A safety warning system for a road merging area

Abstract

The application provides a safety warning system for a road merging area, and belongs to the technical field of intelligent traffic safety warning, and comprises: a time prediction module, which is used for capturing driving information of a first vehicle in a first layer network of a dynamic organization network when it is monitored that the first vehicle drives into the dynamic organization network, and predicting an arrival time of the first vehicle from the first layer network to a second layer network; a situation determination module, which is used for determining a vehicle network topology of the road merging area at the arrival time, calibrating potential conflicts in the vehicle network topology, and analyzing a safety influence situation of the first vehicle; and a safety warning module, which is used for giving an early safety warning to the first vehicle according to the safety influence situation and driving habits of a driver of the first vehicle and driving habits of a driver of a calibrated vehicle, and giving a real-time safety warning to the first vehicle when the first vehicle drives into the second layer network until the first vehicle drives away from the second layer network, so that the timeliness and accuracy of safety warning are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent traffic safety warning, in particular to a safety warning system for a road merging area. BACKGROUND

[0002] The road merging area is a key node for the transition of traffic flow from multiple dispersed streams to a single stream, and its traffic operation state has typical characteristics of dense traffic flow, large speed difference, and complex driving intention. On the one hand, the vehicles on the main road need to maintain the original speed to ensure the traffic efficiency, and on the other hand, the merging vehicles on the ramp or branch road need to complete the operations of acceleration, observation, and merging into the main road traffic, which are prone to potential conflicts such as lateral scraping, rear-end collision, and cross avoidance in the spatial overlap range of the merging area. According to the statistical data of the transportation industry, the traffic accident rate of the road merging area is about 2 to 3 times that of the ordinary road section, and about 60% of the merging accidents are caused by the lack of timely warning, lagging conflict prediction, or lack of targeted control measures. In summary, how to realize the early risk perception and accurate safety warning of the vehicles in the merging area has become a core problem to be solved in the field of intelligent transportation.

[0003] Therefore, the present application provides a safety warning system for a road merging area. SUMMARY

[0004] The present application provides a safety warning system for a road merging area to solve the above technical problems.

[0005] The present application provides a safety warning system for a road merging area, comprising:

[0006] A time prediction module is configured to capture the driving information of a first vehicle in a first layer network of a dynamic organization network in the road merging area in real time when the first vehicle is detected to enter the dynamic organization network, and predict the arrival time of the first vehicle from the first layer network to a second layer network, wherein the coverage range of the dynamic organization network is larger than the area range of the road merging area, and the first layer network and the second layer network constitute the dynamic organization network, and the first layer network is located on the outer side of the second layer network.

[0007] A situation determination module is configured to determine the vehicle network topology of the road merging area at the arrival time based on the capture results of the dynamic organization network, label the potential conflicts in the vehicle network topology, and analyze the safety influence situation of the first vehicle.

[0008] A safety warning module is configured to give an early safety warning to the first vehicle according to the safety influence situation and the driving habits of the driver of the first vehicle and the driving habits of the driver of the reference vehicle, and give a real-time safety warning to the first vehicle when the first vehicle enters the second layer network until the first vehicle leaves the second layer network.

[0009] Preferably, the time prediction module comprises:

[0010] A distribution map acquisition unit is configured to acquire a first vehicle distribution map of a self area of the road merging area based on N times of synchronous and continuous acquisition before a current time, and a second vehicle distribution map of a set road section area based on an entering direction and an exiting direction of the road merging area;

[0011] An analysis unit is configured to fuse the first vehicle distribution map and the second vehicle distribution map of the same acquisition time to obtain an overall distribution map, and analyze vehicle distribution probabilities of each ramp and information pairs of static vehicles and dynamic vehicles existing in each ramp;

[0012] A sequence construction unit is configured to obtain a probability sequence and an information pair sequence of each ramp according to the vehicle distribution probabilities and the information pairs obtained through the N times of continuous acquisition;

[0013] A second layer network construction unit is configured to determine a first radius based on a center point of the road merging area based on the probability sequence and the information pair sequence of each ramp, and turn on standby network devices in a coverage area based on the first radius to construct a second layer network;

[0014] A first layer network construction unit is configured to acquire weather information of a location where the road merging area is located in real time, determine an entering coverage radius of a set road section area based on an entering direction based on a first length of the set road section area based on the entering direction and an entering terrain, determine an exiting coverage radius of a set road section area based on an exiting direction based on weather information and a second length of the set road section area based on the exiting direction and an exiting terrain, and turn on standby network devices in coverage areas of the entering coverage radius and the exiting coverage radius to construct a first layer network;

[0015] Preferably, the first layer network and the second layer network are synchronously updated, and the first layer network and the second layer network predict a next update period according to a change length of a corresponding preset radius based on the first radius and the second radius, respectively.

[0016] Preferably, the second layer network construction unit comprises:

[0017] A feature extraction subunit is configured to extract a peak probability value of the probability sequence of each ramp , a probability change rate and a probability coverage duration Meanwhile, a dynamic vehicle occupancy ratio of each ramp to a sequence of information sequences is extracted a speed difference coefficient and a static vehicle residence duration ;

[0018] a weight determination subunit configured to determine a basic weight wi of each ramp from a priority-merging-weight table according to a ramp function priority and a merging type of a road merging area;

[0019] a normalization subunit configured to respectively perform normalization processing on , , , , , to obtain normalized parameters , , , , , ;

[0020] a basic determination subunit configured to match a basic radius R0 from a standard-radius table according to a design standard of the road merging area;

[0021] a radius calculation subunit configured to calculate an influence coefficient of each ramp to the road merging area , and adjust the basic radius R0;

[0022] ;

[0023] wherein, , , , , , are preset coefficients, and ;

[0024] ;

[0025] wherein, is a first radius; n1 represents the number of ramps.

[0026] Preferably, the time prediction module further comprises:

[0027] a change length determination unit configured to calculate a first change length of the first radius to a corresponding preset radius , and a second change length of the second radius to a corresponding preset radius ;

[0028] a change rate determination unit configured to calculate a first change rate , a second change rate , respectively, based on the first change length , the second change length , wherein, is a current duration after the first radius and the second radius change synchronously;

[0029] a model prediction unit configured to input the first change length , 2, V1, V2 into a cycle prediction model, which is configured to:

[0030] when both the first change length , 2 are within a preset small threshold interval and both V1 and V2 are less than a rate threshold, output a basic update cycle T0;

[0031] when any one of the change lengths exceeds the preset small threshold interval or the change rate reaches the rate threshold, calculate a shortening coefficient , and output an adjusted cycle, wherein, is a preset weight coefficient; is a final change length; is a final change rate; is an average of the sum of the preset radii corresponding to the first radius and the second radius;

[0032] when it is detected that the road merging area has a sudden event, forcibly output a minimum update cycle Tmin.

[0033] Preferably, the cycle prediction model is dynamically optimized through correlation analysis of historical update data and actual risk events, so that the output next update cycle is positively correlated with the dynamic change strength of the traffic flow of the road merging area.

[0034] Preferably, the time prediction module further comprises:

[0035] a path parameter determination unit configured to determine a preset driving path of the first vehicle from a current position to a boundary of the second layer network based on a topological mapping relationship between the first layer network and the second layer network, and extract road feature parameters of the path, the road feature parameters including path length, bend curvature, slope, and lane number change rate, and the driving information including continuous time sequence real-time position coordinates, instantaneous speed, acceleration, steering angle, and lane deviation degree;

[0036] a model analysis unit configured to input the road feature parameters and the driving information into a time prediction model to predict an output arrival time at the boundary of the second layer network.

[0037] Preferably, the situation determination module comprises:

[0038] a parameter extraction unit configured to extract space-time characteristic parameters of all vehicles in the road merging area at the time of arrival based on real-time capture results of the dynamic organization network, the space-time characteristic parameters including predicted position coordinates, travel speed vectors, acceleration change rates, lane occupancy states, and remaining travel times of the vehicles;

[0039] a topology construction unit configured to construct a vehicle network topology based on the space-time characteristic parameters, with the first vehicle as a core node and other vehicles as associated nodes, and edge weights between the nodes being determined by relative distances, speed differences, and trajectory intersection probabilities of the two vehicles;

[0040] a calibration unit configured to calibrate first associated nodes in the vehicle network topology that have spatial intersections with the predicted trajectory of the first vehicle and time differences less than a preset safety threshold, second associated nodes in the vehicle network topology that have no direct trajectory intersections but have lane competition leading to compression of the first vehicle's lane changing space, and third associated nodes in the vehicle network topology that are affected by environmental interference and cause braking delays, to generate conflict calibration judgment parameter groups of each potential level;

[0041] a potential analysis unit configured to perform cluster analysis on the conflict calibration judgment parameter groups of each potential level to obtain calibration clusters of the corresponding potential level, and calculate dispersion degrees of all calibration clusters of the same potential level and matching degrees of calibration results of the corresponding potential level and historical real conflict data;

[0042] a comprehensive analysis unit configured to perform comprehensive analysis on the dispersion degrees and the matching degrees of each potential level to obtain optimal calibration parameter groups of the vehicle network topology, and input the optimal calibration parameter groups into a safety evaluation model to obtain a safety impact situation of the first vehicle.

[0043] Preferably, the safety warning module comprises:

[0044] a habit parameter extraction unit configured to extract driving habit characteristic parameters of a driver of the first vehicle;

[0045] an associated habit extraction unit configured to obtain driving habit characteristic parameters of drivers of calibration vehicles, the calibration vehicles being associated vehicles in the vehicle network topology that are calibrated to have potential conflicts;

[0046] a habit matching calculation unit configured to calculate habit matching degree indexes of the first vehicle and the calibration vehicles;

[0047] a risk level determination unit configured to input the safety impact situation, the habit matching degree indexes of the first vehicle and the calibration vehicles, and calibration types of the calibration vehicles into a comprehensive analysis model to determine a real-time risk dynamic level;

[0048] The safety warning unit is used for selecting a corresponding multi-modal warning mode based on the operation preference of the driver of the first vehicle and performing safety warning according to the real-time risk level.

[0049] Compared with the prior art, the application has the following beneficial effects:

[0050] The dynamic organization network is used for hierarchical monitoring of the merging area and the surrounding area, the vehicle network topology is used for accurate analysis of potential conflicts of multiple vehicles, and the driver driving habit is used for realizing differentiated warning, so that the merging risk can be predicted in advance, and real-time warning can be performed after the vehicle enters the core risk area, the driving safety of the road merging area is effectively improved, and the probability of traffic accidents is reduced.

[0051] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the application. The objects and other advantages of the present application can be achieved and obtained by the structure particularly pointed out in the written description and the accompanying drawings.

[0052] The technical solutions of the present application will be further described in detail below by means of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0053] The accompanying drawings are used to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, used to explain the present application, and do not constitute a limitation on the present application. In the drawings:

[0054] Figure 1 It is a structure diagram of a safety warning system for a road merging area in an embodiment of the present application;

[0055] Figure 2 It is a schematic diagram of an overall distribution diagram in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The preferred embodiments of the present application will be described below in combination with the drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application, and do not limit the present application.

[0057] The present application provides a safety warning system for a road merging area, as shown in Figure 1 The safety warning system comprises:

[0058] a time prediction module configured to capture, in real time, driving information of a first vehicle in a first layer network of a dynamic organization network based on a dynamic organization network of a road merging area when the first vehicle is detected to enter the dynamic organization network, and predict an arrival time of the first vehicle from the first layer network to a second layer network, wherein a coverage range of the dynamic organization network is greater than a range of the road merging area, the first layer network and the second layer network constitute the dynamic organization network, and the first layer network is located outside the second layer network;

[0059] a situation determination module configured to determine a vehicle network topology of the road merging area at the arrival time based on a capture result of the dynamic organization network, label potential conflicts in the vehicle network topology, and analyze a safety influence situation of the first vehicle;

[0060] a safety warning module configured to perform early safety warning on the first vehicle according to the safety influence situation and driving habits of a driver of the first vehicle and a driver of a labeled vehicle, and perform real-time safety warning on the first vehicle when the first vehicle enters the second layer network until the first vehicle leaves the second layer network.

[0061] In this embodiment, the dynamic organization network is a hierarchical monitoring network framework for monitoring the road merging area and surrounding traffic flow, and has a coverage range greater than the actual merging area. The first layer network is an outer monitoring range, i.e., far from the merging area, for capturing vehicles entering in advance. The second layer network is an inner monitoring range, i.e., close to the merging area, for focusing on the core risk area of the merging. Taking the scenario of a city expressway auxiliary road merging into a main road as an example, the first layer network can be set as a road section from 300 meters before the entrance of the auxiliary road to the entrance, and the second layer network is a road section from the entrance to 200 meters after the completion of the merging into the main road.

[0062] The first vehicle is a target vehicle entering the first layer network of the dynamic organization network and needing to be monitored and warned.

[0063] In this embodiment, the vehicle network topology is an abstract representation of the relationship between the first vehicle and other associated vehicles in the merging area, including position, driving direction, speed, etc., by taking the vehicles as network nodes and the mutual relationship between the vehicles as edges.

[0064] The potential conflict labeling is to identify the mutual interaction between the vehicles in the vehicle network topology that may cause a collision or danger, such as the existence of a crossing between the trajectory of the first vehicle and a large truck, and the time difference in reaching the crossing point is less than 2 seconds, which is labeled as a direct trajectory conflict.

[0065] The safety influence situation is to analyze the influence degree of the potential conflict on the safety of the first vehicle, including the probability of conflict occurrence and the severity of the conflict once it occurs.

[0066] In this embodiment, the driving habit is a behavior pattern formed by the driver in long-term driving, including reaction time of operation after encountering danger, lane changing frequency, braking force habit, etc.

[0067] In this embodiment, the early safety warning is a warning based on the prediction result when the first vehicle is still in the first layer network, and the real-time safety warning is continuous monitoring and warning after the first vehicle enters the second layer network until it leaves the second layer network. The beneficial effects of the above technical solutions are: through dynamic organization of network to monitor the merging area and the surrounding area in layers, combining with the vehicle network topology to accurately analyze the potential conflict of multiple vehicles, and integrating the driving habit of the driver to realize differentiated warning, which can not only predict the merging risk in advance, but also warn in real time after the vehicle enters the core risk area, effectively improving the driving safety of the road merging area and reducing the probability of traffic accidents.

[0068] The present application provides a safety warning system for a road merging area, the time prediction module comprising:

[0069] A distribution map acquisition unit is configured to acquire a first vehicle distribution map within the road merging area based on N times of synchronous and continuous acquisition before the current time, and a second vehicle distribution map based on a set road section area of the entering direction and the exiting direction of the road merging area;

[0070] An analysis unit is configured to fuse the first vehicle distribution map and the second vehicle distribution map at the same acquisition time to obtain an overall distribution map, and analyze the vehicle distribution probability of each ramp and the information pair of static vehicles and dynamic vehicles existing in each ramp.

[0071] A sequence construction unit is configured to obtain the probability sequence and the information pair sequence of each ramp based on the vehicle distribution probability and the information pair obtained by continuous N times of acquisition.

[0072] A second layer network construction unit is configured to determine a first radius based on the center point of the road merging area based on the probability sequence and the information pair sequence of each ramp, and turn on the standby network device of the coverage area based on the first radius to construct a second layer network.

[0073] A first layer network construction unit is configured to determine an entering coverage radius based on the set road section area of the entering direction by real-time acquisition of weather information of the location of the road merging area and combination of the first length of the set road section area of the entering direction and the entering terrain, and determine an exiting coverage radius based on the set road section area of the exiting direction based on the weather information and combination of the second length of the set road section area of the exiting direction and the exiting terrain, and turn on the standby network device of the coverage area of the entering coverage radius and the exiting coverage radius to construct a first layer network.

[0074] The first layer network and the second layer network are synchronously updated, and the first layer network and the second layer network respectively predict the next update period according to the change length of the first radius and the second radius from the corresponding preset radius.

[0075] In this embodiment, data is synchronously collected at the same time point for N times forward by 5 seconds / time interval, where N is a preset positive integer, which is set according to the monitoring accuracy requirement, for example, N is 10.

[0076] In this embodiment, the first vehicle distribution map only covers the vehicle position distribution map of the road merging area itself, and records the real-time position coordinates of all vehicles in the area. For example, the road merging area is the intersection section of the main road and the ramp of the highway, which is about 200 meters long and 3 lanes wide, and the first vehicle distribution map records the positions of all vehicles within the 200-meter range.

[0077] In this embodiment, the driving-in direction refers to the upstream section of the merging area, such as the entrance of the ramp or the entrance of the auxiliary road, and the driving-out direction refers to the downstream section of the merging area, such as the extension section after the main road merging. The set section is a fixed length section that is pre-defined according to the traffic monitoring requirement. Taking the highway ramp merging as an example, the set section of the driving-in direction is 300 meters before the entrance of the ramp, and the set section of the driving-out direction is 200 meters after the merging point of the main road.

[0078] In this embodiment, the second vehicle distribution map covers the vehicle position distribution map of the set section of the driving-in direction and the set section of the driving-out direction, and records the real-time position coordinates of all vehicles in the range. Specifically, high-definition cameras and millimeter-wave radars are deployed in the merging area itself, the set section of the driving-in direction and the set section of the driving-out direction, and the vehicle position coordinates based on radar ranging and camera positioning are arranged into a visual distribution map.

[0079] In this embodiment, the fusion processing of the same collection time is to superimpose the vehicle position data of the first vehicle distribution map and the second vehicle distribution map at the same collection time, eliminate data duplication, and form an overall distribution map covering the entire range of the set section of the driving-in direction, the merging area itself and the set section of the driving-out direction, as shown in Figure 2 .

[0080] The vehicle distribution probability of each ramp is the proportion of the number of times a vehicle exists in a certain ramp in the total number of times collected in the last N times of collection, which reflects the traffic congestion degree of the ramp.

[0081] In this embodiment, static vehicles refer to vehicles with speed less than or equal to a preset threshold in the ramp, such as temporarily parked vehicles, broken-down vehicles, etc.; dynamic vehicles refer to normally driving vehicles with speed greater than 5 km / h; and information pair refers to the combined data of the position and speed of each static vehicle and dynamic vehicles within a range of 50 meters around the static vehicle, for example, there is 1 static broken-down vehicle in the ramp, the position of which is 50 meters in front of the ramp and the speed of which is 0 km / h, and there are 2 dynamic sedans around the static vehicle, the position of sedan 1 is 80 meters in front of the ramp and the speed of sedan 1 is 40 km / h, and the position of sedan 2 is 30 meters in front of the ramp and the speed of sedan 2 is 35 km / h, then the information pair is: {broken-down vehicle position / speed, sedan 1 position / speed}, {broken-down vehicle position / speed, sedan 2 position / speed}.

[0082] In this embodiment, the probability sequence is an ordered data set formed by arranging the vehicle distribution probability of each ramp obtained by continuous N times of collection in the order of collection time, for example: {70%, 75%, 80%, 80%, 85%, 85%, 90%, 90%, 85%, 80%}.

[0083] In this embodiment, the information pair sequence is an ordered data set formed by arranging the static-dynamic vehicle information pair of each ramp obtained by continuous N times of collection in the order of collection time, for example: {1st information pair,..., Nth information pair].

[0084] In this embodiment, the first radius is a coverage radius calculated according to the probability sequence and the information pair sequence of each ramp, which determines the monitoring range of the second layer network. In this embodiment, the standby network device is a millimeter wave radar, a high-definition camera, a vehicle-road cooperation sensor, etc. which is pre-deployed around the road and is in a low-power sleep state at ordinary times and is started by remote instruction when the monitoring network needs to be constructed.

[0085] The weather information is real-time collected meteorological data such as precipitation, visibility, and wind force at the location of the merging area, which affects the braking distance and driving speed of vehicles and needs to be included in the calculation of the coverage radius.

[0086] In this embodiment, the driving-in terrain refers to the terrain of the set road section in the driving-in direction of the merging area, such as uphill, downhill, and straight road; and the driving-out terrain refers to the terrain of the set road section in the driving-out direction of the merging area, and the terrain affects the acceleration or deceleration ability of the vehicle and needs to adjust the coverage radius.

[0087] In this embodiment, the driving-in coverage radius is the monitoring radius of the set road section in the driving-in direction; and the driving-out coverage radius is the monitoring radius of the set road section in the driving-out direction, and the driving-in coverage radius = basic driving-in radius x (1 + precipitation influence coefficient + terrain influence coefficient),

[0088] In this embodiment, the driving-out coverage radius = the basic driving-out radius x (1 + the precipitation influence coefficient + the terrain influence coefficient), wherein the terrain influence coefficient is matched from a terrain-influence comparison table by querying the terrain of the driving-in / driving-out section, and the precipitation influence coefficient is matched based on a terrain-weather-influence comparison table, wherein the terrain-influence comparison table and the terrain-weather-influence comparison table are pre-set, and the corresponding influence coefficient can be obtained directly by matching the corresponding comparison table, for example, the terrain is uphill, and the terrain influence coefficient matched from the terrain-influence comparison table is 0.1, for example, the terrain is uphill and the rainfall is light rain, and the precipitation influence coefficient matched from the terrain-weather-influence comparison table is 0.2, and it should be noted that the basic driving-in radius, the basic driving-out radius and the basic radius corresponding to different merging areas are all pre-set in the table, and can be obtained by matching the design standard of the merging area.

[0089] The beneficial effects of the above technical solution are: by acquiring the traffic flow distribution data of the merging area and the upstream and downstream, analyzing the traffic flow characteristics through probability and information pairs, and finally combining the traffic flow, weather and terrain to dynamically build the outer first layer network and the inner second layer network, the adaptive adjustment of the monitoring range of the merging area is realized, the upstream driving-in vehicles are captured in advance, the basis for subsequent prediction of vehicle arrival time and analysis of conflict situation is provided, and the accuracy of safety warning is ensured.

[0090] The application provides a safety warning system for a road merging area, and the second layer network construction unit comprises:

[0091] The feature extraction subunit is used for extracting the peak probability value of the probability sequence of each ramp , the probability change rate and the probability coverage duration , simultaneously, the dynamic vehicle proportion , the speed difference coefficient and the static vehicle retention duration of the information pair sequence of each ramp are extracted.

[0092] In this embodiment, the peak probability value is the maximum distribution probability in the corresponding probability sequence, the probability change rate is the mean value of the probability difference values collected in succession and continuously in the corresponding probability sequence, the probability coverage duration is the cumulative duration of the probability value exceeding 50% in the probability sequence, reflecting the time span of the continuous high flow of the ramp, the dynamic vehicle proportion is the ratio of the number of dynamic vehicles to the total number of vehicles in the information pair sequence, the speed difference coefficient is the absolute value of the difference between the average speed of the dynamic vehicles and the speed limit of the main road of the merging area / the speed limit of the main road of the merging area, and the static vehicle retention duration is the average duration of the continuous presence of static vehicles in the information pair sequence, reflecting the potential congestion risk in the ramp.

[0093] The weight determination sub-unit is configured to determine the basic weight wi of each ramp according to the ramp function priority and the merging type of the road merging area from a priority-merging-weight lookup table.

[0094] In this embodiment, the ramp function priority is the importance level of the ramp function, for example, the importance level of the main entrance ramp of the highway is high.

[0095] In this embodiment, the merging type includes the merging type of the main road ramp of the highway and the merging type of the branch road of the urban trunk road.

[0096] In this embodiment, the priority-merging-weight lookup table is a pre-set table containing the combination of the ramp function priority and the merging type, and the basic weight of the ramp matched with each combination, for example, the high priority and the merging type of the main road ramp of the highway, according to the table query, wi is 0.4, and the sum of all wi under this road merging is 1.

[0097] In this embodiment, the design standard is a pre-set standard followed in road engineering construction, and the standard-radius lookup table is a table pre-stored with the set standard and the corresponding basic radius R0, for example, the merging area of the design speed 120km / h, 3-lane main road and 1-lane ramp, according to the table lookup, R0 is 150m.

[0098] The standardization sub-unit is configured to respectively standardize , , , , , to obtain the standardized parameters , , , , , .

[0099] The basic determination sub-unit is configured to match the basic radius R0 from the standard-radius lookup table according to the design standard of the road merging area.

[0100] The radius calculation sub-unit is configured to calculate the influence coefficient of each ramp on the road merging area, adjust the basic radius R0, and calculate the radius R of each ramp.

[0101] .

[0102] wherein, , 、 、 、 、 is a preset coefficient, ;

[0103] ;

[0104] wherein, is the first radius; n1 represents the number of ramps.

[0105] In this embodiment, 、 、 、 、 、 is obtained by historical merging accident data analysis and simulation experiment calibration.

[0106] In this embodiment, the value of 2 in the formula is determined by historical merging accident data regression analysis and simulation experiment optimization: in a large number of simulation scenarios with different ramp numbers and different risk intensity combinations, when the basic adjustment coefficient is set to 2, the first radius R1 calculated by the formula has the highest matching degree with the optimal monitoring range based on conflict risk.

[0107] The beneficial effects of the above technical solution are: by extracting multi-dimensional features of ramp traffic, combining ramp priority and merging type weighting, and then dynamically calculating the first radius through the standardization and weighting formula, the second layer network coverage range is accurately matched with the actual traffic risk of the merging area, which provides an accurate inner monitoring basis for subsequent vehicle arrival time prediction and conflict situation analysis, and improves the pertinence and accuracy of safety warning.

[0108] The present application provides a safety warning system for a road merging area, wherein the time prediction module further comprises:

[0109] a change length determination unit for calculating a first change length of the first radius corresponding to a preset radius a second change length of the second radius corresponding to a preset radius ;

[0110] In this embodiment, = the first radius - the preset radius of the first radius;

[0111] = the second radius - the preset radius of the second radius;

[0112] and the preset radius is less than or equal to the corresponding calculated radius.

[0113] In this embodiment, the current duration is the time interval from the last radius adjustment to the present.

[0114] In this embodiment, the base update period T0 is the default update period adopted, which is pre-set, for example, T0=30s.

[0115] The change rate determination unit is configured to calculate the first change rate and the second change rate based on the first change length and the second change length respectively, wherein, is the current duration after the first radius and the second radius change synchronously;

[0116] The model prediction unit is configured to input , 2, V1, V2 into the cycle prediction model, which is configured to:

[0117] When , 2 are both in the preset small threshold interval and V1, V2 are both less than the rate threshold, the base update period T0 is outputted.

[0118] When any one of the change lengths exceeds the preset small threshold interval or the change rate reaches the rate threshold, the shortening coefficient is calculated and the adjusted period is outputted, wherein, is the preset weight coefficient; is the final change length; is the final change rate; is the average of the sum of the preset radii corresponding to the first radius and the second radius;

[0119] In this embodiment, The value is 0.8, which reflects the importance of the radius change factor through the calibration of historical merging data.

[0120] ;

[0121] and ;

[0122] ,

[0123] and ;

[0124] In this embodiment, the unexpected situation of the road merging area during the emergency event, for example, traffic accident, vehicle accident, extreme weather mutation, etc.

[0125] In this embodiment, the dynamic optimization of the cycle prediction model is the adjustment of the small threshold, the rate threshold and w0.

[0126] In this embodiment, the minimum update period Tmin is the shortest time window from the appearance of a risk hidden danger to the occurrence of an accident in the road merging area, for example, the shortest time is 5 seconds.

[0127] In this embodiment, the preset micro threshold interval is 0 to 5 meters.

[0128] In this embodiment, the value of the rate threshold is 0.5 meters / second.

[0129] When it is detected that the road merging area has a sudden event, the minimum update period Tmin is forced to be output.

[0130] Preferably, the period prediction model is dynamically optimized through the association analysis of historical update data and actual risk events, so that the output next update period is positively correlated with the dynamic change intensity of the traffic flow of the road merging area.

[0131] In this embodiment, the preset radii of the first radius and the second radius are determined according to the design standards of the merging area, for example, the preset reference of the first radius of the high-speed merging area is 150 meters.

[0132] In this embodiment, the period prediction model is a logic realized by programming multi-condition judgment and formula calculation.

[0133] The beneficial effects of the above technical solution are: by quantifying the change length and change rate of the first radius and the second radius, and combining the period prediction model to dynamically adjust the update period: when the traffic flow is stable, the basic period is used to balance the resources and precision, and when the traffic flow changes greatly or encounters a sudden event, the period is shortened, thereby realizing accurate and timely response to the dynamic traffic flow of the merging area, providing more real-time basic data for subsequent safety warning, and improving the accuracy and timeliness of the warning.

[0134] The present application provides a safety warning system for a road merging area, wherein the time prediction module further comprises:

[0135] A path parameter determination unit is configured to determine a preset driving path of the first vehicle from the current position to the boundary of the second layer network based on the topological mapping relationship between the first layer network and the second layer network, and extract road feature parameters of the path, wherein the road feature parameters include path length, bend curvature, slope, and lane number change rate, and the driving information includes real-time position coordinates, instantaneous speed, acceleration, steering angle, and lane deviation degree in continuous time sequence.

[0136] A model analysis unit is configured to input the road feature parameters and the driving information into a time prediction model to predict the arrival time of reaching the boundary of the second layer network.

[0137] In this embodiment, the topological mapping relationship refers to the corresponding and associated logic of the first layer network and the second layer network based on the spatial layout of the road design CAD drawing, and clearly defines the path space law of the vehicle entering the inner layer from the outer layer, for example, in the high-speed ramp merging scene, the first layer network covers the 300-meter road section before the ramp entrance, and the second layer network covers the core 200-meter area where the ramp intersects with the main road, and the topological mapping relationship is that any position within the 300-meter front ramp can be corresponded to the driving trajectory direction of the second layer network boundary along the ramp direction.

[0138] In this embodiment, the preset driving path is a driving trajectory of the vehicle from the current point in the first layer network to the boundary of the second layer network, which is pre-planned according to the road design and the current position of the first vehicle.

[0139] In this embodiment, the time prediction model is an input of road feature parameters and vehicle driving information, and is learned and mapped by a multi-layer neural network for complex nonlinear relationship, and outputs the predicted time of the first vehicle reaching the boundary of the second layer network, which is obtained by training the neural network model, and the training sample is more than 1000 cases.

[0140] The beneficial effects of the above technical solutions are: based on the spatial correlation of the two-layer monitoring network, the driving path is planned, the road features are extracted, the real-time driving information of the vehicle is collected, these parameters are input into the time prediction model, the time of the first vehicle reaching the boundary of the second layer network is accurately calculated, and a key time benchmark is provided for subsequent conflict situation analysis and safety warning, and the warning accuracy is indirectly improved.

[0141] The present application provides a safety warning system for a road merging area, and the situation determination module comprises:

[0142] The parameter extraction unit is used for extracting the space-time feature parameters of all vehicles in the road merging area according to the real-time capture result of the dynamic organization network, and the space-time feature parameters include the predicted position coordinates, the driving speed vector, the acceleration change rate, the lane occupation state and the remaining passing time of each vehicle.

[0143] The topological construction unit is used for constructing a vehicle network topology according to the space-time feature parameters, taking the first vehicle as a core node and other vehicles as associated nodes, and the edge weight between nodes is determined by the relative distance, speed difference and trajectory intersection probability of two vehicles.

[0144] The calibration unit is used for calibrating the first associated node in the vehicle network topology which has a spatial intersection with the predicted trajectory of the first vehicle and a time difference less than a preset safety threshold, calibrating the second associated node in the vehicle network topology which has no direct trajectory intersection and causes the first vehicle to be compressed in the lane changing space due to lane competition, and calibrating the third associated node in the vehicle network topology which is affected by environmental interference and causes brake delay, to generate conflict calibration parameter groups of each potential level.

[0145] a potential analysis unit configured to perform cluster analysis on the conflict labeling decision parameter set of each potential level to obtain a labeling cluster of the corresponding potential level, and calculate a dispersion of all labeling clusters of the same potential level and a matching degree of the labeling result of the corresponding potential level with historical real conflict data;

[0146] a comprehensive analysis unit configured to perform comprehensive analysis on the dispersion and the matching degree of each potential level to obtain an optimal labeling parameter set of the vehicle network topology, and input the optimal labeling parameter set into a safety evaluation model to obtain a safety influence situation of the first vehicle.

[0147] In this embodiment, the arrival time is the predicted time for the first vehicle to arrive at the boundary of the second layer network from the current position.

[0148] In this embodiment, the predicted position coordinate is the predicted position of the vehicle in the merging area at the arrival time, the travel speed vector includes a vector of speed size and direction, and the lane occupancy state is the lane number and occupancy degree occupied by the vehicle, such as completely occupying the second lane and partially occupying the first lane.

[0149] In this embodiment, the associated node is a vehicle other than the first vehicle in the topology, for example, a large truck or an SUV in the merging area, which is an associated node of the first vehicle.

[0150] In this embodiment, the relative distance is the straight-line distance between the two vehicles, for example, the first vehicle is 30 meters away from the large truck, and the relative distance is 30 meters. The speed difference is the difference between the speeds of the two vehicles, for example, the speed of the first vehicle is 20 m / s, and the speed of the large truck is 15 m / s, so the speed difference is 5 m / s.

[0151] In this embodiment, the trajectory intersection probability is the probability that the predicted trajectories of the two vehicles intersect at a certain time in the future, which is determined by a pre-trained trajectory prediction model, and the model is trained by a neural network based on the driving trajectories of the two vehicles and the trajectory intersection conditions. The training sample is more than 1000 cases, so the trajectory intersection probability can be directly obtained. For example, the trajectory intersection probability of the first vehicle and the SUV is 60%.

[0152] In this embodiment, the edge weight calculation is to normalize and weight the relative distance, the speed difference, and the trajectory intersection probability to obtain the edge weight. For example, the relative distance weight is set to 0.4, the speed difference weight is set to 0.3, and the trajectory intersection probability weight is set to 0.3. If the corresponding normalized value of the relative distance is 0.6, the corresponding normalized value of the speed difference is 0.5, and the corresponding normalized value of the trajectory intersection probability is 0.6, then the weighted calculation can be performed to obtain the edge weight.

[0153] In this embodiment, the first associated node is a node in the vehicle network topology that has a spatial intersection with the first vehicle's predicted trajectory and a time difference less than a preset safety threshold. For example, the first vehicle predicts that it will be at position (100, 50) after 3 seconds, vehicle B also predicts that it will be at position (100, 50) after 3 seconds, and the time difference is only 0.5 seconds, which is less than the safety threshold of 1 second. In this case, vehicle B is the first associated node.

[0154] In this embodiment, the second associated node is a node in the vehicle network topology that has no direct trajectory intersection with the first vehicle, but the first vehicle's lane changing space is compressed due to lane competition. For example, the first vehicle wants to change lanes from the second lane to the first lane, but vehicle C is on the first lane and the lateral distance between vehicle C and the first vehicle is only 1 meter, which is less than the lane changing safety distance of 1.5 meters. In this case, vehicle C is the second associated node.

[0155] In this embodiment, the third associated node is a node in the vehicle network topology that is affected by environmental interference and may cause brake delay. For example, on a rainy day, the road is slippery, and vehicle D is 40 meters away from the first vehicle, but due to the decrease in the road friction coefficient, the braking distance is extended to 50 meters. In this case, vehicle D is the third associated node.

[0156] In this embodiment, the conflict labeling decision parameter set is a set of parameters used to label each associated node, such as spatial intersection coordinates, time difference, lane changing space distance, and environmental interference coefficient. For example, when labeling the first associated node, the parameter set includes the predicted spatial coordinates of the two vehicles (100, 50), the time difference of 0.5 seconds, and the safety threshold of 1 second.

[0157] In this embodiment, the clustering analysis is a grouping of conflict labeling decision parameter sets in the same potential level according to similarity. For example, all parameter sets of the first associated node are clustered according to the size of the time difference and the similarity of the spatial intersection position, resulting in multiple labeling clusters, such as a cluster with a time difference of 0-0.5 seconds and a cluster with a time difference of 0.5-1 second.

[0158] In this embodiment, the labeling cluster is a set of parameters obtained after clustering, representing the same type of conflict characteristics. For example, labeling cluster 1 of the first associated node includes all parameter sets with a time difference of 0-0.5 seconds and a spatial intersection in the middle lane of the main road.

[0159] In this embodiment, the dispersion degree is the dispersion degree of the parameters in the labeling cluster. For example, the time difference in labeling cluster 1 fluctuates within 0-0.5 seconds with a low dispersion degree, while the time difference in another cluster fluctuates within 0-1 second with a high dispersion degree.

[0160] In this embodiment, the matching degree is the degree of coincidence between the calibration results of the calibration cluster and the historical true conflict data, for example, the calibration result of a certain calibration cluster is that the time difference is 0~0.5 seconds and the space intersection is the middle lane of the main road, and the coincidence degree with the actual direct conflict data in this scene in history is 80%, so the matching degree is 80%, specifically: the clustering algorithm K-means is used to cluster the conflict calibration judgment parameter groups of each level; the dispersion of the calibration cluster is calculated by variance and standard deviation; the features of the calibration cluster, such as the time difference range and the spatial position, are compared with the parameter records of the past merging accidents or conflicts stored in the historical conflict database, and the matching degree is calculated.

[0161] In this embodiment, the optimal calibration parameter group is the calibration parameter group that can most accurately reflect the conflict after the dispersion and matching degree of each potential level are comprehensively filtered, for example, by comparing the calibration clusters of each level, the dispersion of a certain parameter group: time difference threshold 0.8 seconds, lane changing space threshold 1.2 meters, and environmental interference coefficient 0.7 is low and the matching degree is high, that is, the optimal calibration parameter group.

[0162] In this embodiment, the safety evaluation model is a pre-trained neural network model that inputs the optimal calibration parameter group and outputs the safety impact situation of the first vehicle. The safety impact situation is the safety risk situation that the first vehicle faces in the merging area, including risk level, collision probability, and impact range, etc. For example, the safety impact situation of the first vehicle is high risk, and there is a possibility of direct trajectory conflict with 2 associated nodes, and the collision probability is 75%.

[0163] The beneficial effects of the above technical solutions are: by extracting the time and space features of the vehicle to construct the network topology, layering the potential conflict nodes, and then analyzing the optimal calibration parameters through clustering and comprehensive analysis, the precise safety impact situation is finally output by inputting the safety evaluation model, which provides comprehensive and quantitative risk basis for subsequent safety warning.

[0164] The present application provides a safety warning system for a road merging area, which comprises a safety warning module, comprising:

[0165] A habit parameter extraction unit is configured to extract driving habit feature parameters of a driver of the first vehicle.

[0166] An associated habit extraction unit is configured to obtain driving habit feature parameters of a driver of a calibration vehicle, wherein the calibration vehicle is an associated vehicle in the vehicle network topology that is calibrated to have potential conflicts.

[0167] A habit matching calculation unit is configured to calculate habit matching degree indexes of the first vehicle and each calibration vehicle.

[0168] a risk level determination unit configured to determine a real-time risk dynamic level by inputting the safety influence situation, the habit matching degree index of the first vehicle and each calibration vehicle, and the calibration type of each calibration vehicle into a comprehensive analysis model;

[0169] a safety warning unit configured to select a corresponding multi-modal warning mode based on the operation preference of the driver of the first vehicle, and perform safety warning according to the real-time risk level.

[0170] In this embodiment, the driving habit characteristic parameters include historical reaction delay time, brake pedal operation force distribution, lane change decision hesitation coefficient, and speed adjustment rate in adverse weather.

[0171] In this embodiment, the calibration vehicle is a related vehicle in a vehicle network topology that is calibrated to have potential conflicts, and is a vehicle that may have conflicts with the first vehicle. In this embodiment, the habit matching degree index is calculated as 1-(reaction time difference weight x normalized reaction difference + operation force deviation weight x normalized force deviation + decision tendency weight x (1-normalized similarity)), wherein after normalization of each index, the reaction difference contributes 0.2, the force deviation contributes 0.15, and the decision tendency contributes 0.15.

[0172] In this embodiment, the comprehensive analysis model is obtained by training a neural network model with safety influence situations in different combinations, habit matching degree indexes of the first vehicle and each calibration vehicle, and calibration types of each calibration vehicle as sample inputs, and expert analysis risk levels for the combinations as sample outputs, and the training sample is more than 1000 cases, so that the real-time risk dynamic level can be directly obtained.

[0173] In this embodiment, the operation preference is the sensitivity and preference of the driver to different warning modes, for example, a driver is more sensitive to steering wheel vibration and tends to ignore voice prompts, and the operation preference is to prefer haptic warning.

[0174] Multi-modal warning mode: a combination of sound, vision, and haptic warning forms, for example, high risk, high frequency beep, fast light flashing, strong steering wheel vibration; medium risk, medium frequency beep, medium light flashing, medium vibration.

[0175] The above technical solution has the beneficial effects that by extracting and matching the driving habits of the driver of the first vehicle and the potential conflict vehicle, and selecting a multi-modal warning mode and adapting the intensity according to the operation preference of the driver, perceptual fatigue caused by single warning is avoided, and the effectiveness of the warning and the acceptance of the driver are improved.

[0176] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A safety warning system for a road merging area, characterized by, Comprising: a time prediction module for predicting the arrival time of a first vehicle from a first layer network to a second layer network based on the dynamic organization network of a road merging area, wherein the coverage range of the dynamic organization network is greater than the area range of the road merging area, the first layer network and the second layer network constitute the dynamic organization network, and the first layer network is located outside the second layer network; a situation determination module for determining the vehicle network topology of the road merging area at the arrival time based on the capture results of the dynamic organization network, labeling potential conflicts in the vehicle network topology, and analyzing the safety impact situation of the first vehicle; a safety warning module for providing early safety warning to the first vehicle according to the safety impact situation and the driving habits of the driver of the first vehicle and the driving habits of the labeled vehicle, and providing real-time safety warning to the first vehicle when the first vehicle enters the second layer network until the first vehicle leaves the second layer network; wherein the time prediction module comprises: a distribution map acquisition unit for acquiring a first vehicle distribution map within the road merging area based on N times of synchronous continuous acquisition before the current time, and a second vehicle distribution map based on the set road section area of the entry direction and the exit direction of the road merging area; an analysis unit for fusing the first vehicle distribution map and the second vehicle distribution map at the same acquisition time to obtain an overall distribution map, and analyzing the vehicle distribution probability of each ramp and the information pair of static vehicles and dynamic vehicles existing in each ramp; a sequence construction unit for obtaining the probability sequence and information pair sequence of each ramp based on the vehicle distribution probability and information pair obtained by continuous N times of acquisition; a second layer network construction unit for determining a first radius based on the center point of the road merging area based on the probability sequence and information pair sequence of each ramp, turning on the standby network device based on the coverage area of the first radius, and constructing the second layer network; a first layer network construction unit for determining the entry coverage radius based on the set road section area of the entry direction by real-time acquisition of weather information of the location of the road merging area and combination of the first length of the set road section area of the entry direction and the entry terrain, and determining the exit coverage radius based on the set road section area of the exit direction based on the weather information and combination of the second length of the set road section area of the exit direction and the exit terrain, and turning on the standby network device of the coverage area of the entry coverage radius and the exit coverage radius to construct the first layer network; wherein the first layer network and the second layer network are synchronously updated, and the first layer network and the second layer network predict the next update period according to the change length of the first radius and the second radius respectively and the corresponding preset radius; wherein the time prediction module further comprises: A path parameter determination unit is configured to determine a preset driving path of the first vehicle from a current position to a boundary of the second layer network based on a topological mapping relationship between the first layer network and the second layer network, and extract road feature parameters of the path, the road feature parameters including path length, bend curvature, slope, and lane number change rate, and the driving information including continuous time sequence real-time position coordinates, instantaneous speed, acceleration, steering angle, and lane deviation degree; A model analysis unit is configured to input the road feature parameters and the driving information into a time prediction model to predict an arrival time of the first vehicle to the boundary of the second layer network.

2. The safety warning system for road merge areas according to claim 1, characterized in that, The second layer network construction unit includes: a feature extraction subunit configured to extract a peak probability value of a probability sequence of each ramp , a probability change rate , and a probability coverage duration Meanwhile, the feature extraction subunit is configured to extract a dynamic vehicle proportion of a sequence of information sequences of each ramp , a speed difference coefficient , and a static vehicle retention duration ; A weight determination subunit is configured to determine a basic weight wi of each ramp based on a ramp function priority and a merging type of a road merging area from a priority-merging-weight reference table; a standardization subunit for respectively performing standardization processing on 、 、 、 、 、 to obtain standardization parameters 、 、 、 、 、 ; A basic determination subunit is configured to match a basic radius R0 from a standard-radius reference table based on a design standard of the road merging area. A radius calculation subunit is configured to calculate an influence coefficient of each ramp on the road confluence region The base radius R0 is adjusted ； wherein , , , , , is a preset coefficient, and ; ； wherein, is the first radius; n1 represents the number of loops.

3. The safety warning system for road confluence areas according to claim 1, characterized in that, The time prediction module further includes: a change length determination unit configured to calculate a first change length of the first radius from a corresponding preset radius a second change length of the second radius from a corresponding preset radius ; a change rate determination unit configured to calculate a first change rate and a second change rate based on the first change length and the second change length, respectively , the first change length and the second change length , the first change rate and the second change rate , the first change rate and the second change rate , wherein is a current duration after the first radius and the second radius change synchronously a model prediction unit configured to , 2. V1, V2 are input into a periodicity prediction model configured to: When , 2 are both in the preset micro threshold interval and V1, V2 are both less than the rate threshold, output the basic update period T0; When the length of any one change exceeds a preset micro threshold interval or the rate of change reaches a rate threshold, a shortening coefficient is calculated and the adjusted period is output, wherein is a preset weight coefficient; is the final length of change; is the final rate of change; is the average of the sum of the preset radii corresponding to the first radius and the second radius, respectively; When it is detected that the road merging area has a sudden event, a minimum update period Tmin is forced to be output.

4. The safety warning system for road confluence areas according to claim 3, characterized in that, The cycle prediction model is dynamically optimized through association analysis of historical update data and actual risk events, so that the output next update period is positively correlated with the dynamic change strength of the traffic flow of the road merging area.

5. The safety warning system for road confluence areas according to claim 1, characterized in that, The situation determination module includes: A parameter extraction unit is configured to extract time-space feature parameters of all vehicles in the road merging area at the arrival time based on real-time capture results of the dynamic organization network, the time-space feature parameters including predicted position coordinates, driving speed vectors, acceleration change rates, lane occupation states, and remaining passage times of the vehicles; A topological construction unit is configured to construct a vehicle network topology based on the time-space feature parameters, take the first vehicle as a core node, take other vehicles as associated nodes, and determine edge weights between the nodes based on relative distances, speed differences, and trajectory intersection probabilities of the two vehicles; A calibration unit is configured to calibrate first associated nodes in the vehicle network topology that have spatial intersections with a predicted trajectory of the first vehicle and time differences less than a preset safety threshold, calibrate second associated nodes in the vehicle network topology that have no direct trajectory intersections and have first vehicle lane change spaces compressed due to lane competition, and calibrate third associated nodes in the vehicle network topology that are affected by environmental interference and cause braking delays, to generate conflict calibration judgment parameter groups of each potential level; A potential analysis unit is configured to perform cluster analysis on the conflict calibration judgment parameter groups of each potential level to obtain calibration clusters of the corresponding potential level, and calculate dispersion degrees of all calibration clusters of the same potential level and matching degrees of calibration results of the corresponding potential level and historical real conflict data; A comprehensive analysis unit is configured to comprehensively analyze the dispersion degrees and the matching degrees of each potential level to obtain optimal calibration parameter groups of the vehicle network topology, and input the optimal calibration parameter groups into a safety evaluation model to obtain a safety influence situation of the first vehicle.

6. The safety warning system for road confluence areas according to claim 1, characterized in that, The safety warning module includes: A habit parameter extraction unit is configured to extract driving habit feature parameters of a driver of the first vehicle. The association habit extraction unit is configured to obtain a driving habit characteristic parameter of a driver of the calibration vehicle, the calibration vehicle being an associated vehicle in the vehicle network topology that is calibrated to exist potential conflicts; The habit matching calculation unit is configured to calculate a habit matching degree index of the first vehicle and each calibration vehicle; The risk level determination unit is configured to input the safety impact situation, the habit matching degree index of the first vehicle and each calibration vehicle, and the calibration type of each calibration vehicle into a comprehensive analysis model to determine a real-time risk dynamic level; The safety warning unit is configured to select a corresponding multi-modal warning mode based on the operation preference of the driver of the first vehicle, and perform safety warning according to the real-time risk level.

Images