Traffic engineering BIM and VR technology combined interactive design and simulation system

Abstract

The invention relates to a traffic engineering BIM (Building Information Modeling) and VR (Virtual Reality) technology combined interactive design and simulation system, which relates to the field of traffic engineering and comprises an interactive configuration module used for selecting simulation configuration to obtain user feedback information; the route mining module is used for performing networking frequent mining to obtain a mobile route case set; the virtual driving module is used for starting simulated driving and obtaining a three-dimensional dynamic time sequence diagram of a vehicle and heart rate monitoring time sequence information of a user; the safety scoring module is used for receiving a user driving safety score through the user side; the risk evaluation module is used for obtaining a driving risk evaluation value set; and the safety identification module is used for performing safety identification and then sending the information to the traffic engineering design end. Through combination of BIM and VR technologies, virtual simulation acquisition and analysis of user driving behaviors are realized, so that the driving risk of a design scheme is evaluated in a targeted manner, traffic engineering optimization design is guided, and the safety of traffic engineering design is improved.

Description

Technical Field

[0001] The present invention relates to the field of traffic engineering, and in particular to an interactive design and simulation system combining BIM and VR technology for traffic engineering. Background Art

[0002] Urban traffic engineering construction is increasing day by day. In order to ensure traffic safety, traffic engineering design needs to fully consider factors such as road environment, traffic flow, and the resulting user driving behavior patterns and potential driving risks. However, in the existing technology, traffic engineering design generally lacks universal user driving behavior analysis and design scheme risk assessment methods. It is difficult for designers to fully obtain user driving behavior data and accurately evaluate the safety performance of the design scheme, resulting in insufficient safety of the final traffic engineering.

[0003] For example, in complex sections such as road intersections, multi-lane lane changes, toll booths and ramps, traffic accidents and safety hazards are easily caused due to differences in user driving behavior and defects in the design scheme itself. For another example, traffic projects on mountain roads, roads near water and cliffs, and in severe weather conditions have a high accident rate due to the lack of targeted driving risk assessment and safety design. Therefore, how to conduct user driving behavior analysis and design scheme risk assessment in traffic engineering design, and on this basis evaluate the safety performance of traffic engineering design schemes, and then guide the optimization design of traffic engineering, so as to effectively improve traffic operation safety, is an urgent problem to be solved in this field. Summary of the invention

[0004] In order to solve the technical problem that traffic engineering design in the prior art lacks universal user driving behavior analysis and design scheme risk assessment, resulting in insufficient safety of traffic engineering design, the present invention provides an interactive design and simulation system combining traffic engineering BIM and VR technology to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] The present invention provides an interactive design and simulation system combining BIM and VR technology for traffic engineering, including: an interactive configuration module, which is used to display the BIM model of the traffic engineering and the simulation configuration menu on the user side and prompt the user to select the simulation configuration, and obtain user feedback information, wherein the user feedback information includes the BIM model three-dimensional area and the simulated transportation tool type; a route mining module, which is used to mine the networking frequency of the BIM model three-dimensional area according to the simulated transportation tool type, and obtain a mobile route use case set; a virtual driving module, which is used to send the mobile route use case set and the BIM model three-dimensional area to the associated VR of the simulated transportation tool type. The driving simulation chamber reminds the user to enter the associated VR driving simulation chamber to start simulated driving, obtain the three-dimensional dynamic timing diagram of the vehicle and the user's heart rate monitoring timing information; the safety scoring module is used to receive the user's driving safety score through the user end; the risk assessment module is used to perform a safety assessment based on the three-dimensional dynamic timing diagram of the vehicle, the user's heart rate monitoring timing information and the user's driving safety score through the driving risk assessment network, and obtain a driving risk evaluation value set; the safety identification module is used to perform a safety identification on the three-dimensional area of ​​the BIM model and send it to the traffic engineering design end when the driving risk evaluation value sets are all less than the driving risk evaluation threshold.

[0007] The beneficial effects of the present invention are:

[0008] Through the interactive configuration module, when the user terminal displays the traffic engineering BIM model and simulation configuration menu and prompts the user to select the simulation configuration, the user feedback information is obtained, including the BIM model three-dimensional area and the simulated vehicle type, and the interaction between the user and the system is realized. The user can select the road area and simulated vehicle to be evaluated, and the system obtains the user's configuration information. Through the route mining module, the BIM model three-dimensional area is frequently mined according to the simulated vehicle type, and the mobile route use case set is obtained. The mobile route use case set covering common driving scenarios is mined to provide a data basis for subsequent simulated driving analysis. Through the virtual driving module, the mobile route use case set and the BIM model three-dimensional area are sent to the associated VR driving simulation warehouse of the simulated vehicle type, and the user is reminded to enter the associated VR driving simulation warehouse to start simulated driving, and the three-dimensional dynamic timing diagram of the vehicle and the user's heart rate monitoring timing information are obtained, and comprehensive driving behavior data are obtained. Through the safety scoring module, the user terminal receives the user's driving safety score as a reference data for driving risk assessment. Through the risk assessment module, a safety assessment is performed based on the driving risk assessment network, the three-dimensional dynamic timing diagram of the vehicle, the user's heart rate monitoring timing information and the user's driving safety score, and a driving risk evaluation value set is obtained to quantitatively evaluate the driving safety risk level of the road section design scheme. Through the safety identification module, when the driving risk evaluation value set is less than the driving risk evaluation threshold, the three-dimensional area of ​​the BIM model is safety marked and sent to the traffic engineering design end, so as to determine whether the safety of the road section design meets the standard, and provide identification feedback for the design scheme with good safety performance to guide the optimization design of the traffic engineering.

[0009] Through the combined application of BIM and VR technology, a virtual driving environment close to reality is constructed, and comprehensive collection and accurate analysis of user driving behavior are achieved. On this basis, the safety performance of traffic engineering design solutions is comprehensively evaluated, providing quantitative basis and improvement direction for design optimization, effectively improving the quality of traffic engineering design. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of the structure of an interactive design and simulation system combining BIM and VR technology for traffic engineering provided by the present invention;

[0011] Figure 2 A schematic diagram of the structure of an electronic device provided by the present invention;

[0012] Figure 3 A schematic diagram of the structure of a computer-readable storage medium provided by the present invention.

[0013] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0014] Interaction configuration module 11, route mining module 12, virtual driving module 13, safety scoring module 14, risk assessment module 15, safety identification module 16, electronic device 200, memory 210, processor 220, first computer program 211, computer readable storage medium 300, second computer program 311. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0016] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0017] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0018] Embodiment 1:

[0019] like Figure 1 As shown, the embodiment of the present invention provides an interactive design and simulation system combining BIM and VR technology for traffic engineering, including:

[0020] The interactive configuration module 11 is used to display the traffic engineering BIM model and simulation configuration menu on the user side and prompt the user to select the simulation configuration, and obtain user feedback information, wherein the user feedback information includes the BIM model three-dimensional area and the simulated transportation tool type.

[0021] Specifically, the interactive configuration module 11 is mainly responsible for interacting with the user and receiving the user's configuration selection of traffic simulation parameters. Specifically, first, through computer graphics rendering technology, the three-dimensional BIM model of the entire traffic project is presented on the user's display device (such as a monitor or touch screen). The user can use human-computer interaction devices such as a mouse and a stylus to observe and understand the spatial layout and detailed features of the BIM model in all directions through operations such as rotation, translation, and zooming. While the BIM model is displayed, the interactive configuration module 11 will also provide a simulation configuration menu on the interface. The simulation configuration menu lists the various parameters that require user decision-making during the simulation process in the form of lists, drop-down boxes, radio buttons, etc., such as traffic flow, vehicle speed distribution, signal timing schemes, etc. Users can select and modify the configuration items on the menu according to actual needs and analysis objectives to customize personalized simulation plans.

[0022] There are two configuration items on the simulation configuration menu, one is the simulation area selection, and the other is the vehicle type. For the simulation area selection, the user needs to select a three-dimensional area on the displayed BIM model, usually a section of road, intersection or square on the road network, to determine the spatial scope of the subsequent simulation. The user can use the mouse to drag or touch the finger to select a three-dimensional area boundary on the three-dimensional BIM model. The interactive configuration module 11 will automatically intercept the digital description information of the corresponding area in the three-dimensional BIM model for subsequent simulation calculations. For the vehicle type, the user needs to specify one or several vehicles in the list of alternative models (such as cars, buses, trucks, etc.) as the object of simulation. Due to the different kinematic characteristics and requirements for road conditions of different models, they need to be treated separately in the simulation. After completing the necessary parameter settings, the user clicks the OK button to complete the configuration process. The interactive configuration module 11 will summarize the user's settings to form user feedback information, including the selected three-dimensional area of ​​the BIM model and the type of simulated vehicle.

[0023] Through the interactive configuration module 11, the user's decision intention and the system's simulation capabilities are effectively obtained. The user can follow simple and intuitive operation steps to flexibly configure parameters such as the 3D BIM model area and the target vehicle model, thereby obtaining a simulation effect that better meets actual needs. This interactive configuration method lowers the user's usage threshold, improves the system's usability and practicality, and lays the foundation for subsequent refined traffic simulation analysis.

[0024] The route mining module 12 is used to mine the networking frequency of the three-dimensional area of ​​the BIM model according to the simulated transportation type to obtain a set of mobile route use cases.

[0025] Specifically, the main function of the route mining module 12 is to discover and extract typical travel routes that meet specific types of vehicles from the BIM model data. Specifically, the route mining module 12 receives the user-selected three-dimensional area and vehicle type transmitted by the interactive configuration module 11, and performs pattern matching and frequency mining accordingly.

[0026] First, the route mining module 12 retrieves the road network information related to the type of vehicle in the BIM model database according to the type of simulated vehicle specified by the user. Among them, the road network information includes attribute parameters such as the geometry, slope, curvature of the road, as well as the spatial position and descriptive information of road ancillary facilities such as traffic signs, signal lights, and lane lines. By analyzing these data, the route mining module 12 can infer the road sections and traffic rules suitable for the type of vehicle. Next, the route mining module 12 focuses on the target area selected by the user on the three-dimensional BIM model, that is, the three-dimensional area of ​​the BIM model. The route mining module 12 extracts the road network topology within the three-dimensional area of ​​the BIM model and constructs a directed graph model consisting of nodes (such as intersections) and edges (such as road sections). On this basis, the route mining module 12 applies frequent pattern mining algorithms (such as Apriori, FP-growth, etc.) to discover frequently occurring path patterns from massive historical traffic trajectory data. These frequent patterns represent the typical driving routes of vehicles in the area, reflecting the driver's usual choices in time and space.

[0027] In the frequent pattern mining process, the route mining module 12 needs to measure and compare the similarities of different routes. Among them, the route mining module 12 uses sequence similarity measurement methods such as dynamic time warping and longest common subsequence to measure the similarity of different routes in space and time, so as to summarize representative frequent route patterns. Afterwards, the route mining module 12 outputs a series of typical routes that conform to specific types of transportation and regional environments to form a mobile route use case set, in which each route not only contains the spatial trajectory coordinates of the vehicle's movement, but also comes with driving state parameters such as speed, acceleration, and steering. These real route use case data can be used to support subsequent traffic simulation and virtual scene construction, allowing users to simulate and experience the vehicle driving process under different working conditions in a realistic virtual environment.

[0028] Through the route mining module 12, data mining is used to extract and summarize the general rules and typical patterns of driving from massive and complex BIM models and traffic trajectory data. This automated and intelligent route mining method can greatly improve the pertinence and accuracy of traffic simulation and provide data support for subsequent virtual driving training, traffic planning and design, and other applications.

[0029] The virtual driving module 13 is used to send the mobile route use case set and the BIM model three-dimensional area to the associated VR driving simulation cabin of the simulated vehicle type, and remind the user to enter the associated VR driving simulation cabin to start simulated driving, and obtain the three-dimensional dynamic timing diagram of the vehicle and the user's heart rate monitoring timing information.

[0030] Specifically, the virtual driving module 13 is responsible for constructing an immersive virtual driving environment and collecting user driving behavior data. Its input includes a mobile route use case set provided by the route mining module 12 and a user-selected BIM model three-dimensional area provided by the interactive configuration module 11.

[0031] After receiving the above input data, the virtual driving module 13 first constructs a virtual three-dimensional driving scene based on the mobile route use case set and the BIM model three-dimensional area. The scene needs to reproduce the topography, road network, transportation facilities and other elements of the target area as realistically as possible, and also generate moving vehicle objects in the virtual scene according to the path shape and vehicle speed parameters in the mobile route use case set. Among them, the entire virtual scene construction process uses professional three-dimensional modeling software and VR engines, such as Unity3D, Unreal Engine, etc.

[0032] After the virtual driving scene is constructed, the virtual driving module 13 associates it with a physical VR driving simulation chamber. The so-called VR driving simulation chamber is an immersive driving simulator equipped with hardware such as a multi-degree-of-freedom motion platform, a steering wheel, and an accelerator and brake pedal. The associated VR driving simulation chamber of the simulated vehicle type refers to a simulator whose hardware configuration and degrees of freedom of motion match the actual driving experience of the target vehicle model. The virtual driving module 13 connects the virtual driving scene with the display system, motion control system, etc. of the VR driving simulation chamber through a software interface, so that the user's operations in the simulation chamber can be reflected in real time to the motion state of the vehicle in the virtual scene.

[0033] When everything is ready, the virtual driving module 13 sends a prompt message to the user, guiding him to enter the VR driving simulation cabin and start simulated driving. In the simulation cabin, the user controls the virtual vehicle to drive on the virtual route through hardware devices such as steering wheels and pedals, and observes the changes in the virtual scene through the display screen of the simulation cabin. During the driving process, the virtual driving module 13 will continuously record the following two types of data: one is the motion trajectory, speed, acceleration, steering angle and other state parameters of the virtual vehicle, forming a three-dimensional dynamic time series diagram of the vehicle; the other is the user's physiological response data, especially the heart rate change curve, forming the user's heart rate monitoring time series information. Both types of data are sampled and stored in the form of time series for subsequent driving behavior analysis and risk assessment.

[0034] Through the virtual driving module 13, VR virtual reality technology is used to combine the BIM model of traffic engineering and typical route use cases to build an immersive virtual driving environment, and through linkage with the physical driving simulator, users can experience the road traffic conditions in the target area in an immersive way. At the same time, the time-series driving data and physiological response data collected by the virtual driving module 13 provide a detailed basis for evaluating the safety of traffic planning and design. This method of combining virtual driving with real-time data collection greatly improves the authenticity and data richness of traffic simulation.

[0035] The safety scoring module 14 is used to receive the user's driving safety score through the user terminal.

[0036] Specifically, the safety scoring module 14 is used to collect users' subjective evaluation of their driving experience in the virtual driving environment. This module is directly oriented to end users and obtains driving safety scoring data fed back by users through a human-computer interaction interface.

[0037] After the user completes the virtual driving task, the safety scoring module 14 will automatically pop up a scoring interface to guide the user to conduct a safety assessment of the driving process just now. The interface can adopt the common five-star rating or percentage rating form, allowing users to intuitively express their subjective feelings about the safety and comfort of the virtual road environment and traffic conditions. Considering that the user's safety assessment may be affected by many factors, the scoring interface can also list several common evaluation dimensions, such as road linear design, traffic flow, and field of vision, to help users score comprehensively, thereby obtaining the user's driving safety score from the user's subjective perspective.

[0038] Through the safety scoring module 14, the user is guided to score and provide feedback on the virtual driving, and the user's driving safety score is obtained, so as to evaluate the safety performance of road traffic planning and design from the user's subjective perspective, and improve the pertinence and effectiveness of road design and traffic management.

[0039] The risk assessment module 15 is used to perform safety assessment based on the three-dimensional dynamic timing diagram of the vehicle, the user's heart rate monitoring timing information and the user's driving safety score through the driving risk assessment network to obtain a driving risk evaluation value set.

[0040] Specifically, the risk assessment module 15 receives three types of data: first, the three-dimensional dynamic time sequence diagram of the vehicle collected by the virtual driving module 13; second, the user's heart rate monitoring time sequence information recorded synchronously during the virtual driving process; and third, the user's subjective driving safety score collected by the safety scoring module 14. The goal of the risk assessment module 15 is to comprehensively utilize these three types of data to objectively evaluate the safety risk level of the virtual driving scene.

[0041] In order to realize the fusion analysis of multi-source heterogeneous data, a driving risk assessment network is designed in the risk assessment module 15. The network uses deep learning technology to extract features and fuse information of different types of time series data and scoring data through models such as convolutional neural networks and long short-term memory networks. For example, the three-dimensional dynamic time series diagram of the vehicle is first processed by a convolutional neural network to extract feature sequences such as speed, acceleration, and steering during the vehicle's movement; the user's heart rate monitoring time series information is modeled through a long short-term memory network to explore the correlation pattern between the user's physiological response and driving risk; the user's driving safety score is used as an important subjective judgment basis and is input into the fully connected network together with the first two types of objective data, and a comprehensive driving risk assessment result is obtained through weighted fusion.

[0042] The training of the driving risk assessment network requires a large amount of sample data. To this end, the risk assessment module 15 needs to connect data with the virtual driving module 13 and the safety scoring module 14, and regularly obtain the latest driving behavior data and user scoring data from these two modules. At the same time, in order to improve the diversity and coverage of samples, the risk assessment module 15 can also access external traffic accident databases, driver physiological response data sets, etc. as a supplement to training data. In the specific training process, the driving risk assessment network adopts an end-to-end supervised learning paradigm, takes the known traffic accident risk level in the real world as the training target, updates the network weight parameters through the back propagation algorithm, and continuously improves the accuracy of risk assessment. After the training of the driving risk assessment network is completed, it can be used to predict the safety risk of new virtual driving data. The risk assessment module 15 inputs the three types of data collected during the virtual driving process into the trained driving risk assessment network, and automatically obtains a driving risk evaluation value. The evaluation value is a real number between 0 and 1, and the larger the value, the higher the driving risk. By batch evaluating driving data of different road sections and different time periods, the risk assessment module 15 finally outputs a driving risk evaluation value set to comprehensively describe the safety risk distribution of road traffic in the three-dimensional area of ​​the BIM model.

[0043] Through the risk assessment module 15, the driving risk assessment network is used to automatically assess the safety risk level of road traffic, which greatly improves the accuracy and practicality of driving risk assessment and provides support for improving the safety level of road traffic.

[0044] The safety identification module 16 is used to carry out safety identification on the three-dimensional area of ​​the BIM model and send it to the traffic engineering design end when the driving risk evaluation value set is less than the driving risk evaluation threshold.

[0045] Specifically, the safety identification module 16 first sets a driving risk assessment threshold. This threshold is given empirically by experts in the transportation field based on the safety standards of the transportation authority or with reference to the statistical distribution of a large amount of historical traffic accident data. When all driving risk assessment values ​​in the driving risk assessment value set are less than this threshold, it means that the overall road traffic safety level in the three-dimensional area of ​​the BIM model is high and can meet the requirements of safe passage; conversely, if there are driving risk assessment values ​​that exceed the threshold, it indicates that there are still certain safety hazards in the area, which require further investigation and rectification. Therefore, when it is determined that the driving risk assessment value sets are all less than the driving risk assessment threshold, a safety mark is made in the three-dimensional area of ​​the BIM model and sent to the traffic engineering design end, indicating that the three-dimensional area of ​​the BIM model meets the requirements.

[0046] Through the safety identification module 16, when the driving risk assessment value sets are all less than the driving risk assessment threshold, the three-dimensional area of ​​the BIM model is safety marked and sent to the traffic engineering design end to provide decision support for the traffic engineering design, thereby improving the safety of the traffic engineering design.

[0047] Furthermore, the embodiment of the present application also includes:

[0048] Taking the BIM model three-dimensional area as a background constraint and the simulated vehicle type as a foreground constraint, collecting a set of compliant driving samples through networking, wherein the compliant driving sample set includes a moving route record data set;

[0049] Performing frequent coefficient identification on the mobile route record data set to obtain a route frequent coefficient set;

[0050] The moving route use case set whose number of moving route use cases is greater than or equal to a frequent coefficient threshold value in the route frequent coefficient set is sorted out from the moving route record data set.

[0051] In a feasible implementation, when the route mining module 12 automatically extracts frequently occurring typical driving routes from the three-dimensional area of ​​the BIM model according to the type of simulated vehicle specified by the user to form a mobile route use case set, first, the route mining module 12 collects real sample data of compliant driving behavior on a large scale on the Internet. In the collection process, two constraints are fully considered: one is to use the three-dimensional area of ​​the BIM model as a background constraint to be highly consistent with the background elements such as the layout of traffic facilities, road surface structure, and surrounding terrain of the real road environment, so as to ensure that the collected driving behavior data matches the virtual environment; the other is to use the type of simulated vehicle as a foreground constraint, that is, the collected samples must be the real driving data of the type of simulated vehicle selected by the user (such as cars, trucks, etc.) to reflect the driving characteristics of different models. Based on the large-scale collection on the Internet, the route mining module 12 obtains a huge set of compliant driving samples, which contains a large number of real vehicle mobile route GPS record data to form a mobile route record data set.

[0052] Then, the route mining module 12 performs frequent pattern mining on the collected massive mobile route record data. Specifically, the route mining module 12 uses frequent pattern mining algorithms, such as Apriori, FP-growth, etc., to count the frequency of each mobile route in the total sample set, and calculate its corresponding route frequency coefficient. The route frequency coefficient is an indicator to measure the frequency of the route. The higher the coefficient, the more frequently the route appears in real traffic. Through frequent pattern mining, the route mining module 12 extracts a set of route frequency coefficients from the original massive record data to characterize the frequent characteristics of different routes. Afterwards, the route mining module 12 selects route records whose frequency coefficients are greater than or equal to the frequency coefficient threshold from the obtained route frequency coefficient set according to the preset frequency coefficient threshold, and forms the final mobile route use case set. The setting of the frequency coefficient threshold needs to comprehensively consider the actual characteristics of road traffic and the statistical requirements of the simulation experiment, and is given by traffic engineering experts based on experience and data analysis results. The mobile route use case set obtained after screening reflects the most commonly used driving routes of the selected simulated transportation type under specific background constraints. It is highly representative and can be used for subsequent traffic virtual simulation.

[0053] Through the route mining module 12, the function of automatically mining typical routes from real driving behavior data is realized, and statistically representative driving routes are efficiently extracted from massive heterogeneous data, which not only improves the coverage and richness of route samples, but also reduces the cost of manual calibration, providing high-quality data support for subsequent traffic virtual simulation.

[0054] Furthermore, the embodiment of the present application also includes:

[0055] Performing pairwise distance marking on the moving route record data sets to obtain a moving route distance calibration data set;

[0056] Extracting first moving route record data according to the moving route record data set;

[0057] Based on the moving route distance calibration data set, counting the number of routes whose moving route distance deviation between the moving route record data set and the first moving route record data set is less than a moving route distance deviation threshold, and counting the ratio of the number of routes to the total number of routes in the moving route record data set, and setting it as a first moving route frequency coefficient;

[0058] The first moving route frequency coefficient is added to the route frequency coefficient set.

[0059] In a preferred embodiment, when the frequent coefficients of the mobile route record data set are marked, first, the route mining module 12 performs pairwise distance calculation and marking on the routes in the mobile route record data set. Specifically, the route mining module 12 traverses each mobile route in the mobile route record data set, and calculates the distance measurement value between it and other routes by means of Euclidean distance, DTW distance, Fréchet distance, etc. Through pairwise distance calculation, the route mining module 12 obtains a mobile route distance calibration data set, which records the distance measurement value between any two routes. Then, the route mining module 12 traverses the mobile route record data set, extracts a route from it each time, and marks it as the first mobile route record data. The record data will be used as a reference benchmark for subsequent calculation of the occurrence frequency of other routes.

[0060] Subsequently, the route mining module 12 uses the obtained mobile route distance calibration data set to count the number of routes that are close to the first mobile route record data. Specifically, the route mining module 12 traverses the mobile route distance calibration data set to determine whether the distance between the first mobile route record data and other routes is less than the preset mobile route distance deviation threshold. If the distance measurement value between the two routes is less than the threshold, it means that the two routes are very close in spatial shape and movement trend, and can be regarded as different instances of the same typical route. Through this step, the route mining module 12 obtains a group of route records that are close to the first mobile route record data. By dividing the number of route records in this group by the total number of routes in the mobile route record data set, the first mobile route frequency coefficient can be obtained, which represents the relative frequency of the first mobile route record data in real traffic. Afterwards, the route mining module 12 adds the calculated first mobile route frequency coefficient to the route frequency coefficient set to provide a basis for subsequent frequent route screening.

[0061] Through the route mining module 12, the frequent coefficient identification of the mobile route record data set is realized, and the frequently appearing typical routes are automatically extracted from the original route data, and their occurrence frequency is quantified. Compared with the traditional manual calibration method, it can process massive route data, comprehensively mine frequent patterns in real traffic, and control the clustering granularity of frequent routes through the mobile route distance deviation threshold, which has strong adaptability.

[0062] Furthermore, the embodiment of the present application also includes:

[0063] Construct a moving route distance calibration function:

[0064]

[0065] in, Represents the first dimension coordinate of the i-th point on path P1, Represents the second-dimensional coordinate of the i-th point on path P1, Represents the third-dimensional coordinate of the i-th point on path P1, Characterizes the first dimension coordinate of the jth point on path P2, Characterizes the second-dimensional coordinate of the j-th point on path P2, Represents the third-dimensional coordinate of the j-th point on path P2, is the path P2 with The nearest point, Represents the direction vector between the jth point on path P2 and the point after it, Represents the direction vector between the i-th point on path P1 and the point after it, θ ij represents the angle difference between two direction vectors determined by calculating the dot product and modulus of the two direction vectors. ΔL represents the length deviation between path P1 and path P2. ij Characterization and The Euclidean distance, M represents the total number of points on path P2, N represents the total number of points on path P1, λ represents the weight coefficient of direction difference, which is used to adjust the contribution of direction difference in the total deviation, μ represents the weight coefficient of path length deviation, which is used to adjust the contribution of path length deviation in the total deviation, w ij Characterization and Importance weight in path deviation calculation;

[0066] The moving route distance calibration function is used to perform distance marking on the moving route record data sets in pairs to obtain a moving route distance calibration data set.

[0067] In a preferred embodiment, in the frequent coefficient identification, the route mining module 12 performs pairwise distance calculation and calibration on the routes in the mobile route record data set. In order to accurately quantify the similarity between different routes, a mobile route distance calibration function is proposed to comprehensively consider the characteristics of the route in multiple aspects such as spatial shape, movement direction, length difference, etc., and construct a distance measurement model between routes.

[0068] The specific form of the moving route distance calibration function is: Construct the moving route distance calibration function:

[0069]

[0070] Where D(P1, P2) represents the overall deviation distance between path P1 and path P2, which is obtained by summing the three sub-distance metrics according to the weights, namely: Euclidean distance d ij 、Angle difference |θ ij | and length deviation ΔL. In the calculation of the overall deviation distance D(P1, P2), it is necessary to traverse all sampling points on path P1 and path P2, and for each pair of sampling points and Calculate the three sub-distance metrics between them and multiply them by the corresponding weight coefficients λ, μ and w ij , and finally sum the weighted sub-distances of all sampling point pairs. Among them, λ is used to balance the contribution of angle difference in the total deviation, μ is used to balance the contribution of length deviation in the total deviation, and w ij It indicates the importance of the i-th sampling point and the j-th sampling point in the deviation calculation. By setting these weight parameters, the influence of different features on the overall route deviation can be flexibly adjusted to improve the adaptability of the distance metric.

[0071] Among the three sub-distance metrics, the Euclidean distance d ij Represents the straight-line distance between two sampling points in three-dimensional space, which is used to characterize the spatial shape difference of the path. Angle difference θ ij represents the angle between the path tangent vectors at two sampling points, which is used to characterize the difference in the movement direction of the path. Specifically, θ ij By calculating the two tangent vectors and The dot product of and The length deviation ΔL represents the difference in the total length of the two paths and is used to characterize the scale difference of the paths.

[0072] Based on the above mobile route distance calibration function, the route mining module 12 can calculate the deviation distance between any two routes in the mobile route record data set. In specific implementation, the route mining module 12 traverses all route pairs in the mobile route record data set, extracts the sampling point coordinates and tangent vector features of each pair of routes, substitutes them into the distance calibration function to solve the deviation distance D (P1, P2) between the route pairs, and stores the results in the mobile route distance calibration data set. Through exhaustive calculation, a complete mobile route distance calibration data set is finally obtained, and the similarity between any two routes in the mobile route record data set is comprehensively evaluated.

[0073] By constructing a mobile route distance calibration function, a measurement model that comprehensively considers spatial shape, movement direction, and length difference is provided to measure the similarity between routes. By designing three sub-distance metrics, namely Euclidean distance, angle difference, and length deviation, and introducing flexible weight parameters, the deviation characteristics between routes can be fully characterized, providing a data basis for subsequent frequent route mining.

[0074] Furthermore, the embodiment of the present application also includes:

[0075] Collecting a first sample of compliant driving to be selected through networking, wherein the first sample of compliant driving to be selected includes a driving scene record model and a vehicle record type;

[0076] When the vehicle record type is the same as the simulated vehicle type, the BIM model stereoscopic area is input into a first feature extraction channel of a background comparison network to obtain a first feature extraction result, and the driving scene record model is input into a second feature extraction channel of a background comparison network to obtain a second feature extraction result, wherein the first feature extraction channel and the second feature extraction channel are twin convolutional neural networks;

[0077] Inputting the first feature extraction result and the second feature extraction result into a feature comparison channel to obtain a feature deviation value;

[0078] When the feature deviation value is less than or equal to a feature deviation threshold, the first candidate compliant driving sample is added to the compliant driving sample set.

[0079] In a preferred embodiment, in the networked frequency mining, the route mining module 12 needs to collect real driving samples that meet specific background constraints and foreground constraints on a large scale on the Internet. In order to automatically screen out valid samples that meet the above two constraints, a method for collecting a set of compliant driving samples based on deep learning is proposed.

[0080] First, the route mining module 12 collects a candidate compliance driving sample from the Internet through web crawlers and other technologies as the first candidate compliance driving sample. The first candidate compliance driving sample contains two parts: one is the driving scene record model, that is, the three-dimensional modeling data of the real road environment; the other is the vehicle record type, that is, the vehicle type that generates the driving record. Then, it is determined whether the vehicle record type is the same as the simulated vehicle type. If they are the same, it means that the first candidate compliance driving sample meets the foreground constraint. At this time, the route mining module 12 uses the pre-trained background comparison network to determine whether the driving scene of the first candidate compliance driving sample matches the three-dimensional area of ​​the BIM model. Among them, the background comparison network adopts the structure of a twin convolutional neural network, which includes two parallel feature extraction channels, namely the first feature extraction channel and the second feature extraction channel. Among them, the first feature extraction channel is responsible for feature extraction of the three-dimensional area of ​​the BIM model, and obtains the first feature vector reflecting the environmental characteristics of the three-dimensional area of ​​the BIM model as the first feature extraction result; the second feature extraction channel is responsible for feature extraction of the driving scene record model, and obtains the second feature vector reflecting the characteristics of the real road environment as the second feature extraction result.

[0081] After obtaining the first feature extraction result and the second feature extraction result, the route mining module 12 inputs the first feature extraction result and the second feature extraction result into the feature comparison channel of the background comparison network to calculate the feature deviation value between the two features. Among them, the feature deviation calculation method of the feature comparison channel can use Euclidean distance, cosine similarity, vector inner product and the like. For example, when using Euclidean distance, the deviation value can be expressed as:

[0082]

[0083] Among them, Feature1 and Feature2 represent the first eigenvector and the second eigenvector respectively. i and Feature2 i They represent the i-th element of the two feature vectors Feature1 and Feature2 respectively. The smaller the value of Distance is, the closer the two feature vectors are, which means that the driving scene recording model of the first candidate compliance driving sample is more matched with the three-dimensional area of ​​the BIM model.

[0084] Afterwards, the route mining module 12 compares the feature deviation values ​​of the first feature extraction result and the second feature extraction result with a preset feature deviation value threshold. If the feature deviation value is less than or equal to the threshold, it is considered that the driving scene recording model of the first candidate compliant driving sample is sufficiently similar to the three-dimensional area of ​​the BIM model and satisfies the background constraint condition. At this time, the route mining module 12 includes the first candidate compliant driving sample into the compliant driving sample set for subsequent frequent route mining.

[0085] By using deep learning technology to automatically screen real driving samples that meet background constraints and foreground constraints, the efficiency and accuracy of sample collection have been greatly improved. Among them, the background comparison network composed of twin convolutional neural networks can effectively capture the similarity between BIM models and driving scenes, the feature comparison channel can flexibly select deviation measurement methods, and the setting of matching thresholds can ensure the quality of samples while taking into account the richness of samples, so that effective samples can be accurately collected from massive heterogeneous network data, laying a data foundation for the subsequent mining of frequent routes.

[0086] Furthermore, the route mining module 12 further includes: when the vehicle record type is different from the simulated vehicle type, or the feature deviation value is greater than the feature deviation threshold, updating the first candidate compliant driving sample.

[0087] In a feasible implementation, when the vehicle record type of the first candidate compliant driving sample is different from the simulated vehicle type, or the feature deviation value between its driving scene and the three-dimensional area of ​​the BIM model is greater than the set feature deviation threshold, the sample will be regarded as an invalid sample and will not be added to the compliant driving sample set. At this time, another candidate compliant driving sample is collected from the Internet as the new first candidate compliant driving sample, and as the object of the next background constraint and foreground constraint screening, so as to continuously enrich the compliant driving sample set.

[0088] Furthermore, the driving risk assessment network includes a balance risk assessment channel, an emotional risk assessment channel and a fully connected channel. Therefore, through the driving risk assessment network, a safety assessment is performed based on the three-dimensional dynamic timing diagram of the vehicle, the user's heart rate monitoring timing information and the user's driving safety score to obtain a driving risk evaluation value set, including:

[0089] Processing the three-dimensional dynamic timing diagram of the vehicle through the balance risk assessment channel to obtain a balance risk assessment value;

[0090] The emotional risk assessment channel is used to count the proportion of time domain steps greater than or equal to the heart rate threshold in the user's heart rate monitoring time series information, and set it as an emotional risk assessment value;

[0091] A first weight is set for the balance risk assessment value, a second weight is set for the emotional risk assessment value, and a third weight is set for the user driving safety score, wherein the sum of the second weight and the third weight is less than the first weight;

[0092] Through the fully connected channel, weighted fitting is performed on the normalized data of the balance risk assessment value, the emotional risk assessment value and the user's driving safety score to obtain a driving risk assessment value, which is added to the driving risk assessment value set.

[0093] In a preferred embodiment, the driving risk assessment network consists of three processing channels, namely, a balance risk assessment channel, an emotional risk assessment channel, and a fully connected channel. Among them, the balance risk assessment channel is responsible for analyzing the balance stability of the vehicle during motion, the emotional risk assessment channel is responsible for analyzing the impact of the driver's emotional fluctuations on driving safety, and the fully connected channel comprehensively considers the objective risk assessment and subjective safety score to generate the final driving risk evaluation value.

[0094] Specifically, the balance risk assessment channel takes the three-dimensional dynamic time series diagram of the vehicle as input, and uses algorithms such as spatiotemporal convolutional neural networks to extract balance stability-related features such as acceleration, angular velocity, and tilt angle during vehicle movement, and calculates the assessment value reflecting the vehicle imbalance risk based on this. The higher the assessment value, the greater the possibility that the vehicle will be in an unstable state such as rollover or loss of control during driving, and the higher the driving risk. The emotional risk assessment channel takes the user's heart rate monitoring time series information as input, focusing on analyzing the driver's emotional stress level during driving. By counting the proportion of time when the heart rate exceeds the normal threshold, this channel can quantitatively assess the degree of driver emotional fluctuation. The higher the proportion of heart rate arrhythmia, the more unstable the driver's emotions are, and the greater the potential threat to driving safety. In the final stage of risk assessment, the fully connected channel comprehensively integrates the balance risk assessment value, the emotional risk assessment value, and the driving safety score scored by the user, and performs weighted fusion to obtain a normalized driving risk evaluation value. In order to balance the importance of various indicators, the channel adopts a weight distribution strategy: the balanced risk assessment value is assigned the highest first weight, while the emotional risk assessment value and the subjective score have relatively low weights, which are the second weight and the third weight respectively, and the sum of the two does not exceed the first weight. This weight setting reflects the decisive influence of objective risk factors on driving safety, while also taking into account the reference value of the driver's subjective feelings. Through weighted fusion, the driving risk assessment network can output a comprehensive driving risk assessment value ranging from 0 to 1. The closer the value is to 1, the higher the safety risk in the driving process, and vice versa. The risk assessment module 15 will calculate the risk assessment value for each section of driving data, and summarize the results into a complete set of driving risk assessment values ​​as a basis for quantitatively assessing the driving safety level of the entire area.

[0095] By comprehensively considering risk information in multiple dimensions such as vehicle motion balance, driver emotional stability and subjective safety perception, it is possible to more accurately and comprehensively assess safety hazards during driving, providing an important basis for traffic road design.

[0096] Furthermore, the balance risk assessment channel includes a turbulence amplitude feature attention extraction layer, a turbulence frequency feature attention extraction layer and a balance risk assessment layer. Therefore, the three-dimensional dynamic timing diagram of the vehicle is processed through the balance risk assessment channel to obtain a balance risk assessment value, including:

[0097] Processing the three-dimensional dynamic time sequence diagram of the vehicle through the bump amplitude feature attention extraction layer to extract the bump amplitude time sequence information;

[0098] Processing the three-dimensional dynamic time sequence diagram of the vehicle through the turbulence frequency feature attention extraction layer to extract turbulence frequency time sequence information;

[0099] Through the balance risk assessment layer, statistics are given on the time domain step ratios in which the turbulence amplitude in the turbulence amplitude time series information is greater than or equal to the turbulence amplitude threshold, or / and the turbulence frequency in the turbulence frequency time series information is greater than or equal to the turbulence frequency threshold, and the statistics are set as the balance risk assessment value.

[0100] In a preferred embodiment, in the driving risk assessment network, the balance risk assessment channel is a channel specifically used to analyze the driving stability of the vehicle. In order to more finely characterize the impact of the vehicle's bumpy state on driving safety, a bump amplitude feature attention extraction layer, a bump frequency feature attention extraction layer and a balance risk assessment layer are designed in the balance risk assessment channel. The bump amplitude feature attention extraction layer and the bump frequency feature attention extraction layer extract the time series characteristics of the vehicle bump process from the two perspectives of vertical displacement amplitude and frequency, respectively, and on this basis, the balance risk assessment layer comprehensively evaluates the risk level of driving balance.

[0101] Specifically, the bump amplitude feature attention extraction layer takes the three-dimensional dynamic time sequence diagram of the vehicle as input and focuses on analyzing the dynamic change law of the vehicle vertical displacement amplitude. By analyzing the three-dimensional dynamic time sequence diagram of the vehicle frame by frame, this layer can accurately extract the bump amplitude information of the vehicle body at each moment and organize it into a complete bump amplitude time sequence signal. In the extraction process, this layer introduces the attention mechanism, and adaptively adjusts the weight of the bump information at different times and in different regions through learning, so that the extracted time series features can focus more on the key bump events that have the greatest impact on the balance risk. At the same time, the bump frequency feature attention extraction layer focuses on analyzing the frequency characteristics of vehicle bumps. By performing time-frequency domain transformation and multi-scale analysis on the three-dimensional dynamic time sequence diagram of the vehicle, this layer can automatically identify the dominant frequency components of the vehicle vertical displacement and reveal the periodic law of vehicle bumps. Similarly, this layer also uses the attention mechanism to optimize the extraction quality of frequency features, so that the extraction results can more accurately reflect the frequency domain characteristics of balance risk. After completing the extraction of the bump amplitude and frequency characteristics, the balance risk assessment channel passes the extraction results to the balance risk assessment layer, which comprehensively analyzes the impact of the vehicle's bump state on driving safety. Specifically, the balance risk assessment layer presets two key risk judgment thresholds, namely the bump amplitude threshold and the bump frequency threshold. The former reflects the dangerous limit of the vehicle body bump displacement, while the latter reflects the dangerous frequency range of vehicle bumps. The balance risk assessment layer compares the thresholds of the bump amplitude time series signal and the frequency time series signal, and calculates the time proportion of dangerous bump events that exceed the safety limit during the entire driving process. This ratio value is the balance risk assessment value that measures the vehicle's driving stability. The higher the value, the greater the risk of the vehicle losing control or overturning on a bumpy road, and the worse the balance stability.

[0102] Through the two complementary feature perspectives of bump amplitude and frequency, the vehicle's balance control state under complex road conditions can be accurately described, and potential risk of loss of control can be discovered in a timely manner, providing an important basis for traffic safety warning and auxiliary decision-making. At the same time, the attention mechanism in deep learning is used to optimize the extraction and representation of key risk information, which is more intelligent, efficient, and more adaptive and robust than traditional artificial feature methods.

[0103] An interactive design and simulation system combining BIM and VR technology for traffic engineering provided by an embodiment of the present invention has at least the following technical effects:

[0104] The interactive configuration module is used to display the traffic engineering BIM model and simulation configuration menu on the user side and prompt the user to select the simulation configuration, obtain user feedback information, wherein the user feedback information includes the BIM model three-dimensional area and the simulated vehicle type, realize the interaction between the user and the system, let the user select the road area and the simulated vehicle type to be evaluated, and the system obtains the user's simulation configuration requirement information based on this, and provides conditions for subsequent simulation evaluation. The route mining module is used to mine the BIM model three-dimensional area for networking frequency according to the simulated vehicle type, obtain the mobile route use case set, and automatically mine and generate a mobile route use case set covering a variety of typical driving conditions according to the vehicle type and BIM road model selected by the user, so as to construct a comprehensive driving behavior evaluation sample space. The virtual driving module is used to send the mobile route use case set and the BIM model three-dimensional area to the associated VR driving simulation warehouse of the simulated vehicle type, and remind the user to enter the associated VR driving simulation warehouse to start simulated driving, obtain the three-dimensional dynamic time sequence diagram of the vehicle and the user's heart rate monitoring time sequence information, and obtain a data set for driving behavior analysis. The safety scoring module is used to receive the user's driving safety score through the user end, so as to obtain the user's subjective driving safety score for different routes as a reference for measuring the driving experience. The risk assessment module is used to conduct a safety assessment based on the three-dimensional dynamic timing diagram of the vehicle, the user's heart rate monitoring timing information and the user's driving safety score through the driving risk assessment network, obtain a driving risk evaluation value set, and quantitatively evaluate the driving safety risk level of the current traffic engineering design scheme. The safety identification module is used to conduct a safety identification of the three-dimensional area of ​​the BIM model and send it to the traffic engineering design end when the driving risk evaluation value set is less than the driving risk evaluation threshold. If the requirements are met, the scheme is safety identified and fed back to the designer to complete the safety verification, thereby achieving a targeted evaluation of the driving risk of the design scheme, guiding the optimization design of traffic engineering, and improving the safety of traffic engineering design.

[0105] Embodiment 2:

[0106] See also Figure 2 , Figure 2 Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 2 As shown, an electronic device 200 provided by an embodiment of the present invention includes a memory 210, a processor 220, and a first computer program 211 stored in the memory 210 and executable on the processor 220. When the processor 220 executes the first computer program 211, an interactive design and simulation system combining BIM and VR technology for traffic engineering is implemented.

[0107] Embodiment three:

[0108] See also Figure 3 , Figure 3A schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 3 As shown, this embodiment provides a computer-readable storage medium 300 on which a second computer program 311 is stored. When the second computer program 311 is executed by a processor, an interactive design and simulation system combining BIM and VR technology for traffic engineering is implemented.

[0109] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0110] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, 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.

[0112] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising 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.

[0113] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0114] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0115] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. An interactive design and simulation system combining BIM and VR technology for traffic engineering, characterized in that: include: An interactive configuration module, wherein the interactive configuration module is used to display the traffic engineering BIM model and the simulation configuration menu on the user side and prompt the user to select the simulation configuration, and obtain user feedback information, wherein the user feedback information includes the three-dimensional area of ​​the BIM model and the type of simulated transportation tool; A route mining module, the route mining module is used to mine the networking frequency of the three-dimensional area of ​​the BIM model according to the type of the simulated transportation tool to obtain a set of mobile route use cases; A virtual driving module, the virtual driving module is used to send the mobile route use case set and the BIM model three-dimensional area to the associated VR driving simulation chamber of the simulated vehicle type, and remind the user to enter the associated VR driving simulation chamber to start simulated driving, and obtain a three-dimensional dynamic timing diagram of the vehicle and user heart rate monitoring timing information; A safety scoring module, the safety scoring module is used to receive a user's driving safety score through a user terminal; A risk assessment module, the risk assessment module is used to perform a safety assessment based on the three-dimensional dynamic timing diagram of the vehicle, the user's heart rate monitoring timing information and the user's driving safety score through a driving risk assessment network to obtain a driving risk evaluation value set; A safety identification module is used to carry out safety identification on the three-dimensional area of ​​the BIM model and send it to the traffic engineering design end when the driving risk evaluation value set is less than the driving risk evaluation threshold.

2. The system according to claim 1, characterized in that The execution steps of the route mining module include: Taking the BIM model three-dimensional area as a background constraint and the simulated vehicle type as a foreground constraint, collecting a set of compliant driving samples through networking, wherein the compliant driving sample set includes a moving route record data set; Performing frequent coefficient identification on the mobile route record data set to obtain a route frequent coefficient set; The moving route use case set whose number of moving route use cases is greater than or equal to a frequent coefficient threshold value in the route frequent coefficient set is sorted out from the moving route record data set.

3. The system according to claim 2, characterized in that The execution steps of the route mining module also include: Performing pairwise distance marking on the moving route record data sets to obtain a moving route distance calibration data set; Extracting first moving route record data according to the moving route record data set; Based on the moving route distance calibration data set, counting the number of routes whose moving route distance deviation between the moving route record data set and the first moving route record data set is less than a moving route distance deviation threshold, and counting the ratio of the number of routes to the total number of routes in the moving route record data set, and setting it as a first moving route frequency coefficient; The first moving route frequency coefficient is added to the route frequency coefficient set.

4. The system according to claim 3, characterized in that The execution steps of the route mining module also include: Construct a moving route distance calibration function: in, Represents the first dimension coordinate of the i-th point on path P1, Represents the second-dimensional coordinate of the i-th point on path P1, Represents the third-dimensional coordinate of the i-th point on path P1, Characterizes the first dimension coordinate of the jth point on path P2, Characterizes the second-dimensional coordinate of the j-th point on path P2, Represents the third-dimensional coordinate of the j-th point on path P2, is the path P2 with The nearest point, Represents the direction vector between the jth point on path P2 and the point after it, Represents the direction vector between the i-th point on path P1 and the point after it, θ ij represents the angle difference between two direction vectors determined by calculating the dot product and modulus of the two direction vectors. ΔL represents the length deviation between path P1 and path P2. ij Characterization and The Euclidean distance, M represents the total number of points on path P2, N represents the total number of points on path P1, λ represents the weight coefficient of direction difference, which is used to adjust the contribution of direction difference in the total deviation, μ represents the weight coefficient of path length deviation, which is used to adjust the contribution of path length deviation in the total deviation, w ij Characterization and Importance weight in path deviation calculation; The moving route distance calibration function is used to perform distance marking on the moving route record data sets in pairs to obtain a moving route distance calibration data set.

5. The system according to claim 2, characterized in that The execution steps of the route mining module also include: Collecting a first sample of compliant driving to be selected through networking, wherein the first sample of compliant driving to be selected includes a driving scene record model and a vehicle record type; When the vehicle record type is the same as the simulated vehicle type, the BIM model stereoscopic area is input into a first feature extraction channel of a background comparison network to obtain a first feature extraction result, and the driving scene record model is input into a second feature extraction channel of a background comparison network to obtain a second feature extraction result, wherein the first feature extraction channel and the second feature extraction channel are twin convolutional neural networks; Inputting the first feature extraction result and the second feature extraction result into a feature comparison channel to obtain a feature deviation value; When the feature deviation value is less than or equal to a feature deviation threshold, the first candidate compliant driving sample is added to the compliant driving sample set.

6. The system according to claim 5, characterized in that The execution step of the route mining module further includes: when the vehicle record type is different from the simulated vehicle type, or the feature deviation value is greater than the feature deviation threshold, updating the first to-be-selected compliant driving sample.

7. The system according to claim 1, characterized in that The driving risk assessment network includes a balance risk assessment channel, an emotional risk assessment channel and a fully connected channel. The execution steps of the risk assessment module include: Processing the three-dimensional dynamic timing diagram of the vehicle through the balance risk assessment channel to obtain a balance risk assessment value; The emotional risk assessment channel is used to count the proportion of time domain steps greater than or equal to the heart rate threshold in the user's heart rate monitoring time series information, and set it as an emotional risk assessment value; A first weight is set for the balance risk assessment value, a second weight is set for the emotional risk assessment value, and a third weight is set for the user driving safety score, wherein the sum of the second weight and the third weight is less than the first weight; Through the fully connected channel, weighted fitting is performed on the normalized data of the balance risk assessment value, the emotional risk assessment value and the user's driving safety score to obtain a driving risk assessment value, which is added to the driving risk assessment value set.

8. The system according to claim 7, characterized in that The balance risk assessment channel includes a turbulence amplitude feature attention extraction layer, a turbulence frequency feature attention extraction layer and a balance risk assessment layer, and the execution steps of the risk assessment module also include: Processing the three-dimensional dynamic time sequence diagram of the vehicle through the bump amplitude feature attention extraction layer to extract the bump amplitude time sequence information; Processing the three-dimensional dynamic time sequence diagram of the vehicle through the turbulence frequency feature attention extraction layer to extract turbulence frequency time sequence information; Through the balance risk assessment layer, statistics are given on the time domain step ratios in which the turbulence amplitude in the turbulence amplitude time series information is greater than or equal to the turbulence amplitude threshold, or / and the turbulence frequency in the turbulence frequency time series information is greater than or equal to the turbulence frequency threshold, and the statistics are set as the balance risk assessment value.

Images