Traffic accident prediction method and device based on multi-factor influence and medium

Abstract

The invention discloses a traffic accident prediction method and device based on multi-factor influence, and a medium, belongs to the technical field of traffic accident prediction, and is used for solving the problem that road safety and accident early warning are not accurate enough in a complex road condition or a complex road environment in an existing urban road, and the traffic accident prediction efficiency is high. And early warning information is difficult to effectively provide for the driver. The method comprises the following steps: performing multi-factor correlation screening processing on all traffic road sections to obtain an accident risk road section area; performing network structure training processing on historical traffic accident time sequence events in the accident risk road section area under the condition that related time characteristics are matched with traffic accident factors to obtain an urban traffic accident prediction model; performing traffic accident risk prediction on the accident risk road section area in the current time period to obtain traffic accident risk information; and road section division of traffic flow diversion positions is carried out on the accident road traffic network corresponding to the accident risk road section area, and an early warning reminding position is determined.

Description

Technical Field

[0001] The present application relates to the field of traffic accident prediction, and in particular to a method, device and medium for predicting traffic accidents under the influence of multiple factors. Background Art

[0002] Traffic accidents on the road are often random events, which are not only restricted by the status of various elements in the road traffic system, but also affected by various accidental factors of society (such as the number of cars, etc.) and nature (such as real-time weather, etc.). However, the randomness of accidents is also dominated by internal laws, such as: complex road factors, driving environment factors, traffic flow factors, human and social activities factors, etc.

[0003] On existing urban roads, the driving status of vehicles is often affected by many factors, especially on complex roads, in bad weather or roads prone to accidents. It is difficult for drivers to predict complex traffic conditions through visual conditions, and the traffic situation in cities is relatively complex, often accompanied by complex road environments such as traffic construction, temporary line changes and road reconstruction. It is even more difficult to predict the type of traffic accidents that may occur, which is not conducive to the driver's driving judgment and risk avoidance. It is also difficult to obtain information that needs to be warned in advance, such as: information on the proneness of traffic accidents, complex accident information on traffic sections, and accident hazard levels, which is not conducive to the driving safety of drivers on urban roads and traffic management on urban roads. Summary of the invention

[0004] The embodiments of the present application provide a traffic accident prediction method, device and medium based on the influence of multiple factors, which are used to solve the following technical problems: in existing urban roads, when encountering complex road conditions or complex road environments, road safety and accident warnings are not accurate enough, and it is difficult to effectively provide warning information to drivers.

[0005] The present application embodiment adopts the following technical solutions:

[0006] On the one hand, an embodiment of the present application provides a method for predicting traffic accidents based on multiple factors, including: based on a road traffic network data set obtained by an urban traffic management system, performing multi-factor correlation screening processing based on accident risk sections on all traffic sections to obtain an accident risk section area; performing network structure training processing on historical traffic accident time series events in the accident risk section area under the conditions of matching relevant time characteristics with traffic accident factors to obtain an urban traffic accident prediction model; using the urban traffic accident prediction model, predicting traffic accident risks on the accident risk section area in the current time period to obtain traffic accident risk information; based on the traffic accident risk information, dividing the accident road traffic network corresponding to the accident risk section area into sections at the traffic diversion position to determine an early warning reminder position; and outputting the traffic accident risk information through the early warning reminder position to complete the traffic accident risk prediction.

[0007] The embodiments of the present application can improve road safety and the accuracy of accident warnings when encountering complex urban road conditions or complex road environments, and effectively provide warning information to drivers. Through multi-factor correlation screening and processing, combined with historical traffic accident data, the risk of traffic accidents can be predicted more accurately. The prediction model can help traffic management departments to reasonably allocate police force and rescue resources and improve the response efficiency in emergency situations. The road section division of the traffic diversion position can guide vehicles to avoid high-risk sections to reduce the occurrence of accidents. The determination of the warning reminder location can provide accident risk information to the driver in a timely manner to help the driver make safe driving decisions. Early prediction and warning can reduce casualties and property losses caused by traffic accidents. By outputting traffic accident risk information, the public's awareness of traffic accident risks can be improved and the formation of safe driving habits can be promoted. By avoiding high-risk sections, traffic congestion can also be reduced and road traffic efficiency can be improved. At the same time, the method can continuously update and optimize the prediction model according to real-time data to improve the adaptability and accuracy of the system.

[0008] In a feasible implementation, based on the road traffic network dataset acquired by the urban traffic management system, all traffic sections are screened for multi-factor correlation based on accident risk sections to obtain accident risk section areas, specifically including: extracting historical traffic accident data based on the traffic accident database in the preset urban traffic management system; and based on the historical traffic accident data, determining the road traffic network dataset where various types of traffic accidents occurred; wherein the historical traffic accident data at least includes: the location of the accident, the time period of the accident, the location of the accident, the type of vehicle, the cause of the accident, the human and social environment, and the accident road environment; performing multi-factor identification and classification processing on the external causes of traffic accidents on the road traffic network dataset to obtain traffic accident independent variable data for each traffic section; wherein the traffic accident The traffic accident independent variable data include: accident time, accident space, road conditions, road environment, traffic flow, weather factors, and specific accident conditions; the road traffic network data set is subjected to multi-factor classification processing related to the intrinsic performance of traffic accidents to obtain traffic accident dependent variable data for each traffic section; wherein, the traffic accident dependent variable data include: accident form, accident severity, accident casualties, and accident duration; through the preset Pearson correlation coefficient algorithm, the correlation between the traffic accident independent variable data and the traffic accident dependent variable data is calculated, and traffic sections with strong correlation are screened out; the area occupied by the strongly correlated traffic sections is determined as the accident risk section area; wherein, the accident risk section area is a section area with inherent accident risk occurrence rules and trend prediction.

[0009] In a feasible implementation manner, before performing network structure training processing on the historical traffic accident time series events in the accident risk section area under the matching of relevant time characteristics and traffic accident factors to obtain the urban traffic accident prediction model, the method also includes: identifying the historical traffic accident events in the accident risk section area, and classifying the historical traffic accident events based on the accident occurrence time nodes under long and short time factors to obtain the historical traffic accident time series events; wherein, the long time factors include: annual time nodes, monthly time nodes, daily time nodes and quarterly time nodes; the short time factors include: planning time period range, morning and evening peak time range and cultural activity time period range; extracting the time characteristics and the corresponding traffic accident factors of each of the historical traffic accident time series events in the accident risk section area; wherein, the traffic accident factors include: accident form, accident severity, accident casualties and accident duration; matching the time characteristics with the traffic accident factors one by one, and integrating the traffic accident factors into the time characteristics to obtain a time series accident feature sequence.

[0010] In a feasible implementation, the historical traffic accident time series events in the accident risk section area are subjected to network structure training processing under the matching of relevant time characteristics and traffic accident factors to obtain an urban traffic accident prediction model, which specifically includes: combining a preset Gaussian white noise with the time series accident feature sequence to obtain a new time series accident feature sequence; wherein the Gaussian white noise is a normally distributed noise; using the empirical mode decomposition technology, the new time series accident feature sequence is repeatedly decomposed based on modal components and residual components to obtain a final residual value and multiple time series accident subsequences; arranging the multiple time series accident subsequences; and using the particle swarm algorithm to perform a prediction on the time series accident feature sequence. The method is used for global search, the number of hidden layer units and the learning rate of the neural network model are optimized, and a neural network structure is constructed according to the hyperparameter value corresponding to the particle initialization position of the particle swarm algorithm; according to the local optimal particle and the global optimal particle after the iteration of the particle swarm algorithm, the optimal position training of the neural network structure related to the global minimum fitness value is performed to obtain the optimized neural network model; the final residual value and the sorted multiple time series accident subsequences are input into the optimized neural network model, and the optimal difference parameter of the particle swarm algorithm is used to train the optimized neural network model to obtain the urban traffic accident prediction model.

[0011] In a feasible implementation manner, the number of hidden layer units and the learning rate of the neural network model are optimized through a global search of a particle swarm algorithm, and a neural network structure is constructed according to the hyperparameter value corresponding to the particle initialization position of the particle swarm algorithm, specifically including: initializing multiple parameters of the particle swarm algorithm; wherein the multiple parameters include: learning factor, number of iterations, number of hidden unit layers, initial velocity of particles, velocity variation range, and size of particle target search area; determining other parameters of the neural network model; wherein the other parameters include: initial learning rate, learning rate variation range, iteration variation step size, initial number of hidden layer units, and parameter range of number of hidden layer units; based on the multiple parameters of the initialized particle swarm algorithm and according to the root mean square error derivative of the measured value and the predicted value of the training set, determining the fitness function of the particle swarm algorithm; through the fitness function, calculating the corresponding fitness value of the particle initialization position, and determining the hyperparameter value corresponding to the particle initialization position; and building and training the neural network structure based on the particle initialization position and the hyperparameter value.

[0012] In a feasible implementation manner, the urban traffic accident prediction model is used to predict the traffic accident risk of the accident risk section area in the current time period to obtain traffic accident risk information, which specifically includes: obtaining the historical time series accident feature sequence of the accident risk section area in the previous time period; inputting the historical time series accident feature sequence into the urban traffic accident prediction model to obtain the traffic accident risk prediction information for the next time period; obtaining the actual traffic network information of the accident risk section area in the current time period; wherein the actual traffic network information is the actual traffic accident independent variable data in the accident risk section area; performing similarity matching between the actual traffic characteristics in the actual traffic network information and the risk prediction parameter information in the traffic accident risk prediction information to determine the traffic accident risk similarity information; based on the similarity threshold in the traffic accident risk similarity information, determining the traffic accident risk information in the next time period; wherein the traffic accident risk information includes: high-risk traffic accident risk information, dangerous traffic accident risk information, medium-risk traffic accident risk information, low-risk traffic accident risk information and no traffic accident risk information.

[0013] In a feasible implementation manner, based on the traffic accident risk information, the accident road traffic network corresponding to the accident risk section area is divided into sections of traffic diversion positions to determine the early warning reminder position, which specifically includes: through the GIS geographic information platform, the road range of the accident risk section area is radiated and identified to determine the section impact area; based on the danger level of the traffic accident risk information in the accident risk section area, the section impact area is reduced and expanded, and the associated section impact area is determined; wherein, the higher the danger level, the corresponding row area range is expanded; the lower the danger level, the corresponding row area range is reduced; if the associated section impact area is an expanded area, the intersection area contained in the expanded area is determined as a traffic diversion position, and the traffic diversion position is set as a necessary early warning reminder position; if the associated section impact area is a reduced area, the intersection area in the reduced area is determined as a traffic diversion position, and the traffic diversion position is set as a non-essential early warning reminder position; wherein the early warning reminder position includes: a necessary early warning reminder position and a non-essential early warning reminder position.

[0014] In a feasible implementation, the traffic accident risk information is output and processed through the early warning reminder position to complete the traffic accident risk prediction, specifically including: placing a movable display device based on the early warning reminder position; wherein the movable display device is an intelligent robot with mobile and visual information display; the section information contained in the accident risk section area and the corresponding traffic accident risk information are input into the movable display device to remind the driver to complete the advance planning of the vehicle route.

[0015] In the second aspect, an embodiment of the present application also provides a traffic accident prediction device based on the influence of multiple factors, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor so that the at least one processor can execute a traffic accident prediction method based on the influence of multiple factors as described in any of the above-mentioned embodiments.

[0016] In the third aspect, an embodiment of the present application also provides a non-volatile computer storage medium, characterized in that the storage medium is a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores at least one program, each of which includes instructions, and when the instructions are executed by a terminal, the terminal executes a traffic accident prediction method based on multiple factors as described in any of the above-mentioned embodiments.

[0017] The embodiments of the present application provide a method, device and medium for predicting traffic accidents under the influence of multiple factors. Compared with the prior art, the present application has the following beneficial technical effects:

[0018] 1. Improve prediction accuracy: Through multi-factor correlation screening and processing, combined with historical traffic accident data, traffic accident risks can be predicted more accurately.

[0019] 2. Optimize traffic management: Based on the prediction results, the traffic management department can know in advance which sections of the road are likely to have traffic accidents, so as to take corresponding preventive measures to reduce the occurrence of traffic accidents.

[0020] 3. Reasonable allocation of resources: The prediction model can help traffic management departments to reasonably allocate police force and rescue resources and improve the response efficiency in emergency situations.

[0021] 4. Dynamic traffic guidance: By dividing the road sections at the traffic diversion locations, vehicles can be guided to avoid high-risk sections and reduce the occurrence of accidents.

[0022] 5. Establishment of early warning system: The determination of early warning reminder locations can provide accident risk information to drivers in a timely manner and help them make safe driving decisions.

[0023] 6. Reduce accident losses: Early prediction and warning can reduce casualties and property losses caused by traffic accidents.

[0024] 7. Enhance public safety awareness: By outputting traffic accident risk information, the public’s awareness of traffic accident risks can be improved and the formation of safe driving habits can be promoted.

[0025] 8. Data-driven decision-making: This method is based on big data analysis and provides data support for traffic management decisions, making decisions more scientific and reasonable.

[0026] 9. Improve traffic efficiency: By avoiding high-risk sections, traffic congestion can be reduced and road traffic efficiency can be improved.

[0027] 10. Enhance system adaptability: This method can continuously update and optimize the prediction model based on real-time data to improve the adaptability and accuracy of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0029] Figure 1 A flow chart of a traffic accident prediction method based on multiple factors provided in an embodiment of the present application;

[0030] Figure 2 A schematic diagram of the division of accident risk road sections provided in an embodiment of the present application;

[0031] Figure 3 A schematic diagram of the structure of a traffic accident prediction device based on the influence of multiple factors provided in an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0033] The present application embodiment provides a traffic accident prediction method based on multiple factors, such as Figure 1As shown, the traffic accident prediction method based on the influence of multiple factors specifically includes steps S101-S105:

[0034] S101. Based on the road traffic network data set obtained by the urban traffic management system, all traffic sections are screened based on the multi-factor correlation of accident risk sections to obtain accident risk section areas.

[0035] Specifically, it is necessary to first extract historical traffic accident data from the traffic accident database in the preset urban traffic management system. Based on the historical traffic accident data, determine the road traffic network data set where various types of traffic accidents occurred. Among them, the historical traffic accident data at least includes: the location of the accident, the time period of the accident, the location of the accident, the type of vehicle, the cause of the accident, the humanistic and social environment, and the accident road environment.

[0036] Furthermore, the road traffic network data set is subjected to multi-factor identification and classification processing of the external causes of traffic accidents to obtain the traffic accident independent variable data for each traffic section. The traffic accident independent variable data include: accident time, accident space, road conditions, road surroundings, traffic flow, weather factors, and specific accident conditions.

[0037] Furthermore, the road traffic network data set is processed by multi-factor classification of the inherent performance of traffic accidents to obtain traffic accident dependent variable data for each traffic section. The traffic accident dependent variable data include: accident form, accident severity, accident casualties and accident duration.

[0038] Furthermore, it is also necessary to calculate the correlation between the traffic accident independent variable data and the traffic accident dependent variable data through a preset Pearson correlation coefficient algorithm, and screen out traffic sections with strong correlation.

[0039] Furthermore, the area occupied by the highly correlated traffic sections is determined as the accident risk section area, wherein the accident risk section area is a section area with inherent accident risk occurrence rules and trend prediction can be performed.

[0040] In one embodiment, Figure 2A schematic diagram of the division of accident risk road section areas provided in the embodiment of the present application. The occurrence of traffic accidents is the result of the combined effect of many factors. When it occurs on urban roads, it is more complicated. In order to obtain relatively accurate prediction results, advance prediction of accident risk road section areas and data analysis are particularly important. Using the Pearson correlation coefficient algorithm, the correlation analysis between traffic accident independent variable data and traffic accident dependent variable data can be performed in sections or areas with a high risk of traffic accidents. That is to say, for the historical traffic accident situation in a road section area, combined with the three major environmental factors of human conditions, traffic conditions and road environment, a comprehensive regional judgment is made to find the accident risk road section area that can perform data prediction, that is, the area occupied by the strongly correlated traffic section is determined as the accident risk road section area.

[0041] S102, performing network structure training processing on the historical traffic accident time series events in the accident risk section area under the condition that the relevant time characteristics are matched with the traffic accident factors, and obtaining an urban traffic accident prediction model.

[0042] Specifically, we first identify historical traffic accident events in the accident risk section area, and classify the historical traffic accident events based on the time nodes of the accident occurrence under long and short time factors to obtain the historical traffic accident time series events. Among them, the long time factors include: annual time nodes, monthly time nodes, daily time nodes and quarterly time nodes. The short time factors include: planning time range, morning and evening peak time range and cultural activity time range.

[0043] Furthermore, the time characteristics of each historical traffic accident time series event in the accident risk section area and the corresponding traffic accident factors are extracted. Among them, the traffic accident factors include: accident form, accident severity, accident casualties and accident duration.

[0044] Furthermore, the time features are matched with the traffic accident factors one by one, and the traffic accident factors are integrated into the time features to obtain a time series accident feature sequence.

[0045] As a feasible implementation method, the use of ensemble empirical mode decomposition is a processing method that decomposes data signals based solely on the time scale characteristics of the data itself. It can decompose different types of signal data and process irregular and volatile time series. That is, it is necessary to first identify the historical traffic accident events in the accident risk section area, and then extract the time characteristics and corresponding traffic accident factors of each historical traffic accident time series event in the accident risk section area based on the results jointly formed by multiple causes in the historical traffic accident events, thereby forming time characteristics and traffic accident factors that can reflect the characteristics of each traffic accident event, and then combining these two characteristic factors to finally form a time series accident feature sequence for subsequent empirical mode decomposition.

[0046] Furthermore, the preset Gaussian white noise is combined with the time series accident feature sequence to obtain a new time series accident feature sequence. Among them, Gaussian white noise is normally distributed noise. Then, the empirical mode decomposition technology is used to repeatedly decompose the new time series accident feature sequence based on the modal component and the residual component to obtain the final residual value and multiple time series accident subsequences. The multiple time series accident subsequences are arranged and processed.

[0047] Furthermore, the number of hidden layer units and the learning rate of the neural network model are optimized through the global search of the particle swarm algorithm, and the neural network structure is constructed according to the hyperparameter values ​​corresponding to the particle initialization positions of the particle swarm algorithm.

[0048] Furthermore, according to the local optimal particles and global optimal particles after the iteration of the particle swarm algorithm, the optimal position training of the neural network structure related to the global minimum fitness value is carried out to obtain the optimized neural network model. The final residual value and the sorted multiple time series accident subsequences are input into the optimized neural network model, and the optimal difference parameters of the particle swarm algorithm are used to train the optimized neural network model to obtain the urban traffic accident prediction model.

[0049] As a feasible implementation method, in the process of constructing a neural network structure, various parameters of the particle swarm algorithm can be initialized first. Among them, the various parameters include: learning factor, number of iterations, number of hidden unit layers, initial particle velocity, velocity variation range, and particle target search area size. Then determine other parameters of the neural network model. Among them, other parameters include: initial learning rate, learning rate variation range, iteration variation step size, initial number of hidden layer units, and parameter range of number of hidden layer units. Based on the various parameters of the initialized particle swarm algorithm, and according to the root mean square error derivative of the measured value and the predicted value of the training set, determine the fitness function of the particle swarm algorithm. Through the fitness function, the corresponding fitness value is calculated for the particle initialization position, and the hyperparameter value corresponding to the particle initialization position is determined. Finally, based on the particle initialization position and the hyperparameter value, build and train the neural network structure.

[0050] In one embodiment, regarding the urban traffic accident prediction model, it is necessary to decompose the time series accident feature sequence, and the model decomposes the sequence into multiple subsequences and a final residual value item based on the characteristics of the time series itself, and each time series accident subsequence is arranged in order from high to low according to frequency. And the number of hidden layer units and the learning rate of the neural network model are used as the optimization object of the particle swarm algorithm. According to the hyperparameter value corresponding to the particle initialization position, the fitness value corresponding to the particle initialization position is calculated through the fitness function, and the hyperparameter value corresponding to the particle initialization position is determined. Finally, based on the particle initialization position and the hyperparameter value, the neural network structure is built and trained. Finally, it is necessary to continuously change the particle position and calculate the corresponding fitness value. Based on the fitness corresponding to the initial position, the individual optimal position is determined, and the global optimal particle is compared on this basis. The particle position and the speed of movement are continuously changed until the global minimum fitness is found, so as to determine the optimal position of the particle, that is, the optimal parameter combination corresponding to the neural network model, and the optimized neural network model is trained based on the optimal parameter combination.

[0051] S103. Using the urban traffic accident prediction model, traffic accident risk prediction is performed on the accident risk road section area in the current time period to obtain traffic accident risk information.

[0052] Specifically, the historical time series accident feature sequence of the accident risk section area in the previous time period is first obtained from the urban traffic management system, and then the historical time series accident feature sequence is input into the urban traffic accident prediction model to obtain the traffic accident risk prediction information in the next time period.

[0053] Furthermore, the actual traffic network information of the accident risk section area in the current time period is obtained, wherein the actual traffic network information is the actual traffic accident independent variable data in the accident risk section area.

[0054] Furthermore, the actual traffic characteristics in the actual traffic network information are matched with the risk prediction parameter information in the traffic accident risk prediction information to determine the traffic accident risk similarity information. Based on the similarity threshold in the traffic accident risk similarity information, the traffic accident risk information in the next time period is determined. Among them, the traffic accident risk information includes: high-risk traffic accident risk information, dangerous traffic accident risk information, medium-risk traffic accident risk information, low-risk traffic accident risk information, and no traffic accident risk information.

[0055] In one embodiment, by using the actual traffic characteristics in the actual traffic network information and the risk prediction parameter information in the traffic accident risk prediction information to perform similarity matching, the risk prediction parameter information can be corrected. That is to say, the occurrence of traffic accidents generally has certain internal rules, that is, it is predictable. After the traffic accident risk prediction information predicted in the next time period is compared with the actual traffic network information for similarity between information characteristics, if the actual traffic characteristics in the actual traffic network information are getting closer and closer to the risk prediction parameter information in the traffic accident risk prediction information, it means that the higher the similarity threshold between the two, the higher the danger level of the traffic accident risk information, such as high-risk traffic accident risk information and dangerous traffic accident risk information; conversely, the lower the similarity threshold between the two, the lower the danger level of the traffic accident risk information, such as medium-risk traffic accident risk information and low-risk traffic accident risk information, or even no traffic accident risk information, thereby further improving the accuracy of the prediction of the traffic accident risk information, making it more in line with the actual traffic conditions on the scene.

[0056] S104: Based on the traffic accident risk information, the accident road traffic network corresponding to the accident risk section area is divided into sections of traffic diversion locations to determine early warning reminder locations.

[0057] Specifically, the GIS geographic information platform is used to radiate the road range of the accident risk section area and determine the affected area of ​​the section.

[0058] Furthermore, if Figure 2 As shown, based on the danger level of the traffic accident risk information in the accident risk section area, the area of ​​the road section influence is reduced and expanded, and the area of ​​the associated road section influence is determined. Among them, the higher the danger level, the corresponding area is expanded. The lower the danger level, the corresponding area is reduced.

[0059] If the area affected by the associated road section is an area after expansion, the intersection area contained in the expansion area is determined as the traffic diversion location, and the traffic diversion location is set as the necessary warning reminder location. If the area affected by the associated road section is an area after reduction, the intersection area in the reduction area is determined as the traffic diversion location, and the traffic diversion location is set as the unnecessary warning reminder location. That is, the warning reminder location includes: necessary warning reminder locations and unnecessary warning reminder locations.

[0060] As a feasible implementation method, when the danger level of traffic accident risk information is higher, the situation of the road section may be more complicated, the danger situation may be higher, the traffic situation may be more complicated, and the drivers may know less specific information about the accident risk section area. Therefore, the higher the danger level, the corresponding area range is expanded, that is, the accident risk section area is expanded, so that the warning location can be notified in advance, so that more drivers can prepare for danger in advance, pay more attention, and even re-plan the route to reduce the risk of traffic accidents. On the contrary, the lower the danger level, the smaller the area range, which is conducive to the convenient management of actual traffic.

[0061] S105. By warning the location, the traffic accident risk information is output and processed to complete the traffic accident risk prediction.

[0062] Specifically, based on the warning reminder location, the movable display device is placed. The movable display device is an intelligent robot with mobile and visual information display. The road section information contained in the accident risk section area and the corresponding traffic accident risk information are input into the movable display device to remind the driver to complete the advance planning of the vehicle route.

[0063] In addition, the embodiment of the present application also provides a traffic accident prediction device based on multiple factors, such as Figure 3 As shown, the traffic accident prediction device based on the influence of multiple factors specifically includes:

[0064] At least one processor 301; and a memory 302 in communication with the at least one processor 301; wherein the memory 302 stores instructions executable by the at least one processor 301, so that the at least one processor 301 can execute:

[0065] According to the road traffic network data set obtained by the urban traffic management system, all traffic sections are screened based on the multi-factor correlation of accident risk sections to obtain the accident risk section area;

[0066] The historical traffic accident time series events in the accident risk section area are processed by network structure training under the matching of relevant time characteristics and traffic accident factors to obtain the urban traffic accident prediction model;

[0067] Through the urban traffic accident prediction model, the traffic accident risk is predicted for the accident risk section area in the current time period to obtain the traffic accident risk information;

[0068] Based on the traffic accident risk information, the accident road network corresponding to the accident risk section area is divided into sections of traffic diversion locations to determine the warning reminder location;

[0069] Through the early warning reminder location, the traffic accident risk information is output and processed to complete the traffic accident risk prediction.

[0070] The embodiments of the present application can improve road safety and the accuracy of accident warnings when encountering complex urban road conditions or complex road environments, and effectively provide warning information to drivers. Through multi-factor correlation screening and processing, combined with historical traffic accident data, the risk of traffic accidents can be predicted more accurately. The prediction model can help traffic management departments to reasonably allocate police force and rescue resources and improve the response efficiency in emergency situations. The road section division of the traffic diversion position can guide vehicles to avoid high-risk sections to reduce the occurrence of accidents. The determination of the warning reminder location can provide accident risk information to the driver in a timely manner to help the driver make safe driving decisions. Early prediction and warning can reduce casualties and property losses caused by traffic accidents. By outputting traffic accident risk information, the public's awareness of traffic accident risks can be improved and the formation of safe driving habits can be promoted. By avoiding high-risk sections, traffic congestion can also be reduced and road traffic efficiency can be improved. At the same time, the method can continuously update and optimize the prediction model according to real-time data to improve the adaptability and accuracy of the system.

[0071] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0072] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0073] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0074] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0075] 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.

[0076] 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 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0077] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0078] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0079] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0080] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0081] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the specification of the present application.

Claims

1. A traffic accident prediction method based on multiple factors, characterized in that: The method comprises: According to the road traffic network data set obtained by the urban traffic management system, all traffic sections are screened based on the multi-factor correlation of accident risk sections to obtain the accident risk section area; Performing network structure training processing on the historical traffic accident time series events in the accident risk section area under the matching of relevant time characteristics and traffic accident factors to obtain an urban traffic accident prediction model; Using the urban traffic accident prediction model, traffic accident risk prediction is performed on the accident risk road section area in the current time period to obtain traffic accident risk information; Based on the traffic accident risk information, the accident road traffic network corresponding to the accident risk section area is divided into sections of traffic diversion positions to determine the early warning reminder position; Through the early warning reminder position, the traffic accident risk information is output and processed to complete the traffic accident risk prediction.

2. The traffic accident prediction method based on multiple factors according to claim 1 is characterized in that: According to the road traffic network data set obtained by the urban traffic management system, all traffic sections are screened based on the multi-factor correlation of accident risk sections to obtain the accident risk section areas, including: According to the traffic accident database in the preset urban traffic management system, historical traffic accident data are extracted; and based on the historical traffic accident data, a road traffic network data set where various types of traffic accidents occurred is determined; wherein the historical traffic accident data at least includes: the location of the accident, the time period of the accident, the location of the accident, the type of vehicle, the cause of the accident, the human and social environment, and the road environment of the accident; The road traffic network data set is subjected to multi-factor identification and classification processing of external causes of traffic accidents to obtain traffic accident independent variable data under each traffic section; wherein the traffic accident independent variable data includes: accident time, accident space, road conditions, road surrounding environment, traffic flow, weather factors, and specific accident conditions; The road traffic network data set is subjected to multi-factor classification processing related to the intrinsic performance of traffic accidents to obtain traffic accident dependent variable data for each traffic section; wherein the traffic accident dependent variable data includes: accident form, accident severity, accident casualties and accident duration; By using a preset Pearson correlation coefficient algorithm, the correlation between the traffic accident independent variable data and the traffic accident dependent variable data is calculated, and traffic sections with strong correlation are screened out; The area occupied by the strongly correlated traffic sections is determined as the accident risk section area; wherein the accident risk section area is a section area that has an inherent accident risk occurrence law and can be trend predicted.

3. The traffic accident prediction method based on multiple factors according to claim 1 is characterized in that: Before performing network structure training processing on the historical traffic accident time series events in the accident risk road section area under the matching of relevant time characteristics with traffic accident factors to obtain the urban traffic accident prediction model, the method further includes: Identify historical traffic accident events in the accident risk section area, and classify the historical traffic accident events based on the accident occurrence time nodes under long and short time factors to obtain the historical traffic accident time sequence events; wherein the long time factors include: annual time nodes, monthly time nodes, daily time nodes and quarterly time nodes; short time factors include: planning time period range, morning and evening peak time range and cultural activity time range; Extracting the time characteristics of each of the historical traffic accident time series events in the accident risk section area and the corresponding traffic accident factors; wherein the traffic accident factors include: accident form, accident severity, accident casualties and accident duration; The time features are matched with the traffic accident factors one by one, and the traffic accident factors are integrated into the time features to obtain a time series accident feature sequence.

4. The traffic accident prediction method based on multiple factors according to claim 3 is characterized in that: The historical traffic accident time series events in the accident risk section area are processed by network structure training under the matching of relevant time characteristics and traffic accident factors to obtain an urban traffic accident prediction model, specifically including: Combining a preset Gaussian white noise with the time series accident feature sequence to obtain a new time series accident feature sequence; wherein the Gaussian white noise is a normally distributed noise; By using the empirical mode decomposition technology, the new time series accident characteristic sequence is repeatedly decomposed based on the modal component and the residual component to obtain a final residual value and multiple time series accident subsequences; Arrange and process a plurality of the time sequence accident subsequences; The number of hidden layer units and the learning rate of the neural network model are optimized through a global search of the particle swarm algorithm, and a neural network structure is constructed according to the hyperparameter values ​​corresponding to the particle initialization positions of the particle swarm algorithm; According to the local optimal particles and the global optimal particles after the iteration of the particle swarm algorithm, the optimal position training of the neural network structure related to the global minimum fitness value is performed to obtain an optimized neural network model; The final residual value and the sorted multiple time series accident subsequences are input into the optimized neural network model, and the optimal difference parameter of the particle swarm algorithm is used to train the optimized neural network model to obtain the urban traffic accident prediction model.

5. The traffic accident prediction method based on multiple factors according to claim 4 is characterized in that: Through the global search of the particle swarm algorithm, the number of hidden layer units and the learning rate of the neural network model are optimized, and the neural network structure is constructed according to the hyperparameter values ​​corresponding to the particle initialization positions of the particle swarm algorithm, specifically including: Initialize various parameters of the particle swarm algorithm; wherein the various parameters include: learning factor, number of iterations, number of hidden unit layers, initial particle velocity, velocity variation range, and particle target search area size; Determine other parameters of the neural network model; wherein the other parameters include: initial learning rate, learning rate variation range, iterative variation step size, initial number of hidden layer units and parameter range of number of hidden layer units; Based on various parameters of the initialized particle swarm algorithm and according to the root mean square error derivative of the measured value and the predicted value of the training set, a fitness function of the particle swarm algorithm is determined; By using the fitness function, the fitness value corresponding to the particle initialization position is calculated to determine the hyperparameter value corresponding to the particle initialization position; Based on the particle initialization positions and the hyperparameter values, the neural network structure is constructed and trained.

6. The traffic accident prediction method based on multiple factors according to claim 1 is characterized in that: Through the urban traffic accident prediction model, traffic accident risk prediction is performed on the accident risk section area in the current time period to obtain traffic accident risk information, specifically including: Obtaining the historical time series accident feature sequence of the accident risk road section area in the previous time period; Inputting the historical time series accident feature sequence into the urban traffic accident prediction model to obtain traffic accident risk prediction information for the next time period; Acquire the actual traffic network information of the accident risk section area in the current time period; wherein the actual traffic network information is the actual traffic accident independent variable data in the accident risk section area; Performing similarity matching between actual traffic characteristics in the actual traffic network information and risk prediction parameter information in the traffic accident risk prediction information to determine traffic accident risk similarity information; Based on the similarity threshold in the traffic accident risk similarity information, the traffic accident risk information in the next time period is determined; wherein the traffic accident risk information includes: high-risk traffic accident risk information, dangerous traffic accident risk information, medium-risk traffic accident risk information, low-risk traffic accident risk information and no traffic accident risk information.

7. The traffic accident prediction method based on multiple factors according to claim 1 is characterized in that: Based on the traffic accident risk information, the accident road traffic network corresponding to the accident risk section area is divided into sections of traffic flow diversion positions to determine the early warning reminder position, specifically including: Through the GIS geographic information platform, the road range of the accident risk section area is radially identified to determine the affected area of ​​the section; Based on the danger level of the traffic accident risk information in the accident risk road section area, the road section influence area is reduced or expanded, and the associated road section influence area is determined; wherein the higher the danger level, the corresponding row area is expanded; and the lower the danger level, the corresponding row area is reduced; If the associated road section influence area is an area after expansion processing, the intersection area included in the expansion area is determined as a traffic diversion location, and the traffic diversion location is set as a necessary warning reminder location; If the associated road section impact area is a reduced area, the intersection area in the reduced area is determined as a traffic diversion location, and the traffic diversion location is set as a non-essential warning reminder location; The warning reminder positions include necessary warning reminder positions and unnecessary warning reminder positions.

8. The traffic accident prediction method based on multiple factors according to claim 1 is characterized in that: Through the warning reminder position, the traffic accident risk information is output and processed to complete the traffic accident risk prediction, specifically including: Based on the warning reminder position, the movable display device is placed; wherein the movable display device is an intelligent robot capable of moving and visualizing information display; The road section information contained in the accident risk road section area and the corresponding traffic accident risk information are input into the movable display device to remind the driver to complete the advance planning of the vehicle route.

9. A traffic accident prediction device based on multiple factors, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute the traffic accident prediction method based on the influence of multiple factors according to any one of claims 1-8.

10. A non-volatile computer storage medium, characterized in that: The storage medium is a non-volatile computer-readable storage medium, which stores at least one program, each of which includes instructions. When the instructions are executed by the terminal, the terminal executes a traffic accident prediction method based on multiple factors according to any one of claims 1-8.

Images