divergence values ​​are within the value range corresponding to the at least one event type, it is determined that no traffic event occurs to the target vehicle in the plurality of road sections.

17. The communication device according to any one of claims 12 to 16, characterized in that: The processor performs the following operations to determine model parameters in the probabilistic topic model: Acquire an observation data set; wherein the observation data set includes observation data subsets of multiple reference road sections; the observation data subsets include observation data obtained by multiple observations of the reference road section, and the observation data is used to indicate the driving data of the floating vehicle on the reference road section; Obtaining a second probability distribution of each observation data in the observation data subset of any reference road section in the any reference road section; wherein the second probability distribution is used to indicate a third posterior probability of the observation data of the corresponding reference road section under the multiple reference traffic conditions, and the second probability distribution contains the unknown model parameters; Determining a likelihood function of the observation data set according to a second probability distribution of all observation data in the observation data set; The likelihood function is solved by an expectation maximization algorithm to obtain the model parameters.

18. The communication device according to claim 17, characterized in that The processor performs solving the likelihood function using an expectation maximization algorithm to obtain the model parameters, specifically: The likelihood function is solved by using an expectation maximization algorithm to obtain initial model parameters; Performing at least one round of iteration according to the likelihood function to update the initial model parameters; The initial model parameters updated in the last round of the iterative process are used as the model parameters.

19. The communication device according to claim 18, characterized in that The number of reference sections is S, and the first round of iteration process includes: The nth observation data x in the observation data subset for the sth reference road segment sn , determine the kth reference traffic state in the first round of iteration to meet the x sn and the fourth posterior probability under the condition of the unknown model parameters of the first round of iteration process; wherein s is a positive integer not greater than S, n is a positive integer not greater than N, and N is the number of observation data contained in the observation data subset of the s-th reference section; Determine the kth reference traffic state and the x in the first iteration process sn a fifth posterior probability conditional on using the initial model parameters; Determine the expected likelihood function of the model parameters in the first round of iterations according to the fourth posterior probability and the fifth posterior probability corresponding to all the observation data in the observation data set in the first round of iterations; Based on the Lagrange multiplier method, the expected likelihood function of the first round of iteration is solved to obtain the unknown model parameters of the first round of iteration, and the unknown model parameters are used as the initial model parameters updated in the first round of iteration.

20. The communication device according to claim 19, characterized in that The non-first round iteration process includes: Get the kth reference traffic state that satisfies the x sn The sixth posterior probability under the condition of the initial model parameters updated by the previous round of iteration process; According to the sixth posterior probability corresponding to each observation data in the observation data subset of the sth reference road section, the mixing coefficient vector and distribution parameters corresponding to the kth reference traffic state in the initial model parameters updated in the previous round of iterations are updated to obtain the mixing coefficient vector and distribution parameters of the kth reference traffic state updated in this round of iterations; The likelihood function is updated according to the mixing coefficient vector and distribution parameter corresponding to the kth reference traffic state updated in the current iterative process; When the likelihood function after update satisfies the set conditions, the iterative process ends.

21. The communication device according to claim 20, characterized in that The non-first round iteration process further includes: When the likelihood function after the update does not satisfy the set condition, the expected likelihood function of the previous round of iteration is updated according to the mixing coefficient vector and distribution parameter corresponding to the k-th reference traffic state updated in the current round of iteration to obtain the expected likelihood function updated in the current round of iteration; Based on the Lagrange multiplier method, the expected likelihood function of the current iterative process is solved to obtain the initial model parameters updated by the current iterative process.

22. The communication device according to claim 10, characterized in that The setting condition includes at least one of the following: The updated likelihood function converges; The updated logarithmic value of the likelihood function converges.

23. A traffic incident detection device, characterized in that: include: A first acquisition unit, used to acquire driving data of a target vehicle on multiple road sections; An input unit, used to input the driving data of the target vehicle on multiple road sections into a probability topic model to obtain a first probability distribution of the target vehicle on the multiple road sections; wherein the first probability distribution is used to indicate a first posterior probability of the driving data of the corresponding road section under multiple reference traffic conditions; A first determining unit, configured to determine a divergence value of a driving trajectory of the target vehicle on the plurality of road sections according to a first probability distribution of the plurality of road sections; The second determination unit is used to determine whether a traffic incident occurs to the target vehicle in the plurality of road sections according to the divergence value.

24. A processor-readable storage medium, characterized in that: The processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the method according to any one of claims 1 to 11.

Images