Traffic risk prediction method and device, electronic equipment and storage medium

Abstract

The application provides a traffic risk prediction method and device, an electronic device and a storage medium. According to the embodiment of the application, traffic data associated with a to-be-identified road section is acquired, the acquired traffic data is subjected to data recognition, first traffic data is selected from the traffic data, and then risk index information corresponding to a set risk dimension is determined for the selected first traffic data, that is, the first traffic data is converted into information with stronger risk correlation, and the data value of the first traffic data is fully mined. Since the risk index information can strengthen the representation of the traffic risk characteristics of the first traffic data compared with the first traffic data, when the to-be-identified road section is subjected to traffic risk prediction based on the determined risk index information and second traffic data other than the first traffic data in the traffic data, a more accurate traffic risk prediction result can be obtained.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation, and more particularly to a method and apparatus for predicting traffic risks, electronic equipment, and storage medium. Background Technology

[0002] With the gradual increase in the number of motor vehicles, road traffic flow has also increased, leading to a rise in traffic risks and more frequent traffic congestion and accidents. The prevalence of traffic congestion and the frequency of traffic accidents not only affect the public's normal travel, threatening the personal and property safety of travelers, but also further increase the workload of traffic management personnel as these issues require timely and effective handling.

[0003] Predicting traffic risks can effectively alleviate the above problems, and improving the accuracy of traffic risk prediction results is of great significance. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and storage medium for predicting traffic risks, in order to accurately predict traffic risks, help traffic management personnel to proactively manage risks, and reduce the impact of traffic congestion and traffic accidents on travelers.

[0005] In a first aspect, embodiments of this application provide a method for predicting traffic risks, the method comprising:

[0006] Obtain traffic data associated with the road segment to be identified;

[0007] Data identification is performed on the acquired traffic data to select the first traffic data from the traffic data;

[0008] Determine the risk indicator information represented by the selected first traffic data in the set risk dimension;

[0009] Based on the determined risk indicator information and the second traffic data excluding the first traffic data, traffic risk prediction is performed on the road segment to be identified.

[0010] Secondly, embodiments of this application provide another method for predicting traffic risks, the method comprising:

[0011] Determine the target road segment currently being traveled by the vehicle;

[0012] Obtain the traffic risk prediction results of the target road segment and the risk handling strategy determined based on the traffic risk prediction results; the traffic risk prediction results are determined based on the risk indicator information represented by the first traffic data in the set risk dimension and the second traffic data other than the first traffic data in the traffic data, and the first traffic data is selected from the traffic data associated with the target road segment;

[0013] Based on the traffic application, the system provides traffic risk prediction results and risk management strategies.

[0014] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor, when executing the computer program, implements the method described in any of the above-mentioned embodiments.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method described in any of the above-mentioned embodiments.

[0016] Compared with related technologies, this application has the following advantages:

[0017] According to the embodiments of this application, traffic data associated with the road segment to be identified is obtained, and data identification is performed on the obtained traffic data to select first traffic data. Then, the risk indicator information corresponding to the selected first traffic data in the set risk dimension is determined, that is, the first traffic data is converted into information with stronger risk correlation, and the data value of the first traffic data is fully explored. Since the risk indicator information can strengthen the representation of the traffic risk characteristics of the first traffic data compared with the first traffic data, and further based on the determined risk indicator information and the second traffic data other than the first traffic data in the traffic data, a more accurate traffic risk prediction result can be obtained when predicting the traffic risk of the road segment to be identified.

[0018] Based on traffic risk prediction results, corresponding risk management strategies can be further provided, offering a basis for mitigating traffic congestion and the impact of traffic accidents. For example, risk warnings facilitate advance risk control by traffic management personnel or help travelers avoid relevant road sections; providing traffic risk solutions aids in the rapid handling of traffic risks; and providing risk consequence mitigation plans facilitates the rapid restoration of normal traffic flow.

[0019] The first traffic data involved in this application embodiment can be target traffic data that is weakly correlated with traffic risk, such as road network data that records road surface conditions. Since target traffic data such as road network data has a weak ability to represent traffic risk, if such traffic data is converted into risk indicator information corresponding to a set risk dimension, the role of target traffic data in traffic risk prediction can be effectively enhanced, and the accuracy of traffic risk prediction can be greatly improved.

[0020] Furthermore, before conducting risk prediction, continuous road segments with similar road conditions can be aggregated. This allows for unified risk prediction of these similar road segments, which not only reduces the resource consumption of risk prediction but also improves the accuracy of risk prediction by combining multiple continuous road segments as the basis for prediction.

[0021] The traffic data used in this application for risk prediction can be a collection of various data such as road network data, vehicle driving data collected by monitoring equipment, traffic management data, and traffic maintenance data. By obtaining a comprehensive range of relevant data, the basis for traffic risk prediction can be more complete, and the accuracy of traffic risk prediction can be guaranteed.

[0022] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0023] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments according to this application and should not be construed as limiting the scope of this application.

[0024] Figure 1 A schematic diagram of a scenario for a traffic risk prediction scheme provided in an embodiment of this application is shown;

[0025] Figure 2 A flowchart of a traffic risk prediction method provided in an embodiment of this application is shown;

[0026] Figure 3 A schematic diagram illustrating a method for determining a set of roads provided in an embodiment of this application is shown;

[0027] Figure 4 A flowchart of another traffic risk prediction method provided in an embodiment of this application is shown;

[0028] Figure 5A structural block diagram of a traffic risk prediction device provided in an embodiment of this application is shown;

[0029] Figure 6 A structural block diagram of another traffic risk prediction device provided in an embodiment of this application is shown; and

[0030] Figure 7 A block diagram of an electronic device used to implement embodiments of this application is shown. Detailed Implementation

[0031] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the concept or scope of this application. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0032] The technical terms used in this application are first introduced as follows:

[0033] Vehicle Kilometers Travelled (VKT): The total mileage traveled by all vehicles on a specific road segment within a certain period. Specifically, VKT is a conversion of the mileage of a single vehicle for statistical convenience. The calculation method for a single vehicle's mileage is: annual average daily traffic volume × road segment length (km) × time (days). The calculation method for VKT is: mileage of a single vehicle / 10. 6 .

[0034] Accident rate per million vehicle-kilometers: The ratio of the number of traffic accidents or injuries caused by traffic accidents on a certain road segment within a certain period of time to the number of million vehicle-kilometers on that road segment.

[0035] Dangerous driving rate per million vehicle-kilometers: The ratio of the number of times the three sudden and one speeding driving behaviors occur on a certain road segment within a certain period of time to the number of million vehicle-kilometers on that road segment.

[0036] "Three sudden actions and one speed": This is an abbreviation for four dangerous driving behaviors, including sudden braking, sudden acceleration, sharp turns, and speeding.

[0037] OBU (On Board Unit): An on-board unit used to monitor vehicle data. It can communicate with roadside units or other terminals through technologies such as Dedicated Short Range Communication (DSRC) and Cellular Network, transmit vehicle data and receive road traffic data, and has certain data processing and storage capabilities.

[0038] RSU (Road Side Unit): A roadside device used to monitor traffic data. It can collect traffic data in real time and communicate with terminals such as OBU, other roadside units, sensing devices, traffic lights, and electronic signs through a communication network. It has certain data processing and storage capabilities, as well as gateway functions, which can be used to ensure that vehicles can access the communication network.

[0039] This application's embodiments can be applied to predicting potential traffic risks on roads, accurately predicting these risks, and providing traffic risk prediction results to traffic management personnel and travelers. This helps traffic management personnel to proactively manage risks and reduce the impact of traffic congestion and accidents on travelers. An application example of the traffic risk prediction method of this application's embodiments is given below. Figure 1 As shown, when predicting traffic risks on the road a vehicle is traveling on, the road segment where the vehicle is located can be identified as the road segment to be identified. When a driver travels according to a route planned by a navigation application, road segments involved in the navigation route can be identified as road segments to be identified, provided the driver authorizes and agrees to provide data. Further traffic data associated with the road segments to be identified is obtained. This traffic data includes road network data, vehicle driving data, traffic management data, and traffic maintenance data. This traffic data can be real-time data collected by roadside monitoring equipment (such as video surveillance cameras) on the road segments to be identified, data authorized by passengers in the vehicle (such as data provided by the RSU or the passenger's mobile phone), or data obtained from an open-source database according to the identifier of the road segments to be identified, or data authorized by the traffic management platform.

[0040] Understandably, traffic risks are influenced by a wide variety of factors. Traffic data used to predict traffic risks can encompass multiple aspects that affect them, thus improving the accuracy of predictions. For example, in some applications, the location of traffic signs can influence driver behavior. For instance, if an exit sign on a highway is too close to the exit, it may cause drivers to make dangerous driving behaviors such as sudden braking or continuous lane changes when they realize they are about to miss their exit. Therefore, when acquiring traffic data, it is essential to obtain as comprehensive a range of data as possible and to extract the value from the acquired data.

[0041] After acquiring traffic data, a portion of the data is selected as the first traffic data. This first traffic data is then converted into risk indicator information based on its correlation with traffic risk. Compared to the first traffic data, the risk indicator information can enhance the characterization of the traffic risk characteristics inherent in the first traffic data. Simultaneously, the second traffic data (excluding the first traffic data) can be preprocessed, for example, by removing data noise and redundant data through data cleaning, thereby improving the reliability of the second traffic data.

[0042] Furthermore, risk indicator information and secondary traffic information are input into a pre-trained traffic prediction model, and the risk prediction results output by the traffic training model are provided to drivers. Drivers can adjust their driving behavior, travel routes, or travel arrangements according to risk management strategies to avoid traffic congestion and reduce the likelihood of traffic accidents. In scenarios where traffic management personnel predict traffic risks, risk prediction results can be obtained through a similar process and provided to traffic management personnel. After learning of potential traffic risks, traffic management personnel can promptly implement corresponding solutions, proactively managing risks. They can also adjust the resource allocation of roadside monitoring equipment through risk management strategies, allocating more resources to road sections prone to traffic risks to strengthen management and reduce the probability of traffic risks occurring on those road sections.

[0043] The execution entity in this application embodiment can be an application, service, instance, functional module in software form, virtual machine (VM), container, or cloud server, or hardware device with data processing capabilities (such as server or terminal device) or hardware chip (such as CPU, GPU, FPGA, NPU, AI accelerator card, or DPU). The device for implementing traffic risk prediction can be deployed on the computing device of the application provider offering traffic risk prediction services or on a cloud computing platform providing computing power, storage, and network resources. The cloud computing platform can provide services in the following modes: IaaS (Infrastructure-as-a-Service), PaaS (Platform-as-a-Service), SaaS (Software-as-a-Service), or DaaS (Data-as-a-Service). Taking the platform providing SaaS (Software-as-a-Service) as an example, the cloud computing platform can utilize its own computing resources to provide functions such as model training and deploying intelligent transportation applications to realize the traffic risk prediction process. The specific application architecture can be built according to service requirements. For example, the platform can provide building services based on the above model to application users or individuals using platform resources, and further invoke the above model and realize the function of traffic risk prediction based on risk prediction requests submitted by relevant client or server devices.

[0044] The technical solutions of this application and how they solve the aforementioned technical problems are described in detail below with specific embodiments. The following related technologies are optional solutions and can be arbitrarily combined with the technical solutions of the embodiments of this application, all of which fall within the protection scope of the embodiments of this application. Identical or similar concepts or processes may not be described again in some embodiments.

[0045] like Figure 2 The diagram shown is a flowchart of a traffic risk prediction method 200 according to an embodiment of this application. The method 200 may include:

[0046] In step S201, traffic data associated with the road segment to be identified is obtained.

[0047] The road segment to be identified in this application embodiment can be a single road segment or a collection of multiple road segments. A single road segment can be a fixed-length segment or a segment specified by a user (e.g., a driver or traffic management personnel). A collection of multiple road segments can be a set of multiple single road segments or a set of segments resulting from the division of a road. When determining the road segment to be tested, different risk prediction needs can be addressed by identifying the road segment for which risk prediction is required. For example, when predicting traffic risks on a driver's travel route, the road segment where the vehicle is located can be identified as the road segment to be tested. When conducting traffic risk analysis on highways, the highway can be divided into multiple segments based on traffic management needs, and then these segments can be identified as the road segments to be tested.

[0048] After identifying the road segment to be identified, traffic data associated with that segment is acquired. This includes traffic data generated on that segment, such as vehicle driving data, traffic maintenance data, traffic management data, or data inherent to the segment itself, such as road network data describing the road surface structure. Traffic data refers to data that may impact traffic risks, specifically data related to vehicles, drivers, road management, the environment, road structure, and traffic accident statistics. Comprehensive acquisition of various relevant traffic data makes traffic risk prediction more comprehensive and ensures its accuracy. This application embodiment only exemplifies some possible traffic data; the acquisition of traffic data is not limited. Traffic data associated with the road segment can be acquired from the data system associated with it, based on the segment's identification information, or by reading traffic data monitored by roadside monitoring equipment on the road. With the traveler's authorization, traffic data provided by onboard monitoring equipment can also be acquired in real-time as vehicles pass through the road. Understandably, while acquiring traffic data, it's also possible to obtain the spatial identifiers of the locations where the traffic data was generated, as well as the time identifiers of when the traffic data was generated. After acquiring the traffic data associated with the road segment to be identified, the traffic data can be aligned according to the order of the spatial and time identifiers to transform it into a matrix form for easier subsequent data processing. Based on the acquired traffic data, potential traffic risks on the road segment to be identified can be predicted. Furthermore, based on real-time acquired traffic data, real-time prediction of traffic risks on the road segment to be identified can also be performed.

[0049] In one possible implementation, the sources of traffic data for any given type can be different. For example, vehicle speed can be monitored by roadside vehicle monitoring equipment, such as roadside checkpoints, video cameras, roadside radar speed measurement devices, and RSUs; or by in-vehicle monitoring equipment, such as navigation devices and OBUs. After acquiring traffic data from multiple sources, the traffic data can be aligned according to the order of their spatial and temporal identifiers, and then converted into a matrix format for easier subsequent data processing. In an optional implementation, outlier data can be removed after data alignment. For example, after acquiring vehicle speed data monitored by video camera equipment, roadside radar speed measurement equipment, and vehicle-side navigation equipment, and aligning the data, it is found that at the same time point on a certain urban road segment (the segment to be identified), the speeds monitored by the above three data sources are 300 km / h, 90 km / h, and 89 km / h, respectively. Since the probability of vehicles traveling at 300 km / h on urban roads is extremely low, this data can be considered an outlier and thus removed. The average of the data obtained from the other two data sources is then used to determine the vehicle speed of the road segment to be identified at the corresponding time point. For example, practical applications have shown that vehicle speed data monitored by roadside radar speed measurement devices is relatively accurate. Therefore, when acquiring vehicle speed data for the road segment to be detected, the data obtained from the roadside radar speed measurement devices on that segment is directly acquired. However, after aligning the acquired data, it was found that data from a certain monitoring point was missing. In this case, data from other sources can be used to fill the gap. Combining these two examples, it can be seen that compared to data from a single source, traffic data obtained from multiple sources, through mutual support, supplementation, and correction, can provide more accurate information, thereby improving the accuracy of risk prediction.

[0050] In step S202, the acquired traffic data is subjected to data identification in order to select the first traffic data from the traffic data.

[0051] After acquiring traffic data, a portion of the traffic data is first selected by identifying the data, and this selected portion is recorded as the first traffic data. To distinguish between different traffic data, all or part of the remaining traffic data after selecting the first traffic data is recorded as the second traffic data.

[0052] Data identification refers to the process of distinguishing a portion of traffic data based on its different characteristics or patterns. During data identification, data identification rules can be pre-set to filter traffic data, automatically selecting the filtered traffic data as the primary traffic data. In an optional implementation, data identification rules can be set solely based on the data volume, specifying the ratio of the selected primary traffic data to all traffic data, and randomly selecting a certain number of primary traffic data points. In practical applications, data identification or the selection of primary traffic data can also be performed manually based on experience. This application does not limit the specific identification or selection methods.

[0053] In one possible implementation, the second traffic data may include vehicle driving data collected by vehicle monitoring equipment, or at least one of traffic management data and traffic maintenance data. The vehicle monitoring equipment involved may be roadside monitoring equipment or vehicle-mounted monitoring equipment. Vehicle driving data collected by the vehicle monitoring equipment may include vehicle speed, vehicle trajectory, and traffic flow. Traffic management data may include data from basic roadside equipment, such as the location and information of road traffic signs, or the instructions of traffic lights. Road traffic signs may include warning signs (such as signs warning of dangerous locations or slippery roads in rainy weather), prohibitory signs (such as signs prohibiting parking or restricting passage), instruction signs (such as signs indicating the direction of travel in tidal flow lanes), and directional signs (such as signs conveying distance information to highway exits or conveying location information). Traffic maintenance data may include data such as the location of signs announcing road construction zones in the road segment to be identified, the length of the construction road corresponding to the road construction, and the number of lanes occupied by the construction.

[0054] In one possible implementation, the selected first traffic data may include target traffic data that is weakly correlated with traffic risk. All types of traffic data acquired may be factors contributing to traffic risk, but the degree of influence of different types of traffic data on traffic risk formation varies. For some traffic data that does not directly affect the formation of traffic risk, data analysis may conclude that these traffic data have a weak correlation with traffic risk or have no impact on the occurrence of traffic risk. In this case, a subset of data can be selected through data identification, where the correlation between the selected data and traffic risk is less than that between the second type of traffic data and traffic risk, and this selected data is recorded as the target traffic data. Specifically, correlation analysis can be performed on multiple categories of traffic data and traffic risk, for example, the Pearson Correlation Coefficient can be calculated, and the correlation can be determined based on the calculation results. By comparing the correlation between various types of traffic data and traffic risk, one or more types of traffic data with a lower correlation to traffic risk can be identified as the target traffic data, such that the correlation between the target traffic data and traffic risk is less than that between the second type of traffic data and traffic risk.

[0055] The target data should include at least road network data recording road surface conditions. Road network data recording road surface conditions refers to data that can characterize the features of the road itself. This can include data recording road structure, such as road elevation, whether it is uphill or downhill, slope, slope length, and horizontal curve radius, as well as data recording road type, such as whether it is one-way, whether it is a bridge, or whether it is a tunnel. When acquiring this type of traffic data, it can be obtained through an open-source road network database associated with the road under test, or through a data system authorized by the traffic management agency and associated with the road under test, based on the spatial identification information of the road under test; alternatively, it can be obtained by reading data from the design drawings or survey maps of the road under test.

[0056] Road network data recording road surface conditions may not be able to characterize the inherent risks in traffic risk prediction due to their weak correlation with traffic risk. For example, drivers slow down by braking when going downhill. If frequent traffic accidents on a particular downhill section are directly caused by driver deceleration, traffic risk prediction for that section often focuses primarily on vehicle speed and driver behavior, neglecting the impact of road structure on traffic risk formation. Therefore, it is possible to select such traffic data and enhance its representation of traffic risk in subsequent processing to improve the accuracy of traffic risk prediction results.

[0057] In one possible implementation, when performing data identification on the acquired traffic data, the relationship between the traffic data and the numerical changes of location and time information can be identified first. Traffic data whose corresponding values ​​do not change with changes in location and time information are identified as target traffic data. In a possible mathematical application example, target traffic data can be identified using the following mathematical formula:

[0058]

[0059] The design concept is that, since each data point in a sequence with zero variance is identical (meaning the data has no fluctuation), the variance of the sequence can be calculated, and whether the result is zero can be used to determine if there is fluctuation in the data. Applied to the embodiments of this application, x i This is a sequence of components along the time dimension of a matrix generated from the acquired traffic data. In other words, it represents the data values ​​of a certain type of data within the traffic data, corresponding to different time dimensions under the same spatial identifier. EX is the expected value of this component sequence, and n is the number of elements in the component sequence. The result Var is the component sequence x. i The variance. If the result Var = 0, it means x i Each data point is identical, x i The corresponding traffic data does not change over time at a fixed spatial location, therefore x can be used as a reference. i The corresponding traffic data is identified as the target traffic data.

[0060] In step S203, the risk indicator information represented by the selected first traffic data in the set risk dimension is determined.

[0061] Traffic risk can be assessed through multiple risk dimensions. When predicting traffic risk, to strengthen the representation of traffic risk characteristics by primary traffic data, this data can be converted into risk indicator information that is more strongly correlated with traffic risk within the defined dimensions. Risk indicator information is used to characterize the traffic risk of primary traffic data within the defined risk dimensions; specifically, it can be the traffic risk probability fitted based on historical traffic data. When assessing traffic risk from the dimension of traffic accidents, the risk indicator information could be the accident rate per million vehicle-kilometers; when assessing traffic risk from the dimension of dangerous driving, the risk indicator information could be the dangerous driving rate per million vehicle-kilometers. In addition to the aforementioned two risk dimensions, other possible risk dimensions can be determined by combining specific traffic risk prediction needs or by integrating multiple risk dimensions.

[0062] In one possible implementation, when determining the risk indicator information represented by the selected first traffic data in a set risk dimension, a first functional relationship between the selected first traffic data and the risk indicator information can be obtained first. This first functional relationship expresses the mapping relationship between the first traffic data and the traffic risk of the set risk dimension. In an optional implementation, the first functional relationship can be obtained by linearly fitting (e.g., using least squares fitting) the historical first traffic data and the traffic risk probability of the set risk dimension. When the risk indicator information is the accident rate per million vehicle-kilometers, the corresponding first functional relationship can be determined through the following process: First, obtain historical traffic accident data corresponding to multiple road segments over a past period, as well as historical first traffic data corresponding to the first traffic data. The historical traffic accident data can be the number of traffic accidents that occurred on the aforementioned multiple road segments over a past period, or the number of injuries or fatalities caused by traffic accidents. Then, calculate the true value of the accident rate per million vehicle-kilometers using the following mathematical formula:

[0063]

[0064] Where D represents historical traffic accident data, VKT represents the number of vehicle-kilometers per million vehicles over a period of time on the aforementioned road segments, and the calculated result V is the true value of the accident rate per million vehicle-kilometers. Then, the least squares fitting method is used to linearly fit the historical first traffic data and the true value of the accident rate per million vehicle-kilometers, and the functional relationship of the generated fitting function is determined as the first functional relationship.

[0065] After obtaining the first functional relationship, the first traffic data is fitted and calculated based on the first functional relationship to determine the corresponding risk indicator information. That is, the first traffic data is used as the input of the first functional relationship, and the fitting result of the first functional relationship calculation is determined as the risk indicator information corresponding to the first traffic data.

[0066] Since traffic risk indicator information is strongly correlated with traffic risk, after selecting the first traffic data, by determining the risk indicator information represented by the selected first traffic data in the set risk dimension, the first traffic data can be transformed into information with a stronger correlation with traffic risk, thus realizing the full mining of the data value of the first traffic data.

[0067] In one possible implementation, traffic risk probability can include at least one of traffic accident probability and dangerous driving probability. The traffic accident probability and dangerous driving probability refer to the probability of a traffic accident and dangerous driving behavior occurring on a certain road segment, respectively; for example, they could be the accident rate per million vehicle-kilometers and the dangerous driving rate per million vehicle-kilometers. The true value of traffic risk probability is calculated based on traffic data associated with the road segment. For example, the true value of the traffic accident rate can be calculated from the statistics of traffic accidents occurring on the road segment or the number of injuries or fatalities caused by accidents; the true value of the dangerous driving rate can be calculated from the statistics of dangerous driving behaviors such as sudden acceleration, deceleration, and speeding occurring on the road.

[0068] In step S204, based on the determined risk indicator information and the second traffic data excluding the first traffic data, traffic risk prediction is performed on the road segment to be identified.

[0069] After converting the first traffic data into risk indicator information that is more correlated with traffic risk, the risk indicator information and the second traffic data are used to predict the traffic risk of the road segment to be identified.

[0070] In one possible implementation, when predicting traffic risks, a pre-trained risk prediction model can be used. Risk indicator information and second traffic data are used as inputs to the risk prediction model to obtain the traffic risk prediction results for the road segment to be identified. The risk prediction model can consist of one or more machine learning models, trained based on traffic data samples and pre-labeled traffic risks. Traffic risks can be labeled by tagging the traffic data samples. The labels can specifically include traffic events that have occurred on the road segment to be identified (such as chain-reaction collisions, accidents, minor vehicle collisions, congestion, etc.), dangerous driving behaviors that are prone to occur on the road segment to be identified (such as speeding, driving against traffic, sudden braking, fatigued driving, etc.), and abnormal situations that are prone to occur on the road segment to be identified (such as cargo spillage, pedestrians on the roadway, etc.).

[0071] Traffic risk prediction results obtained from risk prediction models can include at least one of the following: whether there is a risk, the type of risk, and the probability of risk. The risk type corresponds to the traffic risks pre-labeled in the risk prediction model, such as chain-reaction rear-end collisions, road obstructions, pedestrians on the roadway, etc., or it can be a generalization of the above traffic risks, such as traffic accident risk or traffic congestion risk. The risk probability refers to the probability that a certain type of traffic risk will occur on the road to be identified. When the risk identification model indicates that the road to be identified may have multiple traffic risks, these risks can be sorted from highest to lowest probability of occurrence. When providing traffic risk prediction results, all acquired traffic risks can be provided, or the traffic risks with higher probabilities can be prioritized.

[0072] In one possible implementation, the risk prediction model can be generated through the following steps.

[0073] To distinguish between road segments and traffic data, road segments involved in the data training set used to generate the risk prediction model are denoted as sample road segments, and traffic data associated with the road segment samples are denoted as traffic data samples. When generating the risk prediction model, firstly, the traffic data samples associated with the road segment samples are obtained. Optionally, the traffic data samples can be preprocessed to improve their reliability. Secondly, data identification is performed on the obtained traffic data samples to select a first traffic data sample, which at least represents the road surface condition. Finally, based on the risk indicator information represented by the first traffic data sample in the defined risk dimension, the second traffic data sample excluding the first traffic data sample, and the traffic risks labeled for the traffic data samples, a risk prediction model for predicting traffic risks is trained.

[0074] The trained risk prediction model can output risk prediction results corresponding to pre-labeled traffic risks when subsequent inputs of risk indicator information and second traffic data. The machine learning models used in the risk prediction model can specifically include XGBoost (eXtreme Gradient Boosting), SVM (Support Vector Machines), and LR (Logistic Regression). XGBoost is an ensemble machine learning algorithm based on a decision tree model and is built upon the gradient boosting framework. It can quickly and accurately solve regression and classification problems. SVM is a binary classification supervised learning model that can be used to solve regression and classification problems. LR is also a supervised learning model that can be used to solve regression and classification problems. This is merely an example of possible models; the specific model selection in this application is not limited. In addition to the models listed above, other machine learning models or deep learning models can also be combined to construct the risk prediction model.

[0075] In one possible implementation, before predicting traffic risk for the road segment to be identified based on the determined risk indicator information and second traffic data (excluding the first traffic data), preprocessing can be performed on the first and / or second traffic data. This preprocessing may include at least one of data cleaning, data error correction, and data imputation, with the aim of improving data reliability to obtain more accurate traffic risk prediction results. In practical applications, other suitable preprocessing methods can also be selected based on the acquired traffic data; this application does not limit the specific preprocessing method used.

[0076] In one possible implementation, a corresponding risk management strategy can be determined based on the traffic risk prediction results. This risk management strategy can include at least one of the following: risk warning based on traffic applications, traffic risk solutions, and risk consequence handling plans. The risk warning refers to the alerting of potential traffic risks on the identified road segment based on the risk prediction results. For example, in a scenario where real-time risk prediction is performed on the identified road segment, if the risk prediction result based on traffic data indicates a 70% probability of congestion on the identified road segment, this traffic risk can be provided to traffic management personnel and travelers via push notifications from traffic applications to achieve risk warning. Simultaneously, risk mitigation suggestions can be provided to the traffic applications used by travelers. For example, if the navigation route used by a traveler includes the identified road segment that may be congested, the traveler can be advised to change their route to avoid the potentially congested segment. Furthermore, risk consequence handling plans can be provided, such as suggesting that traffic management personnel be dispatched to the identified road segment to investigate and resolve any abnormal situations that may lead to congestion.

[0077] Traffic applications used to present traffic risk prediction results can be deployed on terminal devices used by traffic management personnel (such as mobile terminals, computers, etc.) or on terminal devices used by travelers (such as RSUs, in-vehicle navigation systems, mobile terminals). Taking a scenario where traffic management personnel monitor traffic conditions as an example, the traffic application can be a smart traffic application deployed on a cloud desktop based on SaaS services. Traffic management personnel can view the interface of this smart traffic application through a monitoring screen. This interface can display one or more traffic condition videos collected by monitoring video probes deployed at roadside monitoring points, and simultaneously display the traffic risk prediction results for the corresponding road segments in the video footage. Traffic management personnel can be alerted to traffic risks on the road by popping up a risk warning window on the smart traffic application interface. In addition, corresponding traffic risk solutions and risk consequence handling plans can be provided on the smart traffic application interface based on the predicted traffic risks. This application embodiment does not limit the specific provision method.

[0078] It is understandable that traffic accidents or traffic congestion often have a tendency to spread. Therefore, after obtaining traffic risk prediction results, further predictions can be made based on the existing risks if they exist. In one optional implementation, while making risk predictions, traffic risks can be directly identified through traffic data monitoring equipment, and corresponding risk warnings can be provided. Alternatively, the traffic risks identified in real time by the aforementioned monitoring equipment can be used as real-time traffic data and input into the traffic risk model for further risk prediction. For example, when a pedestrian appears on a highway, vehicles may slow down or brake suddenly to avoid the pedestrian, which may further lead to traffic congestion or traffic accidents. Video monitoring cameras deployed at highway toll stations can transmit the collected monitoring videos to a cloud data computing center in real time. After the cloud data computing center identifies a traffic risk, such as a pedestrian appearing at a toll station exit, based on the monitoring video, it can provide this traffic risk to traffic applications and use it as input to the traffic risk prediction model to further predict the traffic risks that this situation may cause.

[0079] In one possible implementation, when determining the road segment to be identified, consecutive road segments with similar traffic risk probabilities can be added to the same road segment set, and each consecutive road segment in the set can be used as the road segment to be identified. This process can be understood as aggregating multiple road segments with similar traffic risk probabilities and physical contiguousness into a single road segment. The purpose of this is to perform unified risk prediction on consecutive similar road segments. Specifically, a first predetermined threshold can be set in advance, and the difference in traffic risk probability values ​​between road segments can be calculated. Road segments with a difference in risk probability values ​​less than the first predetermined threshold are judged as having similar traffic risk probabilities. On the one hand, aggregating road segments reduces the number of calculations during risk prediction, thereby reducing resource consumption. On the other hand, compared to the road segments before aggregation, the traffic data associated with the road segment to be identified combines data from multiple consecutive road segments, resulting in a larger data volume. Predicting traffic risk based on a larger volume of traffic data can improve the accuracy of risk prediction.

[0080] In addition, by treating consecutive road segments in the road segment set as road segments to be identified, traffic management personnel can manage similar consecutive road segments in a unified manner, improving their work efficiency. For example, in scenarios where risk prediction is performed on road segments to be identified to formulate road management methods, since the road segments in the road segment set are consecutive and have similar traffic risk probabilities, traffic management personnel can uniformly allocate corresponding management resources to the road segments based on the risk prediction results of the aggregated road segments. Taking a specific application example, consider a 1000-meter-long road with the road surface characteristics of a long uphill section for the first 800 meters and a downhill section for the last 200 meters. Before road segment aggregation, traffic management personnel divide the 1000-meter road into 10 100-meter segments using a grid method, setting up a video monitoring device for each 100-meter segment to monitor vehicles. Applying the embodiments of this application, the 1000-meter road can be divided into two categories: long uphill sections and downhill sections, and these two types of road segments can be used as road segments to be identified for traffic risk prediction. After risk prediction, the results showed that downhill sections with longer uphill sections had greater traffic risks and a higher probability of such risks. Based on this result, traffic management personnel can reduce the number of monitoring devices deployed on long uphill sections and add monitoring devices on downhill sections to focus on monitoring accident-prone downhill sections.

[0081] In one possible implementation, when adding consecutive road segments with similar traffic risk probability values ​​to the same road segment set, consecutive road segments with similar road surface conditions are first aggregated. When aggregating road surfaces, a second predetermined threshold can be set in advance. Multiple consecutive road segments are obtained based on their spatial identifiers, and their road surface structures are acquired. Road segments with road surface structure similarity exceeding the second predetermined threshold are aggregated. Specifically, a K-kernel clustering method, such as K-means, can be used to aggregate consecutive road segments with similar road surface conditions based on their road surface structures. During the initial aggregation, a K value can be pre-set to obtain K initial road segment sets. If subsequent iterations of the K value are needed to optimize the road segment aggregation results, an update step size for the K value can be pre-set, and the K value can be updated according to this step size. The multiple road segment sets obtained after aggregation are denoted as initial road segment sets, and each initial road segment set includes at least one road segment.

[0082] After obtaining multiple initial road segment sets, road segments within the same initial road segment set whose traffic risk probability values ​​differ by less than a first predetermined value are grouped into the same road segment set. The traffic risk probability value is determined based on historical traffic data; that is, it can be calculated based on traffic risk events that occurred on the road segment over a past period. After obtaining the traffic risk probability values, the initial road segment sets are further divided into road segment sets. Thus, the resulting road segment sets consist of continuous road segments with similar pavement structures and close risk probability values.

[0083] Figure 3 A schematic diagram illustrating one possible implementation of how the road set is determined is shown. For example... Figure 3 As shown, a road is divided into 7 segments using a grid-based system. To determine the road set, the first step is to acquire road network data, dangerous driving behavior data, and traffic accident data for each of the 7 segments. The dangerous driving behavior data can be statistical data on the number of instances of sudden acceleration, deceleration, and speeding on each segment, while the traffic accident data can be statistical data on the number of traffic accidents or injuries / fatalities on each segment. The acquired road network data is then converted into vector form and denoted as pavement structure vectors. The initial road segment set is obtained by clustering similar pavement structure vectors using the K-means algorithm.

[0084] When further dividing the initial road segment set—that is, grouping road segments with similar traffic risk probability values ​​that belong to the same initial road segment set into the same road segment set—one can use the risk dimension of dangerous driving behavior. This involves using the dangerous driving probability calculated from dangerous driving behavior data to further divide road segments in the initial road segment set with similar dangerous driving probability values, thus obtaining the final road segment set. In one possible application example, this division process can be accomplished through the following steps: First, sort the traffic risk probability values ​​corresponding to each road segment. Then, according to the sorting order, determine whether a given road segment and its adjacent road segments belong to the initial road segment set, and further group the two road segments belonging to the same initial road segment set into one category. Similarly, the road segments in the initial road segment set can also be further divided from the risk dimension of traffic accidents, combined with the probability of traffic accidents.

[0085] In one possible implementation, when dividing road segments whose traffic risk probability values ​​differ from the first set threshold and are in the same initial road segment set into the same road segment set, the road segments whose traffic risk probability values ​​differ from the first set threshold and are in the same initial road segment set can first be divided into the same road segment set to obtain a reference road segment set.

[0086] Then, a second functional relationship is fitted between the road surface feature set and the traffic risk probability value in the reference road segment set. After obtaining the reference road segment set, the road surface feature set of the reference road segment set can be obtained based on the road surface features of multiple road segments in the reference road segment set. The road surface features in the road surface feature set can be converted into vector form and expressed as road surface feature vectors. The second functional relationship can be obtained by linearly fitting the above feature vectors with the corresponding traffic risk probability values ​​(e.g., using the least squares fitting method).

[0087] Secondly, the second functional relationship is iteratively updated by changing the number of initial road segment sets. It is understood that the initial road segment set will change with the number of initial road segment sets. For example, when using K-means for the first road segment aggregation, the preset K value is 3, and the update step size is 2. After obtaining 3 initial road segment sets, when it is necessary to update the K value, the K value is updated to 5 according to the update step size, and K-means is continued to aggregate road segments, resulting in 5 initial road segment sets. After the initial road segment sets change, the road segment sets obtained by further subdividing the initial road segments may also change accordingly, and therefore the second functional relationship also needs to be updated accordingly. The purpose of iterative updating is to find a suitable number of initial road segment sets so that the second functional relationship after iteration is more accurate than before iteration. In a possible mathematical application example, the accuracy of the second functional relationship can be measured by the root mean square error (RMSE). During the iterative update process, the calculation results of the root mean square error corresponding to different K values ​​can be recorded, and the iteration is terminated when the minimum root mean square error is obtained.

[0088] Finally, at the end of the iteration, the initial road segment set is determined as the final road segment set. When predicting traffic risk for a road, the traffic risk of each road segment set can be predicted uniformly based on the divided road segment set. Since the final road segment set obtained above is the most accurate result determined by each iteration, a more accurate traffic risk prediction result can be obtained when predicting traffic risk for the road segment set.

[0089] This application also provides another method for predicting traffic risks, such as... Figure 4 The diagram shown is a flowchart of a traffic risk prediction method 400 according to another embodiment of this application. The method 400 may include:

[0090] In step S401, the target road segment where the vehicle is currently traveling is determined.

[0091] In step S402, the traffic risk prediction result of the target road segment and the risk handling strategy determined based on the traffic risk prediction result are obtained. The traffic risk prediction result is determined based on the risk indicator information represented by the first traffic data in the set risk dimension and the second traffic data other than the first traffic data in the traffic data. The first traffic data is selected from the traffic data associated with the target road segment.

[0092] In step S403, the traffic risk prediction results and risk handling strategies are presented based on the traffic application.

[0093] According to the embodiments of this application, real-time traffic risk prediction can be performed on a road segment where a vehicle is traveling. When performing traffic risk prediction, the target road segment is first determined. The target road segment can be the road segment where the vehicle is currently located, or it can be a road segment included in the navigation route indicated by a navigation application of a terminal device associated with the vehicle. After determining the target road segment, according to the traffic risk prediction method provided by method 200, the target road segment is used as the road segment to be identified, traffic risk prediction is performed on the target road segment, and the traffic risk prediction result of the target road segment and the risk handling strategy determined based on the traffic risk prediction result are obtained. Specific risk prediction methods can refer to the embodiments provided by method 200 above, and will not be repeated here.

[0094] After obtaining traffic risk prediction results and risk management strategies for the target road segment, corresponding prompts are made to travelers in vehicles through traffic applications. The prompts can be made by pushing risk predictions and risk management strategies to the terminal device carrying the traffic application, and the terminal device can provide corresponding prompts to users through text messages or voice broadcasts. This application embodiment does not limit the specific prompting method.

[0095] Corresponding to the application scenarios and methods provided in the embodiments of this application, the embodiments of this application also provide a traffic risk prediction device 500. For example... Figure 5 The diagram shown is a structural block diagram of a traffic risk prediction device according to an embodiment of this application. The traffic risk prediction device may include:

[0096] The data acquisition module 501 is used to acquire traffic data associated with the road segment to be identified;

[0097] The data recognition module 502 is used to perform data recognition on the acquired traffic data in order to select the first traffic data from the traffic data;

[0098] Information determination module 503 is used to determine the risk indicator information represented by the selected first traffic data in the set risk dimension;

[0099] The risk prediction module 504 is used to predict the traffic risk of the road segment to be identified based on the determined risk indicator information and the second traffic data excluding the first traffic data in the traffic data.

[0100] In one possible implementation, the first traffic data selected by the data identification module 502 includes target traffic data that is weakly correlated with traffic risk. The correlation between the target traffic data and traffic risk is less than that between the second traffic data and traffic risk. The target traffic data includes at least road network data that records road surface conditions.

[0101] In one possible implementation, the data identification module 502 may include:

[0102] The relationship recognition submodule is used to identify the numerical change relationship between the traffic data and the location and time information.

[0103] The data determination submodule is used to determine the target traffic data as traffic data whose corresponding values ​​do not change with changes in location and time information.

[0104] In one possible implementation, the information determination module 503 can be used to obtain a first functional relationship between the selected first traffic data and the risk indicator information; and to perform fitting calculations on the first traffic data based on the first functional relationship to determine the corresponding risk indicator information.

[0105] In one possible implementation, based on the determined risk indicator information and the second traffic data excluding the first traffic data, the risk prediction module 504 can be specifically used to: input the risk indicator information and the second traffic data excluding the first traffic data into a pre-trained risk prediction model to obtain the traffic risk prediction result of the road segment to be identified, wherein the traffic risk prediction result includes at least one of whether there is a risk, the type of risk, and the probability of risk.

[0106] In one possible implementation, the device 500 may further include:

[0107] A preprocessing module is used to preprocess the first traffic data and / or the second traffic data before performing traffic risk prediction on the road segment to be identified based on the determined risk indicator information and the second traffic data excluding the first traffic data. The preprocessing includes at least one of data cleaning, data error correction and data imputation.

[0108] In one possible implementation, the device 500 may further include:

[0109] The strategy determination module is used to determine the corresponding risk handling strategy based on the traffic risk prediction results. The risk handling strategy includes at least one of the following: risk warning based on traffic applications, traffic risk solutions, and risk consequence handling solutions.

[0110] In one possible implementation, the risk prediction module 504 may further include a model generation submodule, which is used to generate a risk prediction model, specifically including:

[0111] The sample acquisition unit is used to acquire traffic data samples associated with road segment samples;

[0112] A sample identification unit is used to identify the acquired traffic data samples in order to select a first traffic data sample from the traffic data samples, wherein the first traffic data sample is used to characterize at least the road surface condition.

[0113] The model training unit is used to train a risk prediction model for predicting traffic risks based on the risk indicator information represented by the first traffic data sample in a set risk dimension, the second traffic data sample in the traffic sample data excluding the first traffic data sample, and the traffic risks marked for the traffic data sample.

[0114] In one possible implementation, the data acquisition module 501 can be specifically used to determine the road segment to be identified, and the data acquisition module 501 may further include:

[0115] The road segment determination submodule is used to add consecutive road segments whose traffic risk probability values ​​differ from a first set threshold to the same road segment set, and to use the consecutive road segments in the road segment set as road segments to be identified.

[0116] In one possible implementation, the road segment determination submodule can also be specifically used to aggregate continuous road segments whose road surface condition similarity exceeds a second set threshold to obtain multiple initial road segment sets; and to classify road segments whose traffic risk probability value difference is less than a first set threshold and are in the same initial road segment set into the same road segment set, wherein the traffic risk probability value is determined based on historical traffic data.

[0117] In one possible implementation, the road segment determination submodule can be further configured to: classify road segments whose traffic risk probability values ​​differ by less than a first set threshold and are in the same initial road segment set into the same road segment set to obtain a reference road segment set; fit a second functional relationship between the road surface feature set and the traffic risk probability values ​​in the reference road segment set; iteratively update the second functional relationship by changing the number of the initial road segment set; and determine the division result of the initial road segment set as the final divided road segment set when the iteration terminates.

[0118] In one possible implementation, the traffic risk probability value includes at least one of a traffic accident probability value and a dangerous driving probability value, and the true value of the traffic risk probability is calculated based on traffic data associated with the road segment.

[0119] In one possible implementation, the traffic data includes at least one of the following: road network data recording road surface conditions, vehicle driving data collected by vehicle monitoring equipment, traffic management data, and traffic maintenance data.

[0120] Corresponding to the application scenarios and methods provided in the embodiments of this application, the embodiments of this application also provide another traffic risk prediction device. For example... Figure 6 The diagram shown is a structural block diagram of a traffic risk prediction device 600 according to another embodiment of this application. The device 600 may include:

[0121] The road segment determination module 601 is used to determine the target road segment currently being traveled by the vehicle;

[0122] The result acquisition module 602 is used to acquire the traffic risk prediction result of the target road segment and the risk handling strategy determined based on the traffic risk prediction result; the traffic risk prediction result is determined based on the risk indicator information represented by the first traffic data in the set risk dimension and the second traffic data in the traffic data excluding the first traffic data, and the first traffic data is selected from the traffic data associated with the target road segment.

[0123] The result prompting module 603 is used to prompt the traffic risk prediction results and risk handling strategies based on the traffic application.

[0124] The functions of each module in each device in the embodiments of this application can be found in the corresponding description in the above method, and they have corresponding beneficial effects, which will not be repeated here.

[0125] Figure 7 This is a block diagram of an electronic device used to implement embodiments of this application. For example... Figure 7 As shown, the electronic device includes a memory 701 and a processor 702. The memory 701 stores a computer program that can run on the processor 702. When the processor 702 executes the computer program, it implements the method described in the above embodiments. The number of memories 701 and processors 702 can be one or more.

[0126] The electronic device also includes:

[0127] The communication interface 703 is used to communicate with external devices and perform data exchange and transmission.

[0128] If the memory 701, processor 702, and communication interface 703 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 7 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0129] Optionally, in a specific implementation, if the memory 701, processor 702, and communication interface 703 are integrated on a single chip, then the memory 701, processor 702, and communication interface 703 can communicate with each other through an internal interface.

[0130] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method provided in this application.

[0131] This application also provides a chip including a processor for calling and executing instructions stored in a memory, causing a communication device with the chip installed to perform the method provided in this application.

[0132] This application also provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, output interface, processor, and memory are connected through an internal connection path. The processor is used to execute code in the memory. When the code is executed, the processor is used to execute the method provided in the application embodiment.

[0133] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0134] Further, optionally, the aforementioned memory may include read-only memory and random access memory. The memory may be volatile memory or non-volatile memory, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0135] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0136] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0137] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0138] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process. Furthermore, the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functionality involved.

[0139] The logic and / or steps described in the flowchart or otherwise herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0140] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. All or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware, the program being stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiments.

[0141] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. This storage medium can be a read-only memory, a disk, or an optical disk, etc.

[0142] The above description is merely an exemplary embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope described in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting traffic risks, comprising: Obtain traffic data associated with the road segment to be identified, including traffic data of various categories; A correlation analysis is performed on the acquired traffic data of various categories and traffic risks to determine the degree of correlation between each category of traffic data and traffic risks. Based on the degree of correlation, the first traffic data with a correlation degree less than a certain threshold is selected from the traffic data. The first traffic data is fitted and calculated according to a predetermined first functional relationship to convert the first traffic data into risk indicator information that is more correlated with traffic risk; wherein, the first functional relationship is used to characterize the mapping relationship between the first traffic data and the traffic risk of a set risk dimension; based on the determined risk indicator information and the second traffic data other than the first traffic data in the traffic data, traffic risk prediction is performed on the road segment to be identified.

2. The method according to claim 1, wherein, The first traffic data includes at least road network data that records road surface conditions.

3. The method according to any one of claims 1-2, wherein, The traffic risk prediction for the road segment to be identified, based on the determined risk indicator information and second traffic data other than the first traffic data, includes: The risk indicator information and the second traffic data (excluding the first traffic data) are input into a pre-trained risk prediction model to obtain the traffic risk prediction result of the road segment to be identified. The traffic risk prediction result includes at least one of the following: whether there is a risk, the type of risk, and the probability of risk.

4. The method according to any one of claims 1-2, wherein, Before performing traffic risk prediction on the road segment to be identified based on the determined risk indicator information and second traffic data excluding the first traffic data from the traffic data, the method further includes: The first traffic data and / or the second traffic data are preprocessed, and the preprocessing includes at least one of data cleaning, data error correction and data imputation.

5. The method according to any one of claims 1-2, wherein, The method further includes: Based on the traffic risk prediction results, corresponding risk management strategies are determined. The risk management strategies include at least one of the following: risk warning based on traffic applications, traffic risk solutions, and risk consequence handling plans.

6. The method according to claim 3, wherein, The risk prediction model is generated through the following steps: Obtain traffic data samples associated with road segment samples; Data identification is performed on the acquired traffic data samples to select a first traffic data sample from the traffic data samples, the first traffic data sample being used at least to characterize the road surface condition; Based on the risk indicator information represented by the first traffic data sample in the set risk dimension, the second traffic data sample other than the first traffic data sample in the traffic data sample, and the traffic risks marked for the traffic data sample, a risk prediction model for predicting traffic risks is trained.

7. The method according to any one of claims 1-2, wherein, The road segment to be identified is determined through the following steps: Continuous road segments whose traffic risk probability values ​​differ from a first set threshold are added to the same road segment set, and the continuous road segments in the road segment set are respectively used as road segments to be identified.

8. The method according to claim 7, wherein, The step of adding consecutive road segments whose traffic risk probability values ​​differ from a first preset threshold to the same segment set includes: Aggregate consecutive road segments whose road surface condition similarity exceeds a second set threshold to obtain multiple initial road segment sets; Road segments whose traffic risk probability values ​​differ from a first set threshold and are in the same initial road segment set are grouped into the same road segment set, wherein the traffic risk probability values ​​are determined based on historical traffic data.

9. The method according to claim 8, wherein, The step of grouping road segments whose traffic risk probability values ​​differ from a first set threshold and are in the same initial road segment set into the same road segment set includes: Road segments whose traffic risk probability values ​​differ from a first set threshold and are in the same initial road segment set are grouped into the same road segment set to obtain a reference road segment set. Fit a second functional relationship between the set of road surface features and the probability value of traffic risk in the reference road segment set; The second functional relationship is iteratively updated by changing the number of segments in the initial road segment set; When the iteration terminates, the initial road segment set is determined as the final road segment set.

10. The method according to claim 7, wherein, The traffic risk probability value includes at least one of the traffic accident probability value and the dangerous driving probability value, and the true value of the traffic risk probability is calculated based on traffic data associated with the road segment.

11. The method according to any one of claims 1-2, wherein, The second traffic data includes at least one of the following: vehicle driving data, traffic management data, and traffic maintenance data collected by vehicle monitoring equipment.

12. A method for predicting traffic risks, comprising: Determine the target road segment currently being traveled by the vehicle; Obtain the traffic risk prediction results for the target road segment and the risk management strategy determined based on the traffic risk prediction results; The traffic risk prediction result is determined based on the risk indicator information represented by the first traffic data in a set risk dimension and the second traffic data excluding the first traffic data. The risk indicator information represented by the first traffic data is obtained in the following way: acquiring traffic data associated with the target road segment, the traffic data including multiple categories of traffic data; performing correlation analysis on the acquired multiple categories of traffic data and traffic risk to determine the degree of correlation between each category of traffic data and traffic risk; selecting the first traffic data whose correlation degree is less than a certain threshold from the traffic data based on the degree of correlation; and performing fitting calculation on the first traffic data according to a predetermined first functional relationship to convert the first traffic data into risk indicator information with a stronger correlation to traffic risk; wherein, the first functional relationship is used to represent the mapping relationship between the first traffic data and the traffic risk in the set risk dimension. Based on the traffic application, the system provides traffic risk prediction results and risk management strategies.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor, when executing the computer program, implements the method of any one of claims 1-12.

14. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method of any one of claims 1-12.

Images